LEVERAGING INFORMATION
TECHNOLOGY
3322-6 LIT
Melissa Crnic-Sariffodeen
.
iveypublishing.ca
NOTICE REGARDING COPYRIGHT
This custom course package contains intellectual property that is protected by copyright law. It is
illegal to copy the material within this package without the written consent of the holder(s) of the
copyright. This material has been copied under license or with permission from the copyright holder.
Resale or further copying of anything in this package is strictly prohibited.
Unless otherwise stated, Copyright ©2024, Ivey Business School Foundation. Ivey Business School
is the leader in providing business case studies with a global perspective.
Table Of Contents
Zara: IT for Fast Fashion
4
NASDAQ OMX: The Facebook Debacle
27
Netflix Inc.: The Disruptor Faces Disruption
39
RBC Mobile Wallet
51
Google Car
61
JPMorgan Chase & Co.: Open Banking
91
Digital Transformation at GE: What Went Wrong?
105
Siemens Canada: Digital Transformation
125
TopCoder (A): Developing Software through Crowdsourcing
136
The DAO Hack: A Blockchain Dilemma
156
Carolinas HealthCare System: Consumer Analytics
164
United Breaks Guitars
180
Toybox: Managing Dynamic Digital Projects
193
Lesley Stowe Fine Foods: The ERP Decision
205
Transforming the Business Service Portfolio at Global Consultancy
228
Challenges in Commercial Deployment of AI: Insights from The Rise and Fall of IBM Watson's AI
Medical System
239
The Obamacare Website
256
IPremier Co. (A): Denial of Service Attack
270
9-604-081
ANDREW MCAFEE
VINCENT DESSAIN
ANDERS SJÖMAN
Zara: IT for Fast Fashion
On a beautiful August night in 2003, Xan Salgado Badás and Bruno Sánchez Ocampo settled into
seats at their favorite tapas bar in the Spanish city of La Coruña, ordered pulpo gallego (octopus
Galician style), and resumed their argument.
Salgado was the head of IT for Inditex, a multinational clothing retailer and manufacturer
headquartered in La Coruña (see Exhibit 1 for a map). He was Sánchez’s boss, although the two men
had worked together for so long that their formal reporting relationship mattered little. It certainly
did not keep Sánchez from disagreeing with every point Salgado made this evening as they discussed
the point-of-sale (POS) terminals used by Zara, Inditex’s largest chain of stores. Sánchez was the
technical lead for the POS system, so the matter was close to his heart.
“It’s time to upgrade them,” said Salgado.
“No, it’s not.”
“Yes, it is. It’s risky to let them get so far behind current technology.”
“No, it’s riskier to upgrade them just to ‘stay current.’ The software works fine now; we shouldn’t
touch it.”
“But it runs on DOS, which you know Microsoft doesn’t even support anymore.” 1
“And you know DOS hasn’t been supported for years now, and that hasn’t stopped us or hurt us,”
Sánchez replied. “We have the right to keep using the operating system—where’s the problem?”
“One problem is that the hardware vendor for our POS terminals could upgrade their machines,
or some peripheral for them, so that they’re not DOS-compatible anymore. Then where would we
be? We’d be explaining why Zara can’t open any new stores because we don’t have POS software
that works with our POS hardware. Do you want that job?”
1 All computers have an operating system (OS), which is a specialized program that “sits between” the hardware (i.e., the
screen, keyboard, disk drive, processor, etc.) and the software (also called “applications” or “programs”) that users want to run
on the computer. Microsoft’s MS-DOS (Microsoft Disk Operating System), or DOS, was a widely installed operating system in
early personal computers. In 1985, Microsoft launched the Windows OS to replace DOS.
________________________________________________________________________________________________________________
Professor Andrew McAfee, Executive Director of the HBS Europe Research Center Vincent Dessain, and Research Associate Anders Sjöman
prepared this case. HBS cases are developed solely as the basis for class discussion. Cases are not intended to serve as endorsements, sources of
primary data, or illustrations of effective or ineffective management.
Copyright © 2004, 2006, 2007 President and Fellows of Harvard College. To order copies or request permission to reproduce materials, call 1-800545-7685, write Harvard Business School Publishing, Boston, MA 02163, or go to http://www.hbsp.harvard.edu. No part of this publication may
be reproduced, stored in a retrieval system, used in a spreadsheet, or transmitted in any form or by any means—electronic, mechanical,
photocopying, recording, or otherwise—without the permission of Harvard Business School.
Page 4 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
REV: SEPTEMBER 6, 2007
604-081
ZARA: IT for Fast Fashion
“And let me remind you why we shouldn’t tamper with POS unless we absolutely have to,”
Sanchez continued. “Everything about it works! Sales get recorded in stores around the world and
transmitted to us here every day like clockwork. I wrote most of the POS application, and I’m the
one a store manager calls when there’s a problem they can’t fix. Do you know how many of those
calls I got last week?”
“I know that the application is stable, but that doesn’t . . .”
Sánchez didn’t let him finish: “None! That’s how many! Let me ask you another question—do
you know how many stores we opened last week?”
“No, but what’s that got to . . .”
“Exactly! That’s because opening a store requires no IT involvement! You don’t have to send
someone to Dubai, or Argentina, or Russia, or wherever. The store manager just unpacks the POS
terminals, inserts a couple disks in each, plugs a modem into a phone line, and starts selling clothes.
Why on Earth would you want to mess with that?”
“Because I’m worried about DOS,” Salgado said. “And because I think it might be time to
upgrade the POS application itself. We could add functionality, we could add networking capability,
we could . . .”
“We could mess it up in the process. We could turn it from an application that we never have to
worry about into a real headache, for us and the stores.”
“But the store managers are asking for POS to include more . . .”
“Store managers are always asking for something more from us. But which do you think they’d
rather have: a basic POS application that always works, or a fancy buggy one?”
“Of course they’d rather have a stable application,” Salgado replied, “but we’ve been hearing
more and more lately that they want to be able to look up inventory balances in their stores, other
stores . . .”
“Great. So instead of selling clothes they’ll be counting them all the time, trying to make sure their
online inventory figures are 100% accurate. Bad idea.
“POS is not broken,” Sánchez finished definitively. “Why are we trying to fix it?”
Instead of replying, Salgado sat back and thought. He was quite familiar with Sánchez’s
arguments, because Salgado himself had made them many times. The two of them, in fact, often
switched sides in this debate; it was how they made sure that they raised all the relevant issues and
examined the whole argument, instead of getting entrenched in one point of view. This thorough
approach, however, had only deepened Salgado’s confusion over what to do about POS. He
wondered what approach would be most in keeping with how Zara developed and exploited its
overall computing infrastructure.
2 “Porting” is the work of rewriting an existing piece of software so that it is compatible with another operating system.
2
Page 5 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
“Of course not. So let’s buy a bunch of the current terminals so if that happens we’ll have plenty
of breathing room while we port the POS application to a new OS.2 But I don’t even think we need to
do that. The terminal vendor has assured us they’re not going to make any drastic changes, and
we’re a big customer.
ZARA: IT for Fast Fashion
604-081
Zara Business Model
Zara was founded by Amancio Ortega, who in 2003 was still its largest shareholder and the richest
man in Spain.3 Ortega had started in 1963 with clothing factories. Over time, he came to believe that
retailing and manufacturing needed to be closely linked in the apparel industry, where consumer
demand was notoriously hard to forecast. So he integrated forward, opening the first Zara store in La
Coruña in 1975.
Two important events occurred in 1985. First, Inditex (Industria de Diseño Textil) was formed as
a holding company atop Zara, other retail chains (see Exhibit 2 for a list of them as of 2003), and a
network of internally owned suppliers. Second, José María Castellano Ríos joined the company.
Castellano had worked as an IT manager and shared Ortega’s belief that computers were critically
important in enabling the kind of business that they wanted to build. Castellano became Inditex’s
CEO in 1997.
Speed and Decision Making
In addition to their affinity for information technology, Ortega and Castellano shared two other
beliefs about the company. First, Zara needed to be able to respond very quickly to the demands of
target customers, who were young, fashion-conscious city dwellers. Their tastes in clothing changed
rapidly, were very hard to predict, and were also hard to influence. Other companies in the apparel
industry had shown that marketing and advertising campaigns could be effective at convincing a
consumer to buy their clothes. History had also demonstrated, however, that “fashion misses” were
common even with extensive advertising and that new styles could appear suddenly (based, for
example, on what a rock star wore during a televised awards show), surge in popularity, then quickly
fade. Zara wanted to be able to produce and deliver such styles while they were still hot, rather than
relying on the persuasiveness of its marketing to push clothes it had made some time ago.
Second, Ortega, Castellano, and the other senior managers wanted to take advantage of the
intelligence and trust the judgment of employees throughout the company, instead of relying on a
small set of decision makers. Store managers at Zara, for example, were given much more
responsibility than those at other large clothing chains. In addition to dealing with customers,
employees, contractors, and landlords, Zara store managers decided what garments would be on sale
at their stores. They placed orders for the items they thought would sell, rather than simply
accepting and displaying what headquarters decided to send them.
Similarly, a group of people at La Coruña called “commercials” had great discretion in deciding
what clothes would be designed and produced. In sharp contrast with those of other chains, Zara’s
collections were not conceptualized and designed by a small, elite team. Instead, collections were
created, then extended and modified over time, by teams of commercials, each dedicated to a section
of the store (Men, Women, or Children) and, within that, to a specific collection (“Basic” and
“Sports,” for example, were both collections within Women). Teams usually consisted of two
designers and two product managers, who purchased material, placed production orders with the
factories, and set prices.
3 This section draws on Pankaj Ghemawat and José Luis Nueno, “Zara: Fast Fashion,” HBS Case No. 703-497 (Boston: Harvard
Business School Publishing, 2003).
3
Page 6 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
The original business idea was very simple. Link customer demand to manufacturing, and link
manufacturing to distribution. That is the idea we still live by.
— José María Castellano Ríos, Inditex CEO
ZARA: IT for Fast Fashion
Another group of commercials, called store product managers, sat in close proximity to the
product teams and served as La Coruña’s main interface with Zara stores around the world. They
traveled extensively, observing what residents were wearing and talking at length with store
managers to find out what kinds of clothes were selling. Even more importantly, they also tried to
learn what kinds of clothes would sell if Zara made them. Store product managers communicated
what they had seen and heard to the design teams, helping them keep abreast of fast-changing trends
and demands (for a layout of the office in La Coruña where design teams and store product
managers sat, see Exhibit 3). Store product managers could initiate store-to-store transfers when they
saw that garments selling slowly in one area were popular in another.
Other employees within the commercial function also exercised a great deal of autonomy. They
decided, for example, which clothes each store would be able to order. When total orders from stores
exceeded availability for an item in any period, commercials decided which stores would get clothes
and which would not. Commercials’ decisions were not typically reviewed by higher-level
managers. Zara believed that such second-guessing would compromise both the company’s speed
and its emphasis on decentralized decision making.
Marketing, Merchandising, and Advertising
Unlike its main competitors, which were other multinational clothing retailers such as H&M, Gap,
and Benetton, Zara did virtually no advertising. The company placed ads only to promote its twiceyearly sales4 and to announce the opening of a new store. As a result, Zara’s marketing expenditures
averaged 0.3% of revenue, instead of the 3%–4% typical for competitors. (For a financial comparison
of Inditex and its three main competitors, see Exhibit 4.)
While it spent little on ads, Zara spent relatively heavily on its stores. They were always located
in a city’s prime retail district, often on the best-known street. And while Zara’s store managers had
a great deal of freedom in deciding what clothes to stock, they had no discretion about the look and
feel of their stores. Store layouts were completely changed every four to five years, with artwork,
window displays, and sales racks changed more frequently. A 1,500-square-meter pilot store was
kept in La Coruña, where all new store layouts were designed and tested before being rolled out
around the world. After a redesign, a La Coruña-based team traveled to each Zara store to set up the
new configuration.
Individual stores also did not have the freedom to set garment prices; these were determined by
product managers. Prices were established for the Spanish market, denominated in euros (Ä), and
noted on the tag affixed to the garment in La Coruña. Prices for other countries were set at a fixed
percentage of this baseline, taking into account distribution costs and market conditions.
Zara did not try to produce “classics”—clothes that would always be in style. In fact, the
company intended its clothes to have fairly short life spans, both within stores and in customers’
closets. Three implications followed from this approach. First, experienced Zara shoppers knew that
if they saw a garment they liked they should buy it on the spot, because it might not be there on their
next visit (about 75% of the merchandise in the average store was changed over three to four weeks).
Second, shoppers also knew that they should visit the store often, since new styles showed up all the
time. Finally, Zara garments were not designed and manufactured to be highly durable; they were
described as “clothes to be worn 10 times.”
4 All large clothing retailers held these sales to get rid of merchandise in advance of a new collection’s debut. Because of its
proficiency at matching supply to demand, Zara typically sold 15%–20% of its clothes during these sales at an average discount
of 15%. European competitors sold 30%–40% of their clothes this way, at an average discount of 30%.
4
Page 7 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
604-081
ZARA: IT for Fast Fashion
604-081
Financials and Growth
At the beginning of 2003, Inditex operated 1,558 stores in 45 countries, of which nearly 550 were
part of the Zara chain. The group opened on average one store per day across the world. Forty-six
percent of the group’s sales were inside Spain, with France the largest international market. Zara
generated 73.3% of the group’s sales. Of the three departments inside Zara, Women accounted for
60% of sales, with the rest evenly split between Men and the fast-growing Children segment. For its
fiscal-year 2002, Inditex had posted a net income of Ä438 million (about $502 million U.S. dollars) on
revenues of Ä3,974 million (about $4,554 million), continuing a trend of rapid and profitable growth;
the company’s earnings, for example, had more than tripled between 1996 and 2000 (Exhibit 5
provides the group’s financial information, Exhibit shows its geographic expansion, and Exhibit 7
shows growth over time).
Inditex executives felt that ample room for growth existed within its current markets. Italy, for
example, had very few Zara stores, despite the fact that shoppers there were some of the most fashion
conscious in Europe. Zara’s Italian stores were extremely popular, giving Castellano confidence that
the country could one day have a store density similar to that of Spain. And Inditex’s western
European expansion could, he felt, be largely supported with its current infrastructure. This implied
that it would not be necessary to build entirely new production and distribution networks in order to
support future growth.
Operations
To reach its goal of quickly and accurately responding to shifting consumer demands, Zara
established three cyclical processes—ordering, fulfillment, and design and manufacturing. Of these,
ordering (of garments by the stores) was the most regular, precisely defined, and standardized
around the world.
Ordering
Every major section of a Zara store—Men, Women, and Children—placed an order to La Coruña
twice a week. The order encompassed both replenishment of existing items and initial requests for
newly available garments. Stores faced “hard” deadlines for submitting these orders; if they missed
the deadline, La Coruña calculated a replenishment-only order for them, based on what they had
sold since the previous order.
Store managers determined replenishment quantities by walking around the store and
determining what had been selling by counting garments and talking to salespeople. Store personnel
could not look up their inventory balances on any in-store computer, so canvassing the store was the
only way to learn about stock levels.
Managers learned about newly available garments by consulting a handheld computer that was
linked each night, via dial-up modem, to information systems at La Coruña. (See Exhibit 8 for a
5
Page 8 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Zara had decided not to sell clothes over the Internet, for two main reasons. First, the company’s
distribution centers (DCs) were not configured for picking small orders and shipping them to
consumers. Second, it would be complicated to handle returns of merchandise bought online. Zara
managers understood that the retail mail-order industry saw return rates as high as 50%–60%, which
they compared unfavorably with their normal 5% store returns. A Web site—www.zara.com—
existed but served only as a digital display window, showing a few typical garments at any time.
604-081
ZARA: IT for Fast Fashion
To facilitate ordering, the store manager usually divided the offer into segments and “beamed”
each segment to a different handheld using infrared technology. Several people then used these
handhelds to fill in their segment of the offer as they walked through the store, then beamed their
segments back to the store manager. After reviewing them, the manager would send the completed
form, now called “the order,” back to La Coruña.
Fulfillment
Fulfillment, or shipping clothes to stores to satisfy their orders, involved another group of
commercials at La Coruña. Their job was to match up the supply of finished clothes coming from
factories into the DC with the stores’ demands for these items. They worked with two pools of
information: the aggregated orders from all stores, which was finalized soon after the order deadline
had passed, and the total supply of inventory in the DC at the same point in time. Both of these were
at the level of the stock-keeping unit (SKU), which was defined as the combination of garment plus
fabric plus color plus size.
When supply and demand lined up closely for a particular SKU, no decisions were required; the
commercial simply allowed the inventory to be divided up, by computer, among all the stores that
wanted it. If, however, demand for an SKU was greater than supply in any ordering period, the
commercial had to determine which stores would get the available inventory and which would not.
He or she did this by looking at which stores had been most successful at selling the item and which
ones, if any, had been shortchanged on these decisions in the recent past.
These commercials also worked with product managers to determine future production for each
SKU. If there were more demand than supply, of course, production would be increased as quickly
as possible. When supply started to exceed demand, the commercial would decrease replenishment
requests and eventually stop placing new factory orders altogether.
Finally, commercials could also ship items that stores did not order. These were typically new
garments for which Zara wanted to assess demand. They would be sent to stores in targeted
geographies; store managers knew to expect such deliveries periodically, and to offer the clothes for
sale rather than asking where they came from. Store managers also knew to expect that some items
that they had ordered might not arrive because total demand had exceeded supply and commercials
had decided to allocate available SKUs elsewhere.
Deliveries typically showed up at stores one or two days after each order was placed. Stores in
western Europe were replenished by truck from the two Spanish DCs. Latin American stores were
replenished from smaller local DCs.5 More remote stores, such as those in northern Europe and the
Middle East, were replenished by air from the Spanish DCs. Garments did not stay long in a DC; the
goal was to produce, then deliver, only what the stores needed, and only when they needed it. In
fact, there was little inventory anywhere in Zara’s supply chain. Clothes flowed quickly, and without
stopping, from factories to DCs to stores, where they were immediately put on the sales floor; Zara
stores had no “back room” where excess inventory could be kept.
5 Latin American DCs were supplied via bulk shipments from Spanish DCs.
6
Page 9 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
picture of a handheld.) Less than 24 hours before each order deadline a digital order form, called
“the offer,” was transmitted to all stores’ handhelds. The offer included descriptions and pictures of
the newly available items, as well as all replenishment items that were still available to that store.
Each store’s offer was different; offers were developed by a team of commercials and were based on
garment availability, regional sales patterns, predictions about what would sell well in each location,
and other factors. (See Exhibit 9 for a portion of an offer.)
ZARA: IT for Fast Fashion
604-081
Each section of all Zara stores ordered twice a week, but different sections received shipments on
different days. As a result, the DC in La Coruña was active throughout the week but most active on
the days when Women’s orders were shipped to stores, since the Women’s Department accounted for
the greatest share of sales, orders, and SKUs.
Like other large clothing retailers, Zara introduced substantially new design collections at the start
of the fall/winter and spring/summer buying periods. In sharp contrast to the competition,
however, Zara also brought out new items continuously throughout the year, including both changes
to existing garments (for example, a shirt with a new collar or color) and entirely new creations. In a
typical year, Zara introduced approximately 11,000 new items; competitors averaged 2,000–4,000.
Zara’s vertically integrated manufacturing operations enabled this constant introduction of new
items and also ensured short lead times. Production requirements were distributed across a network
of specialized facilities that quickly produced and delivered the required goods. Zara owned a group
of factories in and around La Coruña to do the capital-intensive initial production steps of dyeing
and cutting cloth.6 Cut fabric was sewn into garments at a network of small local workshops in
Galicia and northern Portugal that guaranteed quick turnaround times.7 All finished garments were
sent to a Zara facility, where they were ironed, inspected, given a machine-readable tag, and sent to a
DC.
Using this network, Zara could consistently move a new design from conception through
production and into the DC in as little as three weeks.8 Two days after that, the garment could be on
sales racks in stores around the world. This speed enabled Zara to respond to the fast-changing and
unpredictable tastes of its target customers. As far as Inditex managers were aware, no other large
apparel retailer could match this capability.
A consequence of Zara’s approach to design, fulfillment, and manufacturing was that the
company did not have to rely on accurate long-range sales forecasts. Instead, commercials within
design teams simply made an initial guess about how well a garment would sell, then communicated
this guess to factories in the form of a first-production requirement. It was not critical that this guess
be accurate. Stores’ orders told commercials how well the garment was selling and thus whether
future production should be increased or decreased. And flexible factories with short lead times
could adjust to such changes easily and rapidly. Zara did not have to predict what would be selling
six months, or even one month, in the future; it could continuously sense what customers wanted to
buy and respond “on the fly.”
Information Technology
Approaches and Organization
Zara’s approach to information technology was consistent with its preferences for speed and
decentralized decision making. The company had no chief information officer and no formal
6 Zara bought a large amount of undyed fabric on the external market and also owned some textile fabrication facilities.
7 These workshops were not owned by Inditex.
8 Zara outsourced production of some items with comparatively stable and predictable demand patterns like men’s dress
shirts. Lead time for items produced in China was approximately four months; for items produced in Turkey, lead time was
two months.
7
Page 10 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Design and Manufacturing
604-081
ZARA: IT for Fast Fashion
Salgado and his colleagues shared a preference for writing the applications they needed
themselves rather than buying commercially available software. They felt that the company’s
operations were unique enough that commercial packages would not be suitable. The fact that Zara
did business in so many countries and currencies, for example, meant that standard accounting
packages would have to be heavily modified and extended. Rather than attempting this, the IS
department wrote its own accounting software. Similarly, the applications that supported ordering,
fulfillment, and manufacturing were largely developed internally.10
Application development and other IT activities were the responsibility of an IS department of
approximately 50 people, almost all hailing from Galicia and recruited from local universities.11 They
were divided into three groups: Store Solutions, Logistics Support, and Administrative Systems.
With very few exceptions, all IT support of Inditex stores around the world was done directly from
La Coruña. Staff retention was not seen as a problem; in the last 10 years, only one person had left
the department.
La Coruña
At La Coruña, several information systems were used to support Zara’s operations. Internally
developed applications were used to prepare the offer and distribute it over the Internet to stores
around the world and also to receive orders from all of the stores and aggregate them. Another
application compared the aggregated order to available inventory for each SKU, highlighted
situations where supply and demand were imbalanced, and executed commercials’ decisions about
how to allocate products when demand exceeded supply.
Yet another application kept track of the “theoretical inventory” of each SKU at each store.
Shipments to stores increased this inventory, and sales decreased it. At the end of each business day,
each store transmitted that day’s sales for all SKUs back to La Coruña, using a modem connected to
one of the store’s POS terminals (see Exhibit 11 for a photo of a Zara POS terminal). Of course, if
shipments and sales were not recorded perfectly, stores’ theoretical inventory would become
inaccurate; theft, damage, and other losses would also make theoretical inventory a poor reflection of
reality.
The company had not historically been greatly concerned that theoretical inventory be 100%
accurate for each store and SKU at all times. Theoretical inventory was used to help make allocation
9 Castellano estimated Inditex’s 2002 IT spending to be Ä25 million, or approximately 0.5% of revenue. A 2001 survey of large
North American retailers found their IT spending to average approximately 2% of annual revenue (Gartner, Inc., 2001 IT
Spending and Staffing Survey Results).
10 Zara did use standard commercial applications for office productivity (word processing, e–mail, etc.) and computer-aided
clothing design.
11 IT employees accounted for less than 0.5% of Inditex’s total workforce. Large North American retailers, in contrast, devoted
approximately 2.5% of employees to IT on average (Gartner, Inc., 2001 IT Spending and Staffing Survey Results).
8
Page 11 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
processes for setting an IT budget9 or deciding on specific technology investments or projects.
Instead, Salgado and Castellano sat on a technology steering committee and so got involved early in
discussions of initiatives that might include computerization (for a formal organization chart of
Inditex, see Exhibit 10). As these discussions progressed, Salgado and his colleagues would
determine what new systems, if any, were required and which IS department personnel should work
on them. There was little or no formal justification for IT efforts, nor were cost/benefit analyses
typically conducted for a proposed effort.
ZARA: IT for Fast Fashion
604-081
decisions and for little else.12 Salgado maintained that “Having 100% control is most of the time just
too expensive. Being 95% right is pretty good, and often you don’t need more accuracy.”
Inside Zara’s factories, relatively simple applications were used to plan production. These
applications did not use sophisticated mathematics to generate “optimal” plans and schedules.
Instead, they presented factory managers with quantities and due dates for all production requests.
Managers used this information to load their factories and put jobs in sequence.
The most sophisticated technologies inside Zara factories were usually the large computercontrolled equipment that cut cloth into patterns.13 These machines calculated how to position
patterns so as to minimize scrap and could cut over 100 layers of fabric at a time. Cut fabric was then
sent from Zara factories to external workshops for sewing.
Distribution Centers (DCs)
Zara’s DCs relied on a great deal of automation and computerization. At the La Coruña DC, for
example, miles of automated conveyor belts facilitated the ongoing task of receiving bulk quantities
of each garment from factories then recombining these garments into shipments for each store. (See
Exhibit 12 for a picture of these conveyor belts.) Information systems tracked where each SKU was
stored as it entered the DC, then controlled the conveyor belts to pick them up and drop them off at
the appropriate places. Humans helped with this work, particularly by taking garments off the belts
at the end of their journey through the DC, then putting them on hanger racks or in cardboard boxes
that would be sent to each store. Zara’s IT department wrote the applications that controlled the
DC’s automation, often in collaboration with the vendors of conveyor equipment.
Stores
All Zara stores had identical handhelds—also known as personal digital assistants (PDAs)—and
POS systems. PDAs had been introduced in 1995. At that time, many within the company felt that it
was taking too long and costing too much to fax order forms back and forth to all stores around the
world twice a week. Because of the number of SKUs involved these forms could be well over 15
meters long, so it was time consuming to send and receive them. Unreliable fax machines, paper
shortages, and other similar problems also introduced delays and frustration into the critical ordering
process. Salgado and his colleagues decided to address the situation with IT and began
experimenting with handheld computers that would communicate with La Coruña via modem. They
first used Apple’s Newton device and became one of the largest users of this then-new technology.
After the Newton was discontinued in 1998, Zara switched to another PDA manufacturer.
In 2003, PDAs were used primarily for ordering and also for tasks such as handling garment
returns to DCs and transmitting information from headquarters to all stores. Each store had several
PDAs, allowing redundancy and division of labor during the ordering process. Zara constantly
upgraded stores’ PDAs as devices were discontinued or as technological advances such as color
screens became available.
12 All Zara stores periodically conducted physical audits of their inventory. During an audit SKUs were divided into
categories based on price, then the number of garments in each category were counted. If the total value of the inventory in
the store was close enough to the total value of theoretical inventory, the store passed the audit.
13 These were commercially available machines, and Zara did not greatly modify them.
9
Page 12 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Factories
ZARA: IT for Fast Fashion
The POS terminals in use within every store, in contrast, had remained essentially unchanged for
well over a decade. They still used the DOS operating system, which in 2003 was no longer
supported by Microsoft. Zara continued to use DOS, and the internally developed POS application
that ran on top of it, because this combination had proved to be remarkably stable, effective, and easy
to roll out and maintain over time. Store employees, for example, could turn POS terminals on and
off at any time without worrying about following start-up or shutdown procedures. They could also
set up and maintain the complete POS infrastructure themselves. When opening a new store, the
manager simply inserted two floppy drives into each “blank” POS terminal; the floppies contained
DOS and all required applications. In the event of a serious problem with a POS terminal, a complete
software reinstallation was similarly straightforward. As a result, no IT support was required to
open a new store, nor was it necessary to run a large IT support organization to assist the stores.
Neither the POS terminals nor the PDAs were always connected to Zara’s headquarters or to other
stores. One POS terminal at each store had a modem, which was used at the end of each business
day to transmit comprehensive sales information and other data to La Coruña. POS terminals were
not connected to one another via any in-store network, so employees copied daily sales totals from
each terminal onto a floppy disk, then carried these disks to the one modem-equipped terminal to
accomplish the transition. PDAs also used this terminal’s modem to receive the offer and transmit
the order. Stores did not have any computers beyond the POS terminals and PDAs. Within a store,
POS terminals and PDAs could not share information.
The POS terminals and PDAs did not contain information that could be used when one store
wanted to know if a nearby one had a particular SKU in stock. Store personnel telephoned one
another to answer this question.
Conclusion
Salgado and Sánchez both worried about “getting fancy” with store IT. Would upgrading to a
modern operating system, enhancing the POS application itself, and/or building networks within
and between stores put at risk the robust and scalable infrastructure they had built? They also
worried, however, that Zara was building a bigger and bigger company on top of a more and more
obsolete operating system. What if the hardware vendor for POS terminals changed the machines in
such a way that they could no longer use DOS? This vendor had already made it clear that Zara was
its only customer using the ancient operating system. The vendor also said that it had no plans to
change its machines so that they could no longer run DOS, but Salgado had gotten nowhere when he
had tried to include such assurances in Zara’s contract with the terminal maker.
Did all of this mean that it was now time to port the POS application to another OS such as
Windows, UNIX, or Linux? And as insurance against unpleasant surprises, did it make sense to
purchase enough of the current POS terminals now, so that Zara’s needs would be covered in the
event of a sudden loss of support from the vendor?
If they were going to port the POS application to a new operating system, should they also use it
as an occasion to build new capabilities into the software? One of the few complaints store managers
had about the PDAs was that it was time consuming to use their small screens and styluses to
accomplish returns. An updated POS application could easily incorporate this functionality,
allowing store personnel to use a large screen, keyboard, and mouse to quickly execute returns
transactions.
And why stop there? Modern POS terminals, since they were really modern PCs, could
accommodate even more sophisticated capabilities, especially networks within stores and across the
10
Page 13 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
604-081
604-081
company. Wireless networks were particularly intriguing since they were much cheaper to install
within a store. With a wireless network in place, it would no longer be necessary to carry floppy
disks around the store at the end of each business day to tally up total sales. And if all stores and La
Coruña were permanently connected to the Internet, every location could know the theoretical
inventory of all of its SKUs, as well as the theoretical inventory in all other stores. In this scenario,
stores could request inventory transfers from one another online, eliminating the need for phone calls
to see if an item were in stock. (See Exhibit 13 for some industry-based assumptions on development
costs.)
Salgado and Sánchez saw that a move to change the operating system of Zara’s POS terminals
entailed a number of follow-on decisions. Was now the right time to make them, or should the
company simply continue to use the IT infrastructure that had worked so well for so long? Even the
arrival of the delicious pulpo did not take their minds entirely off the issue, but it did temporarily stop
them from arguing about it.
11
Page 14 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
ZARA: IT for Fast Fashion
604-081
Exhibit 1
ZARA: IT for Fast Fashion
Map of Spain, with La Coruña Indicated
Source: University of Texas, http://www.lib.utexas.edu/maps/europe/spain_sm97.gif, accessed January 20, 2004.
12
Page 15 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
La Coruña
ZARA: IT for Fast Fashion
Inditex Retail Chains (end of 2003)
Zara
• Founded in 1975
• Continuous design based on
customer desires, for women,
men, and children.
Pull and Bear
• Founded in 1991
• Offering casual clothing at
affordable prices
Massimo Dutti
• Acquired in 1995
• Higher fashion for men and women
Stradivarius
• Acquired in 1999
• Youthful urban fashion
Source:
Inditex (formatted by casewriters).
Note:
The retailing chains were organized as separate business units.
Bershka
• Founded in 1998
• Trendy clothing for a younger
market
Oysho
• Founded in 2001
• Lingerie
13
Page 16 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Exhibit 2
604-081
604-081
ZARA: IT for Fast Fashion
Exhibit 3
Physical Organization of a Zara Design Department
The other Zara departments—Men and Children—were organized in a similar way.
Design and Production Teams
(with product responsibility)
Design and Production Teams
(with product responsibility)
!!
!!
!
! !!
! !!
!
!!
!!
Source:
!!
!!
Store Product Managers
(with geographic
responsibility)
Casewriters research.
14
Page 17 of 282
!
! !!
! !!
!
!!
!!
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
The department for Zara Women sat in one open landscape. Design and production teams
consisted normally of two designers and two product managers responsible for a specific collection,
such as knitwear for Zara Women. They interacted with the store product managers, who were in
almost daily contact with stores in their geographical region, for instance France.
ZARA: IT for Fast Fashion
Inditex and Key Competitors (financials in Ä)a
2002
Operating Results (Ämn)
Net Operating Revenues
- Cost of Goods Sold
Gross Margin
- Operating Expenses
Operating Profits
- Nonoperating Expenses
Pretax Income
- Income Tax
- Minority Interests
Net Income
Net Margin
2001
Inditex
Gapb
H&M
Benettonc
3,974
1,955
2,019
1,180
839
224
615
173
4
438
11.02%
13,819 4,972
9,122 2,230
4,697 2,742
3,729 1,840
968
902
202
-40
766
943
309
321
0
0
456
621
3.30% 12.49%
1,992
1,124
867
625
243
194
49
57
2
-10
-0.49%
Inditex
Gapb
H&M
Benettonc
3,250 15,559
1,563 10,904
1,687
4,655
982 4,276
705
379
209
108
496
271
150
280
5
0
341
-9
10.49% -0.06%
4,269
2,064
2,205
1,615
590
-28
618
206
0
412
9.65%
2,098
1,189
909
624
285
43
242
92
2
148
7.05%
Financial Position (Ämn)
Current Assets
Property, Plant, and Equipment
Other Noncurrent Assets
Total Assets
1,146
1,413
455
3,014
5,487
3,611
368
9,467
2,038
668
477
3,184
1,637
706
301
2,643
854
1,228
523
2,605
3,436
4,695
435
8,566
1,468
661
54
2,183
1,558
720
543
2,821
Current Liabilities
Noncurrent Liabilities
Total Liabilities
Shareholders' Equity (book value)
Liabilities and Shareholders' Equity
1,013
240
1,253
1,761
3,014
2,607
3,363
5,969
3,497
9,467
578
90
667
2,085
2,752
546
957
1,503
1,141
2,643
834
285
1,119
1,486
2,605
2,320
2,850
5,170
3,396
8,566
432
101
533
1,650
2,183
956
624
1,580
1,241
2,821
Market Capitalization
Equity—Market Valued
1-Year Change in Market Value (%)
13,981
0
12,320 16,496
0
0
1,144
-1
13,433 12,687 15,564
0
-1
0
2,605
0
Other Statistics
Employees
Number of Countries of Operation
Sales in Home Country (%)
Sales in Home Continent (%)
Number of Store Locationse
Stores in Home Country (%)
Stores in Home Continent (%)
Average Stores Size (square meter)
32,535 169,000 25,674
45
6
14
46%
NA
11%
75%
NA
96%
1,558
3,117
884
59%
88%
14%
85%
91%
95%
NA
NA
NA
7,824
120
31%
69%
5,371
NA
NA
NA
26,724 165,000 22,944
39
6
14
46%
87%
12%
77%
NA
96%
1,284
3,097
771
60%
88%
15%
86%
92%
96%
514
632
1,201
7,666
120
44%
78%
5,456
40%
80%
279
Source: Compiled and calculated by casewriters.
– 2002 numbers: Operating results and financial position from companies' annual reports. Market cap/equity data
from analyst reports. Other statistics from company Web sites, annual reports, or analyst reports.
– 2001 numbers: As for 2002 numbers and based on Pankaj Ghemawat and José Luis Nueno, “Zara: Fast Fashion,”
HBS Case No. 703-497 (Boston: Harvard Business School Publishing, 2003).
aConverted to euros for Gap (original financials in USD) and H&M (from Swedish kronor, SEK).
bGap includes retail chains Gap, Banana Republic, and Old Navy.
cBenetton includes main brands United Colours of Benetton, Sisley, Nordica, and Prince.
dEquity market value for 2002 as of February 18, 2003, and for 2001 as of May 22, 2002.
eIncludes franchise stores.
15
Page 18 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Exhibit 4
604-081
604-081
ZARA: IT for Fast Fashion
Inditex Historical Financials (millions of Euro)
Year
Net Operating Revenues
Cost of Goods Sold
Gross Margin
Operating Expenses
Operating Profits
Non-Operating Expenses
Pre-Tax Income
Income Tax
Minority Interest
Net Income
Net Margin
2002
3,974.0
1,954.9
2,019.1
1,179.8
839.3
224.3
615.0
172.5
4.4
438.1
11.02%
2001
3,249.8
1,563.1
1,686.7
982.3
704.4
209.3
495.1
149.9
4.8
340.4
10.47%
2000
2,614.7
1,277.0
1,337.7
816.2
521.5
152.7
368.8
106.9
2.7
259.2
9.91%
1999
2,035.1
988.4
1,046.7
636.2
410.5
118.1
292.4
86.2
1.5
204.7
10.06%
1998
1,614.7
799.9
814.8
489.2
325.6
96.7
228.9
76.1
-0.2
153.0
9.48%
Inventories
Accounts Receivable
Cash and Cash Equivalents
Total Current Assets
Property, Plant, Equipment
Other Non Current Assets
Total Assets
Asset Turnover
ROA
382.4
237.7
525.9
1,146.0
1,412.6
455.2
3,013.8
1.32
14.54%
353.8
184.2
315.7
853.7
1,336.8
414.5
2,605.0
1.25
13.07%
245.1
145.2
210
600.3
1,339.5
167.8
2,107.6
1.24
12.30%
188.5
121.6
171.8
481.9
1,127.4
163.6
1,772.9
1.15
11.55%
Accounts Payable
Other Current Liabilities
Total Current Liabilities
Non Current Liabilities
Total Liabilities
Equity
Total Liabilities and Equity
Leverage (Equity / Total Assets)
506.2
506.5
1012.7
239.8
1,252.5
1,761.3
3,013.8
1.71
426.3
407.9
834.2
284.5
1,118.7
1,486.2
2,605.0
1.75
323.0
347.3
670.3
266.4
936.7
1,170.9
2,107.6
1.80
ROE
24.9%
22.9%
22.1%
Note:
1997
1,217.4
618.3
599.1
345.5
253.6
1996
1,008.5
521.0
487.5
285.4
202.1
117.4
9.64%
72.7
7.21%
157.7
75
158.8
391.5
880.4
54.4
1,326.3
1.22
11.54%
274.0
635.7
67.5
977.2
1.25
12.01%
190.3
820.3
1.23
8.86%
276.1
275.6
551.7
328.0
879.7
893.2
1,772.9
1.98
215.6
229.1
444.7
208.2
652.9
673.4
1,326.3
1.97
131.4
141.5
272.9
174.4
447.3
529.9
977.2
1.84
234.1
171.3
405.4
414.9
820.3
1.98
22.9%
22.7%
25.0%
20.0%
Inditex fiscal year ended January 31. Fiscal year 2002, for instance, ran from February 1, 2002 to January 31, 2003.
Source: Inditex (Formatted by case writers)
16
Page 19 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Exhibit 5
ZARA: IT for Fast Fashion
604-081
Exhibit 6 Inditex Store Locations by Country and Group (January 31, 2003)
Europe
Andorra
Austria
Belgium
Czech Republic
Cyprus
Denmark
Finland
France
Germany
Greece
Iceland
Ireland
Italy
Luxembourg
Malta
Norway
Poland
Portugal
Spain
Sweden
Switzerland
The Netherlands
Turkey
United Kingdom
Middle East
Bahrain
Israel
Jordan
Kuwait
Lebanon
Qatar
Saudi Arabia
United Arab Emirates
Asia-Pacific
Japan
Singapore
Americas
Argentina
Brazil
Canada
Chile
Dominican Republic
El Salvador
Mexico
United States
Uruguay
Venezuela
TOTAL
Kiddy's
Class
1
4
15
1
3
2
1
71
21
23
1
Pull &
Bear
Massimo
Dutti Bershka Stradivarius Oysho
1
Total
5
10
9
3
1
1
29
8
2
7
75
0
6
16
2
18
6
25
0
2
11
2
4
33
1
10
2
1
73
24
39
1
5
5
3
4
1
4
155
918
2
6
5
8
19
1,325
0
4
25
2
8
5
4
15
18
81
0
6
1
7
0
5
10
9
3
1
1
83
8
2
23
145
531
59
296
250
197
153
72
1,558
1
13
4
2
1
2
7
1
3
3
2
1
5
1
5
3
2
1
2
1
3
1
4
35
200
2
4
8
17
419
7
52
59
1
11
38
200
256
1
14
1
2
1
1
3
2
1
8
4
30
0
6
1
7
0
32
155
2
2
1
2
218
20
135
4
24
0
9
48
145
60
2
168
1
1
1
2
1
4
4
14
14
128
1
1
1
4
4
1
3
2
8
1
0
0
0
0
10
16
19
9
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Zara
Source: Inditex.
17
Page 20 of 282
604-081
Inditex Store Development
Year
Europe
Spain
Portugal
France
Greece
Belgium
Sweden
Malta
Cyprus
Norway
Great Britain
Germany
Switzerland
Netherlands
Poland
Andorra
Austria
Denmark
Czech Rep.
Iceland
Ireland
Italy
Finland
Luxembourg
87 88 89
90
91
92
93
94
95
96
97
98
99
57 70 85
1 2
99 201 266 323 350 391 399 433 489 603
4 11 17 28 38 49 60 74 87 97
1
3
5 13 20 30 36 47 55 59
1
6
8 10 14 17 17
4
8 11 13 17 20
1
3
3
4
6
6
1
1
1
1
2
1
2
4
5
1
1
1
1
3
2
0
769
140
68
29
28
3
2
9
1
11
17
Zara only
01
02
6
29
8
3
1
2
3
6
41
8
4
3
2
5
8
55
8
20
4
3
7
8
27
8
7
4
3
7
2
54
2
71
2
107
1
1
3
3
1
3
1
4
7
6
12
6
14
7
20
7
25
4
1
1
1
3
4
4
11
18
20
27
37
6
16
1
3
1
1
1
10
22
3
3
2
3
1
11
3
1
23
4
4
4
5
0
17
11
1
1
24
4
5
4
15
0
5
14
2
2
33
49
70
1
76
Middle East/Asia
Israel
Lebanon
Turkey
Kuwait
United Arab Emirates
China
Japan
Saudi Arabia
Bahrain
Qatar
Singapore
Jordan
8
83
5
23
9
3
10
1
1
2
145
225
38
67
20
14
0
0
2
0
11
15
0
3
2
1
3
2
1
1
0
0
2
66
8
29
5
7
9
3
10
1
1
2
75
0
35
6
8
1
1
1
0
45
57 71 88 105 218 292 369 430 508 541 622 748 922 1,080 1,284 1,558
507
531
0
0
0
0
0
0
0
0
0
6
Source: Inditex.
18
Page 21 of 282
9
2
5
2
4
0
5
6
1
1
200
35
71
23
15
0
1
3
0
17
21
2
4
4
1
4
2
1
1
0
3
1
2
411
25
5
8
8
18
0
6
15
4
4
1
2
96
0
TOTAL
692
104
64
19
21
5
2
8
1
7
7
02
1
406
57 71 87 104 215 288 365 419 490 521 589 678 819
0
01
918
155
73
39
33
2
4
10
1
19
24
6
2
6
5
2
2
4
1
2
2
3
3
4
1
2
2
1
1
1
1
2
5
3
5
1
2
3
939 1,101 1,317
2
2
Americas
United States
Mexico
Argentina
Venezuela
Canada
Chile
Brazil
Dominican Republic
El Salvador
Uruguay
00
11
2
8
3
4
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Exhibit 7
ZARA: IT for Fast Fashion
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
ZARA: IT for Fast Fashion
Exhibit 8
Source:
604-081
Dell Handhelds (as of fall 2003)
Casewriters.
19
Page 22 of 282
Exhibit 9
Source: Casewriters.
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
604-081
ZARA: IT for Fast Fashion
Order Form
20
Page 23 of 282
ZARA: IT for Fast Fashion
Exhibit 10
604-081
Organization
Antonio Abril, General Counsel and Secretary
Javier Monteoliva, Legal Advisory Services
Javier Chércoles, Corporate Responsibility
José María Castellano
Deputy Chairman and CEO
Borja de la Cierva, Chief Financial Officer
Ignacio Fernández, Tax Advisory
Diego Copado, Corporate Communication
Marcos Lopéz, Capital Markets
Juan Cobián, Internet
Juan Carlos R. Cebrián
Managing Director
Agustín García-Poveda
Deputy General Manager
Administration & Systems
Business Units
Administration: Fernando Aguiar Zara: Jose Toledo
IT: Xan Salgado
Pull & Bear: Pablo del Bado
Human Resources: Jesús Vega
Massimo Dutti: Jorge Pérez
Bershka: Carlos Mato
Stradivarius: Jordi Triquell
Oysho: Sergio Bucher
Business Support Areas
Expansion: Ramón Renón
Real Estate: Fernando Martínez
International (Am, As, Mi East):
Iván Barberá, Alfonso Vázquez
International (Europe):
Luis Blanc, Luis Lara
Logistics: Lorena Alba
Raw Material: José María Vandellós
Manufacturing Plants: Directors
Source: Inditex.
21
Page 24 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Amancio Ortega,
Chairman
604-081
ZARA: IT for Fast Fashion
Source:
Inditex.
Exhibit 12
Source:
Zara Point of Sales Terminal
Inditex Distribution Center (interior)
Inditex, formatted by casewriters.
22
Page 25 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Exhibit 11
ZARA: IT for Fast Fashion
Assumptions for Zara’s Upgrade Decision
Category
Value
Operating System for POS terminals (costs per computer/CPU)
Windows: One-time license cost
Annual maintenance fee
Unix
One-time license cost
Annual maintenance fee
Linux
One-time license cost
Service contracta
Hardware (per store, avg. 5 terminals needed per store)
POS Terminals
Wireless Router (1 per store)
Wireless Ethernet Card (1 per POS terminal)
Connectivity (annual cost per store)
High-speed Internet connection
Ä140
Ä30
Ä160
Ä25
Ä0
Ä10–Ä150
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Exhibit 13
604-081
Ä5,000
Ä180
Ä50
Ä240
Overall programming time required to:
Port existing POS application to new OS
Expand POS application to includeb
1. Lookups of same-store theoretical inventory
2. Lookups of other-store theoretical inventory
3. Inventory transfers
Cost per day of programming time
Time required per store to:
Install new POS terminals with new POS application
Establish wireless network
Train staff on new POS applicationc
Cost per day of installation/training time
15,000 hours
3,000 hours
1,000 hours
1,000 hours
Ä450
16 hours
8 hours
8 hours
Ä2,000
Source: Casewriters’ estimate, based on available industry data.
aDepends highly on IT staff's knowledge in Linux programming and maintenance.
bAssuming three features are developed in sequence, i.e., step 3 builds on step 2, which builds on step 1.
cAssuming new application contains inventory lookup functionality.
23
Page 26 of 282
S
w
9B13E006
Ken Mark wrote this case under the supervision of Professors Deborah Compeau, Craig Dunbar, and Michael R. King solely to
provide material for class discussion. The authors do not intend to illustrate either effective or ineffective handling of a managerial
situation. The authors may have disguised certain names and other identifying information to protect confidentiality.
Richard Ivey School of Business Foundation prohibits any form of reproduction, storage or transmittal without its written permission.
Reproduction of this material is not covered under authorization by any reproduction rights organization. To order copies or request
permission to reproduce materials, contact Ivey Publishing, Richard Ivey School of Business Foundation, c/o Richard Ivey School of
Business, The University of Western Ontario, London, Ontario, Canada, N6A 3K7; phone (519) 661-3208; fax (519) 661-3882; email cases@ivey.uwo.ca.
Copyright © 2013, Richard Ivey School of Business Foundation
Version: 2013-03-05
INTRODUCTION
May 18, 2012 should have been a great day for NASDAQ OMX. CEO Robert Greifeld had travelled to
Silicon Valley to “join Facebook executives in remotely ringing the market’s opening bell”2 as part of the
highly anticipated initial public offering (IPO) for the social media company. Having beat out New York
Stock Exchange (NYSE) Euronext to be the listing exchange for the offering, all eyes were on NASDAQ
as investors (and the media) waited to see what would happen to the Facebook stock.
But as the time for the launch neared, problems in confirming last-minute orders were reported by various
traders. A problem in one of NASDAQ’s software programs ended up causing a 30-minute delay in the
opening of trading and delays of up to three hours in confirming all of the orders and cancellations placed
by traders leading up to the opening.
Two months later, on July 20, NASDAQ management reviewed the compensation plan that it was filing
with the Securities and Exchange Commission (SEC). NASDAQ’s proposal to pay $62 million in cash to
compensate market makers for losses in Facebook share trading resulting from the technical glitches was
much larger than the $40 million initially announced a few weeks ago but still less than what was being
demanded by some institutional investors.
Would the compensation plan satisfy the institutional investors who had participated in the IPO? Would
the SEC view it as an adequate response? But, more fundamentally, how had this happened and what were
the long-term implications of this incident?
1
This case has been written on the basis of published sources only. Consequently, the interpretation and perspectives
presented in this case are not necessarily those of NASDAQ or any of its employees.
2
J. Strasburg, A. Ackerman, and A. Lucchetti, “NASDAQ CEO Lost Touch Amid Facebook Chaos,”
http://online.wsj.com/article/SB10001424052702303753904577454611252477238.html, 2012, accessed March 4, 2013.
Page 27 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
NASDAQ OMX: THE FACEBOOK DEBACLE1
Page 2
9B13E006
The global market for securities trading was huge. In 2011, there were close to 46,000 companies listed on
70 stock exchanges around the world with a combined stock market capitalization of $47.4 trillion. Trading
in these stocks took place on organized stock exchanges and in over-the-counter (OTC) markets. Annual
trading through the electronic order books managed by stock exchanges totaled $63 trillion, with 112
trillion trades with an average transaction size of $8,700.3 The two largest exchanges by turnover value —
the NYSE Euronext and the NASDAQ OMX — were headquartered in the United States and together
accounted for almost 50 per cent of global trading on exchanges in 2011. Exhibit 1 shows the number of
listed companies by exchange.
Stock exchanges were a central part of the global financial system and served many purposes. Companies
looking to raise capital sold shares to investors through IPOs or seasoned equity offerings (SEOs) with the
shares listed on stock exchanges. Trading in these shares allowed investors to place a value on a
company’s future prospects and, by extension, their firm. Investors — both institutional and retail — used
stock exchanges and their electronic infrastructure when buying and selling shares to achieve their
investment objectives. The transparency and liquidity provided by stock exchanges allowed investors to
monitor their portfolios during the day and to reallocate funds among different investments with minimal
trading costs. Dealers and traders at investment banks and stock brokers aimed to profit from differences in
views about the future value of a company.
Technological improvements over the past 40 years had transformed the way stocks were traded on
exchanges. Over the past decade, the pace of trading had increased with market players relying
increasingly on computer algorithms to automatically place orders to buy and sell stock with the execution
taking place in hundreds of a millisecond. This complex and fast-paced system was overseen by a
regulator, with the NASDAQ supervised by the SEC under the 1934 Securities Exchange Act. A glossary
of terms in this case can be found in Exhibit 2.
NASDAQ OMX GROUP
Created as a subsidiary of the National Association of Securities Dealers (NASD) in 1971, the NASDAQ
was the first electronic trading platform for U.S. stocks. NASDAQ was created to allow investors to buy
and sell stocks on a computerized, transparent and fast system that would eliminate the inefficiency of
manual stock transactions. NASDAQ originally functioned as a computer bulletin board for trading 2,500
OTC securities. In 1982, NASDAQ National Market System was launched, offering real-time trade
reporting for the 40 highest volume stocks. Then, in 1985, the NASDAQ-100 was founded as an index of
the largest non-financial companies listed on the NASDAQ based on their market capitalization. During
the 1980s and 1990s, the NASDAQ became the listing venue of choice for fast-growing technology
companies that could not qualify for a NYSE listing, attracting the initial listings for Apple (1980),
Microsoft (1986) and Cisco Systems (1990). By 1994, NASDAQ surpassed the NYSE in yearly share
volume. In 1998, NASDAQ secured its spot as the second largest U.S. stock exchange by acquiring the
American Stock Exchange (AMEX). NASDAQ finally went public in July 2002 with its shares trading
under the symbol “NDAQ.” In August 2006, the SEC finally recognized NASDAQ as a national securities
exchange, completing its transformation into a global leader among national stock exchanges.
3
http://www.world-exchanges.org/files/file/stats%20and%20charts/2011%20WFE%20Market%20Highlights.pdf, accessed
September 15, 2012.
Page 28 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
STOCK EXCHANGES: A MARKET FOR SECURITIES TRADING
Page 3
9B13E006
NASDAQ OMX Group delivered trading, clearing, exchange technology, regulatory, securities listing and
public company services across six continents. It generated $387 million in net income on revenues of $3.4
billion in 2011 with a return on equity of 6.8 per cent. NASDAQ managed trading for 3,400 global listed
companies representing a total of $5.3 trillion in market value. Globally, NASDAQ operated 24 stock
markets, three clearinghouses and five central securities depositories supporting all asset classes and the
entire lifecycle of a trade (Exhibit 3 shows the information processing requirements through the trading
lifecycle). In terms of transactions, stock trading volume had risen exponentially from 20,000 per second in
2007 to 500,000 per second in 2009 and to several millions per second by May 2012.4 Exhibit 4 shows
NASDAQ’s market share in trading. On May 2, 2012, NASDAQ unveiled what it called the world’s
fastest matching software, with “round trip latencies” of less than 40 microseconds.
NASDAQ classified its business into three segments: market services, issuer services and market
technology. Market services accounted for two-thirds of revenues in 2011. This segment generated fees
from trading, clearing and settlement services across several assets classes, including cash equities,
derivatives, debt, commodities, structured products and exchange-traded funds (ETFs). It also charged
market participants a fee to access NASDAQ’s proprietary markets. Issuer services generated around 22
per cent of revenues in 2011 from fees charged to companies that listed on the NASDAQ OMX exchanges.
This segment also managed NASDAQ’s stock market indices. Finally, the market technology segment
accounted for the remaining 11 per cent of revenues. NASDAQ leased its proprietary technology for
trading, clearing and settlement to 70 markets in 50 countries. Its clients were competing exchanges,
alternative-trading systems, banks and securities brokers with marketplace offerings of their own.
INFORMATION TECHNOLOGY IN STOCK EXCHANGES
The introduction of computerized trading “has gradually transformed the market from a humanintermediated market to a computer-mediated market with little human or real-time oversight . . .
Electronic technologies have profoundly altered how exchanges, brokers and dealers arrange most trades.
In some cases, innovative trading systems are so different from traditional ones that most political leaders
and regulators do not fully appreciate how they work and the many benefits that they offer to investors and
to the economy as a whole.”5
Before electronic trading, stock trading was conducted by brokers and market makers who haggled over
prices in person. Compared to the present day, stocks were traded infrequently and in large lot sizes.
NASDAQ launched the first electronic market in 1971, leading the NYSE to invest heavily in its own
electronic infrastructure. By 1978, the U.S. stock exchanges inaugurated the Intermarket Trading System
(ITS), which provided an electronic link between the NYSE and competing exchanges, enabling brokers to
4
http://en.community.dell.com/dell-blogs/direct2dell/b/direct2dell/archive/2012/05/17/it-ensuring-a-strong-financial-heartbeatat-nasdaq.aspx, accessed September 15, 2012.
5
James J. Angel, Lawrence E. Harris and Chester S. Spatt, “Equity Trading in the 21st Century,” Quarterly Journal of
Finance 1.1, 2011, p. 4.
Page 29 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
The next decade saw tremendous global consolidation among national stock exchanges, which merged to
gain economies of scope and scale in trading and listing. In 2006, NASDAQ made a failed hostile bid for
the London Stock Exchange in response to the NYSE's plan to acquire Paris-based Euronext NV. To
cement its global footprint, NASDAQ then agreed to a friendly merger with Sweden and Finland’s OMX
group, which controlled seven Nordic and Baltic stock exchanges. The merger closed in February 2008,
and the combined entity was renamed the NASDAQ OMX Group. Headquartered in New York with 30
global offices, the NASDAQ OMX had a market capitalization of $3.8 billion in mid-2012.
Page 4
9B13E006
Starting around the mid-1990s,6 traders began investing in high-frequency trading systems. Typical traders
looked to buy or sell a security a handful of times a day at most and to hold securities for days, weeks or
months. High-frequency trading systems, on the other hand, utilized algorithms to determine and execute
trading strategies. These systems held investments for a few seconds or less, looking to lock in profit of as
little of a fraction of a cent per share on each trade. High-frequency trading systems placed and cancelled
hundreds of orders per second throughout the trading day and sought to exploit shifts in demand for stocks
or tiny mis-pricings in securities. By 2012, two-thirds of all share trading volume on U.S. markets could be
traced to high-frequency trading systems.7
Modern electronic stock exchanges aimed to deliver increased matching speed and accuracy, reduce the
risk of human error and operate at a reasonable price per trade. Trading software aggregated buy and sell
requests for any of the stocks being traded on the market, then matched buyers with sellers as quickly as
possible. An article in the Quarterly Journal of Finance lists some benefits of computerized trading
systems:
. . . computerized trading systems and high-speed communications networks allowed
exchanges, brokers and dealers to better serve and attract clients. With these innovations,
transaction costs have dropped substantially over the years, and the market structure has
also changed dramatically. The winners first and foremost have been the investors who
now obtain better service at a lower cost from financial intermediaries than previously.
Secondary winners have been the exchanges, brokers and dealers who embraced electronic
trading technologies and whose skills allowed them to profitably implement them. The big
losers have been those intermediaries who did not innovate as successfully and, as a
consequence, became less competitive and ultimately less relevant.8
Bob McDowall, an analyst at TowerGroup added:
Without technology, exchanges could not accommodate the increased transaction flows
that are generated both by the proliferation of end investors and by electronic trading,
algorithms and low latency.9
IT Challenges for Exchanges
While technology supported exchange operations and securities trading in numerous ways, it also created
risks. The situation faced by NASDAQ on May 18 was not the first time that trading software malfunctions
had caused market issues.
6
Terrence Hendershott, Charles M. Jones and Albert J. Menkveld, “Does Algorithmic Trading Improve Liquidity?,”
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1100635, accessed September 15, 2012.
7
http://online.wsj.com/article/SB10000872396390443989204577599243693561670.html, accessed September 15, 2012.
8
Angel, Harris and Spatt, “Equity Trading in the 21st Century,” p. 3, accessed September 15, 2012.
9
http://www.computerweekly.com/news/2240083742/The-evolution-of-stock-market-technology, accessed September 15,
2012.
Page 30 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
access all markets. Then, in 1984, the NYSE introduced the Super DOT 250 to route orders electronically
between member firms and specialist trading posts staffed by designated market makers. With the
increasing availability of computer power, the use of computers to execute trades became popular in the
1980s. So-called “program trading” (now termed algorithmic trading) could execute trades much more
quickly than manual entries and thus improved both their speed and cost-effectiveness.
Page 5
9B13E006
There was substantial selling pressure on the NYSE at the open on Monday with a large
imbalance in the number of sell orders relative to buy orders . . . In this situation, many
specialists did not open for trading during the first hour . . . With stocks not trading, some
of the quotes used to construct market indexes were stale, so the values of these indexes
did not decline as much as they might have otherwise . . . By contrast, the futures market
opened on time with heavy selling. With stale quotes in the cash market and declining
prices in the futures market, a gap was created between the value of stock indexes in the
cash market and in the futures market. Index arbitrage traders reportedly sought to take
advantage of this gap by entering sell-at-market orders on the NYSE. When stocks finally
opened, prices gapped down and the index arbitragers discovered they had sold stocks
considerably below what they had been expecting and tried to cover themselves by buying
in the futures market. . . . As stocks opened notably lower, portfolio insurers’ models
prompted them to resume sales . . . Significant selling continued throughout the remainder
of the day with equity prices declining steeply during the last hour and a half of trading.12
Some 23 years later, a similar event occurred. On May 6, 2010, at about 2:45 p.m. in the afternoon, the
Dow Jones Industrial Average, already down by 300 points for the day, plunged another 600 points, or
over 5 per cent in five minutes. While the market recovered most of the 600 point drop by 3:07 p.m.,
observers were left wondering what had triggered what was later known as the Flash Crash. Some pointed
at the risk measures built into high-frequency trading systems as the reason for the crash. When a fund
company decided to hedge its stock market exposure by selling futures contracts, the steep rise in volume
triggered many trading systems to exit the market, accelerating the decline. As the SEC reported:13
At 2:32 p.m. . . . a large fundamental trader . . . initiated a sell program to sell a total of
75,000 E-Mini14 contracts (valued at approximately $4.1 billion) as a hedge to an existing
equity position . . . This large fundamental trader chose to execute this sell program via an
automated execution algorithm that was programmed to feed orders into the June 2010 EMini market to target an execution rate set to 9% of the trading volume calculated over the
previous minute, but without regard to price or time. . . . the sell pressure was initially
absorbed by high frequency traders [HFTs] and other intermediaries in the futures market;
fundamental buyers in the futures market; and cross-market arbitrageurs who transferred
this sell pressure to the equities market by opportunistically buying E-Mini contracts and
simultaneously selling products like SPY,15 or selling individual equities in the S&P 500.
10
Roberta S. Karmel, “The Rashomon Effect in the After-The-Crash Studies,” The Review of Securities & Commodities
Regulation, 21.12, June 22, 1988, p. 103.
11
Report by the Division of Market Regulation of the U.S. Securities and Exchange Commission, The October 1987 Market
Break, February 1988, pp. 3–11, as cited in Karmel, “The Rashomon Effect,” p. 104.
12
http://www.federalreserve.gov/pubs/feds/2007/200713/200713pap.pdf, accessed September 15, 2012.
13
http://www.sec.gov/news/studies/2010/marketevents-report.pdf, accessed September 15, 2012.
14
The E-mini S&P, known as the E-mini, was a stock market index futures contract.
15
SPY was an exchange-traded fund that tracked the S&P 500.
Page 31 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
On Monday, October 19, 1987, the Dow Jones Industrial Average fell by 508 points, or 22.6 per cent, in
what was later known as “Black Monday.” The Commodity Futures Trading Commission, which worked
with the SEC to regulate trading markets, identified the source of the “wave of selling:” a mutual fund’s
single trade of 17.5 million shares of stock on October 19, 1987.10 The SEC found that “the existence of
futures on stock indexes and the use of various strategies involving ‘program trading’ . . . were a
significant factor in accelerating and exacerbating the declines.”11 Mark Carlson of the board of governors
of the Federal Reserve recounted what happened on Black Monday:
Page 6
9B13E006
The selling pressure from the trading programs pushed the price of the E-Mini down by 3 per cent in four
minutes. Next, as the market was experiencing a lack of liquidity, HFTs began rapidly trading contracts to
each other. Without any fundamental or cross-market arbitrage buyers, the E-Mini prices continued to fall,
pulling down the S&P 500 with it.16
Just two months before the Facebook IPO, a systems malfunction caused BATS Global Markets to
withdraw its own IPO — trading on its own exchange — after a software glitch sent its own shares
plunging from $16 to less than one cent within nine seconds and Apple Inc.’s shares — also trading on the
BATS exchange — down by almost 10 per cent.17 BATS, a registered exchange provider with operations
in both the United States and Europe,18 had been founded in 2005 as an alternative trading system (ATS).
Having converted to a national securities exchange in 2008, by 2012 it was the third largest exchange in
the United States. Interestingly, BATS, like other national stock exchanges, was considered a quasigovernmental entity and thus had “absolute immunity on errors.”19
THE FACEBOOK IPO
With 900 million users around the world, Facebook was the dominant social networking site and was
poised to raise $5 billion in an IPO in May 2012. Launched in 2004, Facebook’s mission was “to make the
world more open and connected.”20 Its key strategic goal was to attract and retain as many active users as
possible so as to remain the top social media destination site on the planet. In 2011, Facebook generated $1
billion in net income from revenues of $3.7 billion. Touted as a “highly anticipated IPO”21 by many in the
media, the final offer price valued the firm at $104 billion. It would trade on the NASDAQ under the
symbol “FB.”
The lead-up to the launch had caused some controversy. Initially expected to be priced in the high $20s to
the mid-$30s per share, overwhelming interest from investors prompted the underwriters to boost the IPO
pricing range to between $28 and $35 per share. On May 11, 2012, CNBC reported that Facebook’s IPO
was “many, many” times oversubscribed, setting the stage for a push to price the shares at the high end of
the previously announced range.22 Buoyed by this enthusiasm, on May 14, 2012, the underwriters raised
the IPO price range to $34 to $38.23 Two days later, on May 16, 2012, Facebook announced that it would
16
http://www.sec.gov/news/studies/2010/marketevents-report.pdf, accessed September 15, 2012.
T. Lauricella, S. Pattterson and D. Benoit, “Trading Firm IPO Fizzles in Seconds,” Wall Street Journal, March 25, 2012,
http://online.wsj.com/article/SB10001424052702304636404577299560502440118.html, accessed September 15, 2012.
18
http://batsglobalmarkets.com, accessed February 15, 2013.
19
http://financialservices.house.gov/uploadedfiles/hhrg-112-ba16-wstate-dmathisson-20120620.pdf, accessed September
15, 2012.
20
Facebook S-1, May 16, 2012, p. 1.
21
http://www.ibtimes.com/facebooks-highly-anticipated-ipo-planned-may-18-wsj-694373, accessed September 15, 2012.
22
“Facebook IPO Said ‘Many, Many’ Times Oversubscribed — CNBC,” Dow Jones News Service, May 9, 2012,
http://www.djnewsplus.com/rssarticle/SB133675456713028774.html, accessed September 15, 2012.
23
“WSJ: Facebook Raises Price Range to $34 to $38 — Source,” Dow Jones News Service, May 14, 2012, accessed
September 15, 2012.
17
Page 32 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
. . . between 2:41 p.m. and 2:44 p.m., HFTs aggressively sold about 2,000 E-Mini
contracts in order to reduce their temporary long positions . . . [and] traded nearly 140,000
E-Mini contracts or over 33% of the total trading volume . . . The Sell Algorithm used by
the large trade responded to the increased volume by increasing the rate at which it was
feeding the orders into the market, even though orders that it already sent to the market
were arguably not yet fully absorbed by fundamental buyers or cross-market arbitrageurs.
Page 7
9B13E006
increase the size of its offering from 337.4 million shares to 421.4 million shares,24 with the additional
shares coming from existing shareholders. This offering was being viewed by some as expensive,
especially in the prevailing economic environment.25 One Morningstar analyst stated that “the valuation at
the proposed offer price leaves limited upside for long-term fundamental investors.”26
Some investors are just captivated with the idea of owning Facebook . . . The bankers are going to
do everything they can to ensure a successful offering. If this offering isn’t successful, it would not
just be bad for Facebook, it would be bad for the entire market.27
IT Problems with the Launch
On May 18, 2012, Facebook’s shares were ready to begin trading on the NASDAQ at 11:00 a.m. Reuters
described the events that followed:
Dead silence.
For nearly 20 minutes on the morning of Facebook Inc's trading debut last Friday, the line
NASDAQ had opened up to keep traders informed about the social media company's $16 billion
IPO had been mute. Well after the stock was supposed to have opened at 11 a.m. New York time,
no one from NASDAQ was talking — and there was still no sign of trading.
Finally, at 11:28 a.m., an unidentified person announced that the shares would open in about two
minutes. NASDAQ also said orders and cancellations were still being processed, according to
several sources listening to the call.
Those crucial 20 minutes created confusion that turned into chaos over the next few hours as
market makers — the brokers who quote bid and offer prices — struggled to figure out what was
happening. They were rebuffed in their attempts to get NASDAQ to halt trading and sort out a
growing number of problems.28
Even after the launch, issues continued to be reported. An issue with confirming transactions was noted at
11:59 a.m. Messages were not being properly delivered to the brokerages that placed orders. NASDAQ’s
updates indicated these had been sent by 1:57 p.m.
NASDAQ initially estimated that up to 30 million Facebook shares were affected and that the incident
would cost the stock exchange $13 million in claims. This figure rose to $40 million on June 6, 201229 and
24
“Facebook Boosts Size of IPO to 421.2M Shares,” Dow Jones News Service, May 16, 2012, accessed September 15,
2012.
http://www.bloomberg.com/news/2012-05-15/facebook-said-set-to-finish-taking-ipo-orders-tomorrow.html,
accessed
September 15, 2012.
26
“WSJ BLOG/MarketBeat: Curb Your Facebook IPO Enthusiasm, Morningstar Says,” Dow Jones News Service, May 11,
2012, accessed September 15,.2012.
27
http://www.bloomberg.com/news/2012-05-15/facebook-said-set-to-finish-taking-ipo-orders-tomorrow.html,
accessed
September 15, 2012.
28
http://www.reuters.com/article/2012/05/26/us-facebook-problems-idUSBRE84P00Y20120526, accessed September 15,
2012.
29
http://money.cnn.com/2012/06/06/investing/nasdaq-facebook/index.htm, accessed September 15, 2012.
25
Page 33 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
As such a highly anticipated offering, the IPO was recognized as critical. Dan Veru, chief investment
officer at Palisade Capital Management, noted:
Page 8
9B13E006
then to $62 million in late July 2012.30 In contrast, losses from one single broker, UBS AG, totaled $356
million. UBS announced plans to sue NASDAQ for damages.31
NASDAQ was relying on a program called IPO Cross, operating since 2006, to calculate an opening price
by aggregating auction bids. IPO Cross had been used in over 400 IPOs prior to the Facebook launch and
had been tested under “a hundred scenarios” for thousands of hours32 to ensure that it would stand up to
very high trading volumes.
Greifeld later described the program as “over-engineered” since it accepted orders right up until the
opening rather than having the orders halt for a short time during the lineup.33 The last step before issuing
the opening price was for IPO Cross to check for any additional orders placed and, if needed, recalculate
the opening price. As demand for Facebook stock had been overwhelming and pricing controversial, many
bids and cancellations continued to be submitted. This requirement to check and recalculate the final price
threw IPO Cross into an infinite loop, delaying the start of trading by 30 minutes.34
In an effort to circumvent the system, NASDAQ officials switched to a backup version of IPO Cross and
kicked off trading in Facebook stock at 11:30 a.m. But this backup version had relied on price calculations
made before 11:11 a.m., ignoring all buy and sell bids from 11:11 a.m. to 11:30 a.m. and resulting in
mixed signals to brokers unaware that orders in the 20 minutes prior to 11:30 a.m. had not been taken into
account. Orders for some 30 million shares were submitted during this time frame with about half
involving the possibility of dispute.35
DEVISING A PLAN
With the settlement plan ready for submission to the SEC, NASDAQ management wondered what, if
anything, they could have done to avoid this issue. Would more testing of the IPO Cross system have
eliminated the risk? NASDAQ also wondered whether the SEC would consider the $62 million proposal
adequate. Would investors? And what steps could NASDAQ take to avoid having this nightmare reoccur?
30
http://money.cnn.com/2012/07/23/investing/nasdaq-facebook/index.htm, accessed September 15, 2012.
http://online.wsj.com/article/SB10000872396390444405804577560220392935282.html, accessed September 15, 2012.
32
http://www.bloomberg.com/news/2012-05-20/nasdaq-ceo-says-poor-design-in-ipo-software-delayed-facebook.html,
accessed September 15, 2012.
33
Strasburg et al.
34
http://www.pcworld.com/article/255911/nasdaqs_facebook_glitch_came_from_race_conditions.html, accessed September
15, 2012.
35
http://www.bloomberg.com/news/2012-05-20/nasdaq-ceo-says-poor-design-in-ipo-software-delayed-facebook.html,
accessed September 15, 2012.
31
Page 34 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
In the days and weeks that followed the launch, explanations of the causes began to surface. By mid-July,
an understanding of the situation had crystallized.
Page 35 of 282
Exchange
14
365
100
83
199
131
2,472
1,796
227
3,654
1,919
5,047
947
243
1,403
421
1,790
1,557
1,270
251
899
1,195
462
753
541
2,279
2,007
5,047
955
243
1,420
421
1,808
1,558
1,271
253
899
1,195
782
785
541
2,291
21,476
Domestic
45
373
106
85
246
428
2,771
2,316
231
3,743
10,344
Total
88
NA
8
NA
17
18
1
1
2
NA
NA
320
32
NA
12
31
8
6
2
47
297
299
520
4
89
Foreign
WFE Total
Exchange
Europe - Africa - Middle East
Amman SE
Athens Exchange
BME Spanish Exchanges
Budapest SE
Casablanca SE
Cyprus SE
Deutsche Börse
Egyptian Exchange
Irish SE
Istanbul SE
Johannesburg SE
Ljubljana SE
London SE Group
Luxembourg SE
Malta SE
Mauritius SE
MICEX
NASDAQ OMX Nordic Exchange
NYSE Euronext (Europe)
Oslo Bl1Irs
Saudi stock Markel - Tadawul
SIX Swiss Exchange
Tel Aviv SE
Warsaw SE
Wiener Börse
Total region
TOTAL LISTED COMPANIES BY EXCHANGE
January 2012
Exhibit 1
45,178
277
280
3,319
51
74
110
755
229
59
264
395
71
2,954
288
21
63
250
775
1,133
236
146
293
611
596
108
13,358
Total
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Source: http://www.world-exchanges.org/focus/2012-01/m-5-4.php, accessed February 25, 2013.
Americas
Bermuda SE
BM&FBOVESPA
Buenos Aires SE
Colombia SE
lima SE
Mexican Exchange
NASDAQ OMX
NYSE Euronext (US)
Santiago SE
TMX Group
Total region
Asia- Pacific
Australian SE
Bombay SE
Bursa Malaysia
Colombo SE
Hong Kong Exchanges
Indonesia SE
Korea Exchange
National stock Exchange India
Osaka SE
Philippine SE
Shanghai SE
Shenzhen SE
Singapore Exchange
Taiwan SE Corp,
The stock Exchange of Thailand
Tokyo SE Group
Total region
Page 9
277
277
3,284
48
73
110
681
228
50
263
350
71
2,354
30
21
62
249
750
980
195
146
246
594
581
89
Domestic
NA
3
35
3
1
0
74
1
9
1
45
0
600
258
0
1
1
25
153
41
NA
47
17
15
19
Foreign
9B13E006
Page 10
9B13E006
Exhibit 2
Alternative Trading System
Electronic Communication
Networks
Intermarket Trading System
Round trip latency
Clearing and settlement services
Exchange traded funds
Algorithmic trading (or “algo”
trading)
High-frequency trading systems
Fundamental trader
E-minis
See Electronic Communications Networks. Some ATS operate
as dark pools “where buyers and sellers are matched
anonymously, without pre-trade display of bids and offers. . .
[T]he trade is publicly reported upon execution.”1
“Electronic Communications Networks, or ECNs . . . are
electronic trading systems that automatically match buy and
sell orders at specified prices. ECNs register with the SEC as
broker-dealers. . . . Those who subscribe to ECNs —
institutional investors, broker-dealers and market-makers —
can place trades directly with an ECN. Individual investors
must currently have an account with a broker-dealer subscriber
before their orders can be routed to an ECN for execution.”2
An “electronic communications network linking the trading
floors of seven registered exchanges to permit trading among
them in stocks listed on either the NYSE or AMEX and one or
more regional exchanges.”3
The time required for a packet of information, sent by the
source, to arrive at the destination and return back to the
source.
The series of steps performed after the trade has been made, to
finalize the trade. These include the settlement of trades,
reporting and monitoring functions.
Investment funds which hold a basket of securities and which
trade on stock exchanges.
The reliance on a program and electronic systems to execute
trades without human intervention.
A type of algorithmic trading that combines a customized
trading strategy, electronic data and electronic execution to
issue large numbers of orders quickly, taking advantage of
arbitrage opportunities in a short time window.
An investment professional or firm relying on equity analysis
to make investment decisions
E-minis are futures markets for stock market indices (e.g.,
S&P 500 E-mini, NASDAQ 100 E-Mini).
1
http://www.goldmansachs.com/media-relations/comments-and-responses/archive/market-structure-folder/alt-tradingsys.html, accessed February 28, 2013.
2
http://www.sec.gov/answers/ecn.htm, accessed February 28, 2013.
3
http://financial-dictionary.thefreedictionary.com/Intermarket+Trading+System, accessed February 28, 2013.
Page 36 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
GLOSSARY OF TERMS
Page 11
9B13E006
Exhibit 3
Source: http://www.arabstockexchanges.org/uaseportal/userfiles/file/Abdallah%20Al-Suweilmy(1).pdf, accessed December
1, 2012.
Page 37 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
INFORMATION PROCESSING IN EXCHANGE TRANSACTIONS
Page 38 of 282
Number
of
trading
days
20
23
21
21
21
21
20
20
22
20
22
21
Average
daily
volume
(millions)
2,358
2,552
2,119
2,126
1,884
1,639
1,850
1,926
1,704
1,733
1,914
1,801
Average
number of
trades
(thousands)
10,126
12,075
9,919
9,928
8,576
7,243
7,910
8,158
7,609
8,012
8,734
7,978
Median
number of
trades
(thousands)
8,174
10,486
9,748
9,624
8,765
7,733
7,746
7,941
7,499
8,053
8,810
7,871
Highest volume
day (millions)
2,432
27-Jul
4,117
8-Aug
2,997
22-Sep
3,169
4-Nov
2,531 30-Nov
2,909
16-Dec
2,126
16-Jan
2,218
29-Feb
2,171 16-Mar
2,035
19-Apr
2,661 18-May
3,476
22-Jun
Largest number of
trades in a day
(millions)
10,785
27-Jul
19,716
8-Aug
14,187
22-Sep
14,419
4-Nov
11,865
1-Nov
9,286
16-Dec
8,918
25-Jan
9,710
1-Feb
8,861
1-Mar
9,708
19-Apr
11,482
18-May
9,685
1-Jun
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Lowest volume
day (thousands)
1,592
5-Jul
1,672 29-Aug
1,630
2-Sep
1,641 10-Nov
753
25-Nov
976
27-Nov
1,696
3-Jan
1,652
13-Feb
1,365 12-Mar
1,402
9-Apr
1,294 25-May
1,408
8-Jun
NASDAQ VOLUME AND TRADE STATISTICS
July 2011 to June 2012
Exhibit 4
Source: The Center for Research in Security Prices database, WRDS, accessed February 20, 2013.
Month
Jul-11
Aug-11
Sep-11
Oct-11
Nov-11
Dec-11
Jan-12
Feb-12
Mar-12
Apr-12
May-12
Jun-12
Page 12
Lowest number
of trades in a
day (millions)
7,300
5-Jul
7,787 29-Aug
8,191
2-Sep
7,883 10-Nov
3,608 25-Nov
4,539 27-Nov
7,164 13-Jan
6,911 24-Feb
6,505 12-Mar
6,699
9-Apr
6,295 25-May
7,079 25-Jun
9B13E006
9B17E016
Chris F. Kemerer and Brian K. Dunn wrote this case solely to provide material for class discussion. The authors do not intend to
illustrate either effective or ineffective handling of a managerial situation. The authors may have disguised certain names and other
identifying information to protect confidentiality.
This publication may not be transmitted, photocopied, digitized, or otherwise reproduced in any form or by any means without the
permission of the copyright holder. Reproduction of this material is not covered under authorization by any reproduction rights
organization. To order copies or request permission to reproduce materials, contact Ivey Publishing, Ivey Business School, Western
University, London, Ontario, Canada, N6G 0N1; (t) 519.661.3208; (e) cases@ivey.ca; www.iveycases.com.
Copyright © 2017, Richard Ivey School of Business Foundation
Version: 2017-11-27
THE TABLES ARE TURNED ON NETFLIX
Reed Hastings, chief executive officer of Netflix Inc. (Netflix), was faced with another round of skeptical
business press as he attempted to grow his firm in 2017. His 1990s start-up business plan, which had
introduced the market to the convenience of home delivery of DVDs through the mail, had eviscerated the
prior market leader, Blockbuster LLC (Blockbuster), forcing it to divest itself of thousands of brick-andmortar video rental stores before finally falling into bankruptcy. Hastings was so successful that Fortune
magazine named him its 2010 “Businessperson of the Year.”2 This meteoric rise, however, seemed a distant
memory as Netflix focused on its transition to the digital delivery of video content. Digital delivery required
mastering new technologies and created the need to acquire or create popular content. Numerous
competitors, including both established mainstream content producers and digital upstarts, were making it
difficult for Netflix to recreate its earlier dominant success. The business press had become critical of
Netflix’s slowing acquisition of subscribers and its accelerating levels of debt, which had reached US$3.4
billion3 by March 2017.
Netflix was faced with the challenge of determining where and how quickly it would invest its capital in order
to continue its growth. Though the company had had early success in creating new, exclusive content (e.g.,
television series House of Cards and Orange Is the New Black), this apparent invincibility appeared to be
fading as more recent shows (e.g., The Get Down, Iron Fist) had been panned by the critics. As Netflix faced
increasing competition to acquire exclusive content, some of which came from companies with a long history
of success in content development, both the cost and the risk of failure seemed to be rising. In addition, former
content suppliers that had previously licensed content to Netflix now viewed Netflix more warily, given its
growth and apparent ambition to become more than just a delivery platform.4 Future deals would undoubtedly
be more expensive, if agreement could even be reached at all. Should Netflix continue to try to be a content
producer, competing with Hollywood’s industry leaders? Should it form a partnership with another media
company or companies to align everyone’s incentives? Should it consider moving into other media content
areas outside of traditional entertainment? Further, there remained the question of how to treat its legacy
DVD-by-mail business, a former cash cow that had been the subject of controversy both in the market and
internally at Netflix. Would the best choice be to sell the franchise and cash out? Netflix needed to decide
where, when, and how to invest so as to ensure its future, lest it suffer the same fate as Blockbuster.
Page 39 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
NETFLIX INC.: THE DISRUPTOR FACES DISRUPTION1
Page 2
9B17E016
To better understand Netflix’s situation, it is useful to understand the industry that existed before it entered
the marketplace. Although Sony Corporation (Sony) developed its Betamax videocassette recorder (VCR) in
1976, for U.S. consumers the key development in the growth of home video entertainment was a 1984 U.S.
Supreme Court 5–4 decision that found that the Betamax did not violate copyright laws.6 This was followed
by Congress affirming the copyright law’s “first sale” doctrine and refusing to pass a law proposed by
Hollywood studios that would have forbidden the renting or re-selling of movie videotapes. In short order,
the focus of VCR users shifted from recording broadcast shows onto blank tapes to buying or renting prerecorded media.7
The video rental industry, in which VCR owners could rent tapes with video content (e.g., movies), was
initially dominated by a variety of independent stores that had sprung up quickly in neighbourhoods
everywhere. In 1985, David Cook opened the first Blockbuster store in Dallas, Texas. Blockbuster brought
to the industry an aggressive strategy based on multiple stores and a central database that connected them.
These data allowed Blockbuster to more accurately forecast demand for videos, and its emerging economies
of scale kept its costs lower than that of its competition. By opening stores larger than those run by momand-pop operators, Blockbuster could offer customers more choices and more copies of popular movies.
By 1993, Blockbuster had grown to over 3,400 stores, an accomplishment it achieved despite some
hesitancy by the movie studios, which preferred the higher margins on individual consumer sales via outlets
like Best Buy, rather than the smaller margins available from Blockbuster’s rentals.
NETFLIX’S DISRUPTIVE INNOVATION8
Netflix Knocks Off Blockbuster
Netflix lore noted that Hastings’s desire to found Netflix started with a $40 Blockbuster late fee that he
incurred in 1997.9 As motivating as that might have been, the shift from bulky VHS videotapes to slimmer,
more durable DVDs was also necessary to make a mail-order model feasible. One limitation for Netflix in
1997 was that DVD players were new and expensive, and therefore had limited U.S. household adoption.
Blockbuster continued its growth of brick-and-mortar stores, and also began integrating DVDs into its
inventory. In 1999, Netflix hired Ted Sarandos from Blockbuster competitor West Coast Video, and he
focused on the content side of the business. Rapid growth in U.S. DVD adoption helped both firms—half
of all U.S. households owned a DVD player by 2003 and over 80 per cent of households adopted this
technology within the first nine years of its introduction. This adoption rate was even faster than that for
VCRs, which had taken 13 years to reach the same level.10
Netflix’s pricing evolved from a traditional fee per rental, just as in video stores, to a monthly subscription
model, at first limited to a fixed number of DVDs, then later evolving into a $20/month price for unlimited
rentals. Significantly, Netflix had no late fees—customers could watch their rentals at their convenience,
whereas late fees were Blockbuster’s primary revenue source. Netflix also touted the convenience of shopping
by mail, eliminating both the trip to the store to rent a title and the trip to return it.
However, Netflix found it challenging to provide an adequate number of high-demand titles to its
customers.11 Its solution was to be an early developer and exploiter of database personalization, using a
customer’s past rental history to suggest other titles they might like. Though keyed from past rental history,
the recommendation system was also biased towards titles that Netflix actually had in stock, and therefore
allowed Netflix to fulfill a greater percentage of orders. In addition, its relatively centralized inventory
permitted it to stock a greater variety of titles, which Blockbuster, with its inventory scattered across
Page 40 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
BACKGROUND: VIDEO RENTAL STORES AND THE RISE OF BLOCKBUSTER5
Page 3
9B17E016
In 2000, Netflix, with its 300,000 subscribers, was still not profitable. Meanwhile, DVD-player adoption
had helped fuel Blockbuster’s growth to some 7,700 stores. Hastings, who earlier in his career had started
but later sold the company Pure Software, had reported that in 2000 he engaged in negotiations with
Blockbuster to sell it interest in Netflix for $50 million: “We offered to sell a 49-per-cent stake and take the
name Blockbuster.com,” he said, but Blockbuster was not interested.12
Blockbuster would attempt to copy Netflix with an online service in 2004, hoping to leverage the
availability of its brick-and-mortar stores with the mail-order option and keep the operations integrated. It
was ultimately unsuccessful with its online offering, however, as were other contemporaneous imitators
such as Wal-Mart.
Netflix grew from 4.2 million subscribers in 2005 to 15 million in 2010 (see Exhibit 2). Although
Blockbuster still had 47 million registered customers, its eventual demise seemed apparent. Netflix had
more than doubled to 32 million subscribers by November of 2013—the month that Blockbuster announced
it was going out of business.
NETFLIX FACES DIGITAL DELIVERY
What Is Digital Delivery?
Information goods (i.e., any good that could be represented digitally) could be delivered over a communication
network such as the Internet. Initially, early Internet content was limited to lower-volume content, such as
text, and then later, images and music. Video content, with its relatively large file sizes, tended to be limited
to physical media, such as film, videotape, and discs. However, with advances in both network speeds and
compression algorithms to deliver content with greater fidelity using less data, other options emerged.
Traditionally, the first step in digital delivery was via downloaded content, such as Apple iPod users
downloading the complete file of any music they purchased from the iTunes online store. These files were
stored on the user’s device, and could be played repeatedly, thus mimicking physical media, such as vinyl
records, audio tapes, and music compact discs, which consumers were used to purchasing and owning.
Streaming content, on the other hand, was content that was played on the consumer’s device but stored on
a different device—typically a server operated by the content owner or licensee. Streaming required a
network connection between the two devices in order to play the content. In this way, it more closely
resembled listening to music on the radio, where the listener needed to have a connection to the content
(similar to radio frequency reception), as opposed to owning local copies.
Because video typically required much more data than music, streaming video required a faster network
connection in order to provide a reasonable viewing experience. As a consequence, some content owners
developed lower-resolution versions of their files in order to reduce their size, which then allowed them to
stream the content. This also avoided the problem of allowing consumers to possess a digital copy of their
content, which could potentially be shared or resold. Furthermore, the very large file sizes of full movies
acted as a partial deterrent to digital piracy (as did law enforcement efforts, such as the 2012 closure of the
popular website Megaupload).13
Page 41 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
thousands of individual retail locations, could not easily support. Netflix developed good relationships with
DVD suppliers, who saw it as another outlet for their product, and as a hedge against the growing nearmonopsonist Blockbuster. Netflix also worked closely on advanced integration with the U.S. Postal Service,
becoming one of its biggest customers.
Page 4
9B17E016
However, seeing the potential of streaming, in 2011 Netflix made what would be seen as an ill-fated, tooearly commitment to the digital delivery model when it announced a decision to split itself between its
traditional DVD business (to be re-named “Qwikster”) and an uncoupled streaming business, which
required separate customer logins. As it was to be operated as two separate companies with no bundled
discounts for those seeking both DVD rentals and streaming availability, the new pricing model would
mean a 60 per cent effective price hike for existing subscribers who wanted to continue both service types.
Predictably, the announcement was met with a loss of 800,000 subscribers, and Netflix’s stock price fell 77
per cent.16 Although Netflix reversed the decision a month later, considerable damage had been done.
This public relations disaster fed into a variety of contemporaneous pessimistic predictions about Netflix’s
future. One noted that Netflix shares were in “free fall” after the third-quarter mishaps, and projected future
losses.17 Another said “customers wasted little time in jumping ship.”18
Despite these misgivings, Netflix recovered and continued to grow after 2011, increasing its subscriber count
from 23.5 million to over 93 million by the end of 2016, a compound annual growth rate of 32 per cent.
REED HASTINGS’S DECISION POINTS
Digital Delivery Challenges
Regardless of the mode or degree of digital delivery adoption, digital delivery presented Netflix with
significant challenges in terms of streaming content acquisition. This was not a problem in the DVD-bymail delivery model, as DVDs could be acquired from distributors and then rented. However, acquiring
streaming content was both riskier and more expensive.
First, licensing content was subject to the terms dictated by the content owner. Netflix learned this lesson
early on, when its original streaming contract with the Starz network expired. Starz would not renew the
contract on terms acceptable to Netflix, leaving Netflix with a hole in its available inventory.19
Second, the home entertainment market was a very crowded one, with 12 streaming competitors to Netflix
that had each surpassed 1 million subscribers.20 One significant competitor was Hulu, founded by The
National Broadcasting Company (NBC) and 21st Century Fox (Fox) in 2007, and later joined by the
Disney–ABC Television Group and Time Warner Inc. It operated on a subscription model, but also offered
advertisement-supported options. Fox streamed content 24 hours a day through Hulu, in some cases even
bypassing its local affiliates.21 As of July 2017, Hulu had the greatest viewer engagement of all streaming
services, measured at 2.9 hours per day (versus Netflix’s 2.2 hours per day),22 and its content included the
shows Casual and The Handmaid’s Tale. Another streaming competitor was Amazon Prime Video.
Amazon initially offered its Amazon Prime Video service as a bonus for subscribers to its fee-based Prime
membership service. Doing so provided the video streaming service a large number of subscribers at launch.
Further, the service benefited from Amazon’s financial power and multiple lines of business (e.g., Amazon
Web Services). Amazon spent more than $100 million on content in the third quarter of 2014 alone, and its
portfolio included the television series Transparent and The Man in the High Castle. In addition, its Fire
TV platform grew with each device it sold (see Exhibit 3 for additional details).
Page 42 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Netflix initially thought the future of digital video delivery would be downloading, not streaming.14 It
created a dedicated device (that later was sold and became the Roku box) for this purpose, but eventually
developed a standalone streaming service, which was released as “Watch Instantly” in 2008.15 Early
attempts at streaming were low quality, and in any event, most potential customers did not have the required
Internet bandwidth.
9B17E016
Traditional web giants that competed with Netflix included both Google LLC/YouTube LLC (YouTube), and
Facebook Inc. (Facebook). While 75 per cent of U.S. Wi-Fi households had a Netflix subscription, YouTube
had quickly grown into second place in 2017 with 53 per cent penetration, and reached more of the coveted
18–34-year-old demographic than any cable network.23 Its content included the series Escape the Night.
Facebook, late to the sector, had also announced plans to offer new proprietary content, and tried to make up
for its late start by offering Hollywood studios better terms for their content, including sharing advertising
revenue and data on viewership, the latter of which was an important asset that had distinguished the success
of digital platforms such as Netflix.24
Of course, Netflix also competed with traditional home entertainment outlets, including Home Box Office
(HBO) and Sling TV, Dish’s live TV service. HBO had an extensive library of shows to watch on demand
via cable, and it had an app (HBO Go). It had a large number of subscribers and a strong, extended track
record of developing very popular original content, such as television series The Sopranos and Game of
Thrones. Sling TV, although it had fewer subscribers than Netflix in 2017, had viewers who were more
engaged, averaging 47 viewing hours per month, versus Netflix’s 28 hours per month.25
Many of these competitors fell into the category of “coopetition” to Netflix, a term that suggests cooperation between competing companies. Netflix licensed content from some of the same companies that
formed Hulu, but at the same time provided a competing service. Further, it relied on companies like Sony,
which had its own streaming service, to carry the Netflix app on Sony’s PlayStation videogame consoles.
Lastly, Netflix relied on the availability of bandwidth from cable companies to provide its services to
customers; many of these same cable companies (e.g., Comcast, Time Warner Inc.) were also content
producers with an interest in streaming services, and with which Netflix competed.
Major news outlets, such as The New York Times and The Wall Street Journal, reported that Netflix was seen
as a rival to firms from which it had previously licensed content, and that therefore deals were being cut
back.26 The Wall Street Journal wrote, “[Netflix] must keep things cordial with Hollywood’s traditional
studios . . . [and] prevent them from turning to its main competitors, Amazon and Hulu.” 27 It only had the
one source of revenue, but in its new market, it was competing with firms that had much more diverse
portfolios and revenue streams.
For example, if Netflix wanted to acquire some popular new show, it most likely would have to borrow
money from the bank; if Amazon wanted to acquire it, it could borrow money from the Amazon Web
Services division (of which, somewhat ironically, Netflix was a customer).28 Apple Inc., another financially
powerful competitor with $260 billion in cash in 2017,29 was also investing in content creation, most
recently poaching two senior executives from Sony TV.30
As an alternative to licensing third-party content, in 2013 Netflix ventured into developing its own original
content. It started out with a “home run” when it acquired House of Cards, starring Kevin Spacey, which went
on to win three Emmys.31 Given its lack of a track record at the time, Netflix needed to commit to buying the
entire season of the show without the option of declining after a pilot episode, as traditional media outlets usually
required. It also did well with Orange Is the New Black, a popular series based on a book, as well as the 1980sstyle adventure series Stranger Things. However, and to illustrate the risks, a more recent effort, Crouching
Tiger, Hidden Dragon: Sword of Destiny, was widely regarded as a flop, and Netflix found itself increasingly
cancelling original series that did not produce sufficient returns, such as Sense8 and The Get Down.32
In addition, there were significant sums at risk in these ventures. Netflix budgeted $6 billion in 2017 for
content—more than twice its total revenue. This level of spending resulted in $3.4 billion in long-term
debt.33 Critics said that its debt was such that it must continue to add more subscribers just to feed the
Page 43 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Page 5
Page 6
9B17E016
business model, or everything could come crashing down.34 Whereas Netflix’s streaming content
obligations were approximately equal to its revenue in 2011, by 2017 a significant gap had emerged.35
One consistent challenge confronted by all successful organizations facing significant technological change
was the allure of their existing incumbent line of business. This took several forms, including financial
investment in the existing capital infrastructure, organizational investment in skills and routines, and the
risk associated with moving away from a consistent, well-understood source of cash flow.
Author Joshua Gans36 termed this challenge “supply side disruption,” about which, drawing on the
academic research of Rebecca Henderson,37 he said,
Demand side disruption involves an established firm missing a certain kind of technological
opportunity, but supply side disruption arises when an established firm becomes incapable of taking
advantage of a technological opportunity. Specifically, when a new competing innovation involves a
distinct set of architectural knowledge, established firms that have focus on being “best in breed” in
terms of component innovation may find it difficult to integrate and build on the new architecture.38
At Netflix, those responsible for the legacy business had successfully argued for its re-investment. The
finely tuned industrial-engineered manual process of opening and stuffing the trademark red mailing
envelopes was replaced by special-purpose robots that could process 3,400 envelopes per hour versus the
680 per hour processed the prior, labour-intensive way.39 This enabled Netflix to reduce the number of
distribution centres from 50 to 33. These efficiencies resulted in cutting 75 per cent of its labour costs while
retaining a high degree of customer service—92 per cent of its customers received next-day delivery
service. Competitors also continued to see promise in the DVD format, as Redbox, with its trademark
automated kiosk vending machines, grew from about 100 locations in 2004 to 34,000 by 2012.40
One argument for continued investment in the legacy business was that, as recently as August 2015, fewer
than half of all U.S. homes had high-speed broadband service, and it could be expected that the currently
un-served lower-population-density areas might be without this service for the foreseeable future. This lack
of connectivity contributed to the continued success of brick-and-mortar video rental stores in these areas.41
This primarily rural phenomenon in the United States could presage markets in international locations with
limited broadband that might also continue to support the DVD format.42 And, when it temporarily split off
its DVD business in 2011, Netflix lost a number of its own senior managers.43
On the other hand, Netflix did not face the sort of business cannibalization challenges often faced by
traditional media businesses with the advent of a disruptive technology. Netflix, for example, did not face
the loss of advertising revenue as a result of a shift to streaming delivery, a potentially significant problem
for some of its competitors.44 Further, despite its increased internal efficiencies, its traditional DVD-based
business faced other costs, such as its dependence on the U.S. Postal Service, which had recently resulted
in a $100-million increase in mailing rates.45
Uncertain Future Public Policies—“Net Neutrality” and Its Variants
In 2003, Columbia University media law professor Tim Wu coined the term “net neutrality” to describe a
government regulatory principle that Internet service providers (ISPs) should enable access to all content and
applications regardless of the source, without favouring or blocking particular products or websites. It gained
Page 44 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
In-House Competition; the Challenge of Incumbency
9B17E016
in visibility as Internet traffic became more and more concentrated over time, with Wired magazine reporting
in 2014 that half of all Internet traffic originated from just 30 organizations, including Google LLC, Facebook,
and Netflix.46 The policy required that ISPs not discriminate (e.g., “throttling back” throughput speeds) or
charge differentially by user, content, website, platform, application, type of attached equipment, or mode of
communication. This controversial policy tended to find favour with Internet firms, especially those that
generated a lot of traffic, like Netflix. It tended to be opposed by ISPs, such as Comcast and Verizon, which
argued that by limiting their decision rights, the effect of the regulation would be to reduce investment in
broadband network services and, therefore, to stifle innovation. In 2015, the Democratic Party administration
in the United States enacted the rules into law, effectively making broadband firms “common carriers,” like
the telephone network. In 2017, the Republican Party administration scaled back the regulation.47
Netflix, similar to other Internet firms, originally lobbied for net neutrality, as streaming video required
significant amounts of bandwidth, and limiting ISPs’ ability to differentially charge for this service
effectively subsidized the Internet firms’ activities. By 2017, at least publicly, Hastings argued that it was
no longer especially important.48 However, the uncertainty about the regulation and its future added risk to
any decision-making in the digital delivery market.
In addition, as the telecommunications sector of the economy grew and prospered, it became an attractive
area for lawmakers to target for new taxes. In 2017, Canadian legislators proposed, but did not pass, a 5 per
cent tax on broadband charges, dubbed by the press the “Netflix tax.”49 These monies would have been
redistributed to traditional communications firms, like newspapers, that were adversely affected by the rise
of the Internet. Traditional media firms lost revenue due to the Internet—particularly classified advertising,
but also subscription revenue—as some of their customers shifted use to their Internet equivalents. If
broadband revenue would be ultimately subject to additional taxes, this could decrease demand for it and
thereby hamper the growth of Netflix-like streaming services.
HASTINGS’S DECISIONS
Hastings faced a significant number of business decisions, and, despite Netflix’s successes, he was selfcritical, and was often quoted as admitting to having made a number of, at least in hindsight, poor decisions
in relation to the technology and home entertainment industries. These included online advertisements on the
website, starting an independent film production company, and buying DVDs out of the Sundance Film
Festival (which turned out to offer limited profitability), in addition to the 2011 Qwikster gaffe.50 In addition,
he had initially believed that the future would be digital downloading rather than digital streaming, and in
2009 Hastings said that there would still be DVDs in 2030, a prediction that looked increasingly unlikely.51
However, despite some of these missteps, Netflix was positioned to compete with some of the biggest firms
in both entertainment and consumer technology. Hastings himself noted that, historically, firms facing
disruption, such as AOL Inc. and the Eastman Kodak Company, failed because they were too cautious.52
But which non-cautious path should Netflix take? Should it continue to create its own content, or revert
back to being a neutral platform? Should it look to form exclusive contracts with content providers and/or
hardware manufacturers to “lock in” its customers? Did it need to acquire competitors and/or upstream or
downstream partners? What was the appropriate role for the legacy DVD division—should it continue to
be operated as a cash cow for as long as it was economically viable, or should it be sold off so as to reinvest
the funds in current operations? Or, was it finally time, as suggested by analysts who argued the firm was
over-valued, to sell the company and cash out while it had a high market capitalization?
Page 45 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Page 7
Page 8
9B17E016
Harvard Business School’s Clayton Christensen has authored a number of books and articles defining and
illustrating what he terms “disruptive innovations,” which are new products or services that re-make whole
industries, generally at the expense of the incumbent firm. This occurs because the incumbent firm does not
recognize the threat posed by the innovation due to two main reasons. First, the disruptive innovation
possesses a different package of performance attributes, not all of which are valued by existing customers;
therefore, the incumbent’s current customers do not initially find it attractive. Second, the performance
attributes of the innovation that existing customers do value, which are often weak in the innovation, improve
at such a rapid rate that the new entrant can later move up and capture the incumbent’s existing customers.
Disruptive innovators typically either enter at the low end of the market, or create entirely new markets.
Both of these scenarios are characterized as markets not well served by current solutions. Incumbents,
therefore, have a tendency not to take serious notice of the innovation, because either they are not losing
current customers to it, or, when they are, they are customers who buy low-margin products. Losing such
customers has the paradoxical effect of increasing the incumbent’s average margin, thus improving its
bottom line. As such, the financial signals that incumbents receive create Christensen’s “dilemma,” whereby
incumbent firms often do not recognize the threat.
In addition, incumbents often have difficulty adopting disruptive innovations themselves because, as these
innovations initially serve the existing customers poorly, to adopt them could mean the loss of existing
revenues through cannibalization. Conversely, new entrants have no current customers to lose, removing
any hesitation to invest in the innovation.
Source: Clayton Christensen, The Innovator’s Dilemma (New York, NY: Harper Business, 2001).
EXHIBIT 2: NETFLIX SUBSCRIBER GROWTH (MILLIONS) OVER TIME
100
90
80
70
60
50
40
30
20
10
0
2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016
Source: “Netflix Subscribers from 2001 to 2011 (in 1,000),” Statista, accessed November 12, 2017,
https://www.statista.com/statistics/272551/subscribers-of-netflix-since-2001/; Jeff Dunn, “Here’s How Huge Netflix Has Gotten
in the Past Decade,” Business Insider, January 19, 2017, accessed July 2017, www.businessinsider.com/netflix-subscriberschart-2017-1.
Page 46 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
EXHIBIT 1: CHRISTENSEN’S DISRUPTIVE INNOVATION CONCEPT
Page 47 of 282
Subscription
Subscription
$30.00/month
(45+ channels)
$10.99/month
$20.00/month
(20+ channels)
Oct. 2014
Nov. 2016
Apr. 2015
Oct. 2007
Mar. 2015
Jun. 2015
(streaming
service)
Jan. 2015
CBS All
Access
DirecTV
Now
HBO Now
Hulu
PlayStation
Vue
Showtime
Sling TV
Live TV channels,
VOD
Showtime content,
including current
season
Live TV channels
(limited major
networks)
n/a
331 movies, 34 TV
shows (June 2017)
n/a
1,108 movies,
1,132 TV shows
(June 2017)
On-demand network
programming (except
current season of
CBS programming)
Subscription
(with or without
ads)
Subscription,
purchase, rental
681 movies, 90 TV
shows (June 2017)
HBO content,
including current
season
Subscription
Dish Network
CBS
1.3 million (April
2017)13
1.5 million (February
2017)12
400,000 (March
2017)11
12 million
(November 2016)10
Joint venture
of Disney, Fox,
Comcast/NBC
Universal, and
Time Warner
Sony/Columbia
2 million (February
2017)9
400,000 (March
2017)8
1.5 million (February
2017)7
n/a
80 million in the
United States (April
2017)6
99 million worldwide
(April 2017),3 50.9
million in the United
States (March
2017)4
Subscribers
9B17E016
Time Warner
AT&T
CBS
Apple
Amazon
Netflix
Ownership
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
$7.99/month
$14.99/month
70 TV shows (June
2017)
Live and on-demand
CBS programming,
some original content,
classic CBS content
n/a
Subscription
$35.00/month
(100+
channels)
n/a
VOD content
Live TV channels
Subscription
$5.99/month
Purchase, rental
n/a
Apr. 2003
(iTunes)
Subscription,
purchase, rental,
ad-supported
(limited content)
Apple TV
36,000 hours (April
2016);5 7,430
movies, 513 TV
shows (June 2017)
On-demand TV, ondemand movies,
exclusive content,
video-on-demand
(VOD) content
(additional fee)
Subscription
$79.00/year
36,000 hours (April
2016);2 4,235
movies, 971 TV
shows (June 2017)
On-demand TV (past
seasons), on-demand
movies, exclusive
content
Sep. 2006
Subscription
Content Amount1
Content
Revenue
Sources
Amazon
Prime
Video
$8.00/month
Subscription
Price
Aug. 1997
Founded
EXHIBIT 3: NETFLIX AND ITS COMPETITORS, 2017
Netflix
Page 9
Page 48 of 282
EXHIBIT 3 (CONTINUED)
9B17E016
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Note: n/a = not available
1
“Movie/TV Show Counts,” JustWatch, accessed July 1, 2017, www.justwatch.com.
2
Mark Fahey, “Netflix vs Amazon: Estimating the Better Deal,” CNBC, April 22, 2016, accessed July 2017, www.cnbc.com/2016/04/22/netflix-vs-amazon-estimating-the-betterdeal.html.
3
Seth Fiegerman, “Netflix Nears 100 Million Subscribers,” CNN, April 17, 2017, accessed July 2017, http://money.cnn.com/2017/04/17/technology/netflixsubscribers/index.html.
4
“Number of Netflix Streaming Subscribers in the United States from 3rd Quarter 2011 to 2nd Quarter 2017 (in Millions),” Statista, accessed July 2017,
https://www.statista.com/statistics/250937/quarterly-number-of-netflix-streaming-subscribers-in-the-us.
5
Mark Fahey, “Netflix vs Amazon: Estimating the Better Deal,” CNBC, April 22, 2016, accessed July 2017, www.cnbc.com/2016/04/22/netflix-vs-amazon-estimating-the-betterdeal.html.
6
Stephanie Pandolph and Jonathan Camhi, “Amazon Prime Subscribers Hit 80 Million,” Business Insider, April 27, 2017, accessed July 2017,
www.businessinsider.com/amazon-prime-subscribers-hit-80-million-2017-4. The figure is for Amazon Prime subscribers, a service that includes Amazon Prime Video.
7
Todd Spangler, “Showtime Hits 1.5 Million Streaming Subscribers, CBS All Access Nears Same Mark,” Variety, February 13, 2017, accessed July 2017,
http://variety.com/2017/digital/news/showtime-cbs-all-access-streaming-1-5-million-subscribers-1201986844.
8
Artie Beaty, “Discovery CEO: DIRECTV Now & Vue Have 400,000 Subscribers,” Streaming Observer, March 29, 2017, accessed July 2017,
https://www.streamingobserver.com/discovery-ceo-directv-now-vue-400000-subscribers.
9
Cynthia Littleton, “HBO Now Grows to More Than 2 Million Domestic Subscribers,” Variety, February 8, 2017, accessed July 2017, http://variety.com/2017/tv/news/hbo-now2-million-subscribers-time-warner-1201981371.
10
Craig Smith, “18 Interesting Hulu Statistics (July 2017),” DMR, July 28, 2017, accessed July 2017, http://expandedramblings.com/index.php/hulu-statistics.
11
Artie Beaty, op. cit.
12
Todd Spangler, op. cit.
13
Daniel Frankel, “Sling TV Ended Q1 with 1.3M Subscribers, Dish Is Reportedly Telling Wall Street,” FierceCable, April 28, 2017, accessed July 2017,
www.fiercecable.com/online-video/sling-tv-ended-q1-1-3m-subs-dish-reportedly-telling-wall-street.
Page 10
Page 11
9B17E016
ENDNOTES
This case has been written on the basis of published sources only. Consequently, the interpretation and perspective
presented in this case are not necessarily those of Netflix Inc. or any of its employees.
2
Patricia Sellers, “The Man Who Became Fortune’s Businessperson of the Year,” Fortune, November 19, 2010, accessed
September 27, 2017, http://fortune.com/2010/11/19/the-man-who-became-fortunes-businessperson-of-the-year.
3
All currency amounts are in USD unless otherwise specified.
4
Joe Flint and Shalini Ramachandran, “Netflix: The Monster That’s Eating Hollywood,” The Wall Street Journal, March 24,
2017, accessed July 2017, https://www.wsj.com/articles/netflix-the-monster-thats-eating-hollywood-1490370059.
5
Ken Auletta, “Outside the Box,” The New Yorker, February 3, 2014, accessed July 2017,
https://www.newyorker.com/magazine/2014/02/03/outside-the-box-2.
6
Eduardo Porter, “Copyright Ruling Rings with Echo of Betamax,” The New York Times, March 26, 2013, accessed July 2017,
www.nytimes.com/2013/03/27/business/in-a-copyright-ruling-the-lingering-legacy-of-the-betamax.html.
7
Richard Roehl and Hal R. Varian, “Circulating Libraries and Video Rental Stores,” First Monday 6, no. 5 (2001); Stan
Liebowitz, Re-Thinking the Network Economy (New York: AMACOM, 2002), 194.
8
See Exhibit 1.
9
“Netflix: How a $40 Late Fee Revolutionized Television,” YouTube video, 12:52, posted by “Business Casual,” accessed July
2017, https://www.youtube.com/watch?v=BrpEHssa_gQ.
10
“VCR
&
DVD
Player
Household
Penetration
(1980-2007),”
accessed
July
2017,
https://www.flickr.com/photos/42182583@N00/2299078244.
11
Emily Steel, “Netflix Refines Its DVD Business, Even as Streaming Unit Booms,” The New York Times, July 27, 2015,
accessed July 2017, https://www.nytimes.com/2015/07/27/business/while-its-streaming-service-booms-netflix-streamlinesold-business.html.
12
Ken Auletta, op. cit.
13
Michael D. Smith and Rahul Telang, Streaming, Sharing, Stealing (Cambridge, MA: The MIT Press, 2016).
14
Joe Nocera, “Can Netflix Survive in the New World It Created?,” The New York Times Magazine, June 15, 2016, accessed
July 2017, https://www.nytimes.com/2016/06/19/magazine/can-netflix-survive-in-the-new-world-it-created.html.
15
“Netflix: One Eye on the Present and Another on the Future,” Knowledge@Wharton, October 28, 2009, accessed July 2017,
http://knowledge.wharton.upenn.edu/article/netflix-one-eye-on-the-present-and-another-on-the-future.
16
Greg Sandoval, “Netflix’s Lost Year: The Inside Story of the Price-Hike Train Wreck,” CNET, July 11, 2012, accessed July
2017, https://www.cnet.com/news/netflixs-lost-year-the-inside-story-of-the-price-hike-train-wreck.
17
Larry Dignan, “Netflix’s Big Collapse: Do You Believe in Streaming, International Expansion?,” ZDNet, October 25, 2011,
accessed July 2017, www.zdnet.com/article/netflixs-big-collapse-do-you-believe-in-streaming-international-expansion.
18
“Netflix’s Tough Transition to Online Movie Streaming,” digitalsurgeons, October 3, 2011, accessed July 2017,
https://www.digitalsurgeons.com/thoughts/strategy/netflixs-tough-transition-to-online-movie-streaming.
19
Ibid.
20
Joe Flint and Deepa Seetharaman, “Facebook Is Going Hollywood, Seeking Scripted TV Programming,” The Wall Street
Journal, June 25, 2017, accessed July 2017, https://www.wsj.com/articles/facebook-is-going-hollywood-seeking-scripted-tvprogramming-1498388401.
21
Shalini Ramachandran, “Fox Tries to Gain Leverage Over Affiliates on Live Streaming,” The Wall Street Journal, June 12,
2017, accessed July 2017, https://www.wsj.com/articles/fox-tries-to-gain-leverage-over-affiliates-on-live-streaming1497261600.
22
Sarah Perez, “U.S. Cord Cutters Watch More Netflix than Amazon Video, Hulu and YouTube Combined,” TechCrunch, July
5, 2017, accessed July 2017, https://techcrunch.com/2017/07/05/u-s-cord-cutters-watch-more-netflix-than-amazon-videohulu-and-youtube-combined.
23
Sarah Perez, “Netflix Reaches 75% of US Streaming Service Viewers, but YouTube Is Catching Up,” TechCrunch, April 10,
2017, accessed July 2017, https://techcrunch.com/2017/04/10/netflix-reaches-75-of-u-s-streaming-service-viewers-butyoutube-is-catching-up.
24
M. Smith and R. Telang, op. cit.
25
Sarah Perez, “Netflix Reaches 75% of US Streaming Service Viewers, but YouTube Is Catching Up,” op. cit.
26
Joe Nocera, op. cit.
27
Joe Flint and Shalini Ramachandran, op. cit.
28
Jay Greene and Laura Stevens, “Wal-Mart to Vendors: Get Off Amazon Cloud,” The Wall Street Journal, June 21, 2017,
accessed July 2017, https://www.wsj.com/articles/wal-mart-to-vendors-get-off-amazons-cloud-1498037402
29
Steve Vassallo, “It’s Time for Apple to Go Hollywood,” The Wall Street Journal, June 20, 2017, accessed July 2017,
https://www.wsj.com/articles/its-time-for-apple-to-go-hollywood-1497998797.
30
Tripp Mickle and Joe Flint, “Apple Poaches Sony TV Executives to Lead Push into Original Content,” The Wall Street
Journal, June 16, 2017, accessed July 2017, https://www.wsj.com/articles/apple-poaches-sony-tv-executives-to-lead-pushinto-original-content-1497616203.
31
Peter Cohan, “Netflix’s Reed Hastings Is the Master of Adaptation,” Forbes, October 22, 2013, accessed July 2017,
https://www.forbes.com/sites/petercohan/2013/10/22/netflixs-reed-hastings-is-the-master-of-adaptation.
32
David Trainer and Kyle Guske II, “Netflix’s Stock Is Worth Only About One-Third of Where It Trades Today,” MarketWatch,
June 18, 2017, accessed July 2017, www.marketwatch.com/story/netflixs-stock-is-worth-only-about-one-third-of-where-ittrades-today-2017-06-15.
Page 49 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
1
Page 12
9B17E016
Trey Williams, “Netflix’s Global Blitz Should Lift Subscriptions 20% This Year,” The Wall Street Journal, March 6, 2017,
accessed July 2017, www.marketwatch.com/story/netflixs-global-blitz-should-lift-subscriptions-20-this-year-2017-03-06.
34
Joe Nocera, op. cit.
35
David Trainer and Kyle Guske II, op. cit.
36
Joshua Gans, The Disruption Dilemma (Cambridge, MA: MIT Press, 2016).
37
Rebecca M. Henderson and Kim B. Clark, “Architectural Innovation: The Reconfiguration of Existing Product Technologies
and the Failure of Established Firms,” Administrative Science Quarterly 35, no. 1 (1990): 9–30.
38
Joshua Gans, op. cit.
39
Emily Steel, op. cit.
40
Kirsten Acuna, “8 Ways Online Streaming Killed the Video Store”, Business Insider, February 5, 2013, accessed July 2017,
www.businessinsider.com/online-streaming-is-making-the-dvd-obsolete-2013-1.
41
Mark Perry, “‘The ‘Netflix Effect’: An Excellent Example of ‘Creative Destruction,’” AEI, August 6, 2015, accessed July 2017,
www.aei.org/publication/the-netflix-effect-is-an-excellent-example-of-creative-destruction; E. G. Smith, “Revenge of the Video
Store,” The Wall Street Journal, November 26, 2016, accessed July 2017.
42
Emily Steel, op. cit.
43
Greg Sandoval, op. cit.
44
Ken Doctor, “The Newsonomics of Netflix and the Digital Shift,” NiemanLab, July 28, 2011, accessed July 2017,
www.niemanlab.org/2011/07/the-newsonomics-of-netflix-and-the-digital-shift.
45
Peter Cohan, op. cit.
46
Robert McMillan, “What Everyone Gets Wrong in the Debate Over Net Neutrality,” Wired, June 23, 2014, accessed July
2017, https://www.wired.com/2014/06/net_neutrality_missing; the article argues that the Internet already was not “neutral,”
with ISPs already providing specific resources to handle high-traffic sites.
47
John McKinnon, “FCC Votes to Scale Down Net Neutrality Rules,” The Wall Street Journal, May 18, 2017, accessed July
2017, https://www.wsj.com/articles/fcc-votes-to-scale-down-net-neutrality-rules-1495124194.
48
Erin Carson, “Net Neutrality May Have Lost Netflix as an Ally,” CNET, May 31, 2017, accessed July 2017,
https://www.cnet.com/news/net-neutrality-netflix-reed-hastings.
49
Daniel Leblanc, “Trudeau Rejects New Internet Tax to Help Fund Media Sector,” The Globe and Mail, June 15, 2017,
accessed July 2017, https://beta.theglobeandmail.com/news/politics/liberal-ndp-mps-call-for-new-tax-on-internet-providersto-help-media-companies/article35315234.
50
James B. Stewart, “Netflix Looks Back on Its Near-Death Spiral,” The New York Times, April 26, 2013, accessed July 2017,
www.nytimes.com/2013/04/27/business/netflix-looks-back-on-its-near-death-spiral.html.
51
Knowledge@Wharton, op. cit.
52
Ken Auletta, op cit.
Page 50 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
33
9B16E008
R. Chandrasekhar wrote this case under the supervision of Professors Nicole Haggerty and Binny Samuel solely to provide material
for class discussion. The authors do not intend to illustrate either effective or ineffective handling of a managerial situation. The
authors may have disguised certain names and other identifying information to protect confidentiality.
This publication may not be transmitted, photocopied, digitized or otherwise reproduced in any form or by any means without the
permission of the copyright holder. Reproduction of this material is not covered under authorization by any reproduction rights
organization. To order copies or request permission to reproduce materials, contact Ivey Publishing, Ivey Business School, Western
University, London, Ontario, Canada, N6G 0N1; (t) 519.661.3208; (e) cases@ivey.ca; www.iveycases.com.
Copyright © 2016, Richard Ivey School of Business Foundation
Version: 2016-06-08
It was mid-October 2012 in downtown Toronto when Linda Mantia, executive vice-president (Digital,
Payments and Cards), Royal Bank of Canada (RBC), and Jeremy Bornstein, head (Emerging Payments),
RBC, met with executives from one of Canada’s leading telecom operators to discuss the launch of
RBC’s first mobile wallet.
The mobile wallet, enabled by the telecom operator’s wireless network, would replace the customer’s
leather wallet containing individual credit and debit cards. The wallet was designed by RBC and was
housed in a smartphone. As per the prevailing industry norm, the wallet relied on the subscriber identity
module (SIM) 1 card, contained in a chip, issued by the telecom operator. The mobile wallet would enable
customers to pay for purchases with their card of choice by simply waving the smartphone at the point of
sale (POS) terminal at a checkout counter.
The imminent launch of the mobile wallet would give RBC a head start over its peers in the banking
industry, which had seen Starbucks, a retailer with no legacy either in financial services or technology,
gain instant traction with its mobile payments app launched in 2011.
Mantia was about to sign on the dotted line and close the deal with the telecom operator when she had
second thoughts. Bornstein shared her discomfort and Mantia sensed it.
The concerns had come into sharp focus because engineers at RBC had been secretly designing, under
Mantia’s leadership, a cloud-based solution that could preempt the need to store the data in a SIM card.
Working at an innocuous location in Toronto, dubbed among water-cooler channels at RBC as the
“Shipyard” because of speculation that big things were being built there, the team was devising a product,
tentatively named RBC Secure Cloud that would store customer data in the bank’s own servers and route
it, as required, through a cloud. The cloud-based solution would be the first of its kind and a departure
from the SIM card-based solution, which was the gold standard in the payments industry.
1
Subscriber identification module (SIM) is a card containing a unique serial number, an international mobile subscriber
identity number, security authentication, information related to the local network, a list of the services the user has access to,
and two passwords.
Page 51 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
RBC: MOBILE WALLET
Page 2
9B16E008
PAYMENTS INDUSTRY
The annual turnover available from the global payments industry was of the order of US$1.170 trillion. 2
This amount consisted of fees charged for processing electronic payments for goods and services. The
industry consisted of three segments: business to business (B2B), customer to customer (C2C), and
business to customer (B2C).
B2B payments were valued at US$550 billion in revenues annually. Nearly 50 per cent of B2B payments
were made by cheque, involving a two- to three-day time lag. Electronic formats like automatic clearing
house (ACH), virtual cards, and payment cards (PCards) were gaining traction. They were designed to
ensure not only faster turnaround for banks but lower costs for businesses. C2C payments, valued at
US$30 billion, consisted of the transfer of funds between individual consumers. While banks were the
preferred medium for consumers, new technologies (like Instant ACH) and mobile apps (such as Venmo
and Square Cash) were facilitating real-time transfers. The bulk of C2C payments consisted of transfer of
funds from migrant workers to their families and between unbanked and under-banked individuals in
different geographies.
The B2C payment market was the largest, at US$591 billion. Its ecosystem consisted of consumers,
payment networks, 3 merchants, 4 and card issuers 5. While the payments industry in general was driven by
changes in technology and regulation, the B2C segment had an additional driver: demographic
composition. Changes in demography were leading to new products like mobile wallets. Millennials had
different payment habits than their parents (known as boomers). Sixty per cent of millennials were
regularly performing mobile financial transactions. 6
Data security was one of the factors influencing the course of the payments industry. The course was
characterized by three phases of technology — Contactless Transaction, Point to Point Encryption and
Tokenization — which were often overlapping. Each phase was marked by a gradual reduction in the
number of players who were part of what was known as the “circle of trust.” The wider the circle, the
higher the risk of transaction security and vice versa.
2
James Schneider et al, “The Future of Finance: Redefining the Way We Pay in the Next Decade,” Goldman Sachs Equity
Research, March 10, 2015. All currency in Canadian dollars unless specified otherwise.
3
Visa and MasterCard are examples of payment networks.
4
Merchants are the sellers of goods and services who have POS terminals.
5
Issuers provide the cards to consumers. They are responsible for front-end processing (routing the electronic transaction
from POS to the network) and back-end processing (handling the information flow needed to convert the electronic record
created at POS into cash for the merchant).
6
David Berman, “TD, BMO Unveil New Mobile Banking Features, Courting Tech-Savvy Clients,” Globe and Mail, March 17,
2015, accessed March 18, 2015, www.theglobeandmail.com/report-on-business/td-bmo-unveil-new-mobile-bankingfeatures-courting-tech-savvy-clients/article23512204/.
Page 52 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Mantia and Bornstein told their telecom counterparts at the meeting that they needed more time. As they
walked away from the brink of closing the deal, they were facing a dilemma. Should RBC go forward,
even if at a later stage, with a SIM card-based platform that had the support of every telecom operator and
payment network but diluted user experience and carried a high cost? Or should RBC launch a cloudbased platform that would reduce costs and improve user experience but force RBC to break rank with
peers in an industry that valued alliances?
Page 3
9B16E008
To help consumers migrate from cash to card, the payment industry added the contactless feature to make
it easier and quicker to pay than with swiping and signing. The feature used the medium of radiofrequency identification to transmit the data stored in the magnetic stripe of a credit card. Early adopters
were high-volume and low-ticket transaction businesses (like gas stations and convenience stores). No
signature or personal identification number verification was required. The flipside of the convenience was
that there was a limit to the number of transactions and their individual value. Data theft was also an area
of concern for consumers. The circle of trust was the widest (see Exhibit 1); it included in its sphere every
player beyond the customer.
Point to Point Encryption
Merchants keen on protecting their clients’ data asked the card-issuing bank to set up a system wherein
payment data was encrypted the moment it hit the POS terminal. The system used an algorithmic
calculation to encrypt the customer information stored in the magnetic strip using indecipherable codes.
The codes were sent to the payment gateway for decryption and then passed to the bank for authorization.
The bank approved or declined the transaction, depending upon the cardholder’s balance. The loop was
completed when the merchant at the POS was notified of acceptance or rejection. The process lasted less
than a second. The circle of trust excluded the merchant but included everyone else beyond the customer
(see Exhibit 2). 7
Tokenization
Created by EMVCo (a consortium of Euro-Pay International, MasterCard International and Visa
International), this marked a further step towards data security. Its principal vehicle was referred to as a
token. Once the cardholder data was verified, a token was automatically generated and sent to a
centralized secure server for storage. Simultaneously, a random unique number was generated and
returned to the POS terminal for use in place of the cardholder data. The server maintained a reference
database that allowed the token number to be exchanged for the real cardholder data if it was required
again for something like a chargeback. The token number could not be monetized but could be used in a
number of business applications as a reliable substitute for the real card data. The circle of trust was
limited to the card issuer, token service provider, and the network (see Exhibit 2).
The payments industry consisted of both innovators (delivering value-added services within existing
systems) and disruptors (replacing existing systems). Innovators (e.g., PayPal and Amazon) brought in
new capabilities, including big-data analytics, to increase the number of electronic transactions.
Disruptors (e.g., Bitcoin) eliminated the need for traditional credit/debit card networks. Merchant
Customer Exchange threatened to be a major disruptor. It was a joint venture of more than 70 retailers,
including Walmart, Target, Best Buy, and CVS together accounting for an estimated US$1 trillion of
annual purchase volume. Formed in 2012, it had a singular mandate of reducing processing fees.
7
Acquirer signs up merchants to accept payment cards for the network and arranges processing services for merchants.
Examples of acquirers include Global Payments, Chase Payments Tech, and Moneris.
Page 53 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Contactless Transaction
Page 4
9B16E008
Mobile Payments
The key tool for mobile payments was the mobile wallet. A mobile wallet let a consumer choose among
credit cards, debit cards and account balances to make a payment through a single access point. It had
four core capabilities: it was linked to multiple payment sources; it held currency; it operated in both
physical (offline) and virtual (online) environments; and it worked with multiple devices.
Mobile payments were in the early stage of development in 2012 and did not match the rigour of
traditional payment methods. They were being developed by non-banking start-ups rather than banking
incumbents. One of the high-profile start-ups was Softcard, a joint venture set up in 2010 by three U.S.
telecom carriers, Verizon, AT&T and T-Mobile. Their aim was to block Google, the only company with a
mobile app at the time, from controlling the mobile payments space.
The start-ups had a higher appetite for risk and lower regulatory oversight. They were piggybacking on
the payments infrastructure put in place by banks without being accountable for transactional safety and
leaving banks to pay the price. They were challenging banks on their own turf.
Many of the technologies under development were not state-of-the-art. For example, Starbucks had
dressed up a gift card as a mobile wallet. It used what was known as a Quick Response code to be read by
2D scanners installed at the POS of each of its stores in North America. The chain operated a closed-loop
system whereby its mobile wallet could be used only in its own stores. In general, the technologies
showed limited evidence of standards, protocols and liability protection. They were therefore vulnerable
to security breaches.
Mobile payments were a complex eco-system in which about 40 enterprises from five different industries
— financial institutions, retailers, telecom carriers, credit card service providers, and device
manufacturers — were vying for place. Each was leveraging its domain expertise in developing a unique
business model.
The mobile payments value chain was a picture of contrasts. Those at the front end (e.g., device
manufacturers) were backward integrating while those at the back end (e.g., carriers) were forward
integrating. U.S. banks were largely blocked out of the space by a consortium of the major telecom
networks, while Canadian banks were readying to dive into it. Mantia said:
Banks have a natural competitive advantage in the mobile payments market. In addition to
moving money (which is all the non-banking competitors can do), they store money on behalf of
clients in the form of deposits and lend money to clients in the form of loans. Banks also have a
higher level of public trust. They have extensive bandwidth in risk management and fraud
mitigation in addition to expertise in payments processing. They have longstanding relationships
with customers. They also have branch networks facilitating local deals with local merchants.
Finally, their technology platforms are so enduring that mobile payments are only a feature away.
A majority of their customers are also smartphone users.
Page 54 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Mobile payments were payments made with a smartphone at a POS. The term excluded e-commerce and
m-commerce transactions even when they were made with a mobile phone because they were made
remotely and not at the POS.
Page 5
9B16E008
RBC: COMPANY BACKGROUND
The vision was being realized through three-pronged strategic goals: holding on to the leadership position
in financial services in Canada; providing capital-markets and wealth-management solutions outside
Canada; and offering, in targeted markets, products and services complementary to its core strengths. The
bank’s strategy for growth and profitability had remained consistent over the years while the structure
was being fine-tuned regularly.
In October 2012, RBC had restructured its business into five segments: personal and commercial Banking
(meeting the banking needs of individuals and businesses), wealth management (serving high net worth
individuals); insurance (offering insurance and reinsurance products through various channels); investor
and treasury services (serving the needs of institutional clients); and capital markets (comprising global
wholesale banking). The business segments were supported by two streams: technology & operations
(forming the technological and operational bedrock for the organization) and functions (including finance,
human resources, risk management, corporate treasury, and internal audit).
Mobile Banking
RBC had been piloting several in-house projects in mobile banking. In 2000, it partnered with Fido to
provide wireless transactional banking services through Fido’s Access Finances feature. Potential clients
could use their Fido handsets to check account balances, transfer funds and pay bills. In 2007, RBC had
teamed up with Visa Canada to pilot a mobile payment service in Ontario that used cellphones, rather
than traditional credit cards, for making Visa purchases.
There were three important developments in 2008 at RBC with regard to mobile banking. RBC teamed up
with Research In Motion and Thomson Reuters to launch the BlackBerry Partners Fund, a venture-capital
fund to invest in mobile applications and services for the BlackBerry platform and other mobile
platforms. It also tested the RBC Mobex Mobile Payment Service, which allowed pilot users to send and
receive money instantly through text messages on mobile phones. RBC started working with Rogers and
Motorola on a specially engineered Motorola feature phone. It was keen on developing an open-loop
payment system that could work globally, unlike the closed-loop payment system developed by retailers
like Starbucks, which could only work in the captive environment of their own stores.
By 2009, RBC had completed development on its first mobile wallet. Cardholders could tap their phones
against merchants’ Visa payWave-compatible POS terminals to pay. The challenge was that there were
few Visa payWave-compliant terminals in Canada. No one was purchasing feature phones anymore
because they did not have options like touch display, multi-tasking, and integrated applications. The
world had moved to smartphones, which provided those options and more. By 2010, however, RBC was
the first Canadian bank to launch fully integrated mobile-banking applications for select smartphones.
By mid-October 2012, RBC had completed development on the traditional SIM solution for Visa credit
cards. It was then that Mantia and Bornstein were meeting with their counterparts at a telecom company
but had walked away from signing the deal with it for launching the solution.
Page 55 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Set up in 1869 as Merchant’s Bank of Halifax and named Royal Bank of Canada in 1901, RBC was
Canada’s largest bank as measured by assets and market capitalization in 2012. It was also the largest
card issuer in Canada, with 6.5 million cards. It employed about 80,000 full- and part-time employees
serving over 15 million clients in Canada, the United States, and 49 other countries. RBC was driven by a
vision of “earning the right to be our clients’ first choice.”
Page 6
9B16E008
The RBC Secure Cloud was an in-house initiative for which its two inventors were soon to apply for a
patent. Its development was being funded by a budget of $10 million. It differed from the SIM card-based
model in the sense that the payment data would be stored in a private cloud but was similar to this model
in the sense that the customer would need a smartphone and an SIM card. A seven kilobyte Java applet
would connect the payment data with the near-field communication (NFC) antenna on an NFC-enabled
phone. The applet would communicate with the mobile banking app and become whatever type of
payment the consumer wanted.
The bank would verify the client’s credentials and pre-issue an electronic token that would authorize any
purchase up to $100. To make a purchase, clients would select the desired debit, credit or loyalty card on
the phone and enter a passcode. Then, they would wave the smartphone close to the POS terminal to send
the token. The smartphone would pull down a new token from the cloud. Tokens were preloaded in case
wireless connectivity was not available; soon after connectivity was restored, the tokens would be
refreshed, readying the phone to make another payment.
The payment data would be stored at the bank’s own data centre instead of on the smartphone and would
be transmitted in an encrypted form to the device, one payment token at a time, on demand. The data
would be decoded by the applet, and then transmitted to the NFC antenna and into the merchant’s POS
device.
The circle of trust with RBC Secure Cloud would be limited to the issuer and the token service provider,
which were the same entity, RBC (see Exhibit 2). Bornstein said:
There are five prerequisites for a mobile wallet to gain traction. They also distinguish the winners
from the losers. First, the mobile wallet must have acceptance at merchant locations. The greater
its endorsement by merchants, the firmer would be its hold on the market. Second, it should have
interoperability. The more compatible a mobile wallet is with the systems of telecom operators
and device manufacturers, the higher would be its penetration. Third, it should be secure. The fear
of security breach is the number one barrier to mobile payments adoption. Fourth, it should fit
seamlessly with related pieces. It should be able to link the core payment program with loyalty,
reward, and incentive plans. Finally, it should provide a platform for marketing data integration.
This is particularly relevant at a time now when big data holds promise for all stakeholders in the
payments industry.
ISSUES FOR MANTIA AND BORNSTEIN TO CONSIDER
The primary issue was whether to run with the herd or break rank. The former meant playing it safe; the
latter meant facing the risks of the unknown. In choosing between the SIM card-based mobile wallet and
the cloud-based mobile wallet, Mantia and Bornstein had to weigh five areas of contrast between them.
First, the SIM card had become the industry standard. The cloud was untested in the payments industry
although it was becoming popular across industries as a way to store data.
Second, the telecom operator would have free access to customer credentials stored in the SIM card. The
free access would turn RBC into a commodity provider of credit. The cloud would not require RBC to
share customer data with either the wireless provider or the retailers and device manufacturers.
Page 56 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
RBC Secure Cloud
9B16E008
Third, the SIM card had limited storage, for which other banks would also be competing for space in it.
The SIM card inserted into each device was owned by the telecom operator, which, in turn, would be
renting out spots to competing banks. The overall space itself was finite. The telecom operators were
guarded about sharing information with card issuers on the installed capacity of a SIM card. Although no
individual bank knew which banks it was competing with for space, each bank had to have enough space
in the SIM card to load the client’s credit cards, debit cards, and co-branded cards. A typical RBC
customer held an average of 2.1 cards. But customers would be inclined, given the option, to load their
SIM card with personal data like their driver’s licence, passport, health card and others. Storage capacity
was not an issue with RBC Secure Cloud. The space in the cloud was potentially infinite.
Fourth, there was the matter of rack rent. RBC had 6.5 million cardholders. The telecom operator charged
$3.50 per spot (of 55 kilobytes) per annum to RBC, as the card issuer, for the space it provided on the
SIM card. Known as rack rent, the tariff was set by Enstream, the Canadian telecom consortium.
Assuming 10 per cent of RBC’s card holders were to migrate to the digital wallet, the rack rent would be,
at a minimum, $2,275,000 per annum for the SIM card-based mobile wallet. For the RBC Secure Cloud,
the cost would also be $2,275,000 (since the rent still had to be paid to the telecom operator) but it would
be a one-time, rather than recurring, payment. There was a major saving with RBC Secure Cloud.
Finally, there was the matter of customer experience. It would take up to 60 minutes to load a single card
(known as provisioning) on to the SIM card. The rate of failure in loading the cards was high. Since the
provisioning had to be done by the user, the time lag affected the user experience. Provisioning was a
particular issue in lifecycle situations (such as theft, fraud, expiry, and shift from one card to another with
the same bank) requiring deletion of an existing card. Because of this, very few customers were likely to
load their mobile wallet with more than one card. And, if they loaded only one card, their choice at the
checkout would be limited, potentially changing a merchant’s payment mix and affecting the merchant’s
business. With RBC Secure Cloud, provisioning could be done in less than two minutes with a 99 per
cent success rate, and it only had to be done once.
Mantia and Bornstein were facing dilemmas at two other levels: the ongoing consolidation among
Canadian telecom carriers and the ongoing fragmentation among mobile device manufacturers.
The SIM card used the medium of a telecom operator’s wireless network. The cloud would reduce the
dependence on a wireless partner, limiting its involvement to an applet that was required to direct traffic.
But it would also require the telecom operator to break rank with its own peers. The telecom operator also
had to have a strong and growing subscriber base. Finding such a partner from among existing players in
Canadian telecom would be difficult (see Exhibit 3).
The range of mobile phone devices used by Canadians was wide but limited, principally, to four operating
systems: Android (developed by Google), BlackBerry (RIM), iOS (Apple), and Microsoft (see Exhibit 4).
The payment system was different for each platform and each phone. The strategy was not as simple as
choosing the largest-selling device because the technology was changing rapidly. Interoperability was a
critical factor in the choice of a device. Mantia said:
Every one of our peers in the telecom and payments industries wants us to sign the deal for the
SIM card-based model and move on. While we are on the verge of going against the tide, there
are two questions for which we have to find definitive answers. Would the network providers
certify RBC Secure Cloud and allow the passage of payment credentials through the RBC applet?
They are bound by rules, systems and protocols and are not comfortable with disruption. Would
Page 57 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Page 7
Page 8
9B16E008
There was also the option of not pursuing the mobile payments at all. Two factors were compelling. First,
credit cards were not expected to disappear soon. Only 24 per cent of payments were predicted to be
made on mobile platforms by 2020. By then, 80 per cent of the merchants would be using NFC, and 60
per cent of consumers would own a smartphone and would be using it to pay for half their purchases.
Cards would continue to remain in use until mobile payments accounted for over 95 per cent of payments.
That would give RBC a big time window. Second, only eight million people in Canada owned
smartphones in September 2011, representing 40 per cent of the mobile market in Canada. The market
was growing at about 10 per cent per annum.
The risk in not pursuing mobile payments was that disruptive solutions could disintermediate RBC from
moving money altogether. If RBC misses the opportunity to keep up with services demanded from
customers with respect to mobile payments, there was an inevitable likelihood that their core service of
storing and lending money could be replaced by disruptive solutions as well.
The Ivey Business School gratefully acknowledges the generous support of Pierre Lapointe, MBA ’83, in
the development of this case.
Page 58 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
the telecom carriers agree to place an RBC applet in the SIM and also ensure that the right SIM is
attached to the right phone? They value alliances and would not want to break rank.
Page 9
9B16E008
Source: Company files.
EXHIBIT 2: SUBSEQUENT DEVELOPMENTS
Ring of Trust 1: For Acquirer P2P Encryption
Ring of Trust 2: For EMVCO Tokenization
Ring of Trust 3: For RBC Secure Cloud
Source: Company files.
Page 59 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
EXHIBIT 1: CONTACTLESS TRANSACTION
Page 10
9B16E008
Bell Wireless
MTS mobility
Rogers Wireless
SaskTel Mobility
TELUS Mobility
Wind Mobile
Videotron
Total Subscribers
Number of Subscribers
2011
2010
7,427,482
7,242,048
496,432
483,754
9,335,000
8,977,000
594,405
569,904
7,340,000
6,971,000
403,000
232,641
287,500
92,600
25,883,819
24,568,947
Q3 2012
7,576,027
494,564
9,432,000
594,405
7,558,000
510,484
378,300
26,543,780
2009
6,833,174
458,478
8,494,000
541,105
6,524,000
22,850,757
Source: Case authors.
EXHIBIT 4: CANADIAN MOBILE DEVICES MARKET
Top Mobile Original Equipment Manufacturers
(September 2011)
Device Manufacturer
Share of mobile
subscribers (%)
Samsung
25.2
LG
20.0
RIM
14.3
Apple
12.0
Nokia
10.1
Top Smartphone Platforms (September 2011)
Operating System
manufacturer
RIM
Apple
Google
Symbian
Microsoft
Source: Case authors.
Page 60 of 282
Share of smartphone
subscribers (%)
35.8
30.1
25.0
4.2
3.2
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
EXHIBIT 3: CANADIAN TELECOM OPERATORS
9 -6 1 4 -0 2 2
REV: MARCH 9, 2015
KARIM R. LAKHANI
CHRISTINE SNIVELY
Google Car
In October 2010, Sebastian Thrun, head of Google X, Google Inc.’s secret project division, wrote on
Google’s official blog, “We have developed technology for cars that can drive themselves,” and thus
confirmed speculation that the search and software company was developing autonomous vehicle
technology.1 While not a traditional manufacturer of automobiles, Google by 2014 had invested
millions of dollars2 in its self-driving cars, which had logged over 700,000 miles of testing,3 and
frequent news reports left the impression that Google cars would soon appear in dealer showrooms.
Sergey Brin, Google cofounder and director of special projects, predicted driverless cars would be
available to the general public by 2017.4
Self-driving cars could revolutionize the way people used and thought about cars. They could
make driving safer by reducing accidents and fatalities, make transportation available to individuals
who could not drive themselves such as the elderly or disabled, and they could free up drivers to
perform other tasks while they sat in their cars. Industry observers speculated that a Google car
linked to Google’s other services could someday be capable of predicting where to drive by accessing
the driver’s Google Calendar.5 It could alert the driver ahead of time of highway traffic by viewing
the Google Maps traffic option, and could also suggest alternate routes; in fact in June 2013, Google
acquired traffic analysis company Waze for $966 million. 6 Automobile data could eventually be
uploaded into a cloud storage system that the government could use to make roads safer. 7
Google appeared to be well positioned to develop automated vehicle technology. It had extensive
experience in software development, a vast collection of data from its Google Maps and Street View
products, an ability to attract top talent, access to capital, and a willingness to invest in what the
company referred to as “moonshoots,” or risky and expensive experiments.8
To reach this goal, Google needed to clear significant hurdles. Legal issues, privacy issues, lack of
public acceptance, and costly technology could all keep self-driving cars off the roads. By 2013, while
many new cars were equipped with automated features such as adaptive cruise control, collision
warning systems, lane departure warnings, and other computerized navigation tools—all parts that
contributed to the sum of driverless cars—industry experts disagreed over a projected timeline for
fully autonomous vehicles. Many also differed on the degree of automation that would be available,
specifically, whether cars would be fully autonomous or equipped with partial self-driving
capability.
Professor Karim R. Lakhani, and Senior Case Researcher James Weber and Associate Case Researcher Christine Snively, both of the Case
Research & Writing Group, prepared this case. This case was developed from published sources. Funding for the development of this case was
provided by Harvard Business School and not by the company. Professor Lakhani has been a past recipient of the Google Faculty Research
Award for his scholarship on innovation contests. HBS cases are developed solely as the basis for class discussion. Cases are not intended to
serve as endorsements, sources of primary data, or illustrations of effective or ineffective management.
Copyright © 2014, 2015 President and Fellows of Harvard College. To order copies or request permission to reproduce materials, call 1-800-5457685, write Harvard Business School Publishing, Boston, MA 02163, or go to www.hbsp.harvard.edu. This publication may not be digitized,
photocopied, or otherwise reproduced, posted, or transmitted, without the permission of Harvard Business School.
Page 61 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
JAMES WEBER
Google Car
The Google management team faced several questions: Should Google continue to invest in the
technology behind self-driving cars? How could Google’s core software-based and search business
benefit from self-driving car technology? As large auto manufacturers began to invest in automotive
technology themselves, could Google compete? Once the technology was refined and truly
consumer-ready, what were the potential business models for this technology? Should Google license
the software as a service, seek partnerships with a manufacturer, or build the cars itself? How could
data gathered from individual drivers complement Google’s other lines of business? Was the
investment of time and resources worth it for Google?
History of the Automobile
As early as the 1700s, European engineers experimented with motor-powered vehicles. By the
1800s, steam, internal combustion engines,a and electric motors to power vehicles had all been
attempted. Steam-powered stagecoaches were common in Britain by the 1820s.9 The first electric
carriage was constructed in 1891 in the U.S., and the first motorized truck with a combustion engine
was exhibited in 1903 in Springfield, Massachusetts. Electric cars were initially popular at the turn of
the 20th century, but their batteries lacked the power life necessary for long-distance driving at high
speeds. Steam-powered automobiles were available through the 1920s, but were expensive and
dangerous. Auto industry pioneers Ransom E. Olds, founder of Olds Motor Vehicle Company, and
Henry Ford, of Ford Motor Company, designed cars with internal combustion engines instead of
steam or electric power.10 In the early 1900s, up to 2,000 firms initially began producing cars, but that
number dropped to 1,000 by 1920, and 44 by 1929.11
Mass production of automobiles began in the U.S. in the early 1900s with the 3 horsepowerb
Oldsmobile. Ford Motor Company introduced the affordable gas-powered Model T in 1908,12 and
U.S. auto production surpassed 3.7 million units by 1923.13 The automobile had a tremendous impact
on American culture in the 20th century. Towns were shaped around highways, suburban
populations grew rapidly, and rural families were no longer isolated. Almost 4 million paved roads
were added,14 and commuting to work and traveling for leisure gained in popularity. Between 1960
and 2012, the number of registered vehicles in the U.S. more than tripled, rising from 74 million to
250 million.15
As suburbs grew, so did the amount of time Americans spent in their cars, creating congestion,
contributing to pollution related to emissions, and stressing the already undercapitalized national
infrastructure. By 2014, the average national commute time was 25.4 minutes each way.16 Over 61%
of workers with hour-long commutes drove without passengers. In one-third of the 24 metro areas
the U.S. Census Bureau surveyed in 2011, at least 10% of the workforce commuted one hour or
more.17 In 2011, the number of car crash fatalities exceeded 32,000.18 Motor vehicle accidents
remained the top cause of death for people between the ages of 3 and 34.19 “America’s roads are the
safest they’ve ever been,” Transportation Secretary Ray LaHood said in 2010, “but they must be
safer.”20 (See Exhibit 1 for accident data.) According to the World Health Organization, over 1.2
million people were killed each year in automobile accidents worldwide. 21
a Internal combustion referred to the placement of fuel (gasoline or diesel) into a cylinder or small container; the fuel released
energy when ignited.
b Horsepower was a unit of power (the rate of work done over time). The exact definition of 1 horsepower was moving 33,000
pounds 1 foot per minute. The horsepower of modern sedans ranged from 110 to 150.
2
Page 62 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
614-022
614-022
In addition to increased safety features over time, the automotive industry also incorporated new
technological innovations into its vehicles. Chevrolet introduced the first car radios in 1922; by the
1960s, radios were common in cars. Air-conditioning was introduced in cars in 1940, but was not
commonly found in vehicles until the late 1960s. Power lock and power window functions eliminated
two additional manual activities. Heated car seats became a feature of luxury vehicles, and minivans
were equipped with DVD players to entertain backseat passengers. In the early 21 st century,
manufacturers began adding Internet access, global positioning system (GPS) units, communication
systems, and other technologies.
Globalization of the Auto Industry
In 2014, while the U.S., Europe, and other mature automobile markets experienced modest
growth, automobile sales in emerging markets continued to grow significantly. Sales in saturated
Western markets typically came from car owners replacing their vehicle, whereas in China and India,
the growing middle class created many first-time buyers. In 2010, for the first time, sales in China,
India, Brazil, Eastern Europe, and other emerging markets accounted for over half of the 73 million
passenger vehicles sold worldwide. 22 Manufacturers had worked to set up local production centers to
serve these growing markets. (See Exhibit 2 for vehicle production data by country.)
U.S. auto sales topped 15 million units in 2013 for the first time since 2007.23 In 2009, China
overtook the U.S. as the biggest auto market by units sold. 24 The global car market was expected to
grow in upcoming years, primarily due to the increased demand in China, where sales were expected
to double by 2019, and elsewhere in emerging markets. 25 In Indonesia, sales reached a recordbreaking 1.2 million units, with Toyota controlling 35% of the market. 26 South Korea sales of over 1.5
million were down 4% from 2011, with domestic brand Hyundai holding a 43% market share.27 In
Southeast Asia, Thailand was the largest car market in 2012, with 1.4 million vehicles sold. 28 China
produced the most vehicles in 2012, with over 15 million passenger cars and over 3 million
commercial vehicles, followed by the U.S., Japan, and Germany.
Historically, automobile makers had done their own final assembly and some of their own parts
manufacturing while outsourcing much of the parts manufacturing to subcontractors. In some
situations, particularly with smaller-volume production runs, automakers chose to have contract
manufacturers do final assembly as well. For example, Magna International, a large contract
manufacturer, produced cars for several clients, including Mercedes-Benz (Daimler AG), Infinity
(Nissan Motor Company), and Mini (BMW).29 In 2013, foreign automakers moving into the Indian
market were choosing contract manufacturers in India such as Hindustan Motors. 30
New Possibilities
Self-driving cars had been a fixture of science fiction for generations: long dreamed about but
technologically out of reach. By 2013, some believed that was changing. “The technology to create
self-driving cars is already here,” said one industry insider. “As sci-fi as it sounds, self-driving cars
that don’t ever crash, reduce traffic congestion and make valet attendants obsolete are coming.” 31
For several decades, car manufacturers had been introducing technologies that had gradually
taken on more of the driving decisions. Cruise control, a setting that allowed the vehicle to maintain a
constant speed, became popular in the 1970s, and Ford introduced Antilock Brake System (ABS) in
1985. Several driver assistance features with sensing capabilities were first introduced in Japanese
models in the mid-1990s before appearing in European and American luxury automobiles. Mitsubishi
refined traditional cruise control in 1995 with the introduction of Adaptive Cruise Control (ACC),
3
Page 63 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Google Car
Google Car
which measured the distance and velocity of preceding vehicles through lidar c or radar sensors,
allowing the car in cruise control to slow down when approaching another vehicle. Soon most luxury
models featured ACC. Night vision systems found in General Motors’ Lincoln Navigator used
sensors to provide assistance when visibility was poor. In 2002, Honda unveiled a lane-keeping assist
system to the Japanese market, which sent warning signals the driver as the vehicle drifted. Blindspot video cameras were introduced in several 2006 Volvo models, and short-range radar technology
was placed in Audi vehicles in 2005, providing blind-spot detection alerts.32 In the mid-2000s, the U.S.
Defense Advanced Research Projects Agency (DARPA) Grand Challenges, which aimed to develop
technology for unmanned operational ground combat vehicles, accelerated self-driving car research
at several universities.33
Many considered self-driving car technology a logical extension of existing driver aids. 34 One
industry observer predicted that cars of the future, loaded with increasing amounts of infrared
sensors, video cameras, and laser-based radars, would soon be able to detect nearby objects,
communicate with one another, “see” the speed of nearby vehicles, and react to their turns or
braking.35
About Google
Stanford University graduate students Sergey Brin and Larry Page founded Google in 1997 after
developing a formula to rank the order of webpage search results by relevancy. In 1999, Brin and
Page raised $30 million to officially launch Google. 36 Eric Schmidt, former CTO of Sun Microsystems
and CEO of Novell, was named Google’s CEO in 2001 (Schmidt was Google’s executive chairman in
2013). The company’s mission statement was “to organize the world’s information and make it
universally accessible and useful.”37 From early on, the company’s philosophy—“Don’t be evil”38—
informed its grounding in and access to endless amounts of individual users’ data. The new company
generated revenue primarily through advertisement sales. Advertisers delivered ads targeted to
specific search queries or Web content.
In 2004, Google’s highly anticipated IPO raised $1.6 billion, allowing the company to grow and
offer more services.39 That year, Google launched its Web-based e-mail service, Gmail. The following
year it acquired the Android operating system, used primarily in touch-screen devices such as tablets
and smartphones. The company purchased the video-streaming website YouTube in 2006, which
generated over 19 million monthly visitors before the acquisition. 40 Google also released its
documents and spreadsheets file-sharing service, Google Docs, in 2006. Google acquired e-mail
security company Postini for $625 million in 2007, advancing Google’s push into business software,
and digital ad firm DoubleClick for $3.2 billion in 2008. Google branched out into mobile computing
with its Nexus One smartphone, which ran the Android operating system. 41
In 2010, Google acquired close to 50 companies valued at over $1.8 billion in total, including
mobile advertising network AdMob, social networking widget maker Slide, antipiracy software firm
Widevine Technologies, and a video compression technology company. Google attempted to acquire
the online provider of coupons, Groupon, for $5 billion, but Groupon rejected the deal. In 2012,
Google moved into the hardware business by acquiring mobile phone manufacturer Motorola for
$12.5 billion.42 This was a first for the search company, whose revenue was advertisement-driven, not
based on unit sales.
c Lidar was a remote sensing technology that measured distances by illuminating a target with a laser and analyzing the
reflected light.
4
Page 64 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
614-022
614-022
By 2013, Google’s products and services were available in over 50 countries and territories, and in
over 100 languages. An analyst described Google as “nothing short of relentless in its efforts to
develop or acquire new services and products in order to stay ahead of such rivals as Yahoo! and
Microsoft.” It branched out to provide an e-mail service, photo sharing, interactive maps, and Web
browsing free of charge, all of which displayed Google advertisements. (See Exhibit 3 for a list of
Google’s product offerings.) Its open-source operating system Android was made available to
developers of mobile phones and tablets for no charge. User data collected by Google through its
services on Web browsers and mobile allowed Google to better target ads to its users, resulting in
higher revenue.43 In 2013, Google’s revenues topped $59 billion, with net income of $12.9 billion, up
from $50 billion in revenue and net income of $10.7 billion in 2012. Google’s growth continued into
2014.44 (See Exhibit 4 for financial statements.)
In January 2014, Google announced the formation of the Open Automotive Alliance (OAA), a
partnership with Audi, Honda, General Motors, Hyundai, and chipmaker Nvidia to introduce
Android software in vehicles by the end of the year.45 Drivers would be able to access Google services
while maintaining focus on the road. Forty new partners joined the alliance by June 2014, with 25
manufacturers, including Audi, Chevrolet, Chrysler, Dodge, and Mazda, scheduled to ship cars with
Android software by the end of 2014.46 “Putting Android in the car will bring drivers apps and
services they already know and love, while enabling automakers to more easily deliver cutting-edge
technology to their customers,”47 explained Patrick Brady, director of Android engineering. In June
2013, Apple revealed its plans to integrate iOS software into control panels in BMW, General Motors,
Honda, and Mercedes-Benz vehicles.48 Microsoft developed similar technology with Fiat that allowed
drivers to sync their phones to Fiat automobiles to make hands-free calls, access music libraries, and
listen to text messages.49
Google and Search
While Google had frequently introduced new products, its Internet search product and
advertising related to it remained at the core. Google reportedly spent over 1 million computing
hours building its search index of webpage contents containing over 100 million gigabytes of data.
Google used “web crawling”d software to view webpages and follow links on those pages. Crawlers
moved from link to link and transferred data about those webpages to Google’s servers in order to
maintain its vast index. When a search was conducted, Google’s algorithms compared search terms in
the index to find the appropriate pages, and ranked the pages based on relevance, similarities to key
search terms, and popularity. Google continuously worked to improve its search function. It made
over 500 changes to its algorithms each year based on user clicks and other usage patterns. If users
selected a link from search results and clicked the back button to view additional results, Google used
that feedback data to determine search result rankings. Google aimed to deliver the most relevant
search results in the shortest period of time.50
Google built its new products around its search core. It gathered data through Gmail, the social
network Twitter, Google’s Chrome browser, YouTube, Google Maps, and most other services to help
refine its search capabilities. By 2013, Google controlled 67% of search market share.51 Total Google
searches in 2012 reached 114 billion. 52 In July 2013 alone, users conducted 12.9 billion Google
searches. Google attracted 1.17 billion unique monthly searchers, while over 1 million advertisers
used Google AdWords. Over 300,000 apps served Google Mobile ads.53 Google earned $38.6 billion in
d Web crawlers moved about the Internet viewing websites and following links to catalog data onto Google’s servers in order
to maintain its vast index. The crawlers took note of new websites, as well as changes to existing ones, and removed dead links
from its servers.
5
Page 65 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Google Car
614-022
Google Car
online ad revenues, about 33% of all online ad revenue worldwide in 2013. Facebook, the second
highest earner in ad revenue, earned only $6.4 billion.54
Google maintained data warehouse centers around the world, housing servers in seven locations
in North and South America, three in Asia, and three in Europe. All data collected by Google was
stored on a physical server in one of its data centers. A large portion of Google’s $3.4 billion in 2011
capital expenditures was believed to have been spent on its data center servers. 55
In the early 2000s, Google began designing its own servers for its data centers in an attempt to
save power and cut costs as it continued to grow. 56 The company was secretive about its involvement
in hardware work, but at Google’s annual stockholder meeting in 2012, CFO Patrick Pichette, while
explaining the company’s $12.5 billion acquisition of mobile-phone maker Motorola, told the
audience, “There’s a bit of a mythology that Google doesn’t know anything about hardware.” 57 He
continued, “We’re big in hardware. Google actually builds servers in a factory that actually probably
makes us one of the largest hardware manufacturers in the world. And so we know hardware.”58
Google partnered with original device manufacturers (ODMs) in Asia that built servers for HewlettPackard, Dell, and IBM to build servers to Google’s specifications. 59 Intel, the world’s largest
producer of server chips, revealed in 2012 that eight server producers accounted for 75% of Intel’s
server chip revenues, and Google ranked fifth.60
Energy efficiency at Google’s data centers had improved over the years. Data-center experts
gauged efficiency by measuring power usage effectiveness (PUE). PUE was determined by dividing
the total electrical power used in a facility by the power needed for the servers and related
networking equipment. In 2005, 22 sample companies’ data centers averaged a PUE of 2.0, meaning
that for each watt used to run the computers, another watt was used on lights, office equipment, air
conditioners, and other operational necessities. 61 By the Q3 2014, Google’s data centers achieved an
overall PUE of 1.12. The ideal PUE value was 1.0.62
Google and Data
Google’s collection of user data gave the company a huge competitive advantage, not only in
generating ad revenue but also in making marketing decisions. Google had insight into what users
wanted and made decisions based on that information. Google collected search data through Web
browser activity as well as search activity on Android smartphones, the most used operating system
in the world, and through its smartphone applications (apps). 63
Google Maps was the most popular app in 2013,64 and the top app used for local searches on
phones and tablets.65 Maps provided directions, search, location services, and traffic information. To
maintain real-time traffic updates, Google Maps used the position and movement of Android
smartphones to record how fast traffic was moving. 66 Maps was created through “algorithms and
elbow grease,” as one member of the Maps team described it.67 Operators began with free public data
sets from governments, along with satellite data, and conflated the two. Google also gathered data
from its Street View project to maintain Maps’ accuracy. It was a large-scale operation, requiring
hundreds of people to map one country. 68 Google Maps worked to merge its real-world information
with data from cities, census operations, and other public sources. “If you look at the offline world,
the real world in which we live, that information is not entirely online,” Manik Gupta, the senior
product manager for Google Maps, said. He continued, “Increasingly as we go about our lives, we
are trying to bridge that gap between what we see in the real world and [the online world], and Maps
6
Page 66 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Google Search Infrastructure
Google Car
614-022
really plays that part.”69 An industry observer noted, “Google’s geographic data may become its most
valuable asset. Not solely because of this data alone, but because location data makes everything else
Google does and knows more valuable.”70
In November 2007, 10 months after Apple released the iPhone, Google unveiled its Android
operating system as a free, open-source platform that developers then incorporated into their own
mobile devices.71 Google provided an alternative to Apple’s operating system while carving out its
position as a mobile search provider. Most Android devices were tied to Google services including
search, Maps, Gmail, and YouTube, among many. By September 2012, over 500 million Android
devices were activated globally.72 The Android OS controlled 84% of the smartphone market share in
Q3 2014, with over 283 million units shipped worldwide,73 up from an 81% market share and 211
million units shipped in Q3 2013.74 Google’s mobile segment generated over $4 billion in advertising
revenue in 2013, 8% of Google’s $50 billion in total revenue.75 Ad revenue from mobile devices came
primarily from Google’s own apps such as Google Search, Google Maps, and YouTube.
In 2010, Google released its own Google-brand line of smartphones called Nexus. These phones
were initially manufactured by partner firms. Each new smartphone launch introduced an updated
Android operating system (OS). In January 2010, mobile device maker HTC released the Nexus One
with Android 2.1; in December 2010, Samsung introduced the Nexus S, which used Android 2.3; and
in November 2011 Samsung launched the Galaxy Nexus, which used Android 4.0.76 Android 4.0 was
the first tablet-compatible version of the OS. In 2013, Google began building and selling its own
mobile devices through its Motorola subsidiary. Motorola released the Moto X smartphone with the
Android operating system later that year.77 Google sold Motorola to Lenovo in January 2014 for $2.91
billion, but retained the firm’s patents and licensed them to Lenovo. 78
Sebastian Thrun and Google X
In 2005, Sebastian Thrun, then a Stanford University professor, led the university’s team of 65
students, professors, engineers, and programmers to victory in the $2 million DARPA Challenge. The
DARPA challenges were initiated in 2003 to spur innovation in autonomous vehicle navigation and
work toward the congressional mandate to make one-third of the military’s land vehicles
autonomous by 2015.79 Teams of hobbyists, university researchers, and robotics and software
professionals applied, and 23 finalists were invited to participate. The Stanford team and its privatesector partners, including Intel and Volkswagen, spent one year developing its vehicle. 80 Challenge
rules mandated that the team vehicles needed to fully demonstrate autonomous behavior over a 132mile driving course.81 The course challenged the vehicles to navigate a complicated environment and
maintain autonomy over an unknown terrain. The hardware and software of the vehicle needed to
manage a bumpy, dusty, and perhaps wet environment; maneuver the vehicle around obstacles; and
perform safely in the absence of an onboard human driver. 82 The team’s vehicle, Stanley, navigated
the course in less than seven hours, traveling at an average speed of 19.1 miles per hour. Google CEO
Larry Page was impressed with Thrun, and soon after the challenge he hired Thrun and members of
his team to work on Street View for Google Maps. Thrun’s role at the company quickly evolved.
In 2010, Thrun co-founded Google X, Google’s secret lab in an undisclosed San Francisco Bay Area
location where Google experimented with “shoot for the stars” 83 ideas that required a great deal of
capital.84 Many of the projects involved robots, which could help Google collect and capture data for
Google Street View.85 Page reportedly often said, “We would rather see a smoking crater in the
7
Page 67 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Google Mobile
614-022
Google Car
Google and Self-Driving Cars
By the time Thrun announced the self-driving car project on the Google blog in October 2010, the
team had already made significant progress. Its small fleet of Toyota Priuses had navigated 140,000
miles. Google’s project benefited from the collection of data previously recorded by Google Maps and
Street View. By 2013, dozens of Google X engineers had developed a fleet of 20 self-driving cars that
logged over 500,000 miles without an incident. 89 Automated cars drove along roads and highways,
from Silicon Valley to Lake Tahoe over 200 miles away, through a variety of terrain, dense traffic, and
streets with pedestrians.90 The test cars logged performance data such as speed, street location, and
obstacles detected by sensors. Google in turn used the collected data to refine the technology.
According to Thrun, Google aimed to reduce accidents and fatalities, shorten commute times, and
improve the quality of life for the blind and disabled. “We’ve always been optimistic about
technology’s ability to advance society, which is why we have pushed so hard to improve the
capabilities of self-driving cars beyond where they are today,” Thrun wrote in a Google blog post.91
Google also sought to “free up people’s time.”92 If cars drove themselves, people could browse the
Internet, log on to Gmail, and post messages on Google+ while being chauffeured. In turn, Google’s
ad revenue would increase.
Google was fairly secretive about its self-driving car program. Regarding its plans for the future,
in a July 2013 statement Google explained, “We want to improve people’s lives by making driving
safer, more enjoyable, and more efficient. We’ve successfully driven over half a million miles across a
wide variety of terrain and road conditions, and we’re very pleased with the performance. We’re
continuing to develop and refine the technology, but we aren’t going to elaborate about specific plans
at this point,” a Google spokesperson said.93
Google Car Technology
Google cars used video cameras, radar sensors, and a lidar to “see” other traffic and detect the
movements of other vehicles. The lidar, a spinning laser device which cost between $75,000 and
$85,000,94 was considered “the heart of the system.”95 It was mounted on the roof of the car and
generated a 3D map of the environment that Google’s software then analyzed. 96 The car combined
information collected by the laser with high-resolution maps of the world.97 According to an industry
expert, arguably the most important outcome of the DARPA field trials was the development of lidar,
and it was used by almost all self-driving systems in addition to Google’s. 98 (See Exhibit 5 for an
image of Google’s self-driving car.)
Four additional radars were placed on the front and back bumpers, which allowed the car to “see”
far enough to recognize fast highway traffic. A camera located near the rearview mirror detected and
interpreted traffic lights. A global positioning satellite (GPS), inertial measurement unit, and wheel
encoder determined the vehicle’s location and tracked its movements. 99 The combination of lidar,
radar, and cameras meant the vehicle was capable of spotting road edges and lane markings, read
signs and traffic lights, and even identified pedestrians. 100 Ultrasonic detectors provided a more
8
Page 68 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
ground than a mediocre result,”86 meaning the company was willing to risk significant capital in
innovative and inspiring projects with no guarantee of success. Along with a team of 12 engineers—
including Chris Urmson, technical team leader of Carnegie Mellon University’s 2007 DARPA Urban
Challenge winning team, and Mike Montemerlo, software lead on Thrun’s 2005 Stanford team at
DARPA87—Page tasked Thrun and his team with designing a self-driving car that could manage the
streets of San Francisco and beyond. 88
Google Car
614-022
Google began testing its self-driving cars by first sending out a human driver in control of a
Google car (initially a Toyota Prius, but Audi and Lexus models were later added to the fleet)103 to
map the route and road conditions. Engineers drove a route at least once to collect data on the
environment, including details not available on Street View, such as distances between objects, before
testing a Google car on the route. 104 (See Exhibit 7 for an image of how Google Cars “saw” objects.)
By relying solely on GPS-based techniques, Urmson said, the location could be off by several
meters.105 The mapping of lane markers and traffic signs allowed the software in the car to become
familiar with the environment and its characteristics.106 Each time a car traveled along a particular
route, it collected additional data used to update the 3D map, used by other autonomous vehicles.
When it was the autonomous vehicle’s turn to drive, it compared data it acquired in real time to the
previously recorded data. This was useful for differentiating pedestrians from stationary objects like
mailboxes.107
Despite impressive technological advances, Google engineers still faced many challenges. Google
cars struggled in snowy or rainy conditions. Poor weather restricted the car’s ability to “see” the road
markers and stay within the lane. The cars also got lost when encountering a change not reflected on
its map, such as a new road or a change in the road. In these instances, the car alerted the driver.
(Because the cars “talked” with one another across Google’s network after one car experienced a
change, the other cars “learned” from it.) Another limitation was that the cars struggled in any
situation in which a human directed traffic—for example, when there was construction or an
accident. Many wondered how a car could be programmed to interpret hand signals. 108
Building Self-Driving Cars
Designing self-driving cars was a different matter than making and selling them. According to
sources, in 2013, Google had tried to interest auto manufacturers in partnering to build the cars, but
no deals had been confirmed. In addition, Google had spoken with contract manufacturers to build
cars to Google specifications. Again, no deals had been announced. 109 Anthony Levandowski,
Google’s product manager for autonomous driving, stated that Google was willing to “make
available to the rest of the auto industry all of the building blocks that we ourselves use—the
Android operating system, search, voice, social, maps, navigation.” 110 While Google’s services would
come at no cost, an industry observer noted, the real cost of implementation for car companies would
be the lidar and other required hardware.111
In August 2013, Google was in talks with Continental AG and Magna International, component
providers to big automakers that also helped with assembly, to manufacture a car under Google’s
direction. In Germany, the Frankfurter Allgemeine Zeitung reported that Google and Continental were
close to a deal.112 Other reports claimed that Google was studying how its vehicles could become part
of robo-taxi systems and pick up passengers on demand,113 echoing Urmson 2011 musing that cars
could become a “shared resource, a service that people would use when needed.” 114 Whether or not
Google would operate such a service itself was unclear.115 Also in August, Google Ventures, the
investment arm of Google, invested $258 million (86% of its $300 million yearly fund) into black car
on-demand ride service Uber,116 and invested additional capital in the company in 2014. Google did
not confirm or deny speculation but had not slowed down its testing. In an interview Levandowski
assured, “Google is not a car manufacturer. Nor does it intend to be one.” 117
9
Page 69 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
accurate mapping of short-range surroundings, for example, when parking. 101 One reporter wrote,
“Because the car’s roof-mounted sensors can see in all directions, it arguably has greater situational
awareness than a human driver.”102 (See Exhibit 6 for a diagram of self-driving car technology.)
Google Car
By 2014, Google began building small two-passenger prototype vehicles with no steering wheel,
brakes, or accelerator. The pod-like vehicles featured a button to start and stop the vehicle and a
screen that displayed the intended route. Google planned to build 100 prototypes and began testing
them in the summer of 2014. The cars did not exceed 25 miles per hour, and could not be driven on
public roads at that time, as there was no way for a passenger to take over vehicle control. In order to
test the prototypes in California, the company was required to add temporary manual controls to the
vehicles. Urmson explained, “It was a big decision for us to go and start building our purpose-built
vehicles [. . .] to explore what it really means to have a self-driving vehicle.”118
Market Adoption
Google faced several issues related to whether consumers wanted to buy self-driving cars. In a
May 2013 multinational study conducted by Cisco Systems on public acceptance of self-driving cars,
57% of respondents worldwide trusted driverless cars and 46% would trust them to drive their
children.119 Acceptance varied greatly between markets, however. In Brazil, India, and China, 95%,
86%, and 70% of responders, respectively, trusted driverless cars. In the U.S. and Russia, the numbers
were 60% and 57%, while only 37% of Germans and 28% of Japanese felt comfortable with the
technology.120 Other surveys found different results. A June 2013 survey by Tyco in the U.S. found
that only 30% of responders would feel comfortable in a self-driving car. More than half (55%)
indicated safety technology as the most important roadblock to autonomous vehicles. Many believed
that safety would need to be proved before the general public accepted autonomous cars.121
Cost was another concern. Google’s cars required $150,000 worth of equipment, but cheaper
components, particularly for the laser radar units, could lower costs. 122 According to one study, 20%
of consumers surveyed said that they would “definitely/probably” be willing to spend as much as
$3,000 for autonomous driving applications. 123
Auto manufacturers remained skeptical about demand. Some felt that car enthusiasts might not
be as receptive to self-driving cars as other drivers would be.124 Others argued that baby boomers,
who equated cars with personal freedom and identity, might be reluctant to give up the wheel. 125
Proponents of self-driving cars believed they would significantly cut down on traffic while driving
within speed limits and following all traffic laws. Drivers would have to accept not having control
over the speed of the vehicle. Other industry watchers, however, pointed to two factors with selfdriving cars that might actually increase traffic. First, because “drivers” could spend time on tasks
other than driving, they might not mind spending time in traffic and therefore might go more places.
Second, self-driving cars might not need any human inside. Rather than look for a parking space, a
“driver” might command the car to drive around the block while the driver ran into a store. A range
of possible driverless car scenarios increased the amount of time each car might spend on the road.
Legal Barriers
Google also faced legal barriers. The company had already lobbied several state governments to
allow the testing and use of self-driving cars on public roadways.126 In 2012, Nevada and Florida
became the first states to approve self-driving cars. At Google headquarters, California governor
Edmund “Jerry” Brown signed the state’s autonomous-vehicles bill into law, thereby permitting
testing in that state. After the National Highway Traffic Safety Association (NHTSA) issued
guidelines for self-driving cars in 2013, the state of Michigan approved the testing of self-driving
vehicles in December 2013127 (See Exhibit 8 for NHTSA classifications.) Colorado shelved a bill in
2013 that would have allowed testing of self-driving cars after opposition from trial lawyers and a
10
Page 70 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
614-022
Google Car
614-022
There were also unanswered liability questions regarding who would be responsible in an
accident. In 2013, California’s Department of Motor Vehicles (DMV) was working on regulations
addressing self-driving cars. “We need to take baby steps. We need to make sure these vehicles can
operate safely,” warned a DMV representative. Another consumer advocate added, “We have a lot of
computer technology in cars already. It doesn’t always work right. There are often problems.”129
Consultancy firm KPMG believed that there was no margin for error, and that consumers “will not
relinquish control until they are certain their vehicles and their mobile environment are 100% safe
and reliable.”130 Other observers believed that these legal issues would be overcome. One stated, “I
am convinced that when the technology is ready, the law and policy will make room. There will be a
point when people say this is safe by some reasonable standard.” 131 Google’s Urmson stated, “There
wasn’t legal protection for the Wright brothers when they made that first plane. They made them,
they went out there, and society eventually realized its value.” 132
Google’s self-driving car also raised privacy concerns. In March 2013, Google was issued a $7
million fine for violating privacy laws after its Street View mapping team collected passwords, e-mail
addresses, and other private information from unencrypted networks as its vehicles captured images
of streets and buildings. Google blamed a “rogue engineer” for the infringement. 133 The nonprofit
group Consumer Watchdog attempted to block California legislation that allowed for self-driving car
testing unless it included language that prevented the collection of data for marketing or other
nonsafety purposes.134 Additionally, self-driving cars would have an “always-on” wireless
connection, similar to mobile phones, making their locations easily trackable. 135
Competitors
Google was not alone in its quest to develop self-driving cars. Auto manufacturers and
universities were researching and testing autonomous vehicles of their own. Carnegie Mellon
University’s Transportation Research Center, partnered with General Motors and funded by the U.S.
Department of Transportation, announced in September 2013 that its self-driving 2011 Cadillac SRX
had demonstrated that it could handle congestion, highway traffic, and lane changes by driving 33
miles from Cranberry, Pennsylvania, to Pittsburgh International Airport. The vehicle used only
automotive-grade radars and lidars embedded around the car, and computers hidden under the
floor.136
In 2012, Stanford’s Dynamic Design Lab partnered with Volkswagen’s Electronic Research Lab to
create a self-driving Audi, designed to navigate a course at high speeds as accurately as a human
driver. In an August track test, the vehicle traveled as fast as 120 miles per hour. 137
Mobileye, a Dutch and Israeli tech company that developed Advanced Driver Assistance Systems
and was valued at $1.5 billion, raised $400 million to develop semiautonomous driving technology in
partnership with General Motors, BMW, and Volvo. Mobileye aimed to develop self-driving
technology for stop-and-go traffic scenarios or for handling long stretches of highway driving that
could be significantly cheaper than Google’s technology. Mobileye’s technology could alert drivers to
road dangers and automatically brake when approaching a vehicle. The ability to read traffic and
street signs was scheduled to be released in 2015.138
In 2014, the San Francisco, California-based startup Cruise Automation worked to develop selfdriving features that could be attached to any vehicles. By June 2014, the technology was only
11
Page 71 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Google representative who testified against it. Google declined to elaborate on its reservations about
the legislation.128
614-022
Google Car
In October 2014, electric car manufacturer Tesla introduced a new dual-motor version of its Model
S electric sedan with autonomous driving features. The increased power allowed the vehicle to
accelerate from zero to 60 miles per hour in 3.2 seconds. The car boasted new autopilot features,
including photo and radar systems that recognized lights, stop signs, and pedestrians, a display to
show other nearby cars, and self-parking capabilities. The new dual-moto Model S sedans were
expected to reach the market in early 2015, with prices ranging from $89,000 to $120,000. As one
industry observer noted, “This is Tesla catching up with the industry on auto-driving features.”140
By 2014, most large automobile manufacturers were exploring driverless car technologies.
Audi In 2010, Audi’s self-driving TTS Coupe navigated 12.4 miles of paved and dirt road and
turned 156 corners to summit Pikes Peak in Colorado.141 Audi demonstrated its self-driving car at the
2013 Consumer Electronics Show. Audi developed a version of lidar smaller than that used by
Google’s Toyota Prius to maintain a more sleek design. The company was working on automatic
parallel parking and developed several vehicles designed to test perpendicular parking and driving
in traffic jams.142 Audi tested a self-driving RS7 model on a Formula One track in Germany in 2014.
The driverless vehicle reached 149 miles per hour and maneuvered the track 30 seconds slower than
professional drivers, testing “how you can control a car at the limit,” 143 as one Audi project manager
explained. The vehicle had a built-in detailed map of the track and followed an optimized path, but
torque and steering were not pre-programmed. The car had to decide how to maintain speed, stay on
track, and make adjustments.144
BMW In August 2011, BMW announced its ConnectedDrive Connect (CDC) system to promote
“driver-assistance” technologies that would lead to more automation. BMW engineers believed selfdriving cars would be ready in 10–15 years. In 2011, BMW’s robotic car traveled 105 miles at highway
speeds from Munich to Nuremberg, Germany. In January 2012, the company released a video of its
CDC technology powering a BMW 5 Series model on the Autobahn (with a driver ready to take over
if necessary). Under CDC, the car braked, accelerated, and passed other vehicles while analyzing
traffic using radar, cameras, laser scanners, and ultrasound distance sensors. 145 In April 2014, at the
Consumer Electronics Show in Las Vegas, Nevada, BMW demonstrated the new capabilities of the 2
Series Coupe, which was able to create a controlled drift in order to demonstrate the precision of its
control systems and how it would recover from a similar situation.146
General Motors
GM planned to introduce a semi-automated Cadillac driving system by 2017.
The early versions of GM’s semi-autonomous vehicle could drive themselves on highways or
interstates. In 2009, GM engineers began working on technology that allowed cars to read lane
markings and cross lanes with radar assistance. Its “Super Cruise” technology would allow the car to
break, accelerate, and stay in a lane without driver assistance, and its vehicle-to-vehicle (V2V)
technology would allow certain Cadillac models to transmit information such as vehicle speed,
direction, and location to other similarly equipped cars. A Cadillac spokesperson described Super
Cruise as “a next logical step from systems like adaptive cruise control.” 147 GM also partnered with
Carnegie Mellon University on the development of its self-driving technologies.148
12
Page 72 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
compatible with Audi A4 and S4 vehicles, but Cruise was experimenting with other models. Cruise’s
features operated as a “highway autopilot.” Drivers manually merged into their desired lane and
pushed a button that instructed the system to take control of the accelerator, brakes, and steering. As
with cruise control, the driver could turn off the features by tapping the gas pedal. The company
hoped to sell the technology for $10,000 per vehicle.139
614-022
Mercedes In September 2013, Mercedes demonstrated its autonomous vehicle at the Frankfurt
Auto Show. One month prior, the Mercedes Benz S500 Intelligent Drive research vehicle retraced the
first 103-kilometer (64-mile) trip taken by a passenger car in 1888.149 The 2014 Mercedes S-Class,
while not considered fully autonomous, was capable of slowing down or speeding up depending on
the movements of vehicles in front of it. Its steering assistance kept the car within the lane. 150 The SClass also contained short- and long-range radars and sensors to detect objects and measure
distances.151 Mercedes announced a partnership in September 2013 with Nokia’s “Here” business, a
competitor of Google Maps. The two companies partnered in developing “Smart Maps” for
Mercedes’s connected cars. Mercedes parent Daimler had announced that self-driving cars would be
on the road by 2020.152
Nissan In 2014, Nissan led competing auto manufacturers in electric car sales with its Leaf
Hatchback and began working on bring a self-driving car to market by 2020. A Nissan executive told
reporters that the company was developing its system in-house but was open to a partnership. “I
don’t preclude the possibility of working with Google, or anyone else for that matter,” the executive
said.153 Nissan’s system would reportedly not need to be linked to an Internet-based data system.
Nissan’s executive vice president for research and development said: “We don’t count on
infrastructure so much. All the technology is in the cars.” 154
Volvo Volvo was developing many new technologies with the aim of reducing traffic deaths to
zero by 2020. Adaptive Cruise Control with steer assist, for example, sensed when a car was drifting
out of a lane and gently steered it back. The Swedish manufacturer was also working on “roboparking,” to instruct a car to park itself, and “car-to-car” communication in which cars would warn
each other about potholes, icy patches, and other poor road conditions.155 Volvo was also developing
a pedestrian detection system that required over 500,000 kilometers of test driving and over 3
terabytes of collected data.156 Management believed that engineering a fully self-driving vehicle was
too large of an investment. Instead, Volvo experimented with “platooning,” a system in which cars
traveled one by one on a highway at the same speed and communicated with one another. Drivers
were responsible for joining and exiting the platoon, but cruising in the platoon required no work on
the part of the driver.157 (See Exhibit 9 for a diagram of a platoon.)
The Road Ahead
Professional service firm KPMG believed that sensor technologies would continue to develop and
converge, “leading to an eventual inflection point beyond which it is likely that the driver will
increasingly be taken out of the loop.”158 The transition was expected by many to be gradual, and
over the coming years new cars were expected to take on more driving tasks. In 2014, the U.S.
Department of Transportation was working to craft new rules that would require all new cars to be
equipped with V2V communication by 2017, meaning vehicles would be able to communicate with
one another using wireless technology and share information on speed and location. NHTSA
believed that this could reduce the severity of many accidents. It had completed a year-long pilot in
Michigan, outfitting over 3,000 vehicles with V2V wireless technology. 159
Many industry experts believed entirely autonomous vehicles were several years away. While
Google’s management team needed to answer key questions before driverless cars would be
available to consumers, a senior member of its development team made the company’s position clear:
“We are concentrating now on getting our vision . . . a fully autonomous car without the need of a
driver. That is our ultimate goal.”160
13
Page 73 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Google Car
Google Car
As Google continued to develop autonomous vehicles, the management team needed to answer
key questions before driverless cars would be available to consumers: First, how did this project
relate to Google’s corporate mission of organizing the world’s information? Was the Google Car
project a worthwhile investment of the company’s resources? Could Google compete against
established auto manufacturers working on self-driving technology, or should it establish a
partnership? Would Google manufacture its own vehicles or license its self-driving software? Google
also needed to consider the many legal issues surrounding this new technology. Finally, how would
an autonomous car fit in to Google’s core search business and its advertising-based revenue model?
14
Page 74 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
614-022
Google Car
Source:
U.S. Vehicle Crash Data (Indexed), 1994–2012
Compiled from National Highway Traffic Safety Administration (NHTSA), Fatality Analysis Reporting System,
http://www-fars.nhtsa.dot.gov/Main/index.aspx, accessed December 2014.
15
Page 75 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Exhibit 1
614-022
Page 76 of 282
Source:
Cars
18.085
4.369
8.189
5.439
4.122
3.138
2.722
1.771
0.965
1.071
1.919
1.719
1.458
1.509
1.128
7.858
65.462
2013
Commercial
4.031
6.697
1.440
0.278
0.398
0.741
0.989
1.280
1.414
1.385
0.252
0.443
0.282
0.088
0.004
2.169
21.891
Cars
15.523
4.105
8.554
5.388
4.167
3.285
2.623
1.810
1.040
0.945
1.968
1.539
1.682
1.464
1.171
7.802
63.074
2012
Commercial
3.748
6.223
1.388
0.260
0.394
0.859
0.718
1.191
1.423
1.484
0.262
0.439
0.284
0.112
0.007
2.225
21.025
Vehicle Production by Country, 2009–2013 (in millions)
Cars
14.485
2.976
7.158
5.871
4.221
3.040
2.519
1.657
0.990
0.537
1.744
1.839
1.931
1.343
1.191
8.387
59.897
2011
Commercial
3.933
5.684
1.240
0.439
0.435
0.887
0.888
1.023
1.144
0.919
0.246
0.534
0.311
0.120
0.007
2.330
20.147
Cars
13.897
2.731
8.310
5.552
3.866
2.831
2.584
1.386
0.967
0.554
1.208
1.913
1.924
1.270
1.069
8.274
58.341
2010
Commercial
4.367
5.031
1.318
0.353
0.405
0.725
0.797
0.956
1.101
1.090
0.194
0.474
0.305
0.123
0.006
2.111
19.362
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Compiled from OICA Production Statistics, http://www.oica.net/category/production-statistics/, accessed December 2013.
China
USA
Japan
Germany
South Korea
India
Brazil
Mexico
Canada
Thailand
Russia
Spain
France
UK
Czech Rep.
Others
Total
Exhibit 2
Cars
10.383
2.195
6.862
4.964
3.158
2.175
2.575
0.942
0.822
0.313
0.599
1.812
1.819
0.999
0.976
7.171
47.772
-16-
2009
Commercial
3.407
3.535
1.071
0.245
0.354
0.466
0.607
0.618
0.668
0.685
0.125
0.357
0.228
0.09
0.006
1.549
14.019
614-022
Google Car
Selected Google Products and Services, 1997–2013
Product/Service
Description
Google Web
Search
A comprehensive online search engine available in 181 countries and 146 languages. In 2000,
Google started to sell advertisements for keyword searches. By 2013, over 3 billion searches
were conducted each day.
1997
Google Toolbar
Google Image
Search
Google Groups
A tool that allowed users to add the Google search engine directly to their browsers.
2000
A tool that allowed users to search directly for images.
2001
Google acquired Deja.com’s Usenet Discussion service and created Google Groups, a platform
for online discussion groups.
2001
Google News
Aggregated headlines from over 50,000 news sources around the world. It transitioned out of beta
in 2006 and included personalized searches and recommendations. By 2012, Google News had
sent 6 billion clicks per month to publishers.
2002
Google Shopping
A search engine for products on shopping websites, originally launched as Froogle in 2002. The
name was changed to Google Shopping in 2012.
2002
Google Blogger
Google acquired Pyra Labs, the creator of Blogger, in 2003. A free tool used to create blogs,
Blogger had more than 300 million visitors every month in 2013.
2003
Google Books
Originally launched as Google Print in 2004, Google Books was a database of over 30 million
full and portions of books that had been scanned and converted into text for users to search and
read for free.
Google acquired Picasa in 2004 and offered the product as freeware for managing and editing
digital photos. Google added Picasa Web Albums in 2006 to enable online sharing and offered 1
GB free online storage, with more storage available at a cost.
2004
Google Alerts
A service developed in 2004 that allowed users to receive automatic e-mail updates of relevant
Google results based on their queries.
2004
Google Scholar
A search engine for scholarly literature across many disciplines and sources.
2004
Google Orkut
A social networking site popular in India and Brazil with 66 million active users worldwide by
2013.
2004
Google Video
Search
Begun in 2005 as a video hosting and searching platform. Google developed partnerships with
major media outlets in 2009, and hosted content users could pay to download in the Google
Video store. In 2012, Google Video stopped hosting video but remained a search engine for
videos.
Google acquired Keyhole Inc. in 2001, the developers of EarthViewer 3D, and launched
EarthViewer 3D as Google Earth in 2005. A program available on desktops, online, and mobile
devices, Google Earth allowed users to view satellite images and maps of the earth and later the
sky, ocean, moon, and Mars. By 2011, Google Earth had been downloaded over 1 billion times.
Google launched Google Maps, an online mapping service with street maps, route planners, and a
search for locations and businesses, in 2005. Google Maps was developed at Where 2
Technologies, which Google acquired in 2004, and was available both online and on mobile
devices.
A free, searchable, Web-based e-mail service. When it launched, Gmail gave each user 1 GB of
free storage, more than 100 times the storage other free Web mail services offered at the time. In
2013, Google gave users 15 GB of storage across Gmail, Google Drive, and Google+ Photos. By
2012, Gmail had over 425 million active users around the world.
2005
A search engine for blogs and blog posts, which was discontinued in 2014.
2005
A website that featured developer tools, application programming interfaces (APIs), and other
resources.
Google acquired Urchin Software Corp., a Web analytics company, in 2005. Google Analytics
was released the same year to measure the impact of websites and marketing campaigns.
A video-sharing website that Google acquired in 2006 for $1.65 billion and that allowed users to
upload, share, and view videos. YouTube, which was localized in 56 countries in 61 languages,
had more than 1 billion unique viewers each month in 2013.
Translated over 70 languages. In 2012, Google Translate had over 200 million active monthly
users, 92% of which came from outside of the U.S.
2005
Picasa
Google Earth
Google Maps
Gmail
Google Blog
Search
Google Code
Google Analytics
YouTube
Google Translate
Year
launched
2004
2005
2005
2005
2005
2006
2006
17
Page 77 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Exhibit 3
614-022
Google Car
An online calendar that allowed users to organize their schedules and share events with other
people.
Allowed users to search millions of patents that were submitted to patent offices in the U.S. since
1790 and in Europe since 1978.
2006
A comprehensive financial website for business and enterprise headlines as well as up–to-date
stock information. In 2013, it had over 1.5 million unique monthly visitors.
2006
Custom Search
A platform that allowed users to customize search by selecting certain websites to search; users
could add their unique searches to their websites with the Google Custom Search box.
2006
Google Trends
A tool that analyzed data aggregated from millions of users, Google Trends helped users
visualize trends and show how many searches were conducted for a certain term.
2006
Android
Google acquired the startup Android Inc. in 2005 and announced Android, the first open platform
for mobile devices, in 2007. By 2013, more than 1 billion Android devices had been activated.
Contained a suite of applications including Gmail, Calendar, Maps, and many other Google
services on the Android operating system.
Allowed users to store and organize their digital photos. Millions of photos were uploaded on
Panoramio, which was available in 42 languages and could be viewed by other users on Google
Earth and Google Maps.
2007
A Web browser launched in 2008 designed to run quickly and crash less often than other
browsers. In 2012, it was the most popular browser in the world, with over 310 million active
users.
A Web application that allowed users to easily build and share either public or private sites with
100 MB free storage and 10 GB free storage for Google Apps users.
2008
Gave users a phone number that could be used across different phones. Calls within the U.S. were
free, and international calls were inexpensive, for instance, at 2 cents a minute for calls to
countries such as China, New Zealand, and the U.K.
A cloud-based product that connected printers to the Web, allowing users to access their printers
from any Web-connected device.
2009
Google created Google TV in 2010 for Android and Chrome. The platform combined TV
programming and the Web into one experience.
2010
Google Offers
A service that offered discounts and coupons for products and services. Google Offers was
launched in beta in 2011 and later integrated with Google Maps and Google Wallet.
2011
Google Wallet
A mobile app that allowed users to store and pay with their credit cards, debit cards, rewards, and
offers. Users could also use Google Wallet to send money to anyone in the U.S. with an e-mail
address.
2011
Google+
Google
Chromebook
Google’s social network, Google+, attracted 359 million active users by 2013.
2011
Along with partners Samsung and Acer, Google launched the Chromebook, a fast, simple, and
affordable notebook, in 2011.
2011
Google Play
A cloud-based platform, Google Play gave users access to media that could be downloaded and
wirelessly synced across multiple devices. Google Play combined and replaced Google Android
Market, Google Music, and Google eBookstore.
2012
Google Drive
An online service that allowed users to create, share, and store files either on their hard drives or
online with 5 GB of free storage. Google Docs, an office suite within Google Drive, allowed
multiple users to collaboratively edit documents that could be synced across multiple devices.
2012
Google Fiber
2012
Google Hangouts
Google Fiber, ultra-high-speed Internet, was launched in 2012 in Kansas City, Kansas, and
Kansas City, Missouri.
A videoconferencing tool that allowed users to chat with up to 10 people at a time. It replaced
Google Talk, Google+ Hangouts, and Messenger to become Google’s single communications
system.
Calico
Google announced Calico, which was still under development, in 2013. Calico focused on health
and well-being.
2013
Project Loon
Google announced Project Loon, still under development, in 2013. Project loon provided balloonpowered Internet access to connect rural and underserved areas and be used for crisis response
communication.
2013
Google Calendar
Google Patent
Search
Google Finance
Google Apps
Panoramio
Google Chrome
Google Sites
Google Voice
Google Cloud
Print
Google TV
Source:
For source information, see endnote 161.161
18
Page 78 of 282
2006
2007
2007
2008
2010
2013
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
614-022
Google Car
Google Revenues by Revenue Source, 2011-2013 (in millions USD)
Google:
Advertising Revenues
Google websites
Google Network Members’ websites
Total advertising revenues
Other revenues
Total Google Revenues (advertising and other)
Motorola Mobile:
Total Motorola Mobile revenues (hardware and other)
Total revenues
Source:
2011
2012
2013
$26,145
10,386
36,531
1,374
37,905
$31,221
12,465
43,686
2,353
46,039
$37,422
13,125
50,547
4,972
55,519
0
37,905
4,136
50,175
4,306
59,825
Google Inc., 2012 Annual Report, accessed December 2014.
Exhibit 4b
Consolidated Income Statement (in millions USD)
Revenues
Google (advertising and other)
Motorola Mobile (hardware and other)
Total revenues
Cost and expenses
Cost of revenues—Google (advertising and other)
Cost of revenues—Motorola Mobile (hardware and other)
Research and development
Sales and marketing
General and administrative
Charge related to the resolution of Department of Justice investigation
Total costs and expenses
Income from operations
Interest and other income, net
Income from continuing operations before income taxes
Provision for income taxes
Net income from continuing operations
Net loss from discontinued operations
Net income
Source:
2011
2012
2013
$37,905
0
$37,905
$46,039
4,136
$50,175
$55,519
4,306
$59,825
13,188
0
5,162
4,589
2,724
500
26,163
11,742
584
12,326
2,589
$9,737
0
$9,737
17,176
3,458
6,793
6,143
3,845
0
37,415
12,760
626
13,386
2,589
$10,788
(51)
$10,737
21,993
3,865
7,952
7,253
4,796
0
45,859
13,966
530
14,496
2,282
12,214
706
$12,920
Google Inc., 2012 Annual Report, accessed December 2014.
19
Page 79 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Exhibit 4a
614-022
614-022
Google Car
Source:
“The Google Self Driving Car Maneuvers through the Streets,” http://www.gettyimages.com/detail/newsphoto/the-google-self-driving-car-maneuvers-through-the-streets-news-photo/144473639; “Gov. Brown Signs
Legislation At Google HQ That Allows Testing Of Autonomous Vehicles,” http://www.gettyimages.com
/detail/news-photo/people-look-at-camera-on-top-of-a-google-self-driving-car-news-photo/152766329,
accessed
October 2013.
Exhibit 6
Source:
Google Car Maneuvering Streets and Close-Up of Lidar, 2012
Diagram of Self-Driving Car
“Look, No Hands,” The Economist, September 12, 2012, http://
www.economist.com/node/21560989, accessed October 2013.
20
Page 80 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Exhibit 5
Google Car
Source:
Image of Google Car Reading the Environment, 2011
Google and the Google logo are registered trademarks of Google Inc., used with permission.
Exhibit 8
National Highway Traffic Safety Administration’s Five Levels of Vehicle Automation
No-Automation (Level 0): The driver is in complete and sole control of the primary vehicle controls—brake, steering,
throttle, and motive power—at all times.
Function-specific Automation (Level 1): Automation at this level involves one or more specific control functions.
Examples include electronic stability control or pre-charged brakes, where the vehicle automatically assists with braking to
enable the driver to regain control of the vehicle or stop faster than possible by acting alone.
Combined Function Automation (Level 2): This level involves automation of at least two primary control functions
designed to work in unison to relieve the driver of control of those functions. An example of combined functions enabling a
Level 2 system is adaptive cruise control in combination with lane centering.
Limited Self-Driving Automation (Level 3): Vehicles at this level of automation enable the driver to cede full control of all
safety-critical functions under certain traffic or environmental conditions and in those conditions to rely heavily on the vehicle
to monitor for changes in those conditions requiring transition back to driver control. The driver is expected to be available for
occasional control, but with sufficiently comfortable transition time. The Google car is an example of limited self-driving
automation.
Full Self-Driving Automation (Level 4): The vehicle is designed to perform all safety-critical driving functions and
monitor roadway conditions for an entire trip. Such a design anticipates that the driver will provide destination or navigation
input, but is not expected to be available for control at any time during the trip. This includes both occupied and unoccupied
vehicles.
Source:
“U.S. Department of Transportation Releases Policy on Automated Vehicle Development, “National Highway
Traffic Safety Administration, May 30, 2013, http://www.nhtsa.gov/About+NHTSA/Press+Releases/U.S.
+Department+of+Transportation+Releases+Policy+on+Automated+Vehicle+Development, accessed September
2013.
21
Page 81 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Exhibit 7
614-022
614-022
Google Car
Source:
Diagram of Volvo’s Platoon System
Erik Coelingh and Stefan Solyom, “All Aboard the Robotic Road Train,” IEEE Spectrum, October 26, 2012,
http://spectrum.ieee.org/green-tech/advanced-cars/all-aboard-the-robotic-road-train, accessed October 2012.
22
Page 82 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Exhibit 9
Google Car
614-022
Endnotes
1 Sebastian Thrun, “What We’re Driving At,” Google Blog, October 9, 2010, http://googleblog.blogspot.com/2010/10/
what-were-driving-at.html, accessed September 2013.
http://www.businessinsider.com/google-self-driving-car-sensor-cost-2012-9, accessed October 2013.
3 Doug Gross, “Google: Self-Driving Cars Are Mastering City Streets,” CNN, April 28, 2014,
http://www.cnn.com/2014/04/28/tech/innovation/google-self-driving-car/, accessed December 2014.
4 Donna Tam, “Google’s Sergey Brin: You’ll ride in robot cars within 5 years,” CNET, September 25, 2012,
http://news.cnet.com/8301-11386_3-57520188-76/googles-sergey-brin-youll-ride-in-robot-cars-within-5-years/, accessed
October 2013.
5 Mahendra Ramsinghani, “Forget Google Glass, I Want A Google Car,” Forbes, May 20, 2013,
http://www.forbes.com/sites/mahendraramsinghani/2013/05/20/forget-google-glass-i-want-a-google-car/, accessed
September 2013.
6 Dan Graziano, “Google finally discloses Waze acquisition price,” BGR, July 26, 2013, http://bgr.com/2013/07/26/google-
waze-acquisition-price/, accessed October 2013.
7 Lucas Mearian, “Self-driving cars could create 1GB of data a second,” Computerworld, July 23, 2013,
http://www.computerworld.com/s/article/print/9240992/Self_driving_cars_could_create_1GB_of_data_a_second?taxonom
yName=Emerging+Technologies&taxonomyId=128, accessed September 2013.
8 Lance Whitney, “Google to profit from self-driving cars by decade's end—analyst,” CNET, July 9, 2013,
http://news.cnet.com/8301-1023_3-57592837-93/google-to-profit-from-self-driving-cars-by-decades-endanalyst/?tag=nl.e703&s_cid=e703&ttag=e703&ftag=, accessed September 2013.
9 Mary Bellis, “The History of the Automobile,” About.com, http://inventors.about.com/library/weekly/aacarssteama.htm,
accessed September 2013.
10 “The History of the Automobile,” University of Colorado, http://l3d.cs.colorado.edu/systems/agentsheets/new-
vista/automobile/, accessed September 2013.
11 “The History of the Automobile.”
12 “The History of the Automobile.”
13 “The Automobile Industry, 1920–1929,” Bryant University,
http://web.bryant.edu/~ehu/h364/materials/cars/cars%20_30.htm, accessed September 2013.
14 “Self-driving cars: the next revolution.”
15 “Self-driving cars: the next revolution.”
16 “Average Commute Times,” WYNC, http://project.wnyc.org/commute-times-us/embed.html#5.00/42.000/-89.500,
accessed November 2014.
17 Janet Loehrke, “Americans’ commutes aren’t getting longer,” USA Today, March 5, 2013,
http://www.usatoday.com/story/news/nation/2013/03/05/americans-commutes-not-getting-longer/1963409/, accessed
September 2013.
18 National Highway Traffic Safety Administration (NHTSA), “Fatality Analysis Reporting System (FARS) Encyclopedia,”
http://www-fars.nhtsa.dot.gov/Main/index.aspx, accessed October 2013.
19 Peter Valdes-Dapena, “Traffic deaths lowest since 1950,” CNN Money, September 9, 2010,
http://money.cnn.com/2010/09/09/autos/nhtsa_traffic_deaths/index.htm, accessed October 2013.
20 Valdes-Dapena, “Traffic deaths lowest since 1950.”
21 Thrun, “What We’re Driving At.”
23
Page 83 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
2 Owen Thomas, “Google’s Self-Driving Cars May Cost More Than a Ferrari,” Business Insider, September 7, 2012,
614-022
Google Car
22 “The Global Auto Industry Shifts Its Focus to Overseas and Emerging Markets,” Standard & Poor’s Ratings Services
CreditWeek, September 25, 2013,
http://www.standardandpoors.com/spf/upload/Events_US/US_CR_Event_Auto_cwart1.pdf, accessed December 2013.
23 Chris Isidore, “Car Sales Make a Strong Comeback in 2013,” CNN Money, January 3, 2014,
24 “Car Sales in China Are Forecast to Rise 7%,” Wall Street Journal, January 11, 2013,
http://online.wsj.com/news/articles/SB10001424127887324081704578235150976448808, accessed November 2013.
25 “Strong China Demand to Boost Global Car Sales by 4.8%, Moody’s Says,” IndustryWeek, September 4, 2013,
http://www.industryweek.com/demographics/strong-china-demand-boost-global-car-sales-48-moodys-says, accessed
November 2013.
26 “Indonesia breaks record in car sales during 2012,” Global Times, January 11, 2013,
http://www.globaltimes.cn/content/755237.shtml, accessed November 2013.
27 “South Korea Car Market lost 4% sales in the 2012. Hyundai Avante on top,” focus2move, January 22, 2013,
http://focus2move.com/item/415-south-korea-car-market-lost-4-sales-in-the-2012-hyundai-avante-on-top, accessed
November 2013.
28 “Detroit of the East,” The Economist, April 4, 2013, http://www.economist.com/blogs/schumpeter/2013/04/thailands-
booming-car-industry, accessed November 2013.
29 Magna Steyr Company website, http://www.magnasteyr.com/capabilities/vehicle-engineering-contract-manufacturing,
accessed November 2013.
30 Chanchal Pal Chauhan, “Foreign Carmakers Take Contract Manufacturing Route to Launch Their Products in India,”
Economic Times, October 1, 2013, via Factiva, accessed November 2013.
31 Alan Ohnsman, “Nissan Sets Goal of Introducing First Self-Driving Cars by 2020,” Bloomberg, August 27, 2013,
http://www.bloomberg.com/news/2013-08-27/nissan-sets-goal-of-bringing-first-self-driving-cars-by-2020.html, accessed
October 2013.
32 Umit Ozguner, Christoph Stiller, and Keith Redmill, “Systems for Safety and Autonomous Behavior in Cars: the DARPA
Grand Challenge Experience,” IEEE Xplore 95, no. 2 (February 2007), accessed September 2013.
33 Adam Fisher, “Inside Google’s Quest to Popularize Self-Driving Cars,” Popular Science, September 18, 2013,
http://www.popsci.com/cars/article/2013-09/google-self-driving-car, accessed October 2013.
34 “How does a self-driving car work?” The Economist, April 29, 2013, http://www.economist.com/blogs/economist-
explains/2013/04/economist-explains-how-self-driving-car-works-driverless, accessed September 2013.
35 Lucas Mearian, “Self-driving cars could create 1GB of data a second,” Computerworld, July 23, 2013,
http://www.computerworld.com/s/article/print/9240992/Self_driving_cars_could_create_1GB_of_data_a_second?taxonom
yName=Emerging+Technologies&taxonomyId=128, accessed September 2013.
36 “Google Inc.,” company history, Hoovers, http://www.hoovers.com, accessed October 2013.
37 Google, “Company Overview,” Google website, http://www.google.com/about/company/, accessed October 2013.
38 Google, Investor Relations, “Code of Conduct,” last updated April 25, 2012, http://investor.google.com/corporate/code-of-
conduct.html, accessed October 2013.
39 “Google Inc.,” company history, Hoovers, http://www.hoovers.com, accessed October 2013.
40 “Google buys YouTube for $1.65 billion,” NBC News, October 10, 2006,
http://www.nbcnews.com/id/15196982/ns/business-us_business/t/google-buys-youtube-billion/#.UlV5nhAa5WA,
accessed October 2013.
41 “Google Inc.,” company history.
42 “Google Inc.,” company history.
24
Page 84 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
http://money.cnn.com/2014/01/03/news/companies/car-sales/, accessed November 2014.
Google Car
614-022
43 “How Google Tracks Traffic,” The Connectivist, July 3, 2013, http://www.theconnectivist.com/2013/07/how-google-tracks-
traffic/, accessed October 2013.
44 Google 2013 Annual Report, investor.google.com/pdf/2013_google_annual_report.pdf , accessed November 2014.
http://www.pcworld.com/article/2084480/google-car-companies-to-bring-android-to-cars.html, accessed January 2013.
46 Bogdan Petrovan, “Google, Car Companies to Bring Android to Cars,” PCWorld, January 6, 2014,
http://www.androidauthority.com/google-auto-link-car-os-394598/, accessed December 2014.
47 Essers, “Google, Car Companies to Bring Android to Cars.”
48 Neal E. Boudette and Daisuke Wakabayashi, “Google, Apple Forge Auto Ties,” Wall Street Journal, December 29, 2013,
http://online.wsj.com/news/articles/SB10001424052702304591604579288670734733740, accessed January 2013.
49 Essers, “Google, Car Companies to Bring Android to Cars.”
50 “Data Mining,” Research at Google, http://research.google.com/pubs/DataMining.html, accessed October 2013.
51 “comScore Releases July 2013 U.S. Search Engine Rankings,” comScore, August 14, 2013,
http://www.comscore.com/Insights/Press_Releases/2013/8/comScore_Releases_July_2013_U.S._Search_Engine_Rankings,
accessed October 2013.
52 Danny Sullivan, “Google Still World’s Most Popular Search Engine by Far, But Share of Unique Searchers Dips Slightly,”
Search Engine Land, February 11, 2013, http://searchengineland.com/google-worlds-most-popular-search-engine-148089,
accessed October 2013.
53 Craig Smith, “By the Numbers: A Gigantic List of Google Stats and Facts,” September 2013,
http://expandedramblings.com/index.php/by-the-numbers-a-gigantic-list-of-google-stats-and-facts/, accessed October 2013.
54 Zoe Fox, “Google Earns 33% of Online Ad Revenues,” Mashable, August 28, 2013,
http://mashable.com/2013/08/28/online-ad-revenues/, accessed October 2013.
55 Robert McMillan, “Google: We’re One of the World’s Largest Hardware Makers,” Wired, June 22, 2012,
http://www.wired.com/wiredenterprise/2012/06/google_makes_servers/, accessed November 2013.
56 Cade Metz, “Intel Confirms Decline of Server Giants HP, Dell, and IBM,” Wired, September 12, 2012,
http://www.wired.com/wiredenterprise/2012/09/29853/, accessed November 2013.
57 McMillan, “Google: We’re One of the World’s Largest Hardware Makers.”
58 McMillan, “Google: We’re One of the World’s Largest Hardware Makers.”
59 McMillan, “Google: We’re One of the World’s Largest Hardware Makers.”
60 Metz, “Intel Confirms Decline of Server Giants HP, Dell, and IBM.”
61 Dan Schneider, “Under the Hood at Google and Facebook,” IEEE Spectrum, May 31, 2011,
http://spectrum.ieee.org/telecom/internet/under-the-hood-at-google-and-facebook, accessed November 2013.
62 “Efficiency: How We Do It,” Google Data Centers, http://www.google.com/about/datacenters/efficiency/internal/,
accessed December 2014.
63 “Apple Cedes Market Share in Smartphone Operating System Market as Android Surges and Windows Phone Gains,” IDC,
August 7, 2013, http://www.idc.com/getdoc.jsp?containerId=prUS24257413, accessed October 2013.
64 Jamie Hinks, “Google Maps is the world's most popular smartphone app,” IT ProPortal, August 8, 2013,
http://www.itproportal.com/2013/08/08/google-maps-is-the-worlds-most-popular-smartphone-app/, accessed October
2013.
65 “Mobile Stakes Its Claim on Local Search,” eMarketer, April 11, 2013, accessed October 2013.
66 “How Google Tracks Traffic,” The Connectivist, July 3, 2013, http://www.theconnectivist.com/2013/07/how-google-tracks-
traffic/, accessed October 2013.
25
Page 85 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
45 Loek Essers, “Google, Car Companies to Bring Android to Cars,” PCWorld, January 6, 2014,
614-022
Google Car
67 “Google I/O 2013—Project Ground Truth: Accurate Maps Via Algorithms and Elbow Grease,” Google Developers YouTube
page, May 16, 2013, http://www.youtube.com/watch?v=FsbLEtS0uls#t=294, accessed October 2013.
68 “Google I/O 2013—Project Ground Truth: Accurate Maps Via Algorithms and Elbow Grease.”
September 6, 2012, http://www.theatlantic.com/technology/archive/2012/09/how-google-builds-its-maps-and-what-itmeans-for-the-future-of-everything/261913/, accessed October 2013.
70 Alexis C. Madrigal, “How Google Builds Its Maps—and What It Means for the Future of Everything.”
71 Tom Krazit, “Why Google isn’t worried about Android revenue,” GigaOm, April 1, 2012,
http://gigaom.com/2012/04/01/why-google-isnt-worried-about-androids-revenue/, accessed October 2013.
72 Google, 2012 Annual Report, http://investor.google.com/pdf/2012_google_annual_report.pdf, accessed October 2013.
73 “Smartphone OS Market Share, Q3 2014,” IDC, http://www.idc.com/prodserv/smartphone-os-market-share.jsp, accessed
December 2014.
74 “Android Pushes Past 80% Market Share While Windows Phone Shipments Leap 156.0% Year over Year in the Third
Quarter, According to IDC,” IDC, November 12, 2013, http://www.idc.com/getdoc.jsp?containerId=prUS24442013, accessed
December 2014.
75 Google, 2013 Annual Report.
76 Nathan Olivarez-Giles, “Nexus Smartphones: Who Wins, Who Loses If Google Launches Android 5.0 in Multiple
Handsets?” Wired, May 21, 2012, http://www.wired.com/gadgetlab/2012/05/nexus-smartphones-who-wins-and-loses-ifgoogle-launches-android-5/, accessed October 2013.
77 Thomas Halleck, “Motorola Moto X Release Date Arrives for Verizon, T-Mobile, Sprint: Google Phone Now Offered on Most
U.S. Carriers,” September 10, 2013, International Business Times, http://www.ibtimes.com/motorola-moto-x-release-datearrives-verizon-t-mobile-sprint-google-phone-now-offered-most-us, accessed October 2013.
78 Christopher Mims, “Why Google Just Sold Motorola to Lenovo for $3 Billion,” Quartz, January 30, 2014,
http://qz.com/172207/why-google-just-sold-motorola-to-lenovo-for-3-billion/, accessed December 2014.
79 Steve Russell, “DARPA Grand Challenge Winner: Stanley the Robot!” Popular Mechanics, January 9, 2006,
http://www.popularmechanics.com/technology/engineering/robots/2169012, accessed November 2013.
80 Steve Russell, “DARPA Grand Challenge Winner: Stanley the Robot!”
81 DARPA Grand Challenge 2005 Rules, http://archive.darpa.mil/grandchallenge05/Rules_8oct04.pdf, accessed October
2013.
82 Umit Ozguner, Christoph Stiller, and Keith Redmill, “Systems for Safety and Autonomous Behavior in Cars: the DARPA
Grand Challenge Experience,” IEEE Xplore 95, no. 2 (February 2007), accessed September 2013.
83 Claire Cain Miller and Nick Bilton, “Google’s Lab of Wildest Dreams,” New York Times, November 13, 2011,
http://www.nytimes.com/2011/11/14/technology/at-google-x-a-top-secret-lab-dreaming-up-thefuture.html?pagewanted=all, accessed September 2013.
84 Brad Stone, “Inside Google’s Secret Lab,” Bloomberg BusinessWeek, May 22, 2013,
http://www.businessweek.com/printer/articles/120106-inside-googles-secret-lab, accessed August 2013.
85 Cain Miller and Bilton, “Google’s Lab of Wildest Dreams.”
86 Mahendra Ramsinghani, “Forget Google Glass, I Want a Google Car,” Forbes, May 20, 2013,
http://www.forbes.com/sites/mahendraramsinghani/2013/05/20/forget-google-glass-i-want-a-google-car/, accessed
September 2013.
87 Thrun, “What We’re Driving At.”
88 Andy Kessler, “Sebastian Thrun: What’s Next for Silicon Valley?” Wall Street Journal, June 15, 2012,
http://online.wsj.com/article/SB10001424052702303807404577434891291657730.html, accessed September 2013.
26
Page 86 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
69 Alexis C. Madrigal, “How Google Builds Its Maps—and What It Means for the Future of Everything,” The Atlantic,
Google Car
614-022
89 Fisher, “Inside Google’s Quest.”
90 Steve Rosenbush, “Under Pressure, Google May Slow Rollout of Driverless Car Technology,” Wall Street Journal, July 18,
91 Thrun, “What We’re Driving At.”
92 MG Siegler, “World-Changing Awesome Aside, How Will the Self-Driving Google Car Make Money?” TechCrunch, October
9, 2012, http://techcrunch.com/2010/10/09/google-car/, accessed October 2013.
93 Rosenbush, “Under Pressure, Google May Slow Rollout of Driverless Car Technology.”
94 Fisher, “Inside Google’s Quest.”
95 Fisher, “Inside Google’s Quest.”
96 “How does a self-driving car work?”
97 Erico Guizzo, “How Google’s Self-Driving Car Works,” IEEE Spectrum, October 18, 2011,
http://spectrum.ieee.org/automaton/robotics/artificial-intelligence/how-google-self-driving-car-works#, accessed
September 2013.
98 Fisher, “Inside Google’s Quest.”
99 Guizzo, “How Google’s Self-Driving Car Works.”
100 “How does a self-driving car work?”
101 “How does a self-driving car work?”
102 “How does a self-driving car work?”
103 Davon Lavrinc, “Exclusive: Google Expands Its Autonomous Fleet with Hybrid Lexus RX450h,” Wired, April 16, 2012,
http://www.wired.com/autopia/2012/04/google-autonomous-lexus-rx450h/, accessed October 2013.
104 Guizzo, “How Google's Self-Driving Car Works.”
105 Guizzo, “How Google’s Self-Driving Car Works.”
106 Thrun, “What We’re Driving At.”
107 Guizzo, “How Google’s Self-Driving Car Works.”
108 Henry Blodget, “Here Are Some of the Problems Google Is Having with Its Self-Driving Cars,” March 3, 2013,
BusinessInsider, http://www.businessinsider.com/google-self-driving-car-problems-2013-3, accessed September 2013.
109 Amir Efrati, “Exclusive: Google’s Designing Its Own Self-Driving Car, Considers Robo Taxi,” Jessica Lessin, August 23, 2013,
http://jessicalessin.com/2013/08/23/exclusive-google-designing-its-own-self-driving-car-considers-robo-taxi-2/, accessed
September 2013.
110 Fisher, “Inside Google’s Quest.”
111 Fisher, “Inside Google’s Quest.”
112 Efrati, “Exclusive: Google’s Designing Its Own Self-Driving Car.”
113 Efrati, “Exclusive: Google’s Designing Its Own Self-Driving Car.”
114 Guizzo, “How Google's Self-Driving Car Works.”
115 Efrati, “Exclusive: Google’s Designing Its Own Self-Driving Car.”
116 Alex Wilhelm, “Google Ventures Puts $258M into Uber, Its Largest Deal Ever,” TechCrunch, August 22, 2013,
http://techcrunch.com/2013/08/22/google-ventures-puts-258m-into-uber-its-largest-deal-ever/, accessed September 2013.
117 Fisher, “Inside Google’s Quest.”
27
Page 87 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
2013, http://blogs.wsj.com/cio/2013/07/18/under-pressure-google-may-slow-rollout-of-driverless-car-technology/, accessed
September 2013.
614-022
Google Car
118 “Just Press Go: Designing a Self-Driving Vehicle,” Google Blog, May 27, 2014,
http://googleblog.blogspot.com/2014/05/just-press-go-designing-self-driving.html, accessed December 2014.
119 Frederic Lardinois, “Study: 57% of Consumers Worldwide Say They Would Trust Driverless Cars, 46% Would Let Their
120 Lardinois, “Study: 57% of Consumers Worldwide Say They Would Trust Driverless Cars.”
121 Mary Gannon, “Consumers not quite ready for driverless cars,” Connector Tips, June 27, 2013,
http://www.connectortips.com/consumers-not-quite-ready-for-driverless-cars/, accessed September 2013.
122 Gene Munster, Douglas J. Clinton, Matthew E. Lebo, “Self-Driving Cars a $200+ Billion Opportunity for Google,”
PiperJaffray, July 9, 2013, via Thomson ONE, accessed August 2013.
123 J. D. Power and Associates study cited in “Self-driving cars: the next revolution,” KPMG Center for Automotive Research,
2012, https://www.kpmg.com/US/en/IssuesAndInsights/ArticlesPublications/ Documents/self-driving-cars-nextrevolution.pdf, accessed September 2013.
124 “Self-driving cars: the next revolution.”
125 “Self-driving cars: the next revolution.”
126 Efrati, “Exclusive: Google’s Designing Its Own Self-Driving Car.”
127 Jonathan Oosting, “Michigan gives green light to autonomous vehicle testing despite concerns from Google,” Mlive,
December 13, 2013, http://www.mlive.com/politics/index.ssf/2013/12/michigan_gives_green_light_to.html, accessed
February 2015.
128 Monte Whaley, “Colorado driverless car bill shelved until further notice,” The Denver Post, February 5, 2013,
http://www.denverpost.com/ci_22526956/colorado-driverless-car-bill-shelved-until-further-notice, accessed September 2013.
129 Nannette Miranda, “DMV forced to make rules for Google’s car,” ABC Local, June 23, 2013,
http://abclocal.go.com/kfsn/story?section=news/national_world&id=9147124, accessed September 2013.
130 “Self-driving cars: the next revolution.”
131 Bryant Walker Smith, Stanford Law professor and organizer of the Challenges and Opportunities of Road Vehicle
Automation conference, as cited in Steve Rosenbush, “Under Pressure, Google May Slow Rollout of Driverless Car
Technology,” Wall Street Journal, July 18, 2013, http://blogs.wsj.com/cio/2013/07/18/under-pressure-google-may-slowrollout-of-driverless-car-technology/, accessed September 2013.
132 Fisher, “Inside Google's Quest to Popularize Self-Driving Cars,” Popular Science, September 18, 2013,
http://www.popsci.com/cars/article/2013-09/google-self-driving-car, accessed October 2013.
133 David Streitfeld “Google Concedes That Drive-By Prying Violated Privacy,” New York Times, March 12, 2013,
http://www.nytimes.com/2013/03/13/technology/google-pays-fine-over-street-view-privacybreach.html?pagewanted=all&_r=0, accessed October 2013.
134 James Niccolai, “Self-driving cars a reality for ‘ordinary people’ within 5 years, says Google's Sergey Brin,” Computer World,
September 25, 2012, http://www.computerworld.com/s/article/9231707/
Self_driving_cars_a_reality_for_39_ordinary_people_39_within_5_years_says_Google_39_s_Sergey_Brin, accessed October
2013.
135 Timothy B. Lee, “Self-driving cars are a privacy nightmare. And it’s totally worth it,” Washington Post, May 21, 2013,
http://www.washingtonpost.com/blogs/wonkblog/wp/2013/05/21/self-driving-cars-are-a-privacy-nightmare-and-itstotally-worth-it/, accessed October 2013.
136 Byron Spice, “Press Release: Carnegie Mellon Creates Practical Self-Driving Car Using Automotive-Grade Radars and
Other Sensors,” CMU News, September 4, 2013,
https://www.cmu.edu/news/stories/archives/2013/september/sept4_selfdrivingcar.html, accessed October 2013.
137 “Shelley, Stanford’s robotic racecar, hits the track,” Stanford News, August 13, 2012,
http://news.stanford.edu/news/2012/august/shelley-autonomous-car-081312.html, accessed October 2013.
28
Page 88 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Kids Ride in Them,” Tech Crunch, May 14, 2013, http://techcrunch.com/2013/05/14/study-57-of-consumers-worldwide-saythey-would-trust-driverless-cars-46-would-let-their-kids-ride-in-them/, accessed September 2013.
Google Car
614-022
138 Jeremy Hsu, “Mobileye Raises $400 Million for Cheaper Self-Driving Car Tech,” IEEE Spectrum, July 10, 2013,
http://spectrum.ieee.org/tech-talk/robotics/robotics-software/mobileye-raises-400-million-for-cheaper-self-driving-car-tech,
accessed September 2013.
139 Parmy Olson, “Driverless Cars for $10,000? This Startup is Challenging Google with a Simple Sensor,” Forbes,
140 Andy Fixmer and Brittany Levine, “Elon Musk Reveals Tesla D Can Park Itself, Greet Drivers,” Mashable, October 9, 2014,
http://mashable.com/2014/10/09/tesla-d-event/, accessed December 2014.
141 “How does a self-driving car work?”
142 Mark Hachman, “CES 2013: Audi Demonstrates Its Self-Driving Car,” Popular Science, January 9, 2013,
http://www.popsci.com/cars/article/2013-01/ces-2013-audi-demonstrates-its-self-driving-car, accessed September 2013.
143 Alex Davies, “Audi’s Self-Driving Car Hits 150 MPH on an F1 Track,” Wired, October 24, 2014,
http://www.wired.com/2014/10/audis-self-driving-car-hits-150-mph-f1-track/, accessed December 2014.
144 Davies, “Audi’s Self-Driving Car Hits 150 MPH on an F1 Track.”
145 Nancy Owano, “BMW shows hands-free driving on Autobahn,” Phys Org, January 24, 2012, http://phys.org/news/2012-
01-bmw-hands-free-autobahn-video.html, accessed September 2013.
146 Heather Kelly, “Driverless Car Tech Gets Serious at CES,” CNN, April 7, 2014,
http://www.cnn.com/2014/01/09/tech/innovation/self-driving-cars-ces/, accessed December 2014.
147 “Cadillac Preps Self-Driving Tech for 2017,” Cars.com, September 8, 2014,
http://blogs.cars.com/kickingtires/2014/09/cadillac-preps-self-driving-tech-for-2017.html, accessed December 2014.
148 Angela Greiling Keane, “Google’s Self-Driving Cars Get Boost from U.S. Agency,” Bloomberg, May 30, 2012,
http://www.bloomberg.com/news/2013-05-30/google-s-self-driving-cars-get-boost-from-u-s-agency.html, accessed October
2013.
149 Bill Howard, “Frankfurt Auto Show: Mercedes shows off fully autonomous S-Class, production cars coming by 2020,”
Extreme Tech, September 16, 2013, http://www.extremetech.com/extreme/166598-frankfurt-auto-show-mercedes-shows-offfully-autonomous-s-class-production-cars-coming-by-2020, accessed September 2013.
150 Efrati, “Exclusive: Google’s Designing Its Own Self-Driving Car.”
151 Bill Howard, “Frankfurt Auto Show: Mercedes shows off fully autonomous S-Class, production cars coming by 2020,”
Extreme Tech, September 16, 2013, http://www.extremetech.com/extreme/166598-frankfurt-auto-show-mercedes-shows-offfully-autonomous-s-class-production-cars-coming-by-2020, accessed September 2013.
152 David Meyer, “Nokia is working on self-driving cars with Mercedes-Benz,” GigaOm, September 10, 2013,
http://gigaom.com/2013/09/10/nokia-is-working-on-self-driving-cars-with-mercedes/, accessed September 2013.
153 Ohnsman, “Nissan Sets Goal of Introducing First Self-Driving Cars by 2020.”
154 Ohnsman, “Nissan Sets Goal of Introducing First Self-Driving Cars by 2020.”
155 Jeremy Laird, “Volvo breaks new ground with amazing new self-driving car tech,” Tech Radar, July 22, 2013,
http://www.techradar.com/us/news/car-tech/volvo-aims-for-zero-deaths-with-autonomous-car-tech-1166533, accessed
September 2013.
156 Erik Coelingh and Stefan Solyom, “All Aboard the Robotic Road Train” http://spectrum.ieee.org/green-tech/advanced-
cars/all-aboard-the-robotic-road-train, IEEE Spectrum, November 2012, accessed September 2013.
157 Coelingh and Solyom, “All Aboard the Robotic Road Train.”
158 “Self-driving cars: the next revolution.”
159 Elvina Nawaguna, “U.S. May Mandate ‘Talking’ Cars by Early 2017,” Reuters, February 3, 2014,
http://www.reuters.com/article/2014/02/03/us-autos-technology-rules-idUSBREA1218M20140203, accessed December 2014.
160 Rosenbush, “Under Pressure, Google May Slow Rollout of Driverless Car Technology.”
29
Page 89 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
http://www.forbes.com/sites/parmyolson/2014/06/23/startup-driverless-car-sensors-google/, accessed December 2014.
614-022
Google Car
http://247wallst.com/media/2012/04/24/will-google-dump-google-finance/; “On the alert(s),” Google Blog, September 14,
2004, http://googleblog.blogspot.com/2004/09/on-alerts.html; “Mapping Your Way,” Google Blog, February 8, 2005,
http://googleblog.blogspot.com/2005/02/mapping-your-way.html; “A new year for Google Video,” Google Blog, January 9,
2006, http://googleblog.blogspot.com/2006/01/new-year-for-google-video.html; “Eureka! Your own search engine has
landed!” Google Blog, October 23, 2008, http://googleblog.blogspot.com/2006/10/eureka-your-own-search-engine-has.html;
“A picture’s worth a thousand clicks,” Google Blog, May 30, 2007, http://googleblog.blogspot.com/2007/05/pictures-worththousand-clicks.html; “Google Sites now open to everyone,” Google Blog, May 21, 2008,
http://googleblog.blogspot.com/2008/05/google-sites-now-open-to-everyone.html; “New Blog Search tools: Feeds, Hot
Queries and Latest Posts,” Google Blog, July 2, 2009, http://googleblog.blogspot.com/2009/07/new-blog-search-tools-feedshot-queries.html; “Here Comes Google TV,” Google Blog, October 4, 2010, http://googleblog.blogspot.com/2010/10/herecomes-google-tv.html; “Google Apps Highlights,” Google Blog, January 28, 2011,
http://googleblog.blogspot.com/2011/01/google-apps-highlights-1282011.html; “Google Offers beta launching in New York
City and the Bay Area,” Google Blog, July 12, 2011, http://googleblog.blogspot.com/2011/07/google-offers-beta-launchingin-new.html; “Introducing Google Play: All your entertainment, anywhere you go,” Google Blog, March 6, 2012,
http://googleblog.blogspot.com/2012/03/introducing-google-play-all-your.html; “Introducing Google Drive . . . yes, really,”
Google Blog, April 24, 2012, http://googleblog.blogspot.com/2012/04/introducing-google-drive-yes-really.html; “Chrome &
Apps @ Google I/O: Your web, everywhere,” Google Blog, June 28, 2012, http://googleblog.blogspot.com/2012/06/chromeapps-google-io-your-web.html; “Happy 5th Birthday Google Code!” Google Blog, March 16, 2010,
http://googlecode.blogspot.com/2010/03/happy-5th-birthday-google-code.html; “Google Acquires Picasa,” Google News,
July 13, 2004, http://googlepress.blogspot.com/2004/07/google-acquires-picasa.html; Sarah Perez, “What on Earth Is Google
Doing with Orkut?” Tech Crunch, February 20, 2012, http://techcrunch.com/2012/02/20/what-on-earth-is-google-doing-withorkut/; “Find out how our translations are created,” About Google Translate, http://translate.google.com/about/, “Our
history in depth,” Google Blog, http://www.google.com/intl/en/about/company/history/; Jeff Bertolucci, “Google Voice: 10
Reasons to Check It Out,” PC World, June 23, 2010,
http://www.pcworld.com/article/199611/Google_Voice_10_Reasons_to_Check_It_Out_.html; “About Google Patents,”
https://support.google.com/faqs/answer/2539193?hl=en; all sources accessed October 2013.
30
Page 90 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
161 Compiled from “Will Google Dump Google Finance?” 24/7 Wall Street, April 24, 2012,
9B21M060
Jashan Puniya and Robert D. Austin wrote this case solely to provide material for class discussion. The authors do not intend to
illustrate either effective or ineffective handling of a managerial situation. The authors may have disguised certain names and other
identifying information to protect confidentiality.
This publication may not be transmitted, photocopied, digitized, or otherwise reproduced in any form or by any means without the
permission of the copyright holder. Reproduction of this material is not covered under authorization by any reproduction rights
organization. To order copies or request permission to reproduce materials, contact Ivey Publishing, Ivey Business School, Western
University, London, Ontario, Canada, N6G 0N1; (t) 519.661.3208; (e) cases@ivey.ca; www.iveycases.com. Our goal is to publish
materials of the highest quality; submit any errata to publishcases@ivey.ca. i1v2e5y5pubs
Copyright © 2021, Ivey Business School Foundation
Version: 2021-05-25
In late 2020, observers following the open banking activities of JPMorgan Chase & Co. (JPMC) might have
felt confused. On the one hand, JPMC seemed to be embracing open banking. The company had hired
Sairam Rangachari in 2017 to be its global head of Open Banking, Treasury Services and head of its
application programming interface (API) strategy. He launched JPMC’s open banking initiative in 2018,
and by mid-2019 the company had released more than two dozen APIs for its Corporate Treasury
customers.2 In September of 2020, an entire downloadable section of the JPMC website was prominently
titled “The Open Banking Transformation.”3
On the other hand, JPMC and Wells Fargo, another banking giant, were known in 2020 as leaders of the
resistance to open banking.4 At issuethe crux of open banking disputeswas whether, and how, large
institutions such as JPMC would allow the consumer data housed within their computer systems to be
accessed by third-party firms, many of them start-ups that wanted to use the data to provide new financial
services financial technology companies (fintechs). US law made it clear that consumers owned their own
financial data, including their data housed inside bank systems. Consumers had the legal right, then, to
authorize third parties to access their data residing inside bank systems. And many consumers who wanted to
use leading-edge, application (app)-based financial services did just that, agreeing in writing to let fintechs
access their banking data, often by sharing their bank login credentials with the fintechs. But some banks,
JPMC included, wanted to set terms for permitting this kind of access and sometimes actively obstructed it.5
In 2015, for example, JPMC blocked access by Mint, an online financial management service owned by
Intuit Inc., prohibiting Mint users from accessing their Chase bank accounts through the Mint software.
This caused significant dissatisfaction among some Chase customers.6 When JPMC’s chief executive
officer (CEO), Jamie Dimon, met that year with the head of the Consumer Financial Protection Bureau
(CFPB), he defended the company’s actions by citing cybersecurity and consumer privacy risks.7 Mint and
other fintechs dismissed such explanations; they saw the bank’s actions as attempts to limit competition.8
Indeed, some pundits suggested that the banks had good reason to worry. The swarm of agile, innovative
fintechs building new services on top of the operational frameworks maintained by traditional banks put
the latter at risk of becoming custodians of the industry’s “dumb pipes.”9
However, there was also evidence that concerns such as those raised by Dimon might be justified. In August of
the same year, the Wall Street Journal revealed that data aggregation service Yodlee, Inc. (Yodlee) was selling
transaction data to hedge funds looking for a trading edge.10 Although Yodlee claimed the data was anonymized
Page 91 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
JPMORGAN CHASE & CO.: OPEN BANKING1
Page 2
9B21M060
“The consumer has already decided,” countered John Pitts, an executive at the fintech Plaid Inc. (Plaid).
He explained, “They have already voted with their thumbs that [the service fintechs can provide] is
something they want.” Others argued that uses of data criticized by consumer advocates were the key to
services consumers wanted and intended to agree to. For example, Jason Gross, CEO and co-founder of
Petal Card, Inc., which described itself as “a credit card company started by people who were sick of credit
card companies,” said his company’s access to consumer datawhich critics considered
intrusiveallowed the company to keep better track of borrower activity, thereby allowing the company to
provide credit at a lower rate than would be possible by relying on traditional credit reporting.13
As for cybersecurity breaches, fintechs argued that they, too, could be good at protecting a customer’s data.
Some pointed out that fintechs had never experienced a breach of the magnitude of the 2014 JPMC breach,
in which thieves stole data on millions of customers.14
Clearly, open banking presented JPMC with big questions. What should its approach be to it? If consumers
wanted open banking and other banks went along, it might be impossible to resist. And yet the path to open
banking seemed fraught with risk. Could liability for fintech misbehaviour or mistake rebound on JPMC?
Or might open banking represent new and significant opportunities for banks? In either case, how should
an established bank like JPMC proceed?
THE RETAIL BANKING INDUSTRY
Retail banking (also known as consumer banking or personal banking) provided financial services to the
general public, as opposed to companies or organizations. These financial services included, but were not
limited to, chequing/savings accounts, mortgages, debit/credit cards, loans, automated teller machines, and
retirement accounts. Banks cross-sold other financial products to their retail banking customers, providing
services such as wealth management and investment advice.15
According to the Federal Deposit Insurance Corporation, banking generated US$233 billion16 in net income
during 2019, the industry’s second most profitable year ever.17 The industry was highly consolidated,
dominated by a few key players. It was difficult for new firms to enter the market; high fixed costs and
stringent regulations presented high barriers to entry.18 Compliance with regulations and banking legalities
weighed disproportionately on smaller firms, which had to dedicate proportionally greater resources
towards compliance relative to large firms.19 Switching banks entailed high switching costs, so consumers
rarely did it.20 These factors combined meant that banks had become used to not having to compete fiercely
for the bulk of their customers.
Retail banks generated income through loans, a system known as fractional reserve banking. Banks stored
cash deposits from clients and used the deposits to make loans. The Federal Reserve required a certain
percentage of deposits to be kept on hand; such “reserve requirements” were a safety measure to ensure
that requests for withdrawals would not exceed deposits at any point in time. Banks charged interest on
Page 92 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
and could not be used to identify individuals, researchers from Massachusetts Institute of Technology showed
“they could unmask roughly 90 per cent of people in a data[set].”11 Similar concerns persisted and had become
heightened in 2020. “Consumers don’t actually read privacy policies,” argued Christina Tetreault of Consumer
Reports, Inc. She cited the example of a policy at a financial institution that conveyed to the company the right
to access health information from consumers’ mobile phone apps. Most consumers okayed such policies,
Tetreault suggested, without reading the fine print. As a result, third parties too often achieved access to
consumer data that Tetreault characterized as “way outside reasonable expectations.”12
Page 3
9B21M060
In 2020, after years of low interest rates, business conditions had turned against banks. Annual revenue
growth had slowed.22 Banks looked to cost-cutting and innovation to maintain profit margins. One potential
avenue to both was online bankingproviding services through the Internet and mobile apps, which
reduced the need for customers to travel to physical bank branches, and that suggested possibilities for new
consumer services. While this avenue promised increased convenience and satisfaction for customers, it
also placed banks in closer competition with nimbler fintech companies, who aspired to provide the same
and additional services with more ease and often at lower prices.23
Banks cross-sold investment products, asset/wealth management services, and payment solutions to generate
additional revenues, but many of these products were also prone to disruption and margin compression. With
the rise of low-cost passive investments like exchange-traded funds, banks were forced to reduce fees for
managing their investment funds. Fintech companies also pushed into these spaces. Robo-advisors, which
provided automated, sometimes artificial intelligence (AI)-powered advice, eliminated the need for human
intervention and investment advice that banks (and others) had historically provided. Common in many
fintech offerings were optimized, passive strategies that were inexpensive and had low balance requirements
in comparison with banking services, and that, consequently, meant fintechs could offer prices for equivalent
financial services at discounts as great as 70 per cent.24 Against this backdrop of growing competition for
banks, the rise of open banking seemed likely to make the competition even more intense.
JPMORGAN CHASE & CO.
JPMC was an American financial services company offering a multitude of services in retail banking,
investment banking, asset management, and wealth management. It was ranked by S&P Global as the
largest bank in the United States in 2019 and the 6th largest bank in the world by total assets.25
Like other banks in the current unfavourable business conditions, JPMC turned to technology and
innovation to maintain growth. The company had spent billions on technology in recent years.26 It had more
than 40,000 technologists working in the bank, human capital that was dedicated to programming, system
engineering, app design, and other such functions. JPMC also possessed 31 data centres, 67,000 physical
servers, 27,920 databases, and a global network dedicated to serving clients.27
JPMC had a dedicated digital group “focused on product and platform design and innovation.”28 The group
worked on various initiatives, one of which was the consumer digital initiative. Members of this group
worked on adding new functionalities to the company’s mobile app, and redesigned the website for a more
personalized and simpler user experience. JPMC had over 20 million active Chase Mobile customers, and
had experienced a general shift of its customers toward online and mobile banking.29
THE EMERGING FINANCIAL TECHNOLOGY COMPANIES
Investments in new fintechs had been growing considerably since the 2008 financial crisis. The disaffection
and reduced consumer confidence in large financial institutions that resulted from the 2008 crisis opened
doors for new entrants who aimed to provide financial services at lower price points and with greater
convenience. In 2015, the value of global fintech investments spiked by 75 per cent to $22.3 billion,30 and
Page 93 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
loans at a higher rate than the rate they paid to customers on their deposits. The differential between the
two interest rates was how a bank generated income.21 Thus, profitability in retail banks was highly subject
to economic conditions; lower interest rates compressed margins.
Page 4
9B21M060
The services that fintech companies provided were varied and often served specific niches as opposed to
offering a full suite of solutions. Acorn, for instance, was software that rounded up a client’s purchase to
the nearest dollar, and then took that extra change and put it into an investment account.33 Services such as
this hardly threatened to replace banking entirely. But they did interfere with banks’ efforts to cross-sell
other services, such as in-house financial advisors. And the collective efforts of companies, each targeting
individual niches, could, of course, threaten existing or potential bank revenue sources. The nature of the
threat to banks, then, was not that they would be replaced completely, but that the influx of smaller, more
nimble firms might hive off, bit by bit, a great deal of the revenue that banks wanted for themselves.
Uptake of fintech services was growing. Fintech usage was most prevalent among young, high-income
customers. A 2015 EY Global Financial Services Institute report on fintech adoption stated that “one in
every four respondents aged 25 to 34 has used at least two fintech products in the last six months.”34
Adoption had a clear correlation with income, growing steadily as income moved higher and reaching “44
per cent for those with incomes above $150,000.”35 These were high-value customers that incumbent banks
were increasingly losing to these new entrants (see Exhibits 1, 2, 3, and 4 for data on fintech adoption).
OPEN BANKINGTHE IDEA AND THE REALITY
Open banking reflected the idea that consumers would benefit if, at their discretion, they could grant thirdparty service providers access to their banking data. Consumers sharing their data at will with the financial
services providers of their choice would increase competition in the financial services industries, which
would in turn stimulate innovation and lower prices. It could also allow the creation of new value for
consumers arising from both the integration of services and data aggregation from different sources.36
That banks should be legally obliged to turn over their customers’ data to third parties when consumers
authorized it seemed like a reasonable expectation to many consumers and policy makers. The data did,
after all, belong to the consumer. For the banks, though, this meant they had to turn over data that would
help others compete against them. Needless to say, this idea was challenging for many banks. In addition,
the matter was not as simple as it seemed. Intermingled with consumer data was often other information
that the banks considered proprietary, such as fee levels and the pricing of transactions or accounts.37
There could be little doubt that open banking represented a paradigm shift in the financial ecosystem. Prior
to open banking, financial data was siloed, kept in a walled garden that belonged to incumbent banks. This
privileged position shielded banks from disruptive start-ups, as the financial services of these start-ups could
not be rendered without access to the clients’ accounts. However, as the push for open banking continued,
it became an undeniable reality for banks to reckon with.
Outside of the United States, open banking had been enacted formally in Europe, the United Kingdom,
Australia, and elsewhere through regulatory regimes, with the stated intention of increasing competition in
financial services in ways that would benefit consumers.38 For example, in August 2016, the United
Kingdom’s Competition and Markets Authority issued a broad ruling that forced the United Kingdom’s
nine largest banks to share their data in a standardized form.39 In Europe and elsewhere, this consumer
information issue was framed in terms of “data portability,” the idea that consumers should be able to take
their data anywhere they chose, like it was personal property.40
Page 94 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
by 2018, annual investments had reached $31 billion.31 From 2010 to 2015, more than $50 billion had been
invested in almost 2,500 companies.32
9B21M060
In the United States, policy makers were less inclined to influence markets through regulation and tended
to rely instead on competition to enact changes to benefit consumers. The move to open banking in the
United States, then, was more ad hoc than in other countries.41 Even so, open banking had many advocates
pushing for its implementation, and banking regulations trended toward open information. The 2010 DoddFrank Wall Street Reform and Consumer Protection Act had a section on consumers’ right to information,
which stated that consumer transaction data should be made available in an “electronic form usable by
consumers” upon request.42 Remarks by Richard Cordray, director of the CFPB, in a 2016 CFPB hearing
illustrated the increasing degree of enthusiasm for the open banking ideal in the United States:
Impeding access to digital financial records not only blocks innovation from new entrants, it also
reduces the incentives for financial institutions to innovate. Without new companies introducing
consumer-friendly products or services into the market, established companies are likely to feel
less pressure to compete in this manner. And authorizing access to their financial records can make
it easier for consumers to shop for an alternative provider with more favorable pricing, given the
consumer’s usage patterns. To be clear, it is unacceptable for financial institutions to block access
to consumer information as a means of gaining a competitive advantage in the marketplace.43
The CFPB also announced inquiries into how financial data was shared and what technological
developments could be used to advance it going forward.44
THE PRACTICALITIES OF DATA ACCESS
The sharing of data for open banking was accomplished in two main ways—one that was in many ways
problematic but did not require the co-operation of an established bank, and another that was more reliable
and secure but did require the bank’s co-operation.
The first way, known as “credentialed access,” was more widely and colloquially called “screen scraping.”
In screen scraping, consumers shared usernames and passwords with third-party service providers who used
these credentials for automated logins as if they were the consumer. Once the third parties had access, their
automated routines read the HTML used on the bank website to reverse out sought-after consumer data,
such as account balancesliterally “scraping” data from the screens where the data was displayed.45
Screen scraping presented several potential problems. Lila Fakhraie, senior vice-president of Digital
Banking at Wells Fargo, described one of these with the following analogy: “It’s like giving your house
key to a painter and saying ‘Just paint that one wall. That’s all I want’ . . . and now the house painter has
your key forever. They come and go as they please and take things if they want.” Also, the sharing of data
via screen scraping was all-or-nothing: once in, the third party had access to any information that was there
and could do with it what they wanted. There was no way to provide only limited access to certain
information.46 The process was also error-prone. Changes in the layout of the bank’s web page could cause
screen scraping to pick up the wrong information.47
The sole advantage of screen scraping was that it did not require permission from the banks, nor did it
require a third-party firm to reach a formal agreement with a bank. Not surprisingly, banks disapproved of
the practice, largely for this exact reason. They worried, too, that the risks to consumers entailed in screen
scraping might rebound on them.48 Some banks took specific steps to thwart screen scraping, for example
by slightly altering website screen layouts on a regular basis to cause third parties to pick up wrong
numbers.49 When screen scraping routines failed and were then tried again repeatedly, account lockouts
could be inadvertently triggered, requiring users to unlock their bank accounts and replace and re-input
Page 95 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Page 5
Page 6
9B21M060
banking passwords. Such breakdowns generated hassles for customers and prevented third-party firms from
delivering their services.50
The only problem with APIs, from the perspective of fintechs, was that they required an agreement with
and co-operation from banks. In negotiations to reach agreements with smaller fintechs, established banks
had a huge leverage advantage. In effect, using APIs ceded control from fintechs to banks. In addition,
getting APIs designed, built, and installed took time. Banks’ timelines for developing APIs were often
slower than what was ideal for start-ups.
JPMC OPEN BANKING ACTIVITIES
JPMC was on record expressing many reservations regarding the sharing of financial data with third-party
service providers. In his 2015 shareholder letter, Dimon asserted that “far more information is taken than
the third party needs in order to do its job”51 and that “many third parties sell or trade information in a way
customers may not understand, and the third parties, quite often, are doing it for their own economic
benefitnot for the customer’s benefit.”52 The letter condemned the sharing of bank login credentials to
enable screen scraping, insisting that customersnot the bankwould be responsible if money was stolen
from consumer accounts as a result of this kind of activity.
Fintechs vigorously rebutted these charges. Mint, Acorn, Plaid, Penny (a personal finance app), and others
insisted that they did not take any more data than they needed, and that they took good care of the data
consumers authorized them to access. Critics of JPMC argued that the company was using data concerns
as a cover, and that the bank was more concerned with how fintechs might threaten the bank’s bottom line
than with customer data security and privacy.53 A 2016 article in the New York Times appeared to support
this argument when it quoted an unnamed JPMC official expressing as an ideal that “the bank will invent
services that are good enough to keep every customer inside its firewalls.”54
Since 2017, JPMC had pursued a strategy of resisting uncontrolled third-party access (e.g., via screen
scraping), while also inviting third parties to establish individual, formal agreements with the bank.55 For
example, in the aftermath of the 2015 dispute with Mint, JPMC undertook a deal in 2017 with Intuit Inc.,
Mint’s parent company.56 In early 2020, JPMC told fintechs that they would have to abide by the bank’s
access rules and APIs or be blocked.57 This helped the bank reach agreements with fintechs that covered 95
per cent of the customer data traffic coming into the bank.58 Contracts that JPMC had executed on an
individual basis with third parties included the following:





A stipulation that secure access through APIs would be through the use of secure tokens rather than
customer login credentials
A clear designation of what information the third party could access
A clear assignment of responsibility to third parties for any risks they introduced into the process
A requirement for insurance and indemnification against losses by third parties
A requirement that other companies the third party worked with would have to abide by JPMC security
standards59
Page 96 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
The second and superior approach to data sharing was via APIs, which were software programs that acted
as a bridge between bank and third-party systems, providing a direct connection to banking data. APIs could
be designed to allow a third party access to only specific information, that is, what a customer had agreed
to and nothing else (see Exhibit 5).
Page 7
9B21M060
Although these agreements seemed to have brought some order to the banking and fintech ecosystem, the
fintechs were not entirely happy with such arrangements. Plaid’s Pitts noted that if every major bank did
its own bilateral deals with fintechs, it created the potential for more than 10,000 sets of individual
agreements. Moreover, each agreement could vary in its specific terms.61
The Clearing House Payments Company L.L.C., a New York-based standards setting group, had published
a model agreement institutions could use that was consistent with CFPB guidelines on data access, to reduce
the contractual complexity that would arise from thousands of bilateral agreements. Adoption of the
agreement, however, was voluntary.62
A non-profit group called Financial Data Exchange, which included representatives of banks and fintechs,
had proposed and was developing a set of standard APIs. The idea was to replace the huge variety of APIs
that corresponded to bilateral agreements with a much smaller set of APIs used for data access and sharing,
industry-wide. In early 2020, FMR LLC, the parent company of Fidelity Investments, created the spinoff
Akoya, a company jointly owned by Fidelity Investments, The Clearing House Payments Company L.L.C.,
and 11 member banks, with a mission to provide just such a set of standardized APIs.63
Use of the resulting standard APIs and agreements, however, remained voluntary. The banks generally
opposed regulatory influence that might move the industry toward mandatory use of standardized APIs.
There was one exception to banks’ resistance to regulatory requirements within the banking and fintech
ecosystem: the banks expressed great enthusiasm for the regulation of fintech companiesfor pulling them
under the CFPB’s oversight.64
THE RISE OF FINANCIAL DATA AGGREGATORS
Financial data aggregators were a particular type of fintech player gaining greater influence in the open
banking arena. Companies such as Finicity, Fiserv Inc., Mint, MX Technologies Inc., Plaid, Yodlee, and
YNAB helped solve the problem many fintechs and other third-party companies faced in maintaining access
to banking information across multiple financial institutions (including investment firms and other nonbanks). Aggregators positioned themselves between the institutions that contained consumer information
and the third parties who wanted one-stop access to it. They simplified access to consumer data for third
parties by providing a single portal via which they could access consumer data, regardless of its source.65
Initially, aggregators had compiled their information via screen scraping. But as they acquired more and
more trafficand with that, influence and respectthey were able to conduct deals with many banks to
get more direct access, via APIs.66 In 2020, it was increasingly clear that aggregators were here to stay,
destined to be a factor in open banking competition.67
Page 97 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
The JPMC approach had allowed the bank to create a service for consumers called AccountSafe, which
provided a dashboard to show customers which third parties had access to which data via APIs, as well as
allowing customers to adjust that access if they wished.60
Page 8
9B21M060
WHAT TO DO
Giant tech platforms, such as Apple Inc., Amazon.com Inc., and Google, were venturing increasingly into
banking-related activities. Apple Inc., for instance, had teamed up with Goldman Sachs to launch a lowinterest-rate credit card (“the most successful credit card launch in history,” according to Goldman Sachs’
CEO).68 As of the middle of 2019, Amazon.com Inc. had loaned more than $3 billion to merchants,69 and
its web services infrastructure powered the back offices of celebrated challenger banks and provided
cutting-edge solutions in areas such as capital markets and insurance. Google had announced that it would
launch digital bank accounts in 2021.70 An extreme possible endpoint of such developments could be seen
in the example of China’s WeChat; WeChat had become the financial services provider preferred by many
Chinese consumers mainly because of its convenient integration with the many aspects of their daily lives
though a single app (the WeChat app).71
According to research by Professor Pinar Ozcan of Oxford University’s Saïd Business School, banks had
fumbled the mobile payments business to tech platforms (Apple Pay, Amazon Pay, Google Pay) by
engaging in “turf wars” with the ecosystem partners they needed to co-operate with to deliver mobile
payment services, such as telecommunications companies and mobile handset makers. Everyone wanted to
“own” the customer, everyone wanted to be in charge, and no one could agree on security standards. To
prevent something similar from happening with open banking, Ozcan suggested that banks had to get over
some of their more controlling reflexes.72 Might banks be able to counter tech company incursions into their
markets by engaging in enlightened partnership? Might partnering with third parties to build open banking
ecosystems be a route that banks could take to achieve a foothold in the platform economy?
Co-operating with fintechs building on top of the traditional banking framework could potentially create a
much more valuable ecosystem for customers. But it might also threaten the bank financially, at least in the
short term, cutting into its market share and financial returns. And there were (always) the matters of
security and privacy to consider. Could both objectives be realizedhigher value for customers and the
preservation of JPMC’s business success?
Page 98 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Open banking clearly represented an opportunity to provide more personalized customer experiences. But
it also required a break from banking’s traditionally vertically integrated model, which was inflexible in
rapidly addressing consumer needs. What stance should a bank like JPMC take?
Page 9
9B21M060
Source: Adapted from EY [Ernst & Young], Global Fintech Adoption Index 2019, 2019, accessed September 23, 2020,
https://assets.ey.com/content/dam/ey-sites/ey-com/en_gl/topics/banking-and-capital-markets/ey-global-fintech-adoption-index.pdf.
Page 99 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
EXHIBIT 1: SELECT REASONS FOR USING INCUMBENT FINANCIAL INSTITUTIONS, 20172019
Page 10
9B21M060
Note: The bars in this chart show the percentage of respondents who either ‘agree’ or ‘strongly agree’ that they would be
comfortable if their main bank securely shared their financial data with other organizations.
Source: EY [Ernst & Young], Global Fintech Adoption Index 2019, 2019, accessed September 23, 2020,
https://assets.ey.com/content/dam/ey-sites/ey-com/en_gl/topics/banking-and-capital-markets/ey-global-fintech-adoption-index.pdf.
Page 100 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
EXHIBIT 2: PREFERENCES FOR SHARING BANKING DATA, BY FINTECH ADOPTERS AND NONADOPTERS
Page 11
9B21M060
EXHIBIT 3: COMPARISON OF FINTECH CATEGORIES, RANKED BY ADOPTION RATE FROM 2015
TO 2019
1. Money Transfers/
Payments
2. Savings &
Investments
3.
Budgeting/Financial
Planning
4. Insurance
5. Borrowing
Adoption
Rate
18%
2017
Category
Adoption
Rate
50%
2019
Category
17%
1. Money
Transfers/Payments
2. Insurance
8%
3. Savings & Investments
20%
3. Savings &
Investments
34%
8%
4. Budgeting/Financial
Planning
5. Borrowing
10%
4. Budgeting/Financial
Planning
5. Borrowing
29%
6%
24%
10%
1. Money
Transfers/Payments
2. Insurance
Adoption
Rate
75%
48%
27%
Source: EY [Ernst & Young], Global Fintech Adoption Index 2019, 2019, accessed September 23, 2020,
https://assets.ey.com/content/dam/ey-sites/ey-com/en_gl/topics/banking-and-capital-markets/ey-global-fintech-adoption-index.pdf.
EXHIBIT 4: COMPARISON OF FINTECH ADOPTION IN SIX MARKETS, 20152019
Note: These figures show adoption rates per market for the six markets for which a comparison is available.
Source: EY [Ernst & Young], Global Fintech Adoption Index 2019, 2019, accessed September 23, 2020,
https://assets.ey.com/content/dam/ey-sites/ey-com/en_gl/topics/banking-and-capital-markets/ey-global-fintech-adoption-index.pdf.
Page 101 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
2015
Category
Page 12
9B21M060
Note: App = application; API = application programming interface.
Source: Created by the case author using icons by Gregor Cresnar, under Creative Commons licence CCBY, The Noun
Project, accessed May 6, 2021, https://thenounproject.com/grega.cresnar.
Page 102 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
EXHIBIT 5: THE ROLE OF APIs IN FINANCIAL ECOSYSTEMS
Page 13
9B21M060
ENDNOTES
This case has been written on the basis of published sources only. Consequently, the interpretation and perspectives
presented in this case are not necessarily those of JPMorgan Chase & Co. or any of its employees.
2
Norbert Gehrke, “Open Banking Case Study: J.P. Morgan Treasury Services,” Medium, April 21, 2019, accessed September
8, 2020, https://medium.com/tokyo-Fintech/open-banking-case-study-j-p-morgan-treasury-services-a971c7507285.
3
See “Treasury Services: The Open Banking Transformation,” J.P.Morgan, accessed September 8, 2020,
www.jpmorgan.com/global/treasury-services/open-banking.
4
Steve Cocheo, “Fight over Consumer Data Ownership Pits Banks against Fintechs,” The Financial Brand, March 4, 2020,
accessed September 9, 2020, https://thefinancialbrand.com/93646/financial-budget-management-app-data-sharing-Fintechdigital-cfpb/.
5
Ibid.
6
Ibid.
7
Robin Sidel, “Big Banks Lock Horns with Personal-Finance Web Portals,” Wall Street Journal, November 5, 2015, accessed August
23, 2020, www.wsj.com/articles/big-banks-lock-horns-with-personal-finance-web-portals-1446683450?mod=article_inline.
8
Cocheo, op. cit.
9
“Dumb pipes” referred to when fintechs built leading-edge apps and services on the framework of traditional institutions; Ibid.
10
Bradley Hope, “Provider of Personal Finance Tools Tracks Bank Cards, Sells Data to Investors,” Wall Street Journal, August
7, 2015, accessed August 23, 2020, www.wsj.com/articles/provider-of-personal-finance-tools-tracks-bank-cards-sells-data-toinvestors-1438914620.
11
Ibid.
12
Cocheo, op. cit.
13
Ibid.
14
Plaid Inc., Building a Consumer-first Framework for Modern Technologies, Policy Paper, March 1, 2016, accessed August
23, 2020, https://plaid.com/documents/Plaid-Policy-Paper.pdf.
15
Christina Majaski, “Retail Banking,” Investopedia, updated March 30, 2021, accessed September 9, 2020,
www.investopedia.com/terms/r/retailbanking.asp.
16
All currency amounts are in US$ unless otherwise specified.
17
Ken Sweet, “Banks Made $233.1 Billion in Profits in 2019, Regulator Says,” ABC News, February 25, 2020, accessed
September 9, 2020, https://abcnews.go.com/Business/wireStory/banks-made-2331-billion-profits-2019-regulator-69205274.
18
Investopedia, “What Barriers to Entry Exist in the Financial Services Sector?,” Investopedia, June 25, 2019, accessed
August 23, 2020, www.investopedia.com/ask/answers/031015/what-barriers-entry-exist-financial-services-sector.asp.
19
Burak Dolar and Ben Dale, “The Dodd–Frank Act’s Non-Uniform Regulatory Impact on the Banking Industry,” Journal of
Banking Regulation 21, no. 2 (June 25, 2019): 188195.
20
Claire Matthews, “Switching Costs in Banking: The Regulatory Response,” November 2009, Department of Economics and
Finance, Massey University, Palmerston North, New Zealand.
21
Julia Kagan, “Fractional Reserve Banking,” Investopedia, May 4, 2020, accessed August 23, 2020,
www.investopedia.com/terms/f/fractionalreservebanking.asp.
22
McKinsey & Company, Retail Banking Insights: Driving Revenue Growth in Retail Banking, Number 6, March 2015,
accessed August 23, 2020, www.mckinsey.com/~/media/McKinsey/Industries/Financial Services/Our Insights/Driving
revenue growth in retail banking/Driving-revenue-growth-in-retail-banking.pdf.
23
David Berman, “Banks Could Lose 60% of Retail Profit to Tech Startups: Study,” Globe and Mail, September 29, 2015,
accessed August 23, 2020, www.theglobeandmail.com/report-on-business/Fintech-startups-pose-threat-to-traditional-banksretail-profit/article26587892/.
24
Charles Ludden, Kendra Thompson, and Imon Mohsin, “The Rise of Robo-Advice,” Accenture, 2015, accessed August 23,
2020,.
25
Saqib Chaudhry and Francis Garrido, “The World's 100 Largest Banks,” S&P Global Market Intelligence, April 5, 2019,
accessed August 23, 2020, www.spglobal.com/marketintelligence/en/news-insights/trending/t-38wta5twjgrrqccf4_ca2.
26
JPMorgan Chase & Co., Annual Report 2015, April 6, 2016, accessed August 23, 2020,
www.jpmorganchase.com/content/dam/jpmc/jpmorgan-chase-and-co/investor-relations/documents/2015-annualreport.pdf.
27
Ibid.
28
Ibid.
29
Ibid.
30
Julian Skan, James Dickerson, and Luca Gagliardi, “Fintech and the Evolving Landscape,” Accenture, October 11, 2016,
accessed August 23, 2020, www.accenture.com/t20161011T031409Z__w__/pl-en/_acnmedia/PDF-15/Accenture-FintechEvolving-Landscape.pdf.
31
Alex Rolf, “Fintech Investment Reaches Record Levels at $31 Billion in 2018,” Payments, Cards and Mobile, February 7,
2019, accessed September 9, 2020, www.paymentscardsandmobile.com/fintech-investment-reaches-record-levels/.
32
Skan, Dickerson, and Gagliardi, op. cit.
33
“Acorns Helps You Grow Your Money,” Acorn, accessed April 14, 2021, www.acorns.com.
34
Imran Gulamhuseinwala, Thomas Bull, and Steven Lewis, EY Global Financial Services Institute, “Fintech Is Gaining
Traction and Young, High-income Users Are the Early Adopters,” Journal of Financial Perspectives 3, no. 3 (Winter 2015),
accessed April 30, 2021, https://papers.ssrn.com/sol3/Data_Integrity_Notice.cfm?abid=3083976.
35
Ibid.
Page 103 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
1
Page 14
9B21M060
Investopedia Staff, “Open Banking,” Investopedia, updated August 27, 2020, accessed September 9, 2020,
www.investopedia.com/terms/o/open-banking.asp.
37
Cocheo, op. cit.
38
Ibid.
39
“CMA Paves the Way for Open Banking Revolution,” GOV.UK, press release, August 9, 2016, accessed August 23, 2020,
www.gov.uk/government/news/cma-paves-the-way-for-open-banking-revolution.
40
Cocheo, op. cit.
41
Ibid.
42
Richard Cordray, “Prepared Remarks of CFPB Director Richard Cordray at the Field Hearing on Consumer Access to
Financial Records,” Consumer Financial Protection Bureau, November 17, 2016, accessed August 23, 2020,
www.consumerfinance.gov/about-us/newsroom/prepared-remarks-cfpb-director-richard-cordray-field-hearing-consumeraccess-financial-records/.
43
Ibid.
44
Ibid.
45
Cocheo, op. cit.
46
Ibid.
47
Joel Schlesinger, “Open Banking Could Be ‘Very Positive’ for Financial Advice Business,” Globe and Mail, September 20,
2019, accessed August 23, 2020, www.theglobeandmail.com/investing/globe-advisor/advisor-news/article-open-bankingcould-be-very-positive-for-financial-advice-business/.
48
Cocheo, op. cit.
49
Ibid.
50
Ibid.
51
Jamie Dimon, “Dear Fellow Shareholders,” JPMorgan Chase & Co., April 6, 2016, accessed August 23, 2020,
www.jpmorganchase.com/corporate/investor-relations/document/ar2015-ceolettershareholders.pdf.
52
Ibid.
53
Kristin Wong, “Consumers Left in Dark When Banks Cut off Online Finance Services,” CNBC, November 11, 2015, accessed
August 23, 2020, www.cnbc.com/2015/11/11/consumers-left-in-dark-when-banks-cut-off-online-finance-services.html.
54
Ron Lieber, “Jamie Dimon Wants to Protect You from Innovative Start-Ups,” New York Times, May 6, 2016, accessed
August 23, 2020, www.nytimes.com/2016/05/07/your-money/jamie-dimon-wants-to-protect-you-from-innovative-startups.html.
55
Cocheo, op. cit.
56
Ibid.
57
FinTech Futures, “JP Morgan vs Fintechs: Who Really Owns Your Data?,” FinTech Futures: North American Edition, April
1, 2020, accessed March 15, 2021, www.fintechfutures.com/2020/04/jp-morgan-vs-fintechs-who-really-owns-your-data/.
58
Ibid.
59
Ibid.
60
“JPMorgan Chase, Envestnet l Yodlee Sign Agreement to Increase Customers’ Control of Their Data,” Chase  Media
Center, December 5, 2019, accessed April 14, 2021, https://media.chase.com/news/jpmorgan-chase-envestnet-yodlee-signagreement-to-increase-customerss-control-of-their-data.
61
Ibid.
62
Ibid.
63
Alex Hamilton, “Data Firm Akoya Spun Out as Independent Company by FMR,” FinTech Futures: North American Edition,
February 25, 2020, accessed April 14, 2021, www.fintechfutures.com/2020/02/data-firm-akoya-spun-out-as-independentcompany-by-fmr/.
64
John F. Wasik, “Should Fintechs Be Regulated like Banks?,” BAI, July 24, 2019, accessed April 14, 2021,
www.bai.org/banking-strategies/article-detail/should-fintechs-be-regulated-like-banks/.
65
“List of Best Financial Data Aggregation Service Providers,” Forex News Now, accessed March 15, 2021,
www.forexnewsnow.com/providers/financial-data-aggregation-providers/.
66
Ibid.
67
Ibid.
68
Hugh Son, “Goldman Sachs CEO Says Apple Card is the Most Successful Credit Card Launch Ever,” CNBC, October 15,
2019, accessed December 16, 2020, www.cnbc.com/2019/10/15/goldman-sachs-ceo-says-apple-card-is-the-mostsuccessful-credit-card-launch-ever.html.
69
Ron Shevlin, “Amazon’s Impending Invasion of Banking,” Forbes, July 8, 2019, accessed December 16, 2020,
www.forbes.com/sites/ronshevlin/2019/07/08/amazon-invasion/.
70
Kyle Bradshaw, “Google Pay to Launch Digital Bank Accounts in 2021,” 9to5Google, August 3, 2020, accessed December
16, 2020, https://9to5google.com/2020/08/03/google-pay-digital-bank-accounts/.
71
“How China is Changing Your Internet | The New York Times,” August 9, 2016, YouTube video, 5:57,
https://youtu.be/VAesMQ6VtK8.
72
“Platform Disruption: The Case of Open Banking | Pinar Ozcan, University of Oxford,” January 2, 2020, YouTube video,
34:43, https://youtu.be/-7tBZ2RMPU8.
Page 104 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
36
9B19M110
Genevieve Pelow and Professor Robert D. Austin wrote this case solely to provide material for class discussion. The authors do not
intend to illustrate either effective or ineffective handling of a managerial situation. The authors may have disguised certain names
and other identifying information to protect confidentiality.
This publication may not be transmitted, photocopied, digitized, or otherwise reproduced in any form or by any means without the
permission of the copyright holder. Reproduction of this material is not covered under authorization by any reproduction rights
organization. To order copies or request permission to reproduce materials, contact Ivey Publishing, Ivey Business School, Western
University, London, Ontario, Canada, N6G 0N1; (t) 519.661.3208; (e) cases@ivey.ca; www.iveycases.com. Our goal is to publish
materials of the highest quality; submit any errata to publishcases@ivey.ca. i1v2e5y5pubs
Copyright © 2019, Ivey Business School Foundation
Version: 2019-09-13
On October 1, 2018, the General Electric (GE) board of directors installed H. Lawrence Culp as chief executive
officer (CEO), replacing John Flannery, who had been in that position for only 14 months.2 Culp assumed
command of a company in crisis. GE’s stock had declined to nearly US$103 per share from more than $30 per
share two years earlier, wiping out billions of dollars in shareholder value (see Exhibit 1). The quarterly dividend
had been halved in November 2017—only its second reduction since the Great Depression4—and, less than 30
days later, was on its way to another reduction, by 92 per cent to a mere one cent.5 Worries about the company’s
costs structure and cash-generating ability had resulted in a credit downgrade6 (see Exhibit 2 for GE’s financial
statements), despite Flannery having already trimmed the workforce by 30,000.7 The previous 10 months had
seen a procession of high-profile departures, including the chief financial officer, who was the head of the largest
business unit, and half of the board of directors (see Exhibit 3).8
A few months earlier, in late July, reports had emerged that GE intended to sell GE Digital,9 its software and
technology organization, which GE had built up from nothing since 2011. GE Digital had been described as
the central component of a much-heralded transformation of the company’s business, architected by former
CEO Jeffrey Immelt.10 In late February 2019, GE announced that it would streamline operations and focus on
its core competencies—which no longer included GE Digital.11 GE would, ultimately, in 2019, spin off GE
Digital as an independent company that would sell its services to other companies and be expected to survive
on its own revenues (about $1.2 billion at that time), without further funds from GE.12
Although the decision to spin off rather than sell GE Digital was interpreted as “a measured vote of
confidence in the software business,”13 there could be little doubt that GE’s priorities had shifted. In a
startling turnabout, Immelt’s digital transformation vision for the company had been set aside. A mere two
years earlier, many had touted GE as an example of how established companies could pre-emptively
transform their businesses digitally, without waiting to be forced by competition.14 By 2019, however, GE’s
digitalization initiative looked more and more like a warning to others about the perils of being distracted
by digital hype or attempting to change too fast.
A VERY BRIEF HISTORY OF GENERAL ELECTRIC
Thomas Edison established the Edison General Electric Company in 1890. Two years later, it merged with
a competitor, the Thomson-Houston Company, to form General Electric Company.15 From its beginning,
Page 105 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
DIGITAL TRANSFORMATION AT GE: WHAT WENT WRONG?1
Page 2
9B19M110
The company’s culture placed a high value on learning and professional development. In 1956, GE built an
educational campus at Crotonville, New York, in an effort to become “the best-managed company” in the
world.18 In 2017, GE spent more than $1 billion annually on employee training.19 And indeed, for much of
its life, GE enjoyed a reputation as a “superbly managed company.”20
But after 2000, the rapid advance of technology, free trade, and changes in customer behaviour, combined
with changing global power dynamics and other macroeconomic influences such as fluctuating oil prices,
put pressure on industrial companies to become more competitive and differentiate themselves.21
Globalization threatened to commoditize many GE product businesses.22 Annual revenues peaked in 2008
at more than $180 billion, but in 2017, after significant divestitures, GE revenues remained at more than
$120 billion. In 2018, GE had an industrial base of more than $240 billion in service contracts, and market
presence in 130 countries23 organized in eight major business units (see Exhibit 4).24
THE JACK WELCH ERA
Jack Welch served as GE’s CEO from 1981 to 2001. Under his leadership, the company’s revenues grew by
more than 400 per cent, from $25 billion to $130 billion;25 he completed 600 acquisitions, including NBC
Universal, and proposed a $45 billion acquisition of Honeywell International, which ultimately fell through.26
Welch required GE business units to be number one or number two in their respective industries.27 To help
them achieve this goal, he drove streamlining to remove bureaucracy; launched progressive operational
programs, such as Six Sigma;28 and emphasized running the company as if it were a smaller, more dynamic
business.29 His management of the company was widely admired; in 1999, as the century and Welch’s tenure
as CEO neared their ends, Forbes magazine proclaimed him the “Manager of the Century.”30
Welch also had critics.31 Some blamed his employee ranking system, which required dismissing lowerranking employees—referred to colloquially as “rank and yank”32—for creating an overly competitive work
environment that resulted in less collaboration.33 He was criticized, too, for his relentless “beat the quarter”
approach to earnings management.34 Some have argued that the short-term orientation and complex systems
for measuring and comparing internal business unit performance established during the Welch era
contributed to the problems GE was suffering in 2019.35
IMMELT TAKES THE REINS
Jeffery Immelt, an 18-year GE veteran, followed Welch as CEO. Immelt had held leadership positions in
GE’s Plastics, Appliances, and Healthcare businesses, and came to the top job highly regarded. He would
serve as chairman and CEO from 2001 to 2017, leaving his own deep imprint on the company.36
Immelt’s detailed knowledge of GE’s major businesses helped him execute five key strategies that defined
his tenure as CEO. First, Immelt refocused the GE portfolio, divesting non-industrial and slow-growth
businesses, and renewing a focus on high-tech and manufacturing-based products and services. Second, he
made big investments in technology-driven innovation by increasing spending on research and
development, and by concentrating on clean and energy-efficient products, the industrial Internet, and
additive manufacturing. Third, he expanded the company’s presence globally, moving into developed and
Page 106 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
GE prioritized research and development. Over the next 128 years, GE matured into one of the world’s
largest conglomerates, and accumulated more than 150,000 patents for its many consumer and industrial
products16 (e.g., the incandescent light bulb; the X-ray machine; the electric fan; the refrigerator; the jet
engine; and the magnetic resonance imaging, or MRI, machine).17
Page 3
9B19M110
Immelt’s digital vision for the company attracted much interest. He accurately foresaw a world in which
data would be more valuable than the industrial hardware that produced it, and companies that could deploy
technology to monetize data would exert influence over firms with primarily industrial capabilities.
Companies such as IBM, Microsoft, and SAP had already started to approach GE customers with offers to
leverage data generated by GE systems to realize huge efficiencies.39 Immelt believed, though, that GE
would be the best partner to help its customers make the existing systems more efficient, and that the
company’s huge installed base of equipment would provide a big advantage in harvesting value from
industrial data. To realize this outcome, Immelt envisioned transforming GE into something much more
like a tech company than an industrial conglomerate—a “top ten software company.”40
NEW DIGITAL BUSINESS MODELS
Industry pundits had coined the expression “Internet of Things” (IoT) to describe the idea of connecting a
vast assortment of previously unconnected devices—including coffee makers, refrigerators, lighting
systems, televisions, wearable devices (such as watches), and pretty much any product powered by
electricity—to the Internet (and potentially, therefore, to manufacturers, other companies, and each other).41
Connecting such devices to the Internet enabled an entirely new business activity.
An Internet-connected device could report back to its manufacturer how well it was functioning (to trigger
needed service or to suggest improvements for future products); how it was being used by customers (to help a
firm make products more user-friendly); and even what customers were saying about it (to improve marketing
or as a source for new product ideas).42 The functionality of a “smart product” could be updated across the
Internet, adding functionality without anyone needing to physically access the device. Even more important, the
data from connected devices could be used to identify customers’ behavioural patterns, and thus to suggest
targeted advertising and cross-selling opportunities that the consumer might find useful, thereby yielding higher
returns on advertising investments. Still more important, connected devices could become parts of ecosystems
that provided new services to customers; for example, a smart refrigerator that realized the milk was getting low
could automatically post a replenishment order to the shopping list on a consumer’s smart phone.
In 2018, many of the new service possibilities presented by the IoT also represented possible new business
models. New services could be offered to consumers, leading to the facilitation of new consumer
transactions; and data about device usage, location, and supply levels could be analyzed to create marketing
possibilities. The result would be new sources of revenues and profits for companies.
The Industrial Internet of Things (IIoT) extended a similar idea to industrial devices, such as the turbines,
locomotives, and jet engines that GE sold.43 Industrial devices contained embedded sensors that generated
vast amounts of data about how they were functioning and being used. That data could be processed with
advanced analytical techniques to help improve business performance.44
For example, data from sensors on locomotives could be used to optimize braking, enabling train companies
to realize huge reductions in fuel costs and emissions.45 Or, analytics could be used to successfully predict
when an electric turbine was about to fail, enabling it to be proactively serviced, avoiding an unplanned
Page 107 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
emerging markets, diversifying the revenues of what had been a primarily US-oriented company. Fourth,
he modernized GE’s organizational structure, replacing it with a more agile and decentralized structure, in
an effort to encourage entrepreneurial behaviour within the firm.37 Finally, Immelt aimed to shift GE’s
strategy and competitive advantage away from making and selling increasingly commoditizing hardware
toward new, higher-margin business models enabled by “smart, connected products.”38
Page 4
9B19M110
As Immelt and his fellow GE strategists pondered these opportunities, they realized that a key to seizing
them might be controlling the technological standards for information exchange in the IIoT ecosystem.
Their thinking harkened back to some earlier technology examples, such as the ecosystem surrounding the
IBM personal computer (PC) desktop. The IBM PC, released in the early 1980s, featured an operating
system (OS) built by Microsoft, which, at the time, was a small, little-known company. Microsoft charged
IBM a development fee, but allowed it to deploy the OS for free. Because deploying it was free, IBM had
no cost incentive to limit its deployment. Crucially, though, Microsoft retained the sole legal rights to
license the software for use by other companies. Because the operating system governed how software
would interact with the IBM PC, it became an industry standard that Microsoft controlled. Third-party
software developers who created software for the PC needed to use the standard owned by Microsoft, which
made Microsoft the software gatekeeper for the PC ecosystem. As the ecosystem around IBM PCs (and PC
clones) grew, Microsoft leveraged this control into dominance of the PC desktop through the Windows
operating system. Ownership of the technology standard for data exchange gave Microsoft a key strategic
position within the ecosystem. For years, Microsoft leveraged its position to relegate other competitors to
secondary positions and to become one of the most important players in the computer industry.48
Something similar could happen, Immelt knew, in the IIoT space. The source of competitive advantage for
industrial companies was shifting from hardware manufacturing to software and sensors embedded in
machines and analytics to optimize performance.49 As business school professor Karim R. Lakhani put it,
“the data and analysis become worth more than the installed equipment itself.”50 Professors Michael
Cusumano, Annabelle Gawer, and David Yoffie described the threat to GE in their book on platform
strategy this way: “If a third-party software firm developed a winner-take-all platform that captured the
analytics layer, GE could be forced to join the platform and cede a great deal of the value of its equipment,
including its maintenance services, to the owner of the platform.”51 In other words, a tech company could
achieve a dominant position in the IIoT ecosystem, similar to the position Microsoft held in the desktop PC
ecosystem. Tech companies definitely saw the opportunity presented by the Industrial Internet and were
manoeuvring to secure a first-mover advantage (see Exhibit 5 for an outline of key competitors in the
Industrial Internet). Immelt believed GE could occupy the dominant position by leveraging its scale,
industrial expertise, and customer relationships.52
BUILDING AN INDUSTRIAL INTERNET PLATFORM
GE set out to create a platform called Predix that could go head to head with established digital platform
players, such as Amazon, Google, IBM, Microsoft, and SAP. Applications of Predix would include
preventative maintenance, process variance reduction, control system optimization, and manufacturing
productivity in a variety of industries, including aviation, pharmaceuticals, power, mining, manufacturing,
and oil and gas, to list but a few.53 GE would begin by rolling out the platform internally, which would
allow the company to benefit quickly from internal network effects; and the large amount of data GE could
collect from its own hardware would allow the company to rapidly improve its platform.54 As GE began to
extend the platform externally, trying to convince customers and partners to use Predix data standards for
coordinating their industrial activities, the company attempted to leverage a century’s worth of experience
designing, manufacturing, and servicing hardware. GE argued, to anyone who would listen, that its superior
industry experience made Predix a better choice than alternative offerings from technology companies.55
Page 108 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
outage.46 In 2018, the IIoT had potential applications across a range of industrial activities, including
manufacturing, supply chain management, health care, and retail. This potential was relatively undeveloped,
however, which left open the question of who would seize it and benefit most. Estimates suggested that the
revenue opportunities were vast, $225 billion annually by 2020, or even $11.1 trillion by 2025.47
Page 5
9B19M110
GO-TO-MARKET OPTIONS
GE had, in recent years, made the bulk of its sales from selling hardware (“selling boxes”) and from
providing maintenance contracts (“contract service agreements,” or CSAs) on its installed hardware base.
Digitalizing the company, according to Immelt’s vision, suggested new revenue options would arise from
new services.58 But new digital offerings could be constructed in alternative ways, some more radical than
others, compared with the company’s traditional business.
The least radical option would be to bundle new software and analytics services with hardware sales. GE
would sell boxes and maintenance contracts, as it always had, but bundling extra digital services with GE
products would differentiate this option from other manufacturers’ offerings, making GE products more
valuable to customers, thereby providing a sales advantage and the ability to command higher prices—and
generate higher profit margins. In effect, this option called for GE to stay ahead of its rivals by providing
better integrated digital services, thereby resisting the pressures toward commoditization in selling hardware.
A second option called for selling Predix licences to clients, and creating specialized consulting services to
help clients use Predix. In effect, Predix licensing and consulting would become a separate business line
for GE. Predix services would be sold to GE hardware customers separately from hardware, as an add-on
to help consumers optimize their use of GE hardware.
The third, and arguably most radical, option was to move increasingly in the direction of “outcomes-based”
deals. This option would involve approaching existing customers with a value proposition very different
from selling boxes. Sales consultants would work directly with client businesses, help them analyze how
the unique features of their business could benefit from connecting to the Predix platform, and then structure
contracts specific to that client company based on incremental revenue- and profit-sharing formulas.
Such a deal structure transferred the implementation risk to GE. The customer paid only if its business
achieved measured improvements. But shifting the risk was part of the strategy: as a very large company,
GE could bear risk more efficiently than most of its clients—and many of its competitors (which provided
GE with a defensible advantage over its rivals). In exchange for carrying that risk, GE would recover bigger
margins. However, making this vision a reality would require contracts that were based on a detailed
understanding of clients’ businesses, and on reliable and agreed-on metrics for measuring savings59 so that
customers would remain happy with the deal in the long term and GE would be adequately compensated
for having absorbed the risk.
NEW CAPABILITIES REQUIRED
To compete with the likes of Amazon, Google, IBM, Microsoft, and SAP, GE would need to raise its game
substantially in terms of its internal digital capabilities. The company already had information technology
(IT) organizations, of course, but they were largely focused on the needs of specific business units. Immelt
Page 109 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
“Predix’s goal,” suggested professors Cusumano, Gawer, and Yoffie, “was to become an operating system,
designed to serve as an innovation platform for applications development, providing the services that permit
programmers to quickly build apps for the industrial Internet.”56 To do this, they had to solve a “chicken-oregg” problem: Predix was not the dominant platform until others—namely, vendors and customers—accepted
it as a standard. But others were reluctant to invest heavily in Predix until they were confident it would emerge
as the winning standard. GE attempted to gain a necessary critical mass of users by starting pilot projects with
existing clients. But there were debates within GE about how Predix should be offered to clients.57
Page 6
9B19M110
GE Digital would need to architect the Predix platform, create new software offerings, and develop new
analytical capabilities that allowed for customization client by client.62 Ruh would need to build out a new
organization for product management and platform support teams for ecosystem partners. The new business
model and its associated services needed to be launched, marketed, and priced, requiring input from pricing,
legal, marketing, sales, and customer support functions.63 From its beginning, GE Digital was based away
from established GE organizations, in San Ramon, California, near Silicon Valley. Approximately 5,500
people were hired into the new unit between 2012 and 2016. Almost all of these new employees came from
tech companies, not from within GE’s existing IT organizations.64
SALES FORCE CAPABILITIES
Over the years, GE’s technical sales force had been very good at selling “boxes” and maintenance contracts.
To transition to the new digital GE, though, much more was needed. Outcomes-based contracts called on
the technical sales force to dig into the details of how their clients’ businesses worked and analyze their
businesses at a level of detail that would support the creation of revenue-sharing contracts unique to each
business and its needs. For Ruh, “the value of the API [the application programming interface for
connecting the client’s data flows into Predix] need[ed] to be based on the value of the outcomes it [could]
achieve.”65 This focus on outcomes represented a major change in the way a typical sales engagement was
designed. The new breed of GE salesperson would need to be part business consultant, part contract
designer, part software customizer, and part ongoing relationship manager. In many situations, the
individual they were selling to within the client organization would change since the grander scale of
revenue-sharing contracts would require higher levels of approval within a client organization.66
RELATIONSHIPS WITH CUSTOMERS
Both convincing customers to treat the company’s Predix platform as a dominant standard and moving
toward outcomes-based contracts would require a level of trust well beyond what GE had needed to sell
hardware and maintenance contracts. For the new breed of GE salesperson to dig into a client’s business in
sufficient detail to design outcomes-based contracts, clients would need to trust GE enough to allow this
scrutiny. The client would need to open up its processes and accounts, which would run counter to many
companies’ reflexive tendencies to keep their internal knowledge confidential. Client managers could
perhaps be excused for suspecting that a vendor—even one as familiar and trusted as GE—might request
access to internal information to fulfill some ulterior motives—for example, to gain an advantage in sales
negotiations, or to achieve a dominant platform position that might eventually allow it to extract higher
margins (as Microsoft had). To convince client leaders that it was in their own best interests to comply with
what GE wanted to do would require further strengthening GE’s already strong customer relationships.
Also, the level of technical sophistication of GE customers varied widely, across business lines and
geography. Some customers might well be ready to talk with GE about Predix-based transformation;
however, to others, any talk about data becoming more valuable than hardware and the notion of sharing
incremental revenues and profits might be too much to process.
Page 110 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
began working on this project in 2011, hiring a former Cisco executive, Bill Ruh, to develop GE’s Industrial
Internet strategy.60 In 2015, GE Digital was formally established as a separate business unit; a press release
described “a transformative move that brings together all of the digital capabilities from across the company
into one organization.” Ruh was named chief digital officer and head of the new unit.61
Page 7
9B19M110
IMPLEMENTATION DIFFICULTIES
Immelt had thoroughly communicated his vision, and his successor, John Flannery, had pledged to see
things through.67 However, implementation of GE’s digital transformation proved unexpectedly difficult,
in a variety of ways.
Disagreements flared as GE Digital sought to gain adoption of the Predix platform within GE. IT organizations
within each business unit had commitments to the legacy infrastructure and a duty to support the existing
businesses. Ruh explained: “Every one of our products had a different underpinning platform, architecture,
technology and set of vendors.”68 Tensions rose when it became clear that Predix lacked certain features that
managers had wanted—and that the legacy platforms did have. Because the San Ramon staff had been largely
hired from outside GE, they lacked relationship capital within the broader organization, which made it more
difficult to convince long-time GE employees of the value of the new platform. Different business units bought
in, to varying degrees, with the greatest acceptance in units that relied less on the existing infrastructure.69 The
uneven internal rollout translated into a similarly uneven external rollout.70
Confidence in the new platform was undermined by a series of technical problems. Most of these problems
related to unexpectedly complex legacy and integration issues, but some were caused by software glitches
in the platform itself. The mid-course shift in strategy, from building data centres internally to using
Amazon Web Services and Microsoft Azure services to host Predix, generated additional disruption.71
Accumulating difficulties warranted a “time out” in May and June of 2017, which put the rollout further
behind schedule than it already was.72
Salesforce Retraining
GE’s salesforce experienced challenges selling Predix. The salespeople needed to learn to adapt to a new
engagement model. The new approach lengthened sales cycles and increased the need to educate customers
in the new approach. It also increased the need for salespeople to understand their clients’ businesses in
detail, including a customer’s industry, business model, and how the company made money.73
This process implied significant retooling for the sales staff. GE’s salespeople knew how to sell tangible,
physical hardware through a channel or directly to enterprise customers; however, they now needed to sell
software, analytics services, and outcomes to senior business executives.74 In the new model, GE’s
salespeople needed to look “at all the data we have, to help understand what kind of outcomes we could
achieve by working with the business,”75 said Kate Johnson, GE’s chief commercial officer.
GE’s existing sales staff varied in their willingness and ability to move in this new direction. In outcomes-based
selling, Johnson suggested, “The customer expects you to talk about outcomes and values—not the best
widgets.”76 But explaining why GE widgets were the best was what GE’s salespeople had been doing for years.
Limited Customer Enthusiasm
GE’s vision also proved harder than expected to sell to customers. Many had trouble with understanding
GE’s proposition and were held back by their own lack of organizational readiness:
Page 111 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
The Predix Rollout
Page 8
9B19M110
Many GE customers had not made the mental leap to appreciating the possibilities of an Industrial Internet,
lacked the required infrastructure to support GE’s new methodologies, and did not collect the data that was
driving the new approach. Beth Comstock, GE’s chief marketing officer, explained: “We’re trying to sell
them something they don’t know they need.”78 Educating clients was needed to help them see the
opportunities. Even when GE’s education efforts worked and customers woke up to the new possibilities,
they immediately realized that it would require them, as Ruh put it, “to rethink what they’re doing
entirely,”79—a daunting prospect that many could not quickly wrap their heads around.
CONTRIBUTING DIFFICULTIES
Not all of GE’s difficulties in 2019 were attributable to the digitalization efforts. Other factors presented
additional challenges at about the same time.
Accounting Problems
Due to GE’s long-standing emphasis on metrics to evaluate performance, the company’s internal
accounting used a complicated design and, over time, had accumulated additional complexity. The
complexity seemed to reach a critical point toward the end of Immelt’s tenure as CEO, during a series of
analyst surprises and volatile stock price movements. The US Securities and Exchange Commission (SEC)
began investigating GE’s accounting practices in 2017, after GE reported a $6.2 billion after-tax charge and
suddenly had to reserve $15 billion to cover a previously undisclosed, decade-old liability in the insurance
component of GE Capital.80 The SEC expanded its investigations in 2018.81
Doubts were raised about the transparency of the company’s financial disclosures. Material difficulties,
some suggested, were buried in the complexities of the company’s “black box” and in multi-business unit
accounting,82 and quarterly earnings were being smoothed through “window-dressing.”83 The doubts were
exacerbated by Immelt’s unwavering optimism regarding the company’s position and future, which turned
out to be far from reality.84 Some former executives and investors charged that, despite Immelt’s publicized
attempts to remove bureaucracy from the organization, his sanguinity spawned a culture in which people
believed that he did not want to hear discouraging reports.85
GE’s Power Business (and Other Acquisition Difficulties)
GE’s power business—its largest, accounting for more than 23 per cent of revenues in 2018—experienced
a 22 per cent revenue decrease from 2017 to 2018.86 GE purchased Alstom’s power and grid assets in late
2015, for $10.6 billion.87 The U.S. Department of Justice (DOJ) pressured GE to divest the part of Alstom
that serviced turbines made by its competitors,88 and GE also agreed to divest Alstom’s program to build a
state-of-the-art gas turbine, after being pressured by the European Union.89 These concessions weakened
the acquisition’s justification, causing GE insiders to rally against it. But Immelt wanted it, and GE’s
largest-ever acquisition closed.90 The timing was bad, however; it was “a massive investment in natural gas
power plants just as the market for them was contracting.”91
Page 112 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
GE had taken an informal poll of customers about their readiness and adoption of the Industrial
Internet and learned that 63% of customers polled said their machines were connected to networks,
but they were not yet using these data, 13% claimed they used data for competitive advantage, and
63% were not performing any condition-based maintenance.77
Page 9
9B19M110
In addition, the GE Power business lagged its competitors in recognizing the market signals, which led to
overcapacity and significant unsold inventory; cash tied up in working capital forced discounts and
decreased sales profits.93 GE had poorly forecasted the impacts of developments in the renewable energy
market, such as intensified competition that lowered prices in the wind turbine market and a global shift
toward solar power.94 Operational mishaps, faulty pricing strategies, and unexpected warranty costs added
to the malaise.95
A Legacy of Problems at GE Capital
Although GE had sold off the majority of its assets in GE Capital in 2015 (more than $350 billion),96
difficulties in GE Capital, especially during the 2008 financial crisis, contributed to GE’s downward spiral.
Prior to the financial crisis, GE Capital had operated as a “shadow bank”—a business that facilitated the
creation of credit in the financial markets but was not subject to the same strict rules and regulations of a
bank.97 This framework allowed it to take actions—and risks—that banks could not, by providing cheap
capital to GE’s own businesses and customers.98 For a time, this arrangement was very lucrative for GE; at
its peak, during the Welch years, GE Capital accounted for more than 60 per cent of GE profits.99
But this lucrative situation all backfired when the financial crisis hit, and credit market liquidity evaporated.
GE was unable to finance its customers or its own operations, and the Federal Reserve had to provide the
company with an emergency bailout of $139 billion in government-guaranteed debt.100 When the crisis
abated, GE Capital was designated as a systemically important financial institution, and was required to
adhere to strict regulations—an arrangement that was not nearly as lucrative for GE.101
Also haunting GE were issues related to a subprime mortgage origination company, WMC Mortgage
(WMC), which GE Capital acquired in 2004. In 2015, the DOJ began investigating how WMC sold loans
in 2006 and 2007.102 The DOJ found that that WMC had misrepresented the quality of residential mortgagebacked securities and that “failure to disclose material deficiencies in those loans contributed to the financial
crisis.”103 Settling these allegations cost GE a $1.5 billion fine, and WMC filed for bankruptcy.104
LOOKING FORWARD
As recently as three years earlier, GE had seemed to be on the fast track to becoming an Industrial Internet
leader. However, after $4 billion in investments in GE Digital, that vision had narrowed. Layoffs had come
to San Ramon,105 and Bill Ruh had announced his departure.106 Many people asked: What went wrong?
The possibilities were numerous. Perhaps the vision was flawed. Or perhaps the vision was fine, but GE’s
implementation was faulty. Or perhaps GE was just unlucky, in that a perfect storm of adverse developments
arrived at exactly the wrong time. Or perhaps GE gave up too soon; maybe such major change takes far longer
than it expected, and if it had persisted, the company would have eventually prevailed. It was even possible
that the company would eventually, in the fullness of time, succeed with some version of its original digital
vision, once it got clear of the current difficulties; it did, after all, retain ownership of GE Digital.
The debate raged, but one thing was not in doubt: the once-proud GE, in 2019, was scrambling to recover
some semblance of its former greatness. Only time would tell how the story would end.
Page 113 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Alstom was not GE’s only poorly timed acquisition. From 2010 to 2014, the company bought nine firms in
the oil and gas business, only to see oil prices collapse. The company also lost money on acquisitions in the
home mortgage industry (see the next section) and in the homeland security business.92
Page 10
9B19M110
EXHIBIT 1: GENERAL ELECTRIC’S STOCK PRICES, DECEMBER 2016 TO OCTOBER 2018 (IN US$)
35
25
20
15
10
5
0
Source: Created by case authors based on data from Yahoo! Finance, “General Electric Company (GE),” accessed August 9,
2019, https://finance.yahoo.com/quote/GE/history?period1=1480564800&period2=1538362800&interval=1d&filter=history&fr
equency=1d.
Page 114 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
30
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Page 11
9B19M110
EXHIBIT 2: GENERAL ELECTRIC FINANCIAL STATEMENTS, 2016–2018 (IN US$ MILLION)
Page 115 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Page 12
9B19M110
EXHIBIT 2 (CONTINUED)
Source: General Electric, “Financial Statements, 2016–2018,” Capital IQ, accessed August 9, 2019.
Page 116 of 282
9B19M110
EXHIBIT 3: GENERAL ELECTRIC’S MANAGEMENT DEPARTURES
GE Power
CEO
Steve Bolze
Retires after
24-year tenure
at GE, June
14, 2017
GE CEO
GE Digital CCO
Vice-Chairs
GE CFO
Jeffrey Immelt
Kate Johnson
Retires after
16-year tenure
at GE (replaced
by John
Flannery),
October 2,
2017
Ends four-year
tenure at GE,
September 8,
2017
Beth Comstock
and
John Rice
Jeffrey S.
Bornstein
Retire
Retires after
28-year
tenure at GE,
December
31, 2017
GE reduces
board size from
18 to 12, adds
three new
directors (nine
board
departures),
February 26,
2018
GE CEO
GE Digital CEO
John Flannery
Bill Ruh
Retires after
14-month
tenure as CEO
(replaced by
Lawrence
Culp), October
1, 2018
Announces end
of five-year
tenure at GE,
December 13,
2018
Note: GE = General Electric; CEO = chief executive officer; CCO = chief commercial officer; CFO = chief financial officer
Source: Compiled by case authors based on Alywn Scott, “GE Merges Power Units as Executive Who Lost Out on GE CEO Job Retires,” Reuters, June 14, 2017, accessed
March 14, 2019, www.reuters.com/article/us-ge-power/ge-merges-power-units-as-executive-who-lost-out-on-ge-ceo-job-retires-idUSKBN19524Z?utm_source=34553&utm_
medium=partner4; “John Flannery Succeeds Jeff Immelt as Chairman of GE; Lorenzo Simonelli Named Chairman of Baker Hughes GE,” General Electric, press release,
October 2, 2017, accessed April 25, 2019, www.genewsroom.com/press-releases/john-flannery-succeeds-jeff-immelt-chairman-ge-lorenzo-simonelli-named-chairman; “Kate
Johnson, GE Digital Chief Commercial Officer, to Join Microsoft as President, CVP of Microsoft US,” Official Microsoft Blog, July 17, 2017, accessed April 24, 2019,
https://blogs.microsoft.com/blog/2017/07/17/kate-johnson-ge-digital-chief-commercial-officer-join-microsoft-president-cvp-microsoft-us/; “GE Names Jamie Miller as CFO; Jeff
Bornstein to Leave the Company,” General Electric, press release, October 6, 2017, accessed March 14, 2019, www.genewsroom.com/press-releases/ge-names-jamie-millercfo-jeff-bornstein-leave-company-284049; “GE Announces 2018 Board of Directors Slate; Includes Three New Directors,” General Electric, press release, February 26, 2018,
accessed April 25, 2019, www.genewsroom.com/press-releases/ge-announces-2018-board-directors-slate-includes-three-new-directors-284255; “H. Lawrence Culp, Jr.
Named Chairman and CEO of GE,” General Electric, press release, October 1, 2018, accessed April 25, 2019, www.genewsroom.com/press-releases/h-lawrence-culp-jrnamed-chairman-and-ceo-ge-284509; “GE Advances Digital Leadership with Launch of $1.2 Billion Industrial IoT Software Company,” General Electric, press release,
December 13, 2018, accessed April 18, 2019, www.genewsroom.com/press-releases/ge-advances-digital-leadership-launch-12-billion-industrial-iot-software-company.
Page 117 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Page 13
Page 14
9B19M110
EXHIBIT 4: GENERAL ELECTRIC’S BUSINESS UNITS IN 2019
GE Power created energy technologies under the following operating units: Gas Power Systems, Steam
Power Systems, Power Services, Grid Solutions, Power Conversion, Automation & Controls, and GE
Hitachi Nuclear.
GE Lighting provided industrial, commercial, and consumer lighting solutions.
GE Renewable Energy provided equipment for the production of energy from renewable sources under the
following operating units: Wind, Hydro, Hybrid, and Grid.
GE Oil & Gas was a full-stream provider of integrated oilfield products, services, and digital solutions.
GE Healthcare provided a portfolio of products, solution and services used in the diagnosis, treatment, and
monitoring of patients under the following operating units: Healthcare Systems and Life Sciences.
Healthcare Systems comprised Imaging, Ultrasound, Life Care Solutions, and Enterprise Software &
Solutions business units.
GE Transportation provided equipment to the rail, mining, marine, stationary power, and drilling industries.
GE Capital was GE’s financial services, which focused on financing customers in GE’s industrial
businesses, primarily in aviation and energy finance.
Source: General Electric Company, Form 10-K, 4, February 26, 2019, accessed March 15, 2019,
www.sec.gov/Archives/edgar/data/40545/000004054519000014/ge10-k2018.htm; “Powering Forward,” GE Power, accessed
April 23, 2019, www.ge.com/power; “GE Aviation,” GE Aviation, accessed April 23, 2019, www.geaviation.com/; “About GE
Renewable Energy,” GE Renewable Energy, accessed April 23, 2019, www.ge.com/renewableenergy/about-us; General
Electric,
2018
Annual
Report,
8,
2019,
accessed
April
23,
2019,
www.ge.com/investorrelations/sites/default/files/GE_AR18.pdf; “About GE Healthcare Systems,” GE Healthcare, accessed April 23, 2019,
www.gehealthcare.com/en/about/about-ge-healthcare-systems; “Home,” GE Transportation, accessed April 23, 2019,
www.getransportation.com/home; “We’re GE Capital,” GE Capital, accessed April 23, 2019, www.gecapital.com/.
Page 118 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
GE Aviation provided and serviced aircraft engines under the following operating units: Commercial,
Military, and Business and General Aviation.
Page 15
9B19M110
EXHIBIT 5: GENERAL ELECTRIC’S COMPETITORS IN THE INDUSTRIAL INTERNET
IBM
Philips
Philips, a Dutch multinational health care and lighting electronics company with 77,400 employees,
operated in the diagnostic & treatment, connected care & health informatics, personal health, and other
businesses. Philips operated its HealthSuite Digital Platform in its “other” business unit and offered Internet
of Things (IoT) solutions for industrial, commercial, and consumer applications through its lighting business.
Philips partnered with Cisco in 2015 to sell Internet-connected lights to the global office market.
Toshiba
Toshiba, a Japanese multinational company with 141,256 employees, operated in the power systems and
industrial systems industries. Toshiba launched SPINEX, an IoT architecture, in November 2016, and
aimed to expand its IoT-related sales to ¥200 billion (US$2 billion) by 2020, doubling its 2016 sales.
Google
Google, an American multinational technology company with 98,771 employees, produced and supported
products and platforms, including Android, Chrome, Gmail, Google Maps, Google Play, Search, and
YouTube. Google launched its Things platform of IoT products and services in May 2018.
Microsoft
Microsoft, a global technology company with 131,000 employees, developed and supported software,
services, devices, and solutions for individual and enterprise customers. Microsoft Azure was a set of cloud
products and services that, among other functions, provided the infrastructure for customers to develop IoT
capabilities. Microsoft Azure’s IoT capabilities were launched in April 2018. At the same time, Microsoft
announced it would spend $5 billion to further develop its IoT capabilities over the next four years.
Cisco
Cisco, an American multinational technology company with 74,200 employees, designed and sold a broad
range of Internet-enabling technologies and operated in the infrastructure platforms, applications, security,
and other products businesses. Cisco produced switches, a networking hardware used in the IoT, and IoT
and analytics software offerings.
Page 119 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
IBM, a global technology company with 381,100 employees, offered consulting and information technology
(IT) implementation services, cloud, digital, and cognitive offerings, and enterprise systems and software.
In 2013, IBM’s supercomputer, Watson, launched its first commercial application in health care to provide
a range of enterprise artificial intelligence services, applications, and solutions.
Page 16
9B19M110
EXHIBIT 5 (CONTINUED)
Honeywell
Honeywell, a diversified technology and manufacturing company with 114,000 employees, operated in the
aerospace, building technologies, performance materials and technologies, and safety and productivity
solutions businesses. IIoT by Honeywell, a cloud-enabled software service, was launched in October 2016.
Amazon, an American multinational technology company with 647,500 employees, operated an
ecommerce marketplace and provided services to developers and enterprises through its Amazon Web
Services (AWS) segment. Amazon had both consumer and industrial applications in the Industrial Internet
of Things (IIoT). AWS launched its managed IoT cloud platform in October 2015.
Siemens
Siemens, a German conglomerate with 379,000 employees, competed with GE in its power, oil and gas,
transportation, medical products, and renewable power businesses. Siemens entered the IIoT market with
a launch, in March 2016, of MindSphere 3.0, a cloud-based, open IoT operating system connecting
products, plants, systems, and machines. Siemens acquired Mendix, a low-code application development
platform in August 2018.
Note: Each company’s number of employees is stated as listed in each company’s 2018 annual report.
Source: International Business Machines, Form 10-K, 1, 68, 2018, accessed April 25, 2019,
www.sec.gov/Archives/edgar/data/51143/000104746919000712/a2237254z10-k.htm; Bruce Upbin, “IBM’s Watson Gets Its
First Piece of Business in Healthcare,” Forbes, February 8, 2013, accessed April 25, 2019,
www.forbes.com/sites/bruceupbin/2013/02/08/ibms-watson-gets-its-first-piece-of-business-in-healthcare/#1d4d76325402;
“IBM Watson About,” IBM Watson, accessed April 25, 2019, www.ibm.com/watson/about; Philips, Annual Report 2018, 8, 10,
16, 2018, accessed April 24, 2019, www.results.philips.com/publications/ar18?type=annual-report; “Discover Our Professional
and Consumer Philips Lighting Products,” Philips, accessed April 24, 2019, www.lighting.philips.com.au/home; “Philips and
Cisco Form Global Strategic Alliance to Address EUR 1 Billion Office Lighting Market,” Philips, press release, December 9,
2015, accessed April 25, 2019, www.philips.com/a-w/about/news/archive/standard/news/press/2015/20151209-Philips-andCisco-form-global-strategic-alliance-to-address-EUR-1-billion-office-lighting-market.html; “Basic Corporate Data,” Toshiba,
accessed April 25, 2019, www.toshiba.co.jp/worldwide/about/corp_data.html; Toshiba, 2018 Annual Report, 8–9, 2018,
accessed April 25, 2019, www.toshiba-tpsc.co.jp/pdf/english/ir/pdf/AR2018.pdf; “Toshiba Reinforces IoT Business with the
Launch
of
SPINEX,”
Toshiba,
press
release,
November
1,
2016,
accessed
April
25,
2019,
www.toshiba.co.jp/about/press/2016_11/pr0102.htm; Alphabet Inc., Form 10-K, 3, 6, 2018, accessed April 25, 2019,
www.sec.gov/Archives/edgar/data/1652044/000165204418000007/goog10-kq42017.htm#sDFF4659E47B3F771F6E978DB
C5993A59; “Google Cloud IoT,” Google Cloud, accessed April 25, 2019, https://cloud.google.com/solutions/iot/; Anthony
Spadafora, “Google Officially Launches Things, Its IoT Platform,” ITProPortal, May 8, 2018, accessed April 25, 2019,
www.itproportal.com/news/google-officially-launches-things-its-iot-platform/; Microsoft Corporation, Form 10-K, 3, 16, 18,
2018, accessed April 25, 2019, https://microsoft.gcs-web.com/static-files/bae1357b-19a0-4075-99c2-e715aa6919ad; Julia
White, “Microsoft Will Invest $5 Billion in IoT. Here’s Why,” Microsoft, April 4, 2018, accessed April 25, 2019,
https://azure.microsoft.com/en-us/blog/microsoft-will-invest-5-billion-in-iot-here-s-why/; Cisco Systems Inc., 2018 Annual
Report, 20, 22, 2018, accessed April 24, 2019, www.cisco.com/c/dam/en_us/about/annual-report/2018-annual-report-full.pdf;
“Honeywell Industrial Internet of Things—Cloud Software,” Honeywell International Inc., press release, October 7, 2016,
accessed April 24, 2019, www.honeywell.com/en-us/newsroom/news/2016/10/honeywell-launches-cloud-enabled-softwareservice-to-optimize-plant-performance; Amazon.com Inc., Form 10-K, 3–4, 2018, accessed April 24, 2019,
www.sec.gov/Archives/edgar/data/1018724/000101872419000004/amzn-20181231x10k.htm#s52F7FEDACEA351758F06B
0012FFEA2EF;
“AWS
IoT,”
Amazon
Web
Services,
accessed
April
8,
2019,
https://aws.amazon.com/iot/?sc_channel=PS&sc_campaign=acquisition_CA&sc_publisher=google&sc_medium=iot_b&sc_c
ontent=iot_bmm&sc_detail=%2Bamazon%20%2BInternet%20%2Bof%20%2Bthings&sc_category=iot&sc_segment=153184
809251&sc_matchtype=b&sc_country=CA&s_kwcid=; “Amazon.com Announces Third Quarter Sales up 23% to $25.4 Billion,”
Amazon Web Services, press release, October 22, 2015, accessed April 24, 2019, https://press.aboutamazon.com/newsreleases/news-release-details/amazoncom-announces-third-quarter-sales-23-254-billion; Siemens, Annual Report 2018, 4,
2018, accessed April 25, 2019, www.siemens.com/investor/pool/en/investor_relations/Siemens_AR2018.pdf; “This Is
MindSphere!,” Siemens, accessed March 19, 2019, https://new.siemens.com/global/en/products/software/mindsphere.html;
“Siemens Launches MindSphere Open Industry Cloud,” Siemens, press release, March 1, 2016, accessed March 19, 2019,
www.siemens.com/press/pool/de/pressemitteilungen/2016/digitalfactory/pr2016030171dfen.pdf.
Page 120 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Amazon
Page 17
9B19M110
ENDNOTES
This case has been written on the basis of published sources only. Consequently, the interpretation and perspectives
presented in this case are not necessarily those of General Electric or any of its employees.
2
Alwyn Scott and Arunima Banerjee, “GE Shares Jump as CEO John Flannery Replaced by Outsider H. Lawrence Culp,” Globe and
Mail, October 1, 2018, accessed April 19, 2019, www.theglobeandmail.com/business/article-ge-ceo-john-flannery-steps-down/.
3
All currency amounts are in US$ unless otherwise specified.
4
Matt Egan, “GE Cuts Dividend for Second Time Since Great Depression,” CNN Business, November 13, 2017, accessed
April 19, 2019, https://money.cnn.com/2017/11/13/investing/ge-dividend-cut/index.html.
5
Wayne Duggan, “GE Cuts Quarterly Dividend to One Cent,” US News and World Report, October 30, 2108, accessed Juy
28, 2019, https://money.usnews.com/investing/stock-market-news/articles/2018-10-30/ge-cuts-quarterly-dividend-to-1-cent.
6
Fred Imbert, “S&P Downgrades General Electric’s Credit Rating a Day after CEO Is Fired,” October 2, 2018, accessed March
5, 2019, www.cnbc.com/2018/10/02/ge-credit-rating-under-review-for-possible-downgrade-moodys.html.
7
Thomas Gryta, “GE Shed 30,000 Workers Last Year,” Wall Street Journal, February 26, 2019, accessed March 14, 2019,
www.wsj.com/articles/ge-shed-30-000-workers-last-year-11551221265?mod=searchresults&page=1&pos=16.
8
Geoff Colvin, “What the Hell Happened at GE?,” Fortune, May 24, 2018, accessed July 28, 2019,
http://fortune.com/longform/ge-decline-what-the-hell-happened/.
9
Dana Cimilluca, Dana Mattioli, and Thomas Gryta, “GE Puts Digital Assets on the Block,” Wall Street Journal, July 30, 2018,
accessed July 28, 2019, www.wsj.com/articles/ge-puts-digital-assets-on-the-block-1532972822.
10
Jeffrey R. Immelt, “How I Remade GE,” Harvard Business Review, September 2017, accessed February 4, 2019,
https://hbr.org/2017/09/inside-ges-transformation#how-i-remade-ge.
11
Michael Sheetz, “GE CEO Culp Lays Out Focus on 4 Businesses, Aims to Restore Dividend to Inline with Peers,” CNBC,
February 26, 2019, accessed April 18, 2019, www.cnbc.com/2019/02/26/ge-ceo-culp-lays-out-new-focus-on-4-businessesaims-to-restore-dividend-to-inline-with-peers.html.
12
“GE Advances Digital Leadership with Launch of $1.2 Billion Industrial IoT Software Company,” GE Digital, press release,
December 13, 2018, accessed January 15, 2019, www.ge.com/digital/blog/ge-advances-digital-leadership-launch-12-billionindustrial-iot-software-company.
13
Steve Lohr, “G.E. to Spin Off Its Digital Business,” New York Times, December 13, 2018, accessed February 4, 2019,
www.nytimes.com/2018/12/13/business/ge-digital-spinoff.html.
14
“GE’s Digital Transformation Journey,” Wall Street Journal, February 28, 2017, accessed July 28, 2019,
https://deloitte.wsj.com/cio/2017/02/28/ges-digital-transformation-journey/; Molly St. Louis, “GE Is Learning about Leadership
from Silicon Valley,” Inc., December 26, 2016, accessed July 28, 2019, www.inc.com/molly-reynolds/ge-is-learning-aboutleadership-from-silicon-valley.html.
15
“Thomas Edison & the History of Electricity,” General Electric, accessed March 3, 2019, www.ge.com/aboutus/history/thomas-edison.
16
“Company Profile: General Electric,” PatSnap, September 6, 2018, accessed April 8, 2019,
www.patsnap.com/resources/innovation/general-electric.
17
“General Electric’s Greatest Inventions,” Telegraph, accessed April 11, 2019, www.telegraph.co.uk/finance/newsbysector/i
ndustry/8197977/General-Electrics-greatest-inventions-in-pictures.html.
18
Jane Nicholls, “Inside Crotonville: GE’s Corporate Vault Unlocked,” GE Reports, October 29, 2017, accessed July 24, 2019,
www.ge.com/reports/inside-crotonville-ges-corporate-vault-unlocked/.
19
Jon Chesto, “GE’s Secret Weapon Is Its Training Center on Hudson River,” Boston Globe, February 14, 2017, accessed April 11, 2019,
www.bostonglobe.com/business/2017/02/13/hudson-river-secret-corporate-weapon/lu2txUamtMMT94HZ38ztNM/story.html.
20
Colvin, op. cit.; Dan Marcec, “CEO Tenure Rates,” Harvard Law School Forum on Corporate Governance and Financial
Regulation, February 12, 2018, accessed April 11, 2019, https://corpgov.law.harvard.edu/2018/02/12/ceo-tenure-rates/.
21
Deborah Sherry, “How the Digitalisation of Planes, Trains Industries Is Shaping the Future of Automobiles,” LinkedIn,
September 15, 2017, accessed April 19, 2019, www.linkedin.com/pulse/digital-planes-trains-automobiles-deborah-sherry/.
22
Immelt, op. cit.; Thomas Gryta, “GE Begins to Sell Off One of Its Oldest Businesses: Lights,” Wall Street Journal, February
15, 2018, accessed April 26, 2019, www.wsj.com/articles/ge-moves-to-sell-first-part-of-lighting-business-1518735170.
23
“Directory,” General Electric, accessed March 15, 2019, www.ge.com/directory.
24
General Electric Company, Form 10-K, 4, February 26, 2019, accessed March 15, 2019,
www.sec.gov/Archives/edgar/data/40545/000004054519000014/ge10-k2018.htm.
25
Catherine Clifford, “Jack Welch: This Is the No. 1 Key to Success as a Leader,” CNBC, November 17, 2017, accessed
March 3, 2019, www.cnbc.com/2017/11/17/former-ge-ceo-jack-welch-how-to-be-a-great-leader.html.
26
Adam Hartung, “GE Needs a New Strategy and a New CEO,” Forbes, March 28, 2017, accessed April 10, 2019,
www.forbes.com/sites/adamhartung/2017/03/28/ge-needs-a-new-strategy-and-a-new-ceo/#509fcca14ad2; Martha Slud, “GE
Deal Is Classic Welch,” October 23, 2000, CNN Money, accessed August 5, 2019, https://money.cnn.com/2000/10/23/deals/
welch/index.htm.
27
Noel Tichy and Ram Charan, “Speed, Simplicity, Self-Confidence: An Interview with Jack Welch,” Harvard Business Review
67, no. 5 (September/October 1989): 112–120. Available from Ivey Publishing, product no. 89513.
28
Six-Sigma, which GE implemented in 1995, was a set of tools and techniques developed by Motorola for the purpose of
quality management and process improvement.
29
“John F. Welch, Jr.,” General Electric, accessed March 8, 2019, www.ge.com/about-us/leadership/profiles/john-f-welch-jr.
Page 121 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
1
9B19M110
30
Geoffrey Colvin, “The Ultimate Manager in a Time of Hidebound, Formulaic Thinking, General Electric’s Jack Welch Gave
Power to the Worker and the Shareholder. He Built One Hell of a Company in the Process,” Fortune, November 22, 1999,
accessed March 3, 2019, http://archive.fortune.com/magazines/fortune/fortune_archive/1999/11/22/269126/index.htm.
31
Ibid.
32
“Rank and yank” was an organizational management practice used by GE, in which managers annually divided employees
into three categories: the top 20 per cent, the middle 70 per cent, and the bottom 10 per cent. The top tercile received praise
and financial rewards, the middle received coaching and training, and the bottom tercile was dismissed.
33
Jeffrey Pfeffer and Robert I. Sutton, “Evidence-Based Management,” Harvard Business Review 84, no. 1 (January 2006):
62–74. Available from Ivey Publishing, product no. R0601E; Alan Murray, “Should I Rank My Employees?,” accessed April
13, 2019, http://guides.wsj.com/management/recruiting-hiring-and-firing/should-i-rank-my-employees/; Jack Welch, “Jack
Welch: ‘Rank-and-Yank’? That’s Not How It’s Done,” Wall Street Journal, November 14, 2013, accessed April 13, 2019,
www.wsj.com/articles/8216rankandyank8217-that8217s-not-how-it8217s-done-1384473281.
34
Vitaliy Katsenelson, “Welch vs Bezos,” Contrarian Edge, accessed March 8, 2019, https://contrarianedge.com/welch-vs-bezos/.
35
James B. Stewart, “Did the Jack Welch Model Sow Seeds of G.E.’s Decline?,” New York Times, June 15, 2017, accessed
July 26, 2019, www.nytimes.com/2017/06/15/business/ge-jack-welch-immelt.html.
36
“Jeffrey R. Immelt,” General Electric, accessed April 11, 2019, www.ge.com/about-us/leadership/profiles/jeffrey-r-immelt.
37
Immelt, op. cit.
38
Michael E. Porter and James E. Heppelmann, “How Smart, Connected Products Are Transforming Companies,” Harvard
Business Review, October 2015, accessed August 19, 2019, https://hbr.org/2015/10/how-smart-connected-products-aretransforming-companies.
39
Karim R. Lakhani, Marco Iansiti, and Kerry Herman, GE and the Industrial Internet (Boston, MA: Harvard Business
Publishing, 2015), 2. Available from Ivey Publishing, product no. 614032.
40
Rajiv Lal and Scott Johnson, GE Digital (Boston, MA: Harvard Business Publishing, 2017). Available from Ivey Publishing,
product no. 517063.
41
Jacob Morgan, “A Simple Explanation of the Internet of Things,” Forbes, May 13, 2014, accessed July 24, 2019,
www.forbes.com/sites/jacobmorgan/2014/05/13/simple-explanation-internet-things-that-anyone-canunderstand//#431c1b491d.
42
David Goldman, “Your Samsung TV Is Eavesdropping on Your Private Conversations,” CNN Business, February 10, 2015,
accessed on July 24, 2019, https://money.cnn.com/2015/02/09/technology/security/samsung-smart-tv-privacy/index.html.
43
“What Is IIoT?,” Hewlett Packard Enterprise, accessed April 11, 2019, www.hpe.com/ca/en/what-is/industrial-iot.html.
44
“Everything You Need to Know about the Industrial Internet of Things,” GE Digital, accessed March 15, 2019,
www.ge.com/digital/blog/everything-you-need-know-about-industrial-Internet-things.
45
“Trip Optimizer,” GE Transportation, 2016, accessed July 28, 2019, www.ge.com/digital/sites/default/files/download_asset
s/GE-Transportation-Trip-Optimizer-20160824.pdf.
46
Manjish Naik, “When Assets Need Optimized Proactive Maintenance,” GE Digital blog, accessed July 28, 2019,
www.ge.com/digital/blog/when-assets-need-optimized-proactive-maintenance.
47
Steve Lohr, “G.E., The 124-Year-Old Software Start-Up,” New York Times, August 27, 2016, accessed February 4, 2019,
www.nytimes.com/2016/08/28/technology/ge-the-124-year-old-software-start-up.html?module=inline; Jacques Bughin, Michael
Chui, and James Manyika, “An Executive’s Guide to the Internet of Things,” McKinsey Quarterly, August 2015, accessed March
3, 2019, www.mckinsey.com/business-functions/digital-mckinsey/our-insights/an-executives-guide-to-the-Internet-of-things.
48
Michael Cusumano, Annabelle Gawer, and David Yoffie, The Business of Platforms: Strategy in the Age of Digital
Competition, Innovation, and Power (New York, NY: Harper Business, 2019).
49
Immelt, op. cit.
50
Lohr, “G.E., the 124-Year-Old Software Start-Up,” op. cit.
51
Cusumano, Gawer, and Yoffie, op. cit., 161.
52
Vijay Govindarajan and Jeffrey R. Immelt, “The Only Way Manufacturers Can Survive,” MIT Sloan Management Review 60, no. 3
(2019): 24–33. Available from Ivey Publishing, product no. SMR60303; Lohr, “G.E., The 124-Year-Old Software Start-Up,” op. cit.
53
“Customer Stories,” GE Digital, accessed April 11, 2019, www.ge.com/digital/customers.
54
Steve Lohr, “G.E. Makes a Sharp ‘Pivot’ on Digital,” New York Times, April 19, 2018, accessed February 4, 2019,
www.nytimes.com/2018/04/19/business/ge-digital-ambitions.html?module=inline; Immelt, op. cit.
55
Immelt, op. cit.
56
Cusumano, Gawer, and Yoffie, op. cit., 162.
57
Ibid.
58
Lakhani, Iansiti, and Herman, op. cit., 9
59
Ibid.
60
Bill Ruh, “Bill Ruh’s Related Insights,” GE blog, accessed August 1, 2019, www.ge.com/digital/blog/author/bill-ruh.
61
“Creation of GE Digital,” General Electric, press release, September 14, 2015, accessed July 25, 2019,
www.genewsroom.com/press-releases/creation-ge-digital.
62
Greg Cline, “IoT and Analytics: Better Manufacturing Decisions in the Era of Industry 4.0,” Aberdeen Group, August 2017,
accessed April 8, 2019, www.ibm.com/downloads/cas/ZPB2PN2G; “AWS IoT,” Amazon Web Services, accessed April 8,
2019, https://aws.amazon.com/iot/?sc_channel=PS&sc_campaign=acquisition_CA&sc_publisher=google&sc_medium=iot_b
&sc_content=iot_bmm&sc_detail=%2Bamazon%20%2BInternet%20%2Bof%20%2Bthings&sc_category=iot&sc_segment=1
53184809251&sc_matchtype=b&sc_country=CA&s_kwcid=.
Page 122 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Page 18
9B19M110
63
Massimo Russo, George Bene, Sanjay Verma, and Aakash Arora, “How Hardware Makers Can Win in the Software World,”
Boston Consulting Group, May 26, 2016, accessed March 3, 2019, www.bcg.com/publications/2016/marketing-sales-pricinghow-hardware-makers-can-win-software-world.aspx. At first conditioned by industrial reflexes, GE leaned toward building the
platform internally, hosting the software itself. Eventually GE management realized that building out a platform-based
ecosystem by themselves would be expensive, so GE began to partner, basing Predix on Amazon Web Services and Microsoft
Azure cloud technology; Barb Darrow, “Here’s Why GE Shelved Plans to Build Its Own Amazon-Like Cloud,” Fortune,
September 6, 2017, accessed on August 19, 2019, https://fortune.com/2017/09/06/general-electric-cloud-pivot/.
64
Lakhani, Iansiti, and Herman, op. cit., 9.
65
Cliff Saran, “Executive Interview: GE Software Chief Bill Ruh on Value of an Industrial Cloud,” ComputerWeekly, March 6,
2015, accessed April 16, 2019, www.computerweekly.com/news/2240241843/Executive-interview-GEs-software-chief-BillRuh-on-value-of-an-industrial-cloud.
66
Russo, Bene, Verma, and Arora, op. cit.
67
John Flannery, “Our Future Is Digital,” LinkedIn post, September 2017, accessed August 7, 2019,
www.linkedin.com/pulse/our-future-digital-john-flannery/.
68
Lakhani, Iansiti, and Herman, op. cit., 5.
69
Ibid., 11.
70
Peter High, “The CEO of GE Digital on What Is Next for the Industrial Icon,” Forbes, July 23, 2018, accessed April 26, 2019,
www.forbes.com/sites/peterhigh/2018/07/23/the-ceo-of-ge-digital-on-what-is-next-for-the-industrial-icon/#81bb06a3fdb8.
71
Alwyn Scott, “GE Shifts Strategy, Financial Targets for Digital Business after Missteps,” Reuters, August 28, 2017, accessed
April 19, 2019, www.reuters.com/article/us-ge-digital-outlook-insight/ge-shifts-strategy-financial-targets-for-digital-businessafter-missteps-idUSKCN1B80CB.
72
Alwyn Scott, “GE Is Shifting the Strategy for Its $12 Billion Digital Business,” Reuters, August 28, 2017, accessed July 26,
2019, www.businessinsider.com/r-ge-shifts-strategy-financial-targets-for-digital-business-after-missteps-2017-8.
73
Lisa Kelly, “Interview: Kate Johnson, Chief Commercial Officer at GE Digital, on Digitising Manufacturing,”
Computerweekly.com, May 24, 2016, accessed April 16, 2019, www.computerweekly.com/news/450288465/Interview-KateJohnson-chief-commercial-officer-at-GE-Digital-on-digitising-manufacturing.
74
Russo, Bene, Verma, and Arora, op. cit.
75
Kelly, op. cit.
76
Ibid.
77
Lakhani, Iansiti, and Herman, op. cit., 2.
78
Ibid.,19.
79
“Meet Bill Ruh, CEO of GE Digital,” YouTube video, 6:47, published by GE Digital, June 28, 2016, accessed August 19,
2019 www.youtube.com/watch?v=D6jrHqnE4kI.
80
Michael Sheetz, “GE Shares Fall as SEC and DOJ Expand Investigations of Accounting Practices,” CNBC, October 30,
2018, accessed April 16, 2019, www.cnbc.com/2018/10/30/ge-says-sec-expanding-scope-of-ongoing-accountinginvestigation-shares-fall.html; United States District Court Southern District of New York, “Case 1:18-cv-04746-UA,” 11, May
30, 2018, accessed April 26, 2019, www.courthousenews.com/wp-content/uploads/2018/06/Generalelectric.pdf.
81
Thomas Gryta, “SEC Has Opened Probe of GE’s Accounting,” Wall Street Journal, January 24, 2018, accessed April 16,
2019,
www.wsj.com/articles/ge-shows-long-road-ahead-in-restructuring-push-1516797761?mod=article_inline;
“GE
Completes Acquisition of Alstom's Power and Grid Businesses,” General Electric, press release, November 2, 2015, accessed
April 19, 2019, www.ge.com/power/about/alstom-acquisition; General Electric, FQ3 2018 Earnings Call Transcripts, 2018, 5,
accessed January 22, 2019, www.nasdaq.com/aspx/call-transcript.aspx?StoryId=4215813&Title=general-electric-ge-q32018-results-earnings-call-transcript; Michael Rapoport, “How GE Built Up and Wrote Down $22 Billion in Assets,” Wall Street
Journal, March 13, 2019, accessed April 16, 2019, www.wsj.com/articles/how-ge-built-up-and-wrote-down-22-billion-inassets-11552469401?ns=prod/accounts-wsj.
82
Michael Rapoport, “GE Shows How ‘Black Box’ Assets Boost Profits,” Wall Street Journal, November 1, 2017, accessed
April 16, 2019, www.wsj.com/articles/ge-shows-how-black-box-assets-boost-profits-1509549624?mod=article_inline.
83
Geoffrey Colvin and Katie Benner, “GE under Siege,” Fortune, October 15, 2008, accessed April 16, 2019,
http://archive.fortune.com/2008/10/09/news/companies/colvin_ge.fortune/index.htm.
84
Thomas Gryta and Ted Mann, “GE Powered the American Century—Then It Burned Out,” Wall Street Journal, December
14, 2018, accessed April 19, 2019, www.wsj.com/articles/ge-powered-the-american-centurythen-it-burned-out-11544796010.
85
Thomas Gryta, Joann S. Lublin, and David Benoit, “How Jeffrey Immelt’s ‘Success Theater’ Masked the Rot at GE,” Wall
Street Journal, February 21, 2018, accessed April 16, 2019, www.wsj.com/articles/how-jeffrey-immelts-success-theatermasked-the-rot-at-ge-1519231067?mod=article_inline; Matt Egan, “GE’s Legal Troubles Are Mounting,” CNN Business,
February 26, 2018, accessed April 19, 2019, https://money.cnn.com/2018/02/26/investing/ge-legal-problems/index.html.
86
General Electric, 2018 Annual Report, 8, 2019, accessed April 16, 2019, www.ge.com/investorrelations/sites/default/files/GE_AR18.pdf.
87
“GE Completes Acquisition of Alstom’s Power and Grid Businesses,” op. cit.
88
Gryta and Mann, op. cit.
89
Ibid.
90
Ibid; “Meet Bill Ruh, CEO of GE Digital,” op. cit.
91
Drake Bennett, “How GE Went from American Icon to Astonishing Mess,” Bloomberg Businessweek, February 1, 2018,
accessed July 26, 2019, www.bloomberg.com/news/features/2018-02-01/how-ge-went-from-american-icon-to-astonishing-mess.
92
Colvin, “What the Hell Happened at GE?,” op. cit.
Page 123 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Page 19
Page 20
9B19M110
93
Gryta, Lublin, and Benoit, op. cit.
Brooke Sutherland, “GE’s $23 Billion Writedown Stems from a Bad Bet on Fossil Fuels,” Bloomberg Businessweek, October
22, 2018, accessed April 16, 2019, www.bloomberg.com/news/articles/2018-10-22/ge-s-23-billion-writedown-stems-from-a-badbet-on-fossil-fuels; Fraunhofer Institute for Solar Energy Systems, ISE, Photovoltaics Report, 5, March 14, 2019, accessed April
15, 2019, www.ise.fraunhofer.de/content/dam/ise/de/documents/publications/studies/Photovoltaics-Report.pdf.
95
Gryta, Lublin, and Benoit, op. cit.; General Electric, Powering Forward: GE’s Record Setting HA Gas Turbine Ignites a New
Era of Power Generation, September 2018, accessed April 19, 2019, www.ge.com/content/dam/gepowerpgdp/global/en_US/documents/product/gas%20turbines/White%20Paper/GEA33853%20HA_Gas_Turbine_Fleet_RampUp_
Whitepaper_R15.pdf; Russell Stokes, “Making the Best Turbines Is Hard Enough. At GE, We Never Stop Making Them
Better,” LinkedIn, September 19, 2018, accessed April 16, 2019, www.linkedin.com/pulse/making-best-turbines-hard-enoughge-we-never-stop-them-russell-stokes/.
96
General Electric, GE Capital: 2015 Resolution Plan Public Section, 7, 2015, accessed April 23, 2019,
www.fdic.gov/regulations/reform/resplans/plans/gecc-165-1512.pdf?mod=article_inline.
97
Ibid.
98
Gryta and Mann, op. cit.
99
Patrick Jenkins, “GE Capital Tells a Cautionary Tale for Shadow Banks,” Financial Times, April 13, 2015, accessed April
23, 2019, www.ft.com/content/802d0aee-dfa7-11e4-a06a-00144feab7de#axzz3X7uek1Ds; James B. Stewart, “Did the Jack
Welch Model Sow Seeds of G.E.’s Decline?,” New York Times, June 15, 2017, accessed July 26, 2019,
www.nytimes.com/2017/06/15/business/ge-jack-welch-immelt.html.
100
Matt O’Brien, “Financial Reform Is Working, Kind Of: GE Doesn’t Want to Be a Bank Anymore,” Washington Post, April 14,
2015, accessed April 23, 2019, www.washingtonpost.com/news/wonk/wp/2015/04/14/americas-industrial-giant-is-changingits-mind-about-being-a-finance-company/?noredirect=on&utm_term=.bb9373b65336.
101
Gryta and Mann, op. cit.
102
Egan, “GE’s Legal Troubles Are Mounting,” op. cit.
103
Rick Clough, “GE Finalizes $1.5 Billion DOJ Settlement over Old Subprime Unit,” Bloomberg, April 12, 2019, accessed
April 23, 2019, www.bloomberg.com/news/articles/2019-04-12/ge-finalizes-1-5-billion-doj-settlement-over-old-subprime-unit.
104
Jonathan Stempel, “GE’s WMC Mortgage Unit, Felled by Financial Crisis, Files Chapter 11 Bankruptcy,” Reuters, April 23,
2019, accessed April 23, 2019, https://kfgo.com/news/articles/2019/apr/23/ges-wmc-mortgage-unit-felled-by-financial-crisisfiles-chapter-11-bankruptcy/.
105
“GE Hires Bankers to Mull Sale of Digital Assets: WSJ,” Reuters, July 30, 2018, accessed April 11, 2019,
www.reuters.com/article/us-ge-divestiture/ge-hires-bankers-to-mull-sale-of-digital-assets-wsj-idUSKBN1KK24S.
106
Lohr, “G.E. to Spin Off Its Digital Business,” op. cit.
Page 124 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
94
9B20M067
R. Chandrasekar wrote this case under the supervision of Professor Ning Su solely to provide material for class discussion. The
authors do not intend to illustrate either effective or ineffective handling of a managerial situation. The authors may have disguised
certain names and other identifying information to protect confidentiality.
This publication may not be transmitted, photocopied, digitized, or otherwise reproduced in any form or by any means without the
permission of the copyright holder. Reproduction of this material is not covered under authorization by any reproduction rights
organization. To order copies or request permission to reproduce materials, contact Ivey Publishing, Ivey Business School, Western
University, London, Ontario, Canada, N6G 0N1; (t) 519.661.3208; (e) cases@ivey.ca; www.iveycases.com. Our goal is to publish
materials of the highest quality; submit any errata to publishcases@ivey.ca. i1v2e5y5pubs
Copyright © 2020, Ivey Business School Foundation
Version: 2020-04-27
In late July 2018, Faisal Kazi, president and chief executive officer (CEO) of Siemens Canada, was chairing
a meeting with his direct reports at the company’s headquarters in Oakville, Ontario. Siemens Canada was
a fully owned subsidiary of the multinational Siemens AG. The parent company was to unveil, in a few
days, a new growth plan entitled Vision 2020+. It was meant to be a mid-course adjustment of the five-year
growth plan Vision 2020, which the company had launched in October 2014. The adjustment was necessary
because, in a display of employee commitment to the growth plan, the goals of Vision 2020 had been
realized two years ahead of schedule (see Exhibit 1). Kazi explained,
As a core team, we now need to address two issues. First, how can we, at Siemens Canada, get
ready to execute Vision 2020+? What are the additional skills and competencies needed locally to
do so? Second, how can we, at Siemens Canada, add value to the new growth plan? What local
initiatives on our part could enhance the global objectives of Siemens AG?
SIEMENS AG
Siemens AG was founded by Ernst Werner Siemens in October 1847 in the city of Berlin in northeastern
Germany. Ernst Siemens had recently invented a telegraph that replaced the Morse code with a needle that
pointed to the right letter. Known as the “electric pointer telegraph,” the invention led him to set up a
workshop producing pointer telegraphs. He was soon joined by Johann Georg Halske, a precision mechanic,
and together they hired 10 artisans and set up a firm called Telegraphen-Bauanstalt von Siemens & Halske.1
It was a predecessor to Siemens AG, and soon the three brothers of Ernst Siemens came on board. Siemens
AG went on to become a family enterprise “born in Germany, raised in Europe, and at home in the world.”2
The company had seen—and prospered through—the advent of locomotives, electricity, the internal
combustion engine, steam turbines, jet engines, personal computers, and clean energy. Siemens had not
1
Insa Wrede, “No Time to Rest as Siemens Prepares for the Future,” DW, November 15, 2017, accessed March 2, 2019,
www.dw.com/en/no-time-to-rest-as-siemens-prepares-for-the-future/a-41394998.
2
Joe Kaeser, “Ownership Culture—The Code for Sustainable Success,” St. Gallen Business Review, May 13, 2015, accessed
February 17, 2019, www.stgallenbusinessreview.com/ownership-culture-the-code-for-sustainable-success.
Page 125 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
SIEMENS CANADA: DIGITAL TRANSFORMATION
Page 2
9B20M067
Siemens AG had consolidated revenues of €83,049 million4 for the fiscal year ending September 2017 (see
Exhibits 2 and 3). It had 363,000 employees worldwide, of whom 223,000 were directly involved in
manufacturing. Siemens AG had 32 joint ventures and associate companies worldwide in addition to many
subsidiaries: 129 in Germany; 519 in Europe, the Commonwealth of Independent States,5 Africa, and the Middle
East; 165 in the Americas; and 218 in Asia and Australia. The company had six business divisions: Power and
Gas, Energy Management, Building Technologies, Mobility, Digital Factory, and Process Industries and Drives.
Siemens AG also had two autonomous divisions: Healthineers and Renewables (see Exhibit 4).
Siemens AG had been operating in Canada since 1912. The Canadian subsidiary was involved in several
lines of business, including communications systems, power generation, industrial and building automation,
medical technology, railway vehicles, and water treatment systems. It employed 4,500 people in 39 offices
and 14 production facilities across the country. It had revenues of CA$2.2 billion6 in 2018.
Vision 2020
In October 2014, under CEO Joe Kaeser, who had taken over a year earlier, Siemens AG had announced a
five-year growth plan. Vision 2020 was the company’s biggest reorganization in the past 25 years. It
repositioned the company along three businesses that had a common value chain: electrification,
automation, and digitalization. Each business was a mega market with high growth rates, but digitalization
was the fastest growing of the three. Electrification was the company’s traditional competence, and Siemens
was a world leader in automation. The company was now building on that platform to move into the age of
digitalization in order to lead from the front.
Vision 2020 had been triggered by two developments. First, the company’s major competitors were
growing faster and achieving higher profit margins faster than Siemens AG, leading to a realization that
Siemens was not living up to its full potential. Second, its long-standing business domains were under attack
by software companies, such as SAP Software Company, Microsoft Corporation, IBM Corporation, Google
LLC, and Apple Inc., which were venturing into manufacturing technologies.
The growth plan was designed to strengthen Siemens AH so that it could “consistently occupy attractive
growth fields, reinforce its core business, and outpace competitors in efficiency and performance” and
become entrepreneurial by “applying the virtues of a family enterprise to a multinational.”
Vision 2020 had seven overarching goals that were to be monitored regularly: (1) cut costs by €1 billion;
(2) tap growth fields and get underperforming businesses back on track; (3) secure a return of 15 to 20 per
cent on capital employed; (4) position 30 per cent of Division and Business Unit managers outside of
Germany; (5) ensure 20 per cent improvement in the Net Promoter Score; (6) achieve a 75 per cent approval
3
“We Need to See Calm Restored,” Der Spiegel, August 7, 2013, accessed February 16, 2019,
www.spiegel.de/international/business/spiegel-interview-with-siemens-ceo-joe-kaeser-a-915314.html.
4
€ = euro; €1 = US$1.20 on September 8, 2017, accessed March 10, 2020, www.poundsterlinglive.com/best-exchangerates/euro-to-us-dollar-exchange-rate-on-2017-09-08.
5
The Commonwealth of Independent States was a regional intergovernmental organization formed after the dissolution of the
Soviet Union in 1991. Its nine-member states comprised Armenia, Azerbaijan, Belarus, Kazakhstan, Kyrgyzstan, Moldova,
Russia, Tajikistan, and Uzbekistan.
6
All dollar-denominated currency amounts are in CA$ unless specified otherwise.
Page 126 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
only survived technological revolutions but had also led some of these revolutions. Siemens had also missed
opportunities, such as in telecommunications. The company had underestimated the Internet.3
Page 3
9B20M067
rating in the categories of Leadership and Diversity globally; and (7) increase the number of employee
shareholders by 50 per cent.
Vision 2020 had three milestones: drive performance (2015), strengthen the core (2017), and scale up
(2020) (see Exhibit 4).
Vision 2020+ was designed to realign the global-local balance at Siemens AG by providing greater
autonomy to the subsidiaries. Its overarching purpose was to make Siemens nimbler and more profitable in
the digital industrial age and to sharpen customer focus.
Vision 2020+ was founded on two pillars. First, the company would consolidate its product portfolio into
three operating companies—gas and power, smart infrastructure, and digital industries—and into three
strategic companies—Siemens Alstom, Siemens Gamesa, and Siemens Healthineers. Second, the Munich
headquarters (HQ) would devolve decision-making to the operating companies, which would henceforth
be reporting to regional HQs at Houston, Texas, in the United States (for gas and power), at Zug,
Switzerland (for smart infrastructure), and at Nuremberg, Germany (for digital factories). The role of the
Munich office would be limited to finance, governance and markets, legal and compliance, human
resources, and communications. The new structure was to become effective October 2018.
Digitalization: Internal
Siemens formally began its digital journey in 2007 when, instead of outsourcing software skills as in the
past, it started building a pool of in-house talent through acquisitions. By 2018, the company had spent
more than US$10 billion in acquiring small and medium software companies possessing niche capabilities.7
By 2017, it employed 17,500 qualified software engineers who were involved in various streams of activity
in the company’s sprawling operating divisions worldwide.
One of the major activities involved developing software to be embedded into the company’s products. Each
Siemens product had inbuilt sensors, and each sensor had a story to tell. Every moment, the sensors delivered
data to the embedded software both for real-time analytics and for storage in the cloud for later retrieval.
Historically, Siemens was focused on tracking markets because the markets for many of its products were
cyclical, ranging from two to seven years. Since 2007, Siemens was also focused on supporting customer
requirements by means of data because data helped in recognizing changes in the markets early on and
responding better to them. The data helped foresee, for example, how the customers’ customers were
changing; it was a new competitive advantage.
The launch of the Digital Factory division, effective October 2014 as part of Vision 2020, denoted a hightech resurgence at Siemens. For many years, the company had been involved in enabling its industrial
customers to go to market faster with their products; however, doing so without incurring huge costs was a
perennial challenge, both for its customers and for Siemens as a facilitator. Siemens engineers started
looking at a seemingly fictional scenario: enabling an industrial customer to deep dive into manufacturing
7
Sarmad Khan, “Siemens Scouts Market for More Software Business Acquisitions,” The National, September 14, 2018,
accessed February 25, 2019, www.thenational.ae/business/technology/exclusive-siemens-scouts-market-for-more-softwarebusiness-acquisitions-1.770108.
Page 127 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Vision 2020+
Page 4
9B20M067
Digital twins, involves creating an integrated virtual environment of a product or a manufacturing
process into the virtual world, making as many changes or versions to optimize and validate the
best solutions before applying them back into the real world of manufacturing. It involves creating
a simulated model, based on Siemens software(s), and tweaking it in terms of optimizing
engineering, identifying bottlenecks, and monitoring performance metrics like quality, uptime and
load time. The model replaces the conventional destructive testing with non-destructive testing, as
everything can now be perfected virtually prior to real world implementation.
A major prerequisite for full digitization was complete integration. The product needed to be availably fully
digitally; the production line needed to be available fully digitally; and the performance metrics from the
real world had to be monitored fully digitally. It was only in a digital ambience that the product-productionperformance process could be fully integrated to deliver an end to end digital transformation for a customer.
Digital Enterprise was thus a market-led innovation practice that provided a powerful incentive for Siemens
to embark on a journey of digitization both internally in its own operations and externally to help customers.
By 2017, every division of Siemens had a digital enterprise strategy with clearly defined mandates often
implemented by cross-divisional teams. The company was practising what it was preaching to customers.
Siemens had several digital tools in place: It had implemented customer relationship management systems.
It was deploying cloud computing. It was utilizing artificial intelligence tools in recruitment. It had digitized
sales quotations and employee travel/expense statements.
The company had several training tools to ensure that employees stayed tuned to the current trends in
digitalization. It offered regular inhouse webinars on becoming digital. The company had an internal
Facebook page called Siemens Social Network, which served as an information exchange point on topical
issues, including digitalization. Siemens Canada offered an ongoing two-day course in digitalization for its
divisional managers, business unit heads, and all employees. According to Jeff Phillips, a manufacturing
engineering manager at Siemens Canada,
Siemens has several ongoing digital projects. Engineers at the company’s plant at Peterborough,
for example, have been working on two applications for MindSphere. They are known as overall
equipment efficiency (OEE) and cause and effect manager (CEM). OEE is a performance metric
indicating how well the Peterborough plant is using its capital assets. The metric is a multiple of
three coefficients—availability of equipment, performance of equipment, and quality of output. A
drop in OEE means a problem with the production line requiring resolution. CEM is a
supplementary metric flowing from OEE. Another project, for example, that has been in play at
Peterborough is the digitization of the manual components of a typical sales order flow.
Digitalization: External
The stimulus for digitalization at Siemens came from its industrial customers who were keen on reducing
their costs of going to market. Siemens had provided the solution with its digital twin. The customers had
other needs that digitalization could address. Mass production was being gradually replaced by
customization. In mass production, scale meant that, for example, if a factory produced 5,000 high-
Page 128 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
not only without setting up physical infrastructure (e.g., a production line) but also without even having a
finished product on the table. They found the answer in digital twins. According to Shalabh Bakshi, Siemens
chief digital officer,
Page 5
9B20M067
The company’s customers had other questions: How could sensors be deployed better? How could data
points be reconfigured to facilitate remote decisions? How could augmented reality be used to run remote
diagnostics? Siemens was experimenting with different digital technologies in a bid to provide the answers.
According to Vishal Gandhi, a digital enterprise coordinator for process industries in Canada,
Industrial manufacturers are under pressure. They are looking for ways of increasing productivity.
They are under compulsion to move from mass production to customized production. Given that
the innovation cycles are very short nowadays for the manufacturers, they’re looking at how
quickly they can come up with new products. The onus on seeking differentiation in the
marketplace is heavy.
A major limitation for a facilitator such as Siemens AG was that manufacturing environments were not
homogenous. Key performance indicators (KPIs) varied from one customer to another. The stage of life
cycle at which a customer would commence the digital journey also differed. An understanding on the part
of the customer of where they wanted to be in five or 10 years was crucial because digitalization was more
like a marathon than a sprint. Siemens was keen on an approach it called product life-cycle management
(PLM), whereby customers treated digitalization as a journey and not an end. However, this approach did
not always coincide with customer expectations.
Digitalization also meant organizational transformation. Not all people in a client company shared a
common understanding of what was going on. Not all were motivated or incentivized to change the way
they worked. These were the barriers in working with customers on their digitalization.
To overcome these barriers, Siemens had developed an integrated suite it called Teamcenter. It was a PLM
system that connected people and processes across functional silos with a digital thread for innovation.
Combined with the use of a digital twin, Teamcenter generated flexibility in new product development.
Several companies were involved in multi-year implementations of digital solutions provided by Siemens,
including aerospace companies such as Bombardier, automotive companies such as Maserati, and several
companies in the injection moulding business.
A highlight of digitalization at Siemens was an internally developed software product called MindSphere,
a cloud-based operating system that was offered as a platform-as-a-service. It connected Siemens products,
plants, systems, and machines through what was known as the Internet of Things (IoT) and used advanced
analytics to harness the data generated by the IoT. MindSphere had an open architecture, enabling
engineers, both at Siemens and at its customers, to collect data, store it, and then write mobile applications
(apps) of their own to process that data.
Digitalization products were also becoming an independent source of revenue for Siemens. MindSphere was
being marketed as a stand-alone product at a one-time registration fee of €2,000 and an annual subscription
fee of €640. There was also the possibility of a secondary source of revenue by providing MindSphere as an
add-on to a company’s process instrumentation products. MindSphere core value proposition was that it could
monitor processes, track key metrics, and help schedule predictive maintenance.
Page 129 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
temperature blades, the unit costs would be lower, but if it produced only 50 blades, the unit costs would
be higher. With digitalization, the blade would have a digital twin that could be produced in small-volume
prototypes using 3-D printing. Digitalization not only took scalability out of the equation but, by bridging
the gap between scale and scope, enabled the manufacture of different types of blades. An industrial
customer could thus produce a lot size of one and still be viable. This ability represented a leap forward in
manufacturing technology facilitated by digitalization.
Page 6
9B20M067
Ownership Culture
The message played out at the ground level in three ways. First, 144,000 employees of the company
worldwide were shareholders of Siemens. The stated target, as part of Vision 2020, was to increase
employee shareholding by at least 50 per cent, to more than 200,000. Equity stakes ensured that employees
shared the company’s progress year to year through both stock appreciation and dividends. Second,
ownership encouraged employees to act entrepreneurially, not necessarily in making an investment decision
but in their day-to-day interactions, however small. It showed up in employees asking routine questions
such as the following: Is the customer happy? How can I improve my performance? How can I make this
process more efficient? Third, Siemens was giving priority, as in a family enterprise, to financial solidity
over short-term profit, which was evident both in the company’s financial targets and its compensation
systems, which were oriented toward the creation of sustainable value.
The message had gained credibility because it was being reinforced by the company’s legacy of having
“close and trusted partnerships with vendors and customers, often over generations.”10
Innovation
Siemens had begun as a start-up in a backyard in Berlin. Successive leadership teams believed that a piece
of that spark of entrepreneurship resided inside each employee. But it was buried deeply for two reasons.
First, sectors such as utilities and oil and gas were regulated by governments, and the companies operating
in these sectors, such as Siemens, were prone to being conservative. In fact, it was only when Siemens
entered the healthcare business that it could begin to break out of its conventional template and replicate
some of its innovations in health care and other verticals. Second, as it had grown over the decades into a
multinational conglomerate, the company needed to comply with statutory requirements around fiduciary
duties, financial controls, provisions of the Sarbanes–Oxley Act, and corporate governance norms.
Conformity was a test of corporate citizenship.
The leadership team, led by company veteran Joseph Kaeser as CEO, made innovation the nucleus of its
growth agenda. For reasons of size and others, Siemens could not pretend to be a start-up, but it could
become a sought-after partner for start-ups. It could help them become what they aspired to be: a company
as successful as Siemens. That was the core positioning Siemens was seeking through innovation.
A major part of Siemens internal messaging, of late, was tolerance for mistakes. In reiterating that
employees should act as if Siemens was their own company, Kaeser made a distinction between acceptable
and unacceptable mistakes. An acceptable mistake occurred when one could not predict the outcome,
which, after due consideration, could go either way. An unacceptable mistake occurred when one knew for
8
Kaeser, op. cit.
Ibid.
10
Ibid.
9
Page 130 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
A message that was regularly reiterated at various levels of leadership at Siemens was that the employees
owned the company. It was borne out of a conviction that employees were more loyal to culture than to
strategy and that the best way to secure buy-in for strategy was to ensure that culture and strategy were
mutually reinforcing. It was part of the mandate of the global CEO that Siemens should incorporate “the
virtues of a family enterprise.”8 There was also recognition on the part of leadership that “culture change is
a marathon, not a sprint.”9
Page 7
9B20M067
Two models of innovation ran parallel at Siemens, which, as a global company, was oriented along products
manufactured by individual business units. The traditional research and development (R&D) model of
innovation was decentralized to the business units. Several units comprised a business division, each of
which had a global headquarters. The second model of innovation, the open model, was largely conducted
in conjunction with customers by business units. According to Ann Adair, vice-president of strategy at
Siemens Canada,
The bridge that links the R&D model and open innovation model is an entity known as portfolio
owner, which I think is unique to Siemens. Portfolio owners have global span of control and could be
stationed anywhere in the world, depending upon the products in which they would be specialized.
They would be anchoring business decisions around which product in the pipeline should be
commercially scaled and in which part of the world the product under review should be manufactured.
The beginnings of customer-led innovation could be traced to the company’s energy division, which dealt
directly with utility companies as an equipment vendor. Traditionally, a utility company would put together
a capital investment plan, usually on an annual basis, to announce major investments for the period. It would
then appoint general contractors (GCs) to execute projects. Vendors such as Siemens were required to work
with each GC to provide the hardware and software needed for each project. They would be enlisted only
when the utility company had already decided on the crucial factors, such as project financing and the
performance metrics. Siemens was often successful in becoming involved in a project during its early stages
of strategy development.
Customer-led innovation became known internally as Customer Value Co-creation (CVCC). At its core
was the blend of Siemens’s domain knowledge at Siemens and the client’s expertise at creating a product
that would benefit both parties. CVCC had a well-defined process around how Siemens would work with
a customer, understand their business drivers, empathize with their pain points, and get a sense of their KPIs
before arriving, together, at a minimum viable product for scaling up. An issue that often required
deliberation was the ownership of the intellectual property that would be jointly developed.
An example of a successful CVCC was the collaboration between Siemens Canada and NB Power, the
electrical utility in the province of New Brunswick. Siemens Canada provided its expertise in the smart grid
compass to help its long-standing customer develop an energy road map for the future. A result of the
successful initiative was that Siemens Canada also received funding from the federally regulated Strategic
Innovation Fund to conduct a pilot project of smart grid solutions in Atlantic Canada.
Siemens had several innovation tools in place. For example, every year its Energy division held a Dragons’
Den12–style program for three months, where employees could present their new portfolio ideas. The
winners received investment funding to develop their products. The company had a Corporate Technology
group, whose primary role was to scan the industrial environment to recommend focus areas for R&D, but
also funded innovative ideas from within. The group had built up enough credibility to bypass the
company’s signatory decision-making processes to quickly approve new product development projects that
11
“Viewpoints: Joe Kaeser of Siemens,” WSJ video, 32:13, January 12, 2016, accessed February 18, 2019,
www.wsj.com/video/viewpoints-joe-kaeser-of-siemens/856BABEE-6065-43B7-9325-51DEB8DA022B.html.
12
Dragons' Den, CBC, accessed February 18, 2019, www.cbc.ca/dragonsden/m_episodes.
Page 131 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
certain the outcome but did not want to go ahead because the route was untested. Kaeser had also made it
clear that people who made a mistake by trying to do their best would never be in trouble; but if people did
not ask for help, they would be in trouble because someone somewhere in the Siemens universe had likely
not only dealt with the same issue but also would have been eager to help.11
Page 8
9B20M067
In addition, Siemens had the 3I Initiative, which invited, once or twice a year, proposals from around the globe
for scaling up new ideas, new technologies, and new solutions on which the applicants had already started
working. Another program, the Next47, looked for promising start-ups, both internally and externally.
Siemens was also involved in creating an ecosystem of collaboration. For example, if a manufacturing firm
in Ontario wanted to implement automation but was unsure how to get started, Siemens Canada would help
find industry funding, connect the firm with a vocational training college in which Siemens would have
installed its equipment, and help the company manage the transition to automation.
Realigning the Global-Local Balance
With operations spanning more than 200 countries and regions, Siemens was a global enterprise. In 2017,
it generated 87 per cent of its revenue outside Germany. With a range of disparate businesses in its portfolio,
it was also a conglomerate.
CEO Kaeser made it clear, however, that the days of conglomerates were over. Businesses that were not
core to electrification, automation, and digitalization were being spun off. Health care and wind energy
were among the businesses that were being cut loose. It was a sharp contrast to the company’s history,
whereby getting bigger used to be the answer to most of its strategic challenges. As part of Vision 2020,
the “lumbering aircraft carrier,” as the CEO called it, was yielding place to a “fleet of ships,” each thriving
by its own means and moving speedily and purposefully in its focus area.13
The company was also striking a balance between pursuing its global aspirations and meeting local needs.
Country subsidiaries were gaining more leeway in fulfilling the growth priorities set by HQ and identifying
their own growth opportunities. Local growth opportunities were given shape and substance in a document
known internally as the Country Opportunity Plan and reviewed monthly by senior managers at the country
level. When a formal proposal was ready, it was presented by the country’s CEO to senior management at
HQ for approval. A key factor in getting approval was preparing the ground though informal discussions
and information sharing with HQ staff.14
Siemens had two channels for escalating a local solution globally: Corporate Technology located at HQ,
which looked at innovations in different spaces and in different areas of Siemens, and portfolio owners
located in different parts of the globe. The latter reported to product managers who were also located in
different parts of the globe. For example, the Peterborough plant in Canada housed portfolio owners
specializing in level-sensing products. They reported to the product management group based in Karlsruhe,
Germany. Similarly, the product management group for pressure transducers was based in the United States,
and the product management group for flow products was based in Germany.
13
Chris Bryant, “Siemens Launches a Thousand Ships,” Bloomberg Businessweek, August 3, 2017, accessed March 2, 2019,
www.bloomberg.com/news/articles/2017-08-03/siemens-launches-a-thousand-ships.
14
Paul Boothe and Jean-Louis Schaan, Core Manufacturing: Lessons from Four Global Giants in Canada, (London, ON:
Lawrence
National
Centre
for
Policy
and
Management,
2016),
accessed
March
2,
2019,
www.ivey.uwo.ca/cmsmedia/2677918/core-manufacturing-lessons-from-four-global-giants-in-canada.pdf.
Page 132 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
it identified as being worthwhile. Siemens also had a division called Innovation AG, wherein anyone, both
inside and outside the company, could file a business case. The division was insulated from the rest of
Siemens and acted as a typical venture capitalist or a business angel to help potential entrepreneurs start
and grow a business. It had extensive networks outside Siemens and access to resources within Siemens.
Page 9
9B20M067
Part of the Siemens strategy for digitalization was based on Digital Lighthouse Factories, the global
repositories of flagship projects that provided expertise to country subsidiaries that were executing similar
projects. These repositories were the first to be called on within Siemens for help. An example was the
process analytics factory at Karlsruhe, Germany, which was the lead for MindSphere. The Karsruhe unit
coordinated the MindSphere Application Centers (MACs), which were based in 36 locations worldwide,
ranging from Austin, Texas, in the United States, to Shanghai, China. Each location focused on a market
vertical.15 All MACs were doing virtual sprints in developing apps such as the OEE.
FUTURE PLANS OF SIEMENS CANADA
Vision 2020+ led to a devolution of decision-making powers from the corporate office in Munich to the
three regional headquarters: at Houston, Texas, in the United States (for gas and power); at Zug, Switzerland
(for smart infrastructure); and at Nuremberg, Germany (for digital factories). The subsidiaries would report
to one of the three regional HQs rather than to Munich, whose role would be limited to finance, governance
and markets, legal and compliance, human resources, and communications. Kazi clarified,
Our global CEO has outlined the broad direction in which the company will move forward. We will
be looking for growth options within that mandate. We will be watching megatrends. Urbanization,
for example, is one of them. By 2050, 70 per cent of the world’s population is forecast to live in
cities. Suburban transportation is a window of business opportunity for us in Canada. On the
technology front, I am excited by machine learning and artificial intelligence. Data analytics is
another area where we will be looking for new business opportunities. Meanwhile, we need to answer
several key questions. How do we incorporate best practices from the Siemens universe? How do we
generate best practices locally for absorption by subsidiaries worldwide? How do we innovate faster
locally? How do we measure up at Siemens Canada to the entrepreneurial freedom being given to
subsidiaries as part of Vision 2020+?
15
Siemens, “MindSphere Application Centers,” accessed October 17, 2019, https://new.siemens.com/global/en/products/
software/mindsphere/application-centers.html.
Page 133 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
An example of a local initiative triggering a global mandate was the Smart Grid Centre of Competence,
opened in January 2013 in Fredericton, in the Canadian province of New Brunswick. It was borne out of a
successful collaboration between Siemens Canada and NB Power. The team that worked on the NB Power
project also worked on other global projects involving collaboration with utility companies.
Page 10
9B20M067
Note: FY = fiscal year; € = euro; €1 = US$1.20 on September 8, 2017; bn = billion; ROCE = return on capital employed.
Source: Company files.
EXHIBIT 2: SIEMENS AG—CONSOLIDATED INCOME STATEMENT, 2013–2017
Year ending in September (in € millions)
Revenue
Cost of goods sold
Gross profit
Research and development
Selling and general administration
Other operating income
Other operating expenses
Income from investments
Interest income
Interest expenses
Other financial income
Income tax
Income from continuing operations
Income from discontinued operations
Net income
2017
83,049
(58,021)
25,029
(5,164)
(12,225)
647
(595)
43
1,487
(1,051)
135
(2,180)
6,126
53
6,179
2016
79,644
(55,826)
23,819
(4,732)
(11,669)
328
(427)
134
1,314
(989)
(373)
(2,008)
5,396
188
5,584
Note: € = euro; €1 = US$1.20 on September 8, 2017.
Source: Company files.
Page 134 of 282
2015
75,636
(53,789)
21,847
(4,483)
(11,409)
476
(389)
1,235
1,260
(818)
(500)
(1,869)
5,349
2,031
7,380
2014
71,227
(50,869)
20,357
(4,020)
(10,190)
654
(194)
582
1,058
(754)
(177)
(2,014)
5,302
215
5,507
2013
73,445
(53,309)
20,135
(4,048)
(10,869)
500
(424)
510
947
(784)
(154)
(1,634)
4,179
231
4,409
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
EXHIBIT 1: VISION 2020+ AT SIEMENS AG—BUILDING ON VISION 2020
Page 11
9B20M067
Year ending in September (in € millions)
By business division
Power and Gas
Wind Power and Renewables
Energy Management
Building Technologies
Mobility
Digital Factory
Process Industries and Drives
Healthineers
Healthcare
Siemens Gamesa Renewable Energy
Energy
Industry
Infrastructure and Cities
Sub-total: industrial business
Sub-total: financial services
Reconciliation with consolidation
Total revenue
By location
Germany
Rest of Europe, CIS, Africa and Middle East
Americas
Asia, Australia
Total revenue
2017
2016
2015
2014
2013
15,467
–
12,277
6,523
8,099
11,378
8,876
13,789
–
7,922
16,471
–
11,940
6,156
7,825
10,172
9,038
13,535
–
5,976
13,193
5,660
11,922
5,999
7,508
9,956
9,894
–
12,930
–
12,720
5,567
10,708
5,569
7,249
9,201
9,645
–
11,736
–
84,331
921
(2,202)
83,049
81,112
979
(2,447)
79,644
77,062
1,048
(2,475)
75,636
72,396
937
(2,106)
71,227
–
–
–
–
–
–
–
–
12,819
–28,797
16,688
21,894
80,198
1,072
(1,699)
73,445
11,142
32,225
23,516
16,166
83,049
10,739
31,080
22,707
15,118
79,644
11,244
27,555
21,702
15,135
75,636
11,244
23,156
21,702
15,125
71,227
73,445
Note: €1 = US$1.20 on September 08, 2017; CIS = Commonwealth of Independent States.
Source: Company files.
EXHIBIT 4: SIEMENS AG AS OF OCTOBER 1, 2014
Source: Company files.
Page 135 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
EXHIBIT 3: SIEMENS AG—REVENUE BREAKDOWN, 2013–2017
9 -6 1 0 -0 3 2
REV: MAY 31, 2012
KARIM R. LAKHANI
DAVID A. GARVIN
TopCoder (A): Developing Software through
Crowdsourcing
In December 2009, Jack Hughes, CEO and founder of TopCoder Inc., entered his company’s
headquarters in Glastonbury, Connecticut, eager to review a particularly complex software
development project for an energy firm’s dynamic power pricing system. Eight years after founding
TopCoder, Hughes still enjoyed detailed project reviews. He was particularly proud that his
company could produce high-quality software solutions for which his own employees did not have
to write a single line of code. Instead, the firm nurtured a global community of more than 225,000
programmers who competed to design and create software modules for TopCoder clients, a process
that the popular press called crowdsourcing.1
Hughes smiled at the project’s success. The resulting software code was bug-free and operational
on its first day, a rarity in the software industry. Especially impressive to Hughes was that in four
months, 65 participants from 11 countries on six continents had competed in 57 contests to create this
critical pricing system for the client (see Exhibit 1). As of 2009, TopCoder routinely produced
software solutions for over 45 clients, including AOL, Best Buy, Eli Lilly, ESPN, GEICO, and the
Royal Bank of Scotland.
In the past eight years, Hughes had refined TopCoder’s business model to accommodate ongoing
changes in the software industry, while also pursuing its unique competition-based software
development approach. He had transitioned his business from a model that helped other software
firms identify “top coders” to a company that developed custom software through a combination of
traditional IT consulting services and competitions, mobilizing developers world-wide to solve
clients’ problems.
The shift to a greater emphasis on competitions, encompassing all aspects of software
development, however, meant that project volume was a growing issue for TopCoder. Hughes had to
think through how a competition-based business model, which increasingly stressed contests as an
organizing as well as money-making approach, could handle increases in numbers of competitions,
clients, and participants. Hughes considered his own goal: attaining $200 million in revenue from a
high of just over $18 million in 2008. He fundamentally believed that contest demand would spur the
supply of TopCoder participants, who would in turn create high-quality software solutions. But, was
1 Jeff Howe, “The Rise of Crowdsourcing,” Wired Magazine 14.06, June 2006.
________________________________________________________________________________________________________________
Professors Karim R. Lakhani and David A. Garvin and Research Associate Eric Lonstein prepared this case. HBS cases are developed solely as
the basis for class discussion. Cases are not intended to serve as endorsements, sources of primary data, or illustrations of effective or ineffective
management.
Copyright © 2010, 2011, 2012 President and Fellows of Harvard College. To order copies or request permission to reproduce materials, call 1-800545-7685, write Harvard Business School Publishing, Boston, MA 02163, or go to www.hbsp.harvard.edu/educators. This publication may not be
digitized, photocopied, or otherwise reproduced, posted, or transmitted, without the permission of Harvard Business School.
Page 136 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
ERIC LONSTEIN
610-032
TopCoder (A): Developing Software through Crowdsourcing
$200 million in revenue a reasonable goal? Did his assumptions make sense? If so, what would it take
to increase revenues by over an order of magnitude?
Before he founded TopCoder in 2001, Hughes had built a custom software development2
company, Business Data Services, in 1985; the company name changed to Tallan in 1991. Tallan
employed some 600 people before being sold to CMGI in 2000.3 As he was completing the
transaction, Hughes reflected on what he had learned from his experiences at Tallan—the experiences
that would inspire the core tenets of the TopCoder business model. Although Hughes enjoyed his
time at Tallan, the company struggled in some areas. For example, recruitment was an expensive and
frustrating process because finding qualified programmers was time-consuming and talent was
difficult to assess. Due to constantly evolving technologies, programmers’ skill sets often became
obsolete after only a few years of productive service, leading to high levels of employee turnover.
Furthermore, despite Tallan’s goal of maximizing billable hours, Hughes believed there were
opportunities to save clients time and money by, for example, reusing computer programs’ basic
components instead of building each application from scratch.
Drawing upon these and other insights, Hughes set about creating a new kind of organization that
would build a “community” of programmers to help address the issues he had identified. These
programmers would compete—as well as affiliate—by building and using components that had
already been tested and found workable. The idea of reusing software components for new projects
would become the core of the solutions the new company, called TopCoder, provided. Hughes
envisioned the company as a “two-sided platform” for software development. One side of the
platform would be clients, firms that needed software developed, who would work with his staff to
specify programming challenges. The other side would be community members who would compete
in contests to create solutions to the challenges for money and skill ratings. TopCoder would be in the
middle as the platform host, designing and enforcing the rules of engagement between clients and
the community members. Pete Bourdon, TopCoder’s CFO, explained that the company needed to
excel at five core tasks: breaking down large client software projects into components, taking in and
processing client project specifications, determining appropriate contest prizes, having a consistent
and unbiased way of selecting contest winners, and fixing bugs at the back end of development.
Setting out to amass an initial collection of highly skilled programmers, from 2001 to 2003
TopCoder asked established software development companies to sponsor world-wide web-based
programming competitions. The sponsorships increased the popularity and legitimacy of TopCoder’s
competition platform and provided the company with access to talented programmers from around
the world. In return, the sponsors, including Sun Microsystems and Google, used the contests to
advertise and recruit new talent. Tanya Horgan, TopCoder’s vice president of finance, explained that
during the sponsorship phase, TopCoder offered unusually large prizes—as much as $5,000 to
$10,000 per match for tournament winners—to attract competitors and expand the community. In
addition, every contestant that participated received an objective numerical rating for their
2 Custom software development by specialist firms in the global IT consulting and services sector (for example, Accenture and
IBM) was an over $500 billion segment in 2008. (Source: “Global IT Consulting & Other Services: Industry Profile,” Data
Monitor, March 2009.)
3 Clint Boulton, “CMGI Acquires Tallan for $920 Million,” InternetNews.com, February 14, 2000, http://www.internetnews.
com/ec-news/article.php/303771/CMGI-Acquires-Tallan-for-920-Million.htm.
2
Page 137 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Background and Current Operations
TopCoder (A): Developing Software through Crowdsourcing
610-032
performance against the global talent pool, providing a clear signal to TopCoder and others about the
talent in the community.
In 2005, TopCoder began to use its community to develop software components and applications.
Hughes first tested this model by having highly rated community members compete to redesign and
rewrite the firm’s own software platform. The resulting code was higher quality and much less
expensive than TopCoder’s own internally developed solution. Hughes now had positive proof that
complex software systems could be built through competitions.
Initially, TopCoder adopted a model to create solutions for clients by contracting with community
members, running competitions, and providing consulting services. The company broke down the
software development process into seven distinct but interrelated tasks: 1) conceptualization, 2)
specification, 3) architecture, 4) component production, 5) application assembly, 6) certification, and
7) deployment. Most revenue came from consulting services: TopCoder billed clients for the time the
company’s platform managers spent conceptualizing and specifying client problems, setting up
component design and development competitions, assembling components, and delivering finished
solutions.
Shortly after TopCoder started developing software for clients, the company identified reusable
components from the software it was creating and collected the components in a catalog. These
software components became an important part of TopCoder’s value proposition to its clients. Many
of the custom applications could be produced by combining existing catalog components with new
components built through competition. TopCoder had also received eight U.S. patents for various
aspects of running online programming contests in a distributed community setting and had other
patents pending domestically and internationally.
TopCoder’s hybrid consulting model led to large increases in revenues. However, Hughes was
still dissatisfied: “I viewed the hours-based services approach to be a broken, inefficient model.” In
2007 and 2008, TopCoder produced nearly $20 million in revenue, but platform manager costs
remained high (see Exhibit 2 for information on revenue and platform manager costs). Attempting to
alleviate costs, in 2007 TopCoder introduced competition tracks for component architecture and
assembly. With these new competition tracks in place, the work traditionally done by platform
managers would now be done by the community. In 2008, the company also added competitions in
software development tasks, such as conceptualization and specification, as well as deployment and
bug fixing.
By early 2009, TopCoder had moved increasingly away from the hybrid consulting model. It now
focused on completing all tasks in software development through competitions. Instead of paying for
time and materials for TopCoder platform managers, clients paid a monthly platform fee based on
the complexity of their software requirements and the estimated number of competitions they would
run through the TopCoder platform each month. The platform fee also provided clients with
unlimited access to the over 1400 components in TopCoder’s catalog. Roughly 60% of most clients’
projects could be accomplished through reusing components from the catalog. The company coupled
the move from the hybrid consultancy model to a competition model with the reduction of many
platform manager positions, leaving the company with 16 project managers servicing 35 clients by
the end of 2009.
3
Page 138 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
By the end of 2004, the TopCoder community was 50,000 members strong. In its early efforts to
use the community to generate revenue, TopCoder acted as a placement firm, matching top-rated
community members with firms seeking new talent. Hughes, however, was dissatisfied: “I cringed at
the idea of TopCoder becoming a placement firm. That was not my end vision for the company.”
610-032
TopCoder (A): Developing Software through Crowdsourcing
The second type of competition targeted developing software applications for specific client
needs. A TopCoder platform manager initially worked with the client staff to develop a “game plan”
(see Exhibit 3 for a representative game plan) or a project road map for building the software. The
first step typically involved a contest where the general client problem was presented to the
TopCoder community in a conceptualization contest. Here contestants publicly cross-examined the
client staff as to their actual needs and then submitted a business requirements document and highlevel use cases. The client chose the submission or submissions that best represented the client’s
needs as the basis for further development. Then a series of specification contests was held to create
the application’s requirements documents, application wireframes (i.e. the logical flow of the
application), and storyboards (detailed cases of the user experience). Next, the output of the
specification contests was fed into several architecture contests to create the overall system and
component level designs. At this point, the TopCoder platform manager would work with the client
to either select components from the catalog or commission the creation of new components through
design and development competitions. After the component production phase, all the relevant
components were put together through an assembly competition with the objective of creating a
working system. Assembly was then followed by certification and testing contests and then,
eventually, deployment. Throughout the execution of the game plan, TopCoder retained flexibility in
development by running “bug races” to accommodate changing client specifications or unforeseen
errors.
To determine winners and assess quality in client software development, TopCoder used a
community-based peer-review system. In particular, expert and experienced TopCoder community
members were paid to grade and comment on all contest submissions using detailed scorecards,
ultimately picking the contest winners. The winning competitors for each contest then received
monetary prizes, and all participants received updated ratings for their performance. TopCoder also
ran studio contests if an application required logos or graphics; in those cases, clients chose the
winners.
Evolution of the TopCoder Community
Growth and Composition
From 2001 to 2009, TopCoder added an average of 25,000 new computer programmers to its
community each year. After filling out a short online registration form, anybody in the world could
participate in a software development competition; by spring 2009, the TopCoder community had
over 200,000 members (see Exhibit 4 for community growth). Although the size of the overall
community was large, the number of people within that community who actively participated in
contests and posted in forums was much smaller. The majority of community members at TopCoder
registered as members of the community but never competed in any contests. In fact, by 2009, only
35,000 unique individuals had competed in contests. To Mike Lydon, TopCoder’s chief technology
officer, the remaining 82.5% of the community was the “latent pool”: people who were interested
4
Page 139 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
As of late 2009, TopCoder ran two different types of competitions on its platform: algorithm and
client software development. Algorithm competitions served as the primary means for attracting new
members and retaining existing members. These competitions required members to develop creative
software solutions to relatively difficult programming challenges. All members were assessed against
each other through an automated computer scoring system; they then received a TopCoder rating for
their performance. Some algorithm competitions also had cash prizes for winners.
TopCoder (A): Developing Software through Crowdsourcing
610-032
A second group within the TopCoder community comprised those members who at one time
participated in TopCoder contests but then stopped participating. Lydon noted that, after TopCoder
decreased prize values in 2008, many competitors from the United States and Canada left the
TopCoder community. Yet another group included people who participated in TopCoder contests
but did not win. TopCoder saw those competitors as the “long tail”—people who primarily
competed for the sake of learning. One of TopCoder’s main goals was to cultivate the long tail so that
less-skilled competitors could improve over time and increase their levels of contribution. Lastly,
TopCoder’s most valuable group of competitors included the everyday winners. The talent of
TopCoder’s elite programmers was equal to the best in the world, but such members only accounted
for 0.5% of the total TopCoder population.
The core of TopCoder’s community was made up of single, highly competitive males in their 20s.
According to Michael Paweska, a six-year veteran at TopCoder: “To be successful at TopCoder, you
must ask yourself, ‘Are you a competitor?’ You need to be able to thrive on competition; you can’t be
scared of it. You also need the flexibility to work long hours. TopCoder is a bachelor’s sport: the
moment you become involved with someone else, it becomes a point of friction.” TopCoder attracted
competitors from developed nations such as the United States, Canada, South Korea, and Japan, as
well as from emerging economies such as China, Russia, Poland, India, and Ukraine. Wu Yanbo, a
Chinese TopCoder community member studying abroad in Australia, explained that most
competitors in the lower-paid contests were from developing countries. According to Wu, the prizes
were not large enough for many individuals from developed countries to compete, since they could
spend their time better elsewhere.
Justin Gasper, a member since 2001, began experimenting with the TopCoder platform while
working for a traditional software engineering company. After winning significant money with
TopCoder, Gasper decided to quit his job in 2005 and devote 40 to 50 hours a week to TopCoder.
Gasper explained: “TopCoder is my full-time job; I don’t have a day job.” Gasper was one of
TopCoder’s regular winners, a member of the “global elite” of programmers. In architecture
competitions, Gasper won at least second place 95% of the time and had a win percentage of 69.23%.
Competitors at TopCoder could choose which contests and what type of contests to join (see
Exhibit 5 for participation and prize data by contest type).
Profiles and Ratings
Each programmer in the TopCoder community maintained a public profile that displayed his or
her user name, contest history, and basic personal information. Another part of the member profile
displayed a competitor’s numeric rating for each type of contest. The rating system was modeled on
the one used to rank grandmaster chess players engaged in worldwide competition. A “red color
rating,” or a rating of over 2,200, represented elite status within the community and a high skill level.
Yellow, blue, and green color ratings represented descending skill levels. Each competitor’s country
rank, total community rank, success rates for contests, and reliability—or percentage of times the
contestant joined a competition and submitted a passing solution—were featured in their profiles.
TopCoder members could also choose whether or not to display their total earnings on their profiles
(see Exhibit 6 for an example member profile).
5
Page 140 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
enough in the TopCoder platform to register and had the potential to provide TopCoder with
increased development under the right conditions.
610-032
TopCoder (A): Developing Software through Crowdsourcing
Between 2001 and 2009, TopCoder paid out over $20 million in prizes and peer review money to
its community of developers. However, prize money was not evenly distributed throughout the
TopCoder community. The top 5% of prize earners received approximately 80% of the total prize
pool, while the majority of TopCoder community members earned little or no money from
competitions. Some competitors were extremely successful. For example, from 2006 to 2008, Paweska
earned $200,000 to $300,000 per year, while Gasper averaged over $100,000 annually. Wu
commented: “I have to say money is the most attractive thing. The prize is very good compared to the
income of my friends who are working in some local companies in China. Even though the economy
is not very good and TopCoder reduced its prizes, I can still earn around $1000 per month in my
spare time.” TopCoder typically awarded prizes to the top two submissions in each contest, with the
lion’s share of the prize money going to the top performer.
Besides prizes awarded on a contest-by-contest basis, another main source of income for members
was the Digital Run. In the Digital Run system, the top five ranked competitors for each contest were
awarded points based on contest rank and performance. At the end of each month, TopCoder tallied
competitors’ total points and awarded the top point earners thousands of dollars in bonus prizes.
Paweska explained that success in the Digital Run was not all about who was the best programmer
but more about who could handle the most all-nighters. Other competitors, such as Gasper, also
made money through contracted projects that TopCoder assigned.
In addition to their cash earnings, many community members reported that their TopCoder rating
was very important because it provided an objective assessment of ability. Wu commented that it was
not easy to maintain a very high rating as it required familiarity with many kinds of technologies,
quick thinking, the ability to learn independently, a strong work ethic, and attention to detail.
According to Wu, a TopCoder rating could be important for a programmer’s future career. For
example, a high TopCoder rating helped one of Wu’s friends earn a job at Google. Gasper noted that
TopCoder ratings were also symbols of status and prestige for many programmers: “If you have red
ratings, people look up to you.” Indeed, many prestigious software firms asked potential recruits to
get a TopCoder rating before applying for a job. To others, however, the rating system was less
important. Gasper, for example, explained that winning and making money meant more to him than
ratings.
Although there were differences of opinion regarding the importance of ratings, almost all
community members agreed that competing at TopCoder provided numerous opportunities to learn
and improve. In fact, for many programmers, a TopCoder career often began with failure, but postcontest evaluation and peer review of each submission helped them grow and improve. Gasper
noted: “I totally failed in my first competition. But the reviewers were really good at pointing me in
the right direction, saying ‘here’s where you went wrong’ … You can’t fake it because you’re getting
peer reviewed by people who are better at programming than you are. The reviewers don’t care if
they hurt your feelings; they are direct. If they see a bad design, they rip it apart.” Paweska agreed
that getting feedback from reviewers was crucial and added that community members could also
learn from acting as a reviewer for contests. For scientists and developers, Wu believed that
algorithm contests were particularly helpful at sharpening research skills and improving critical
thinking abilities. In all cases, continual learning opportunities from peers were an important reason
for participation.
Gasper described the appeal of working at home on a web-based platform instead of in a
traditional “cubicle farm” setting: “I like the flexibility that TopCoder gives me. I don’t need to drive
6
Page 141 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Motivating Members
TopCoder (A): Developing Software through Crowdsourcing
610-032
Setting one’s hours was convenient but also challenging, as competitors had to actively manage
their individual levels of participation. Gasper constantly balanced effort and reward to maximize
income while still living a sustainable lifestyle. “If there’s something that is way too much work for
the payment, I won’t do it . . . that’s a super power I’ve developed. I know when the spec is clean and
worthwhile to solve. It’s a skill that comes from doing tons and tons of contests.“
A “Community” of Competitors
Wu noted that, although the firm was competitive in sprit, competition at TopCoder was never
disrespectful or nasty and that people liked to help each other, even when they competed in the same
arena. TopCoder forums were the main source for collaboration. In the forums, less-experienced
community members asked for assistance on certain problems and received instant feedback from
more-experienced competitors.
At TopCoder, conversations and relationships extended beyond the scope of software
development. Hughes reflected on a particularly remarkable exhibit of communal strength and
caring for fellow community members outside of software development. “When one of the
community members died,” he said, “the outpouring of support was such that a number of the
community members took all of their winnings for a few weeks and gave it to the deceased’s wife. It
ended up being tens of thousands of dollars.”
Once a year, TopCoder paid for all of the best talent from the community to travel to Las Vegas,
Nevada, to compete in the TopCoder Open (TCO). In addition to serving as a proving ground for the
best programmers in the world, the TCO provided community members with the opportunity to
network professionally and socially.
The TopCoder community had a distinctive culture, with identifiable personalities. Wu explained:
“I believe this community, like all others, has its own culture. Clearly, the members built it up
continuously. When I joined the community, there were already some leading members who were
active in competitions and forums, brought out good suggestions, and started up interesting and
important discussions.” In some cases, the fame of community leaders extended well beyond
TopCoder. For example, Tomasz Czajka, from Poland, achieved “rock star” status and had his picture
plastered on billboards throughout Warsaw after he won the TopCoder Open in 2006.
The Client’s Perspective
Clients came to TopCoder to have high-quality software developed in a cost-effective and timeefficient manner. TopCoder positioned itself to serve both large firms and medium- to small-sized
business that wanted to see systems developed. Keith Moore, a TopCoder client and former senior
vice president at LendingTree.com, believed that, regardless of size, any company could take
advantage of TopCoder, whether it was a five-person operation or large outsourcing vendor. For
many CIOs, the process of software development and talent recruitment was a major headache, and
missed deadlines and large cost overruns were common worries. According to Stephen Laster, the
CIO at Harvard Business School and a TopCoder client, “A typical IT shop will turn over 48% of its
employees every three years. This process is very costly. The same problem exists with our
7
Page 142 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
a half an hour to work each day and can do the same work at home. If I want to take off a day to play
golf, I just do it. I also don’t have to work from 2:00 to 6:00 p.m., my most unproductive hours.”
Sharing similar sentiments, Paweska liked that while working at TopCoder he did not have a
supervisor looking over his shoulder.
610-032
TopCoder (A): Developing Software through Crowdsourcing
outsourcing consultants. When selecting consultant teams, we tried out 60 programmers before
finding our team of 20. With TopCoder, I pay for performance and the CIO sees Nirvana.”
Benefits
Better Ideas
Before sinking thousands of dollars into a project, a client could run a
conceptualization contest through which TopCoder members helped identify bad ideas and generate
better approaches early in the development cycle. When the client introduced a business problem to
the community, members asked hundreds of questions. Nic Perez, a former technical director at
AOL, explained that the community’s questions “gave us insights into problems I didn’t even really
know I had” and “saved us money by doing all of those questions upfront.” Using online forums,
clients answered questions for all competitors only once, avoiding repeated efforts. In some cases,
clients scrapped product ideas entirely after the community raised concerns about the product’s
likely success or usability in the marketplace.
TopCoder’s contest-based development system consistently produced highly creative ideas and
solutions. According to Darren Smith, a solution architect for the e-commerce division at Ferguson
Enterprises, North America’s largest plumbing supplies wholesaler and distributor, “The community
comes back with many options. It really has surprised us. You never know what you are you going to
get. The creative side allows us to go to the marketing management team and say, ‘We could do X, Y,
and Z that we may not have previously considered.’ They’re adding value to our business because
they bring us solutions that quite frankly we may not have considered or were not resourced to
deliver.”
Superior Quality, Cost, Speed, and Flexibility
Clients praised TopCoder’s rigorous
evaluation and documentation process for being well above industry standards. Reflecting on his
experience working on the Google Talk interface to AOL Instant Messenger, Perez stated that
TopCoder and its community had a strong desire to deliver bug-free code and that even the most
complex systems always had fewer than 100 identified bugs. According to Perez, the same sized
projects, developed internally, at AOL would have had five to eight times that number of bugs.
Another TopCoder client, a Web-based startup business, noted that it would have had to pay
$350,000 to a large IT consulting firm, $200,000 to a small IT consulting firm, or $80,000 to individual
contractors to build the company’s website. Using TopCoder, the client only spent $35,000. This same
client proclaimed: “At $35,000, it’s priceless. There is no other game out there.” A different client
noted that based on its experience working with almost every type of software development
company, TopCoder charged approximately half of the fee of a large, tier-one IT consulting firm.
Using the community for parallel problem solving, TopCoder marketed itself as faster than
other software development shops. This was true for back-end bug races and system checks, as
TopCoder took 72 hours to complete the same bug testing that a traditional development firm
finished in 10 business days. However, for other steps in the software development process, reports
on speed were mixed. Some clients said that TopCoder worked at about the same speed as a large IT
consulting firm, while others lauded TopCoder for speed of completion.
8
Page 143 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
As of 2009, TopCoder had developed a strong relationship with existing clients for delivering
high-quality software solutions and superior customer service. After completing their first project
with TopCoder, 82% of clients signed up for a second round of contests. These clients cited several
advantages.
TopCoder (A): Developing Software through Crowdsourcing
610-032
Concerns
Although CIOs were impressed by TopCoder’s technical capabilities and cost-saving potential,
many often had initial reservations about working with TopCoder’s unusual software development
model.
Intellectual Property (IP) and Security
According to Ira Heffan, TopCoder’s chief legal
counsel, “For new clients unfamiliar with TopCoder’s model, IP and security concerns can be an
initial point of resistance. Until they understand the documentation and processes we have in place
with the community members, they see IP and security as potential barriers to working with a
community.” For example, some clients were apprehensive that a TopCoder community member
might divulge proprietary ideas, business plans, or operations to their competitors. In addition, some
clients worried that once a component became an integral part of their IT systems, the community
member who built the component might attempt to prohibit its use or ask the client to pay
considerable royalties. Lastly, some clients were concerned that a solution submitted by a community
member could be stolen, copyrighted, or taken from open-source software projects, thus potentially
opening the door for intellectual property disputes.
TopCoder had in place a number of initiatives targeted at addressing these concerns and reducing
the risk level for clients, and also took steps to communicate its processes. To ease clients’ intellectual
property and security concerns, TopCoder produced a white paper that detailed confidentiality
policies, intellectual property assignment rules, and TopCoder’s modular approach to software
development. In addition, TopCoder allowed clients to keep their company names anonymous
during competitions and helped clients generate test data sets to avoid the exposure of sensitive
information. At the client’s request, before a community member was allowed entry into a
competition, all competitors could be required to sign a standard competition confidentiality
agreement.
The peer-review process was another means to ensure code security and quality. Peer reviewers
were selected and vetted by TopCoder employees based on their superior performance on prior
competitions. TopCoder clients also had the option of running testing competitions at the back end of
software production, serving as an additional means of checking code security and quality.
TopCoder’s compartmentalized software development process also made it difficult for a single
competitor to insert harmful code into a program, since individual contests only addressed one small
piece of the overall program.
Cultural Change
Many clients realized that working with TopCoder would be difficult
culturally for their company. In particular, CIOs believed that internal employees would view
TopCoder as a threat to their job security. One new client observed: “TopCoder is a CIO’s dream but
a programmer’s worst nightmare. I fully expect that if this goes well and if my programmers see
good quality work coming out of TopCoder, fear will quickly explode throughout the building.”
Although using TopCoder could help a company scale and reduce the programming staff costs,
companies still had to retain the “big thinking” people—the employees who could guide the
9
Page 144 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Especially appealing to clients was TopCoder’s ability to supply flexible software development
capacity. In particular, a TopCoder client could expand or reduce its business requirements and
development capabilities without having to hire or fire programmers. According to one client, a basic
in-house computer programmer cost $120,000 a year, after accounting for benefits, sick time, and
vacation. Working with TopCoder, clients did not have to spend as much on employees’ benefits and
downtime.
610-032
TopCoder (A): Developing Software through Crowdsourcing
Coding Challenges
Even if TopCoder did all of a company’s internal development work, the
company still needed to have an internal staff to integrate the deliverable into the client’s existing
systems, review the code for security issues, and adjust and fix code as systems changed over time.
For example, Smith’s team at Ferguson spent a significant amount of time inspecting, testing, and
processing TopCoder’s work to make sure that there were no security threats or bugs. In some
contests, TopCoder clients also spent time evaluating ideas and approaches from multiple winning
solutions.
Another ongoing issue for clients was finding the right types of problems and providing the
appropriate amount of problem detail for the TopCoder community. As Moore described it, “You
want neither too much nor too little detail. You do not want to quell innovation but also want a
solution that makes sense in your system.” Clients discovered that contest participation decreased if
they were unclear about what problems they wanted to solve or presented problems that were too
complex or vast in scope; in those cases, the TopCoder community struggled to produce an
acceptable solution. Clients also found that community members worked best when contests lasted
less than two weeks. If projects took too long to complete, contestants would lose interest and not
make submissions.
Managing TopCoder
The Supply Side
A management job at TopCoder was unique. Along with supervising internal TopCoder
employees, managers at the firm had to oversee a community of over 200,000 members and direct the
process of competition-based software development. According to senior vice president George
Tsipolitis, the key to success was effective process management: “When you’re managing a
community, you are no longer managing individuals, you’re managing a whole. We can’t control
individuals. We can only control the process of their participation.” Clients and employees alike
believed that the sustainable value of the company was dependent on TopCoder’s ability to facilitate
community participation and foster community growth. Lydon described the risks: “From the
beginning, we focused on the community. We knew they could be unforgiving. If you did the wrong
thing, you got crucified.”
Attraction
To run many competitions simultaneously and produce solutions for many clients
at the same time, TopCoder needed to have access to a critical mass of talent and coding capacity.
TopCoder’s primary means of attracting new members into the community was the appeal and
challenge of the algorithm contests. In addition, TopCoder occasionally advertised its online
competition platform by paying for Google keyword searches using terms such as “design contests.”
A third mechanism for attracting talent was “member development days.” Organized by a small
team of TopCoder employees, member development days were held at Chinese and other
international universities. At a member development day, a student representative would post signs
around the school and explain the TopCoder system. A primary goal of these member development
10
Page 145 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
TopCoder development process. The managers at TopCoder clients also had to adjust to a perceived
loss of control over the software development process. Smith commented: “We set the competitions,
but they manage the whole process. Our project management group works with the TopCoder
manager to ensure delivery according to pre-determined service level agreements (SLAs).”
Additionally, some clients found a few community members to be pushy and rude during precompetition question-and-answer sessions.
TopCoder (A): Developing Software through Crowdsourcing
610-032
Norms
As the community grew, TopCoder paid close attention to establishing community
norms. As contest administrator, the company had to maintain the highest standards of contest
integrity, fairness, transparency, and quality. For example, TopCoder personnel strictly monitored
competitions and tolerated no form of cheating. Community members who peeked at other
competitors’ solutions, shared ideas during competition, or used unauthorized code were
immediately eliminated from the contest. Often they were kicked out of the community entirely.
If any uncertainty or disagreement arose about which competitor won a particular contest,
TopCoder would spend extra money to re-run the competition. Another part of contest integrity,
Tsipolitis explained, was TopCoder’s emphasis on maintaining consistency of rules and procedures:
“The second that participants can’t figure out how to win, they’ll stop participating. So we can’t
change the rules of a competition mid-stream.” TopCoder also guaranteed complete contest
transparency by storing all contest and competitor statistics, peer reviews, and solutions in a data
warehouse. The data were publicly available to registered community members, accessed via the
TopCoder website.
Integrity and fairness also extended to TopCoder‘s corporate motivations and community
compensation philosophy. In particular, TopCoder was up-front with the community over its
intention to make money. When TopCoder made a decision to change corporate direction or
competition procedures, Hughes posted the information in the forums and explained the business
reasons behind his decisions. Hughes also believed that, since the company benefited from the
community’s hard work, adequately compensating community members was essential.
Governance
Although TopCoder executives were responsible for final decisions, they
frequently incorporated community members’ views into the process. Lydon explained: “We treat the
community as the driver for everything we do. If we have enough dissent from members, we always
take that into account. The problem is that when we don’t know what to do, our members will also be
split.” Community member Gasper shared a similar perspective: “TopCoder tends to push out ideas
into the forums to get feedback. Seventy-five percent of the time, they listen to the community. But
TopCoder also has its own business interests to consider. Sometimes the community and business
interests don’t line up.”
Similarly, if competitors were unhappy with a peer-review scoring outcome, TopCoder allowed
them to appeal the decision. Over 90% of contests featured at least one appeal. If a member appealed,
peer reviewers had to provide specific reasons why the appeal was accepted or rejected. If
disagreement remained between contestant and reviewer, TopCoder employees often investigated.
Contestants could also appeal directly and privately to TopCoder personnel or post complaints
publicly on the TopCoder forums.
TopCoder managers inevitably made decisions that sometimes disturbed and upset the TopCoder
community. For example, facing a very difficult economic environment in the summer and fall of
2008, TopCoder reduced the contest prize amounts, cut payments to peer reviewers, and reduced the
number of algorithm competitions. During this period, some TopCoder competitors left the
community entirely and others dramatically reduced their participation levels. Gasper argued that
11
Page 146 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
days was to encourage participation in the higher-revenue-producing development and design
contests. During one member development day in China, TopCoder registered over one thousand
new community members. Bourdon noted that TopCoder had achieved critical mass once it crossed
the 200,000 member threshold, as there were now many members with deep and narrow skills over a
range of software development challenges (See Exhibit 7 for the number of participants by contest
type).
610-032
TopCoder (A): Developing Software through Crowdsourcing
the payment cuts also led to many superficial reviews because the best reviewers were no longer
doing the work, which then required additional cycles to achieve acceptable quality.
Another part of the TopCoder managerial role was allocating
community resources and controlling contest participation. Lydon explained, “We have to figure out
how to distribute the number of people who want to participate across the number of contests that
have to be solved.” Unlike a typical software development firm, TopCoder could not assign specific
people to a task or project. As participation manager, TopCoder’s goal was to minimize the costs of
evaluation, stimulate effort through competition, and get at least one solution that was acceptable to
the client. To achieve the ideal number of submissions and participants, TopCoder adjusted the prize
amount, the duration and timing of the contest, the number of other contests running concurrently,
and the problem’s complexity and scope. When deciding on the ideal number of competitors,
TopCoder also considered contestants’ reliability. As a last resort, TopCoder employees reached out
directly to individual community members if other methods did not lead to the desired participation
levels.
Although TopCoder managers could pull many levers to influence contest participation, they
believed it was important not to act like the community’s boss. Hughes explained his community
management philosophy: “We don’t own this community. We want people to be here when they
want to be here. You are just going to get much better results when you let people do what they
really want to do.”
Retention
At the same time, TopCoder executives worked to retain community members and
encourage future contest participation. At least one client raised concerns in this area: “I think that
communities are fickle. Community members could start to ask, why do I need them? For example,
what happens if an imitator comes along and offers twice the prize amount?” To avoid such
problems, TopCoder tried to supply community members with consistent work streams and prize
money. TopCoder also encouraged community members to engage in the community as much as
possible by dedicating significant resources to facilitating forum discussions and inviting contestants
to participate in peer reviews, write problems for contests, and develop TopCoder’s internal systems.
TopCoder community members differed on their level of loyalty to the TopCoder community.
Paweska, appreciative of all the opportunities TopCoder had provided, reflected, “I have some
loyalty. I think it would take a lot for me to leave. Only if there were no projects would I leave.”
Gasper viewed his position in the community in a different light. “I’m not super loyal to TopCoder or
anything,” he said. “[For me] to defect, the payment and work would have to outweigh the payment
and flexibility I have at TopCoder.”
The Demand Side
Platform Managers
The other side of management at TopCoder was guiding clients through
the contest-based software development process. This was the responsibility of the company’s
platform managers, whose job was to induce the appropriate amount of community participation,
make suggestions for contest prize amounts, gather feedback between contests, and provide project
status updates to clients. Before starting the next step in the game plan, platform managers also
adjusted contest requirements based on the work already completed. Once the product was delivered
to the client, TopCoder platform managers were required to act in a support and service role. If there
was a technical problem with a solution, the platform manager often contacted the community
members who developed the component and worked with the community members to fix the issue.
12
Page 147 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Resource Allocation
610-032
Most enterprise-level clients believed the platform manager was pivotal to a project’s success. At a
large client like LendingTree, the platform manager was on site three to four days a week, conducting
daily meetings with the internal teams. A large part of the platform manager’s role was managing
client expectations and serving as a sounding board for client concerns. At the back end of projects,
although the community often assembled a project’s components through competition, the platform
manager was also an expert at combining the small software pieces. The component integration role
saved the client hours of work trying to figure out how all the pieces fit together. At Ferguson, Smith
considered the TopCoder personnel working on site to be an integral part of his team.
TopCoder Direct However, each platform manager added to TopCoder’s overhead costs and
narrowed profit margins. As of 2009, a typical platform manager at TopCoder cost $100,000 a year
including benefits. To Tsipolitis, the platform manager’s time was not always well spent. “Our project
managers spend a lot of time babysitting,” he said.
To avoid a potentially large increase in expenses as TopCoder added clients and projects, Hughes
came up with the concept of “TopCoder Direct,” in which the client used the company’s platform
with little to no intervention from its employees. Under this self-service model, platform managers
would educate clients on how to use the TopCoder platform to manage the contest-based software
development process themselves. Hughes envisioned shifting the platform manager’s responsibilities
to an experienced community member or an external consultant familiar with TopCoder’s platform—
someone who would serve as a “co-pilot” to assist the client staff. With co-pilots taking the role of
platform managers, Hughes estimated that the platform manager’s weekly time on a project would
shrink from 40 hours to two, thus saving the client and TopCoder considerable time and money.
The Future
As of December 2009, no competitors had elected to copy TopCoder’s business model by offering
full-service software development through a competition-based approach. Instead, companies such
as RentACoder, Elance, and oDesk served as online liaisons between clients and freelance software
developers. Unlike TopCoder, whose clients only paid for solutions, clients at these firms used a “buy
talent” approach: they selected one or more programmers to solve their problem. More similar to
TopCoder, uTest used crowdsourcing to find bugs and check the functional usability of web, mobile,
desktop, and gaming applications, but did not engage in software development. According to
Hughes, this lack of direct competition reflected the technical difficulties and costs associated with
building a full-fledged community and platform.
After a significant downturn in the global economy in 2008 and 2009, Hughes believed that
TopCoder was primed for growth. Sales staff were forecasting aggressive targets for the volume of
competitions and revenues, and several strategic partnerships were under consideration. However,
significant challenges and uncertainty remained. In particular, Hughes wondered whether the
community, as well as the company, could grow to meet increasing demand.
Stakeholders had divergent views. Mike Morris, vice president of sales, saw unlimited potential:
“If sales grow at a linear rate, membership grows at an exponential rate. The supply of community
members is not going to limit growth. If you throw enough money out there, you will get enough
programmers.” Community members Paweska and Wu agreed that offering more money per contest
would increase participation among existing members. Paweska also believed, however, that holding
many more contests than usual in a given week would result in inexperienced competitors competing
actively for the prizes, possibly reducing code quality. Furthermore, Lydon noted that that during
TopCoder’s last large scale-up in 2007, review quality suffered during a transitional period of a few
13
Page 148 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
TopCoder (A): Developing Software through Crowdsourcing
610-032
TopCoder (A): Developing Software through Crowdsourcing
Hughes also worried about client service. If the number of TopCoder clients expanded
significantly, TopCoder’s staff might face increasing difficulties responding to all those clients’
questions and concerns. For large clients, expansion might require adding more platform managers,
but Rob Hughes, TopCoder’s COO, was concerned that too many platform managers might make the
firm appear to be like any other large IT consulting company, with the risk of losing its unique
business model.
Even if Hughes succeeded at growing TopCoder, he was unsure about the company’s competitive
position. In particular, Hughes wondered if community members would stick with TopCoder if a
new competition-based software development company emerged. What would happen if a company
like Accenture started to develop software in the same way as TopCoder? Would the TopCoder
community remain intact?
14
Page 149 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
months. As more contests became available, the usual reviewers wanted to compete in the contests,
rather than review them, leaving TopCoder scrambling to find replacements. In addition, a few
clients worried that as the number of avenues of competition at TopCoder grew, attracting the same
group of competitors would prove much more difficult, reducing contest consistency and continuity,
which were especially critical for addressing legacy systems.
Page 150 of 282
TopCoder Members Involved in Creating a Power Pricing System for an Energy Company
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Source: Company documents; developed via a TopCoder Studio competition.
Exhibit 1
610-032
-15-
Page 151 of 282
1.23
Cost of Platform
Managers ($MM)a
1.24
51
4.50
32
1.12
44
3.80
34
1.30
52
5.35
25
1.36
52
5.80
24
1.19
46
5.50
38
Q2
1.13
44
4.85
37
Q3
0.91
36
2.60
47
Q4
0.53
20
2.25
46
Q1
2009
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
a Includes platform managers’ salaries, benefits, and other expenses.
Source: Company statistics.
51
4.66
32
Number of
Platform
Managers
Total Revenue
($MM)
Q4
Q1
Q3
Q1
Q2
2008
2007
Number of Clients, Revenue, Number of Platform Managers, and Platform Manager Costs by Quarter
Number of Clients
Exhibit 2
0.49
19
1.92
47
Q2
0.46
18
1.82
38
Q3
610-032
0.40
16
2.45
35
Q4
-16-
Phase
Exhibit 3
Source:
Company documents.
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x
x x x x x x x x
x x x x x x x x x
Updates
Bug Races
x x x x x x x x x
x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x
x x x x x x x x x x
x x x x x x x x x x
x x x x x x x x
x x x x x x x x x x
x x x x x x x x x x
x x x x x x x x
x x x x x x x x x x
x x x x x x x
Deployment
Deployment
x x x x x x x x x x
x x x x x x x x x
x x x x x x x x x x
x x x x x
Testing
Test Scenarios
Test Cases
Bug Hunt
Prototype Assembly
<Functional> Module
Module Specification
Module Architecture
Component Design
Component Development
Catalog Components
Module Assembly
Application Build
System Architecture
System Architecture
Component Design
Component Development
Catalog Components
System Assembly
x x x x x
4,000.00
4,000.00
6,900.00
6,600.00
4,900.00
4,500.00
2,300.00
2,050.00
4,900.00
$ 4,200.00
$ 74,800.00
Total
$ 4,000.00
$ 3,900.00
$ 3,900.00
$ 2,750.00
$ 4,900.00
$
$
$
$
$
$
$
$
$
$
$
$
$ 4,500.00
$ 3,000.00
$ 3,500.00
30
20
2
2
1
1
1
1
3
3
0
1
1
1
1
0
1
0
1
1
1
-17-
Estimated
Costs
610-032
# of
Timeline
Contests
4/27
5/4
5/11
5/18
5/25
6/1
6/8
6/15
6/22
6/29
7/6
7/13
7/20
7/27
8/3
8/10
M T WT F M T WT F M T WT F M T WT F M T WT F M T WT F M T WT F M T WT F M T WT F M T WT F M T WT F M T WT F M T WT F M T WT F M T WT F M T WT F
Sample Game Plan
Conceptualization
Logo - Tournament
Concept Contest - Tournament
Wireframes - Tournament
Storyboard - Tournament
Page 152 of 282
610-032
TopCoder (A): Developing Software through Crowdsourcing
Exhibit 4
Community Growth
250,000
150,000
100,000
Total
Community
Members
50,000
0
Source:
Company statistics.
Exhibit 5
and 2009
Average Contest Registration, Submission, and Prize Amount for Client Contests in 2008
2008
Contest Type
2009
Number of
Registrants
per Contest
Submissions
per Contest
Prize
Amount per
Contesta
Number of
Registrants
per Contest
Submissions
per Contest
Prize
Amount per
Contesta
Conceptualization
n/ab
n/ab
n/ab
17.57
3.60
$1,314
Specification
n/ab
n/ab
n/ab
14.20
1.94
$1,017
Architecture
16.3
1.64
$1,590
19.34
1.75
$1,095
Component
Design
9.83
2.85
$899
16.26
1.94
$559
Component
Development
15.4
2.69
$733
25.59
2.56
$465
Assembly
16.09
1.12
$1,628
18.38
1.18
$913
Studio
27.55
14.57
$795
27.57
20.04
$1,015
Source: Company statistics.
a Prize per contest – Prize for first and second places and reserve for Digital Run.
b n/a – Data not available for most of 2008.
18
Page 153 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
200,000
TopCoder (A): Developing Software through Crowdsourcing
Example Community Profile
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Exhibit 6
610-032
Source:
http://www.topcoder.com/tc?module=MemberProfile&cr=287614, accessed December 23, 2009.
19
Page 154 of 282
Page 155 of 282
1,370
33
58
1,319
500c
5,565
157
316
5,287
500c
730
362
287
81
1c
2,146
65c
45
88
1,930
273
8,517
223c
225
451
7,525
1,532
1,224
17c
615
484
105
20
2,867
81
1c
61
135
7
2,638
281
2007
2008
10,072
453
4c
243
605
47
8,994
1,588
1,825
118
4c
698
780
86
113
30
3,198
66
7c
5c
4
38
91
10
2,945
253
11,487
279
11c
7c
29
144
434
59
10,433
1,621
2,023
451
9c
16c
65
488
733
191
66
29
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
d Represents the unique number of participants during a given time period. Columns are not additive.
c Partial year data—contest track officially did not officially start until the middle or end of the year.
b Programming contests that run from 3–30 days.
a 75-minute programming contest.
Source: Company statistics.
Totald
Studio
Design Contests
Conceptualization
Specification
Architecture
Component Design
Component Dev.
Assembly
Client Software
Development Contests
Single Rounda
Marathon Matchb
2006
2009
-20-
2,911
102
6
6
9
20
41
20
2,558
314
11,122
429
34
32
36
93
204
104
9,616
2,150
1,881
456
70
71
145
300
337
416
51
35
Average Submitters Contests Average Submitters Contests Average Submitters Contests Average Submitters Contests Average Submitters Contests
per Month per Year per Year per Month per Year per Year per Month per Year per Year per Month per Year per Year per Month per Year per Year
2005
Number of Unique Participants by Contest Type per Year and Month/Total Number of Official Contests per Year
Algorithm Contests
Contest Type
Exhibit 7
610-032
9B20E017
Helen Lang and Robert D. Austin wrote this case solely to provide material for class discussion. The authors do not intend to illustrate
either effective or ineffective handling of a managerial situation. The authors may have disguised certain names and other identifying
information to protect confidentiality.
This publication may not be transmitted, photocopied, digitized, or otherwise reproduced in any form or by any means without the
permission of the copyright holder. Reproduction of this material is not covered under authorization by any reproduction rights
organization. To order copies or request permission to reproduce materials, contact Ivey Publishing, Ivey Business School, Western
University, London, Ontario, Canada, N6G 0N1; (t) 519.661.3208; (e) cases@ivey.ca; www.iveycases.com. Our goal is to publish
materials of the highest quality; submit any errata to publishcases@ivey.ca. i1v2e5y5pubs
Copyright © 2020, Ivey Business School Foundation
Version: 2020-09-29
The strength of blockchain tech is that it is a ledger, a statement of truth. That ledger is
only as good as its resistance to censorship, change, demands or attack. 2
Bruce Fenton, board member of Bitcoin Foundation
On Friday, June 17, 2016, Stephan Tual, a founder of the Decentralized Autonomous Organization (DAO), 3
noticed something that interrupted his months-long effort to build an experimental blockchain-based
venture capital organization. An anonymous individual or group (referred to hereafter as the “Attacker”),
who was also a DAO investor (investors were not required to disclose their identities), had made an
unauthorized withdrawal of 3.6 million ether (ETH), roughly US$55 million, 4 which accounted for
approximately one-third of the total amount raised by the DAOfunds that mostly did not belong to him
(or her, or them). 5 The Attacker had accomplished this by identifying and exploiting a previously
undetected flaw in a DAO smart contract. This brazen act was in complete contradiction to the spirit of the
DAO. It was, in essence, an act of theft. Unfortunately, though, things were not quite as simple as that.
The terms of the DAO membership agreement included the following: “Nothing in this explanation of terms
or in any other document or communication may modify or add any additional obligations or guarantees
beyond those set forth in the DAO’s code.” 6 This odd legal clause was central to the DAO experiment. The
idea was to eliminate, to the fullest extent possible, all audit and transaction coststo eliminate lawyers
and other humans from contract negotiation and dispute resolutionby replacing them with a blockchain
and smart contracts. 7 To accomplish this, the smart contracts, the rules by which the organization operated
as written in software code, were the contractsthe only contractsthat legally governed how the DAO
operated. It followed, then, that anything the software code would do was, by definition, in accordance with
the organization’s governing legal rules. The flaw in the code was effectively a loophole in a contract. Using
the flaw to withdraw funds was therefore arguably legal.
The Attacker made this argument after the hack had been discovered. In a message to other members of the
DAO (see Exhibit 1), the Attacker wrote:
I have carefully examined the code of the DAO and decided to participate after finding the feature
where splitting is rewarded with additional ether. I have made use of this feature and have rightfully
claimed 3,641,694 ether, and would like to thank the DAO for this reward . . . I am disappointed
by those who are characterizing the use of this intentional feature as “theft.” I am making use of
Page 156 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
THE DAO HACK: A BLOCKCHAIN DILEMMA 1
Page 2
9B20E017
this explicitly coded feature as per the smart contract terms and my law firm has advised me that
my action is fully compliant with United States criminal and tort law. 8
But the decision whether to do so was also not simple. Undoing the withdrawal would require reversing a
transaction already registered on the blockchain“forking the chain” in blockchain parlance. But the
credibility of the blockchain, the reason that it was viewed as a viable alternative to trusted intermediaries like
banks, was based on the assurance that it could not be forked. Undoing the withdrawal, then, would undermine
the credibility of the blockchain. The Attacker made exactly this argument, in his (or her, or their) message:
A . . . fork would amount to seizure of my legitimate and rightful ether, claimed legally through the
terms of a smart contract. Such fork would permanently and irrevocably ruin all confidence in not
only Ethereum but also in the field of smart contracts and blockchain technology. 9
The DAO held 15 per cent of all ETH in the Ethereum network, so a loss of confidence in the DAO could
undermine the integrity of the ether currency itself. 10 A collapse of ETH cryptocurrency was a real
possibility; two hours after the DAO hack became known, ETH had fallen 38 per cent, from $21 to $11.13. 11
The choices before the co-founders were stark. They could let the Attacker take possession of the 3.6 million
ETH that did not belong to him (or her, or them), or they could fork the blockchain, which would likely
undermine the integrity of the DAO and possibly the ETH cryptocurrency.
BACKGROUND
The DAO was incorporated in Germany by Christoph Jentzsch, Simon Jentzsch, and Stephan Tual as an
experimental venture capital firm based on the Ethereum blockchain. At that time, ETH was the second largest
digital currency in terms of market capitalization; only Bitcoin was larger. 12 Launched on April 30, 2016, the
DAO raised funds from investors for 28 days (see Exhibit 2). 13 The organization was completely virtual and
did not have owners or employees, only members who had bought into it (investors). Its operations were
carried out entirely by smart contracts implemented in software. It was designed to be stateless, not tied to
any particular nation state, which raised issues of how national financial regulators might deal with it. 14
All three founders had strong technical backgrounds. Christoph Jentzch had studied theoretical physics and
had worked optimizing software solutions for high-performance computing. His brother, Simon, had
worked for 20 years as a project manager, software developer, and software architect, helping to implement
enterprise solutions. Tual had 25 years of enterprise information technology (IT) experience; he had become
a recognized expert in smart contracts and regularly spoke at conferences on this topic. 15 These three had
previously founded Slock.it, a blockchain and Internet-of-Things solution company in 2015.
The goal of a “decentralized autonomous organization” was “to codify the rules and decision-making
apparatus of an organization, eliminating the need for documents and people in governing, creating a
structure with decentralized control.” 16 The founders and other volunteers wrote computer code that would
serve as the smart contract infrastructure that would run the organization. During the initial funding period,
participants purchased “DAO tokens” with ETH currency. These tokens provided investors with voting
rights in the DAO. An investor had voting power proportionate to the amount of his or her investment.
Once the funds had all been raised, the DAO investors circulated and scrutinized funding proposals from
Page 157 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
As troubling as all this seemed, when Tual and his two DAO co-founders disclosed to the other members
of the DAO what had happened, they were also able to convey at least some good news: the Attacker did
not yet actually have the money. The withdrawal process had a built-in delay of 34 daysthere was time
to intervene, to stop the theft.
Page 3
9B20E017
entrepreneurs and voted on which to fund. They also voted to authorize distribution of earnings from the
fund, as allocated by smart contract software.
According to the DAO website, created to promote the organization at the time of its initial funding, its
purpose was, “To blaze a new path in business for the betterment of its members, existing simultaneously
nowhere and everywhere and operating solely with the steadfast iron will of unstoppable code.” 18 Once
announced, the DAO drew much interest and attention, and raised the equivalent of more than $161 million
from investors, 19 making it the most successful example of crowdfunding to that point in history. 20
A SHORT BLOCKCHAIN PRIMER
A blockchain was a database that was “consensually shared, replicated, and synchronized.” 21 It was a
relatively new concept, having been introduced in 2008 as the technology that ran Bitcoin. 22 Today,
blockchain was used for many other purposes besides cryptocurrency. 23
Blockchain was composed of three components: the block, chain, and network. The block was a list of
transactions within the network over a period of time. Once a block had the maximum number of
transactions, it was chained to the previous block. If a change was made to any of the blocks, users could
see that the original block had been altered, and it was therefore not to be considered trustworthy. 24 The
information that the block recorded typically included the timestamp and transaction data, which was
generated automatically. 25 This information was publicly visible to all the users in the chain, creating
complete transparency. Moreover, copies of the chain of blocks resided in many different locations, which
were kept in sync as the blockchain grew (as transactions were added). The fact that the blockchain existed
in multiple locations provided a further source of security as it made a change to one or a few copies
obviously different from all the others.
A blockchain could, then, record transactions between parties without a need for a third party, such as a
bank, to maintain an “official” transaction record. Because an intermediary was no longer needed, costs
were reduced, and transacting parties were independent of any influence of intermediaries.
Operating a blockchain required that certain challenges be met. The ledger had to update information in
real time. Communications between the copies of the blockchain had to be maintained, to keep all versions
in sync. And the blockchain had to be protected from efforts to manipulate it. A variety of complex
communication, encryption, and security technologies served these purposes.
A smart contract was a contract written in code that operated on the blockchain in the normal course of the
business of a blockchain-based process. Smart contracts could be executed by a triggering event, as defined
in the code of the contract. 26 The code itself expressed the terms of legal contracts, so developers had to be
careful when writing it.
FORESHADOWING THE HACK
On May 26, 2016, a group of computer scientists, experts on the Ethereum network, published a paper that
identified risks for a DAO-like entity. 27 The paper suggested ways to mitigate the risks but recommended
Page 158 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
The DAO was decentralized in that it was run, collectively, by its members. It was autonomous, in that its
operations were fully determined smart contracts written into code, which were also legally binding
contracts. 17 It was an organization only in the loosest sense of that word.
Page 4
9B20E017
not going forward with investments in such an entity without addressing the risks. Gün Sirer, an associate
professor of Computer Science at Cornell University, read this paper and sent it to the DAO founders. 28
On June 12, less than a week before the hack, Tual publicly announced that there were “recursive call bugs”
within the DAO code, which led to the vulnerability that Vessenes had described. 30 Tual reassured members
that no DAO funds were at risk and promised a fix. However, this fix required a change to DAO code,
which, in accordance with the DAO’s governing procedures, required investors to vote to authorize it.
Voting required a two-week voting period, and a majority of the token holders would have to vote. 31 Up to
this point, the Ethereum network had no other known bugs and had never been hacked, so many investors
were not that concerned about a potential attack. 32
However, on June 17, before the voting period had ended, the Attacker initiated a withdrawal, exploiting the
known vulnerability in the code. It was possible, perhaps likely, that the Attacker had been inspired by the
announcement of the vote to investigate the vulnerability and conceive the attack. Regardless of that detail, once
the withdrawal was triggered, Tual, his partners, and the other members of the DAO faced a tough decision.
OPTION 1: PERMIT THE WITHDRAWAL
The first option for Tual and the other members of the DAO was to allow the Attacker to proceed with the
withdrawal and collect the funds. This option would result in major losses for many DAO members but
would not undermine the integrity of the blockchain. Investors’ funds would be sacrificed for the good of
the blockchain and ETH. The principle that the code was the contract, and thus that everything in the code
should be considered to be legitimate, would be upheld.
OPTION 2: FORK THE BLOCKCHAIN 33
The second option was for Tual and his partners to intervene at the behest of members of the DAOto
“fork” the blockchain.
Two different variations of forking (“soft” and “hard”) were discussed. 34 Both would require rule changes
and thus code changes.
A soft fork would require a rule change that applied to future transactions—one that would prevent the
Attacker from completing the withdrawal. This change required “changing the deal” on which the DAO
was founded, the process by which the blockchain was legitimately updated, but it did not require changes
to the blockchain itself.
A hard fork would be a more extreme intervention, in that it would alter transactions already recorded in
the blockchain; overwrite history and reverse the undesirable transactions; and in the process, belie the idea
that the blockchain could not be altered retroactively.
A fork would allow investors to recover their money, but either option could undermine the integrity of the
DAO’s legal basis, as well as confidence in the organization, the blockchain, andpotentially, by
extensionthe ETH currency. If they could do it once, after assuring everyone that it would never be done,
then they could do it again. Such an action would, in effect, make a mockery of the idea that the DAO
Page 159 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
One particularly prominent risk highlighted by the paper had to do with processes for withdrawing funds
from the DAO. The withdrawal process included waiting periods and voting by the members to authorize
withdrawals. But on June 9, Peter Vessenes, a blockchain expert, pointed out a vulnerability in the process
that would potentially allow users to withdraw twice the amount of their actual balances. 29
Page 5
9B20E017
“operated solely with the steadfast iron will of unstoppable code,” 35 as was stated in its purpose for existing.
The aim had been to remove people from the process of administering transactions, but voting to alter how
the DAO worked to address a specific transaction put people right back into the mix.
Much debate went on within the Ethereum community and beyond about whether DAO members should
vote for or against a fork. Although the members of the DAO were anonymous by design, making it
impossible to know who they all were, it is probably safe to conclude that people who would step forward
to participate in this kind of early blockchain experiment were disproportionately committed to the idea
that blockchains could operate autonomously and replace trusted third parties. 36 But they likely also did not
want to lose money.
The larger debate precipitated by the DAO hack concerned what these events meant for the viability of the
vision of blockchain advocates. Advocates had mapped out futures in which blockchain-based systems
provided a means by which electronic transactions could be facilitated without depending on “trust.” They
pointed to a world to come in which banks and credit card companies, among other organizations, would
be viewed as totally unnecessary and overly expensive. But the DAO hack raised the spectre of a different
world, one in which such descriptions seemed like idealistic hype. Might it be true, as computer security
expert Bruce Schneier would write in 2019, partly in reaction to the DAO hack, that “[t]here’s no good
reason to trust blockchain technologies”?37
What did the DAO hack really mean for the future of blockchain? Perhaps the hack proved how difficult
promised visions would be to actually achieve.
Was the hack just an example of growing pains, something that would eventually be worked out? Given
the publicity around the event, it seemed unlikely that anyone would ever again make the same mistake the
DAO had made in its smart contracts. Was it just a matter of eliminating this mistake, and possibly even a
small number of mistakes yet to be discovered, to reach a point where all smart code loopholes were
eliminated? Or would there always be a loophole lurking somewhere or an attack that some clever hacker
could execute that no one had thought of before?
Or, maybe the hack signified that blockchain would be good for some things but not others; that blockchain
would be useful but not the be-all and end-all forecasted by blockchain enthusiasts. If this were the case, what
kinds of things would it be good for? And in what conditions or applications would it not be a good solution?
The message from the Attacker to the DAO and members of the Ethereum community also included a threat
that suggested the possibility of another difficulty for blockchain. The Attacker’s message closed with the
following words: “I reserve all rights to take any and all legal action against any accomplices of illegitimate
theft, freezing, or seizure of my legitimate ether, and am actively working with my law firm.” 38 The entire
purpose of entities like the DAO was to achieve separation from the usual sources of transaction costs, such
as the costs of legal disputes. But could this succeed? The Attacker’s threat implied that blockchain-based
solutions might not, in reality, be able to escape the complexities of legal claims and clever lawyers. Worse,
inserting people into the process to change the rules, as would be required to instigate a fork in the DAO
case, might open the door that led to such possibilities.
Tual, his co-founders, and members of the DAO needed to decide what to do. A vote would be necessary,
and the rules required that it happen in a certain way and in a certain time frame. It would all need to be
completed and decided before the Attacker’s withdrawal transaction could be completed.
Page 160 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
THE DECISION AND THE LARGER DEBATE
Page 6
9B20E017
EXHIBIT 1: EXCERPTS FROM THE ATTACKER’S MESSAGE
To the DAO and the Ethereum community,
I am disappointed by those who are characterizing the use of this intentional feature as “theft”. I am making
use of this explicitly coded feature as per the smart contract terms and my law firm has advised me that my
action is fully compliant with United States criminal and tort law. For reference please review the terms of
the DAO:
“The terms of The DAO Creation are set forth in the smart contract code existing on the Ethereum
blockchain . . . The DAO’s code controls and sets forth all terms of The DAO Creation.”
A soft or hard fork would amount to seizure of my legitimate and rightful ether, claimed legally through the
terms of a smart contract. Such fork would permanently and irrevocably ruin all confidence in not only
Ethereum but also in the field of smart contracts and blockchain technology. Many large Ethereum holders
will dump their ether, and developers, researchers, and companies will leave Ethereum. Make no mistake:
any fork, soft or hard, will further damage Ethereum and destroy its reputation and appeal.
I reserve all rights to take any and all legal action against any accomplices of illegitimate theft, freezing, or
seizure of my legitimate ether, and am actively working with my law firm.
Yours truly,
"The Attacker"
Source: excerpted from David Siegel, “Understanding the DAO Attack,” CoinDesk, June 27, 2016, accessed February 3, 2020,
www.coindesk.com/understanding-dao-hack-journalists.
EXHIBIT 2: TIMELINE OF EVENTS
Note: M = million; DAO = decentralized autonomous organization
Source: Created by the case authors.
Page 161 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
I have carefully examined the code of The DAO and decided to participate after finding the feature where
splitting is rewarded with additional ether. I have made use of this feature and have rightfully claimed
3,641,694 ether, and would like to thank the DAO for this reward. It is my understanding that the DAO code
contains this feature to promote decentralization and encourage the creation of “child DAOs.”
Page 7
9B20E017
1
This case has been written on the basis of published sources only. Consequently, the interpretation and perspectives presented
in this case are not necessarily those of The Decentralized Autonomous Organization (DAO) or any of its employees.
2
Nathaniel Popper, “A Hacking More Than $50 Million Dashes Hopes in the World of Virtual Currency,” New York Times,
June 17, 2016, accessed February 4, 2020, www.nytimes.com/2016/06/18/business/dealbook/hacker-may-have-removedmore-than-50-million-from-experimental-cybercurrency-project.html.
3
Ibid.
4
Currency exchange rate is at the time of the hack (June 17, 2016); all figures are in US$ unless specified otherwise.
5
Evan D. Wolff, Matthew B. Welling, and Tyler A. O’Connor, “The DAO Hack Provides Lessons for Companies Using
Blockchain and Distributed Ledger Technology,” Crowell Moring, June 27, 2016, accessed February 3, 2020,
www.crowell.com/NewsEvents/AlertsNewsletters/all/The-DAO-Hack-Provides-Lessons-for-Companies-Using-Blockchainand-Distributed-Ledger-Technology.
6
Usha R. Rodrigues, “Law and the Blockchain,” Iowa Law Review 104, no. 2 (2019): 1–65, accessed July 10, 2020,
https://ilr.law.uiowa.edu/print/volume-104-issue-2/law-and-the-blockchain/.
7
“Tokenized Networks: What is a DAO?” Blockchain Hub Berlin, July 2019, accessed July 10, 2020,
https://blockchainhub.net/dao-decentralized-autonomous-organization/.
8
Anonymous, 2016.
9
Ibid.
10
David Siegel, “Understanding the DAO Attack,” CoinDesk, June 27, 2016, accessed February 3, 2020,
www.coindesk.com/understanding-dao-hack-journalists.
11
Jessica Sier, “The DAO Hack: $US50 Million Lost,” AFR Online, June 19, 2016, accessed February 2, 2020,
https://advance-lexis-com.proxy1.lib.uwo.ca/api/document?collection=news&id=urn:contentItem:5K21-YS51-DY19-C26F00000-00&context=1516831.
12
Siegel, op. cit.
13
“Report of Investigation Pursuant to Section 21(a) of the Securities Exchange Act of 1934: The DAO,” Securities and
Exchange Commission: Securities Exchange Act of 1934, July 25, 2017, accessed February 3, 2020,
www.sec.gov/litigation/investreport/34-81207.pdf.
14
“The DAO of Accrue: A New, Automated Investment Fund Has Attracted Stacks of Digital Money,” Economist, May 19,
2016, accessed June 7, 2020, www.economist.com/finance-and-economics/2016/05/19/the-dao-of-accrue.
15
“Stephan Tual,” LinkedIn, accessed February 20, 2020, www.linkedin.com/in/stephantual/?originalSubdomain=uk.
16
Michael Juntao Yuan, Building Blockchain Apps (Boston, MA: Addison-Wesley Professional, 2019), 134.
17
“Report of Investigation Pursuant to Section 21(a),” op. cit.
18
Ibid.
19
Brady Dale, “The DAO: How the Employeeless Company Has Already Made a Boatload of Money,” Observer, May 20,
2016, accessed February 4, 2020, https://observer.com/2016/05/dao-decenteralized-autonomous-organizatons/.
20
“Crowdfunding,” Merriam-Webster, accessed February 23, 2020, www.merriam-webster.com/dictionary/crowdfunding.
21
National Archives and Records Administration, “Blockchain White Paper,” National Archives: Office of the chief Records Officer,
February 2019, accessed February 5, 2020, www.archives.gov/files/records-mgmt/policy/nara-blockchain-whitepaper.pdf.
22
“The Great Chain of Being Sure about Things,” Economist, October 31, 2015, accessed February 17, 2020,
www.economist.com/briefing/2015/10/31/the-great-chain-of-being-sure-about-things.
23
Bernard Marr, “A Very Brief History of Blockchain Technology Everyone Should Read,” Forbes, February 16, 2018,
accessed February 8, 2020, www.forbes.com/sites/bernardmarr/2018/02/16/a-very-brief-history-of-blockchain-technologyeveryone-should-read/#a69ebc87bc47.
24
National Archives and Records Administration, op. cit.
25
Julija Golosova and Andrejs Romanovs, “The Advantages and Disadvantages of the Blockchain Technology,” Research
Gate,
November
2018,
accessed
February
10,
2020,
www.researchgate.net/publication/330028734_The_Advantages_and_Disadvantages_of_the_Blockchain_Technology.
26
National Archives and Records Administration, op. cit.
27
Dino Mark, Vlad Zamfir, and Emin Gün Sirer, “A Call for a Temporary Moratorium on ‘The DAO,’” May 30, 2016, accessed
February 3, 2020, https://docs.google.com/document/d/10kTyCmGPhvZy94F7VWyS-dQ4lsBacR2dUgGTtV98C40/edit#.
28
Evan D. Wolff, Matthew B. Welling, and Tyler A. O’Connor, “The DAO Hack Provides Lessons for Companies Using
Blockchain and Distributed Ledger Technology,” Crowell Moring, June 27, 2016, accessed February 3, 2020,
www.crowell.com/NewsEvents/AlertsNewsletters/all/The-DAO-Hack-Provides-Lessons-for-Companies-Using-Blockchainand-Distributed-Ledger-Technology.
29
Peter Vessenes, “More Ethereum Attacks: Race-To-Empty is the Real Deal,” Vessenes, June 9, 2016, accessed February
10, 2020, https://vessenes.com/more-ethereum-attacks-race-to-empty-is-the-real-deal/.
30
Stephan Tual, “No DAO Funds at Risk Following the Ethereum Smart Contract ‘Recursive Call’ Bug Discovery,” Slock.it
Blog, June 12, 2016, accessed February 17, 2020, https://blog.slock.it/no-dao-funds-at-risk-following-the-ethereum-smartcontract-recursive-call-bug-discovery-29f482d348b#.mbfqikiyo.
31
Christoph Jentzsch, “The History of the DAO and Lessons Learned,” Slock.it Blog, August 24, 2016, accessed March 3,
2020, https://blog.slock.it/the-history-of-the-dao-and-lessons-learned-d06740f8cfa5.
Page 162 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
ENDNOTES
Page 8
9B20E017
32
Siegel, op. cit.
Osman Gazi Güçlütürk, “The DAO Hack Explained: Unfortunate Take-off of Smart Contracts,” Medium, August 1, 2018,
accessed March 7, 2020, https://medium.com/@ogucluturk/the-dao-hack-explained-unfortunate-take-off-of-smart-contracts2bd8c8db3562.
34
“The Dao, the Hack, the Soft Fork and the Hard Fork,” CryptoCompare, March 12, 2019, accessed July 10, 2020
www.cryptocompare.com/coins/guides/the-dao-the-hack-the-soft-fork-and-the-hard-fork/.
35
David Gerard. Ch. 10: Smart Contracts, Stupid Humans. Excerpt “The DAO: The Steadfast Iron Will of Unstoppable Code.”
In Attack of the 50 Foot Blockchain (108–110), https://davidgerard.co.uk/blockchain/the-dao/.
36
For a discussion of the motivations for participation in a decentralized autonomous organization, see “Tokenized Networks,”
op. cit.
37
Bruce Schneier, “There’s No Good Reason to Trust Blockchain Technology,” Wired, February 6, 2019, accessed February
3, 2020, www.wired.com/story/theres-no-good-reason-to-trust-blockchain-technology/.
38
Siegel, op. cit.
Page 163 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
33
9 -5 1 5 -0 6 0
JOHN A. QUELCH
MARGARET L. RODRIGUE Z
Carolinas HealthCare System: Consumer Analytics
In 2014, Dr. Michael Dulin, chief clinical officer for analytics and outcomes research and head of
the Dickson Advanced Analytics (DA2) group at Carolinas HealthCare System (CHS), was preparing
for a planning meeting with Carol Lovin, executive vice president and chief strategy officer at CHS.
In the three years since DA2 was formed, the team had successfully unified all analytics talent and
resources into one group that served CHS. Rapid increases in computing power and decreases in data
storage costs had enabled DA2’s data architects to build predictive models incorporating complex
clinical, financial, demographic, and claims data that would have been impossible to create only a
few years before. Although DA2 had blazed the trail for applied analytics in healthcare, other players
in the value chain were making increased investments in their own modeling capabilities.
Healthcare payers, such as Humana and UnitedHealth, were increasingly making analytics the
focus of a strategic shift toward consumer-centric healthcare, going so far as to create targeted
communications strategies for different patient segments and engaging behavioral health companies
to provide exercise, nutrition, and other programs that would reduce the healthcare costs of their
highest-risk patients. While many agreed that analytics could help the healthcare industry reduce
costs and increase access to care, CHS recognized that privacy protections on patient data, as well as
competitive rivalries, restricted the sharing of data among the various healthcare stakeholders.
Dulin also noted the entry of consumer tech companies into the healthcare space; in 2014, both
Apple and Google announced features in their new mobile operating systems that aggregated and
tracked the output from various health wearables (like heart-rate monitors or step counters), as well
as electronic medical record (EMR) data. Apple’s HealthKit could even incorporate the results of lab
tests into the dashboard (with the user’s permission). Although the tech giants did not yet have
access to claims or clinical data, they could potentially enter the field by acquiring an EMR company.
Their expertise in analytics, access to demographic and location data, as well as the broad consumer
adoption of their devices, led Dulin to consider which industry players consumers would trust to
integrate their healthcare data in the future and what role DA2 could play.
Professor John A. Quelch and Research Associate Margaret L. Rodriguez prepared this case. It was reviewed and approved before publication by
a company designate. Funding for the development of this case was provided by Harvard Business School and not by the company. . Professor
Quelch is the Charles Edward Wilson Professor of Business Administration at the Harvard Business School and Professor in Health Policy and
Management at the Harvard T.H. Chan School of Public Health. HBS cases are developed solely as the basis for class discussion. Cases a re not
intended to serve as endorsements, sources of primary data, or illustrations of effective or ineffective management.
Copyright © 2015 President and Fellows of Harvard College. To order copies or request permission to reproduce materials, call 1-800-545-7685,
write Harvard Business School Publishing, Boston, MA 02163, or go to www.hbsp.harvard.edu. This publication may not be digitized,
photocopied, or otherwise reproduced, posted, or transmitted, without the permission of Harvard Business School.
Page 164 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
REV: SEPTEMBER 8, 2015
515-060
Carolinas HealthCare System: Consumer Analytics
In 2014, the Carolinas HealthCare System, headquartered in Charlotte, North Carolina, owned
and managed hospitals and acute care facilities that served over 2.2 million patientsa per year across
three states (North Carolina, South Carolina, and Georgia). CHS was one of the oldest healthcare
systems in the U.S.; its origins could be traced to the state’s first civilian hospital, Charlotte Home and
Hospital, established in 1876. The Charlotte Home and Hospital operated until 1940 (although the
name was later changed to St. Peter’s Hospital), when it was replaced by a new facility, the Charlotte
Memorial Hospital, in a different location. During World War II, the hospital’s financial difficulties
led Rush S. Dickson, a local businessman, to lobby the city and county governments for larger
reimbursements for emergency and indigent patients, and solicited financial support from local
nonprofits and corporations. In 1943, the Charlotte-Mecklenburg Hospital Authority was organized
under the North Carolina Hospital Authorities Act, which provided for oversight mechanisms for
Charlotte Memorial Hospital (including rules governing the construction of new facilities, funding,
and management of day-to-day operations). The act also provided the legal and financial frameworks
to support patients who could not afford to pay for healthcare services.
In 1990, the name of the Charlotte Memorial Hospital was changed to Carolinas Medical Center
(CMC) to reflect the hospital’s increasing focus on education. That year, the facility was designated
an “Academic Medical Center Teaching Hospital” by the state of North Carolina (one of only five
hospitals in North Carolina to receive the designation). Five years later, the authority changed the
name of the growing hospital network to CHS. In 2007, CHS opened the Levine Children’s Hospital,
which housed more than 30 medical specialties. In 2010, CHS announced a 10-year, $500 million
investment to advance cancer treatment strategies and research through the creation of the Levine
Cancer Institute. In 2010, CMC (now a part of CHS) was designated the Charlotte Campus of the
University of North Carolina (UNC) School of Medicine and hosted third- and fourth-year medical
students.
By 2014, CHS had become the biggest healthcare provider in North Carolina, with more than
61,000 full-time and part-time employees and an annual budget of over $7.7 billion (see Exhibit 1 for
selected financial data). Its medical education and research center included over 300 residents and
fellows pursuing a variety of medical specialties and had established research relationships with
Oxford (stroke), UNC (dementia), Duke University, and many other academic centers across the U.S.
CHS operated 900 care locations and 7,494 licensed beds in three states, including 39 hospitals (21 of
which were managed by CHS, and 18 of which were owned), as well as additional virtual care
services. Roughly 75% of patients were located in North Carolina, and CHS spent $20 million each
year on community outreach in the greater Charlotte area alone. CHS tracked patient satisfaction
with mailed surveys or follow-ups within days of an appointment or discharge. Satisfaction was
measured by a patient’s likelihood to recommend CHS. As part of its role as a public healthcare
system, CHS provided healthcare services to underserved patients and communities. CHS offered
financial support to patients without insurance (or who were underinsured), subsidies for Medicare
and Medicaid recipients, as well as funding for its education, behavioral health, and community
health clinics. CHS gave medical supplies and equipment to nonprofits valued at over $1.5 million in
2013 (see Exhibit 2 for a full list of CHS charitable expenditures for 2013). Roughly 62% of annual
revenue came from Medicare and Medicaid patients.
a Patients were considered active if they had engaged with one of the CHS sites (including primary care facility, hospital,
worksite clinic, a virtual visit, or a trip to a clinic inside of CVS) at least once in the prior 18 months. Roughly 13,000 patients
fell off the active rolls each month, many due to relocating out of state, death, or attrition.
2
Page 165 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Carolinas HealthCare System Background
Carolinas HealthCare System: Consumer Analytics
515-060
Health Provider Industry Background
In 2012, healthcare expenditures in the U.S. totaled over $7,600 per capita versus an average of
$2,800 per capita among OECD countries.1 The majority of healthcare expenses were paid for by
government-sponsored coverage, such as Medicare and Medicaid, b or by health insurance companies
that sold plans to employers and individuals (either to those who purchased via healthcare exchanges
or those who were eligible for Medicare Advantagec). Government reimbursements per Medicare
patient to cover healthcare had declined over time (see Exhibit 5); however, the 2010 Affordable Care
Act (ACA) offered providers who were organized as accountable care organizations (ACOs) d a share
of the cost savings generated in the delivery of care to Medicare patients, so long as minimum quality
thresholds were met.
Key changes to the U.S. healthcare landscape over the decade prior had influenced healthcare
providers like CHS and other healthcare stakeholders to revisit their care delivery models:

Fee-for-value instead of fee-for-service After the ACA’s passage, hospital compensation was
determined in part by the quality of outcomes, rather than simply on a fee-for-service basis as
before. Hospitals faced penalties for high readmission rates and hospital-acquired conditions,
but could also receive financial rewards for exceeding clinical quality outcome or patient
satisfaction benchmarks. Many healthcare providers sought to quickly build capabilities in
analytics and measurement in order to track quality improvements, and to shift the
organizational focus toward the continuous improvement of patient care.

Physician shortages In the U.S., there were roughly 2.5 physicians per 1,000 people (versus
an average of 3.3 per 1,000 among comparable OECD countries).2 The ACA was expected to
exacerbate the supply-demand shortfall in the future, since it increased the population
covered by health insurance. In 2014 alone, nearly 32 million new people entered the
healthcare system.3 The Association of American Medical Colleges estimated that, by 2025, the
U.S. would face a shortage of over 130,000 doctors. 4

Digitization of healthcare In 2011, the U.S. Centers for Medicare and Medicaid Services
(CMS) established an incentive system for doctors’ offices and hospitals to switch from paper
to EMRs. Hospitals that served Medicare patients could receive up to a $2 million incentive
for adopting EMRs.5 Although patients were free to view and request corrections to the data
in their EMRs, the platforms made it difficult for providers to extract and model the data held
b Medicare was the federal health insurance program offered to seniors aged 65 and older. Medicaid was offered to low-
income individuals who were unable to obtain healthcare via the exchanges or an employer.
c Insurers offered Medicare Advantage plans to eligible patients who chose to receive their benefits via the insurer’s network.
d Qualified ACOs agreed to be accountable for the overall care of their Medicare patients, obtain adequate participation of
primary care physicians, create processes around evidence-based medicine, report on quality and costs, and coordinate care.
3
Page 166 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
By 2014, the vision of CHS had remained unchanged for two decades: healthcare, education, and
research. That year, Lovin led an initiative to renew the strategic road map to guide CHS’s future
growth. She worked with her executive team colleagues to craft a strategy that provided for
personalized, high-quality service across a single, unified enterprise. “Our customers are the
consumers and patients first, payers second,” said Lovin. The team developed a list of strategic
priorities (see Exhibit 3) and performance measures (see Exhibit 4) to guide the organization toward
its goals.
515-060
Carolinas HealthCare System: Consumer Analytics

Shift to outpatient In an effort to cut costs and increase access to care, providers encouraged
patients to seek care via outpatient facilities, or even via virtual checkups, rather than at highcost treatment locations like emergency departments (EDs). Outpatient care comprised 51%
of health expenditures in the U.S., whereas the average among OECD countries was 33%.6 In
2013, CHS chief executive Michael Tarwater observed: “More than 90% of our patient
encounters now take place in a setting other than the bedside of an inpatient hospital room.” 7

New entrants The shift to outpatient care led consumers to seek more convenient and
inexpensive healthcare services. Retailers (including CVS, Walmart, Target, and Kroger)
began opening healthcare clinics staffed by nurse practitioners in their stores as early as 2000.
By 2014, there were 1,600 walk-in clinics in the U.S., and the number was expected to reach
nearly 3,000 by 2015.8 Cost of care at the clinics for three common illnesses averaged $110,
versus $166 at doctors’ offices and $570 in EDs.9
HIPAA Regulations
In 1996, the U.S. House of Representatives passed the Health Insurance Portability and
Accountability Act (HIPAA). HIPAA provided for both the portability of employer-provided health
insurance (which enabled an individual to keep the same health insurance between jobs) and the
establishment of the first set of national security and confidentiality standards for patient health
data.10 The U.S. Department of Health and Human Services (HHS) established a privacy rule to
protect individually identifiable patient data, while still enabling stakeholders to access the data
needed to provide care. Data protected under HIPAA included physical or mental health conditions
(including those that occurred in the past), healthcare provided, payments made for healthcare
received, and demographic information that could be used to identify the individual (see Exhibit 6
for examples).
Health plans (payers), healthcare providers, and healthcare “clearinghouses” (which included
billing services, community health management information systems, value-added networks, and
other business associates) were all subject to the privacy standards outlined in HIPAA. Healthcare
organizations had to notify patients of their privacy rights (including acceptable use of personally
identifiable information) and obtain signed authorization from patients for any use of individual data
beyond treatment, payment, and healthcare operations.11 Restrictions on data use could be waived if
data were “de-identified,” either by the formal assessment of a statistician to prove individual
anonymity was retained, or by the removal of indicators used to identify the individual and his or her
relatives, employer, or household members.12 De-identified data became propriety to the company
that held it.
HIPAA supported use of patient data to perform analysis necessary to make improvements to
healthcare systems, including quality reviews, utilization reviews, and population reviews (often for
a given condition, such as diabetes). Such information could be shared with other entities also subject
to HIPAA, such as payers. Employers like CHS, who both provided healthcare and self-funded
insurance to their employees, were not permitted to view their employees’ disaggregated healthcare
data; in addition, employee information collected through the human resources department was kept
4
Page 167 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
in the EMRs. In 2014, Google and Apple each announced healthcare dashboards that would
aggregate data from wearable devices, including scales, running apps, and sleep trackers, into
a consumer-friendly dashboard. Apple even partnered with several EMR companies to make
medical records and lab results available to consumers on their iPhones via the single
dashboard.
Carolinas HealthCare System: Consumer Analytics
515-060
separately from employee health data. Those who contravened the HIPAA privacy rule could be
subject to civil and/or criminal charges and fines of over $1 million. 13
Following CMC’s designation as a teaching hospital in 1990, CHS established partnerships with 11
academic research centers. There was no medical school located in Charlotte at the time, so one of the
independent centers was designated a branch of UNC. The Dickson Institute for Health Studies, as
the center was known, provided education and training facilities to 300 residents, nurses, and
graduate students. It partnered with UNC–Charlotte and UNC–Chapel Hill to provide health-datafocused projects to PhD candidates. The Dickson Institute initially focused on improving acute care
quality, but the mission was later broadened to cover an array of healthcare projects under the
leadership of Dr. Roger Ray, executive vice president and chief physician executive at CHS. Although
the Dickson Institute conducted research, data analysis, and public reporting of key metrics for CHS,
Ray noted that its activities did not influence the majority of day-to-day operations that occurred
within the CHS network.
Beginning in the 2000s, CHS embarked on a visioning and process-development project to
determine what data analytics capabilities would be integral to CHS’s operations in the future. It
determined it would need to develop a distributed data system and create a corporate data
warehouse and decided to coalesce analytics personnel who were currently working in small silos
throughout the organization to achieve the vision. CHS Information Services leadership anticipated
that cost of data storage would plummet, based upon their experience implementing the EMR system
at CHS in 2006, so the team decided to build generous data storage to support the new analytics
team. Prior to CHS’s adoption of EMR, most of the data it collected was financial data generated
through transactions. The EMR rollout served as a proof-of-concept that patient data and financial
data could be combined to provide decision support. With increasing computing power, CHS and
other providers could collect, store, and model a variety of clinical data (including unstructured data)
that it could not have assembled previously.
CHS hired consultants to advise the organization on the creation of a unified analytics group
through the development of a high-level road map. Lovin was interviewed by the consultant group
and asked to provide executive leadership for the initiative, and she recruited Dulin to help execute
on CHS’s vision of creating a unified, data-driven system. Dulin was trained in electrical and
biomedical engineering and worked as a quality control specialist for a microchip manufacturer
before attending medical school. At the time, there were many groups within CHS that handled
analytics, but most were tied to a particular business, function, or geography with no integration.
CHS decided to differentiate on the basis of its analytics capability and made investments to raise the
analytics “IQ” of the organization. CHS leadership and others believed the system could move
beyond its current analytics-related key performance indicators (KPIs) to include data in key
decisions that would change patient care and save money over the long term. The new analytics
group could have adopted a hybrid model structure wherein descriptive analysis linked to
performance metrics was performed internally and more rigorous analytics were outsourced, but
CHS instead chose to develop both its foundational descriptive analytics and more advanced
predictive and prescriptive models in-house.
Creation of DA2
Dickson Advanced Analytics (DA2) was launched in 2011 with an annual budget of $14 million. It
was initially composed of 70 people sourced from the disparate internal analytics groups that
5
Page 168 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Dickson Advanced Analytics (DA2)
515-060
Carolinas HealthCare System: Consumer Analytics
The success criterion for DA2 was to improve outcomes, rather than increase the size of CHS. As
Ray explained, “Healthcare is a massive cottage industry.” He estimated that one-third of clinical
work was “nonstandard,” meaning that it diverged from a care plan and/or choices were made in
the absence of evidence. Deploying analytics could help improve outcomes for work that was
previously considered nonstandard. DA2 also played an important role in the communication
strategy, since improving quality of outcomes often required engaging the patient to change his or
her behavior. Many at CHS believed the key to DA2’s success was a continuing commitment to build
strong relationships with the physicians and nurses. The data DA2 used was collected at many points
of care through the CHS networks; in addition, any recommendations and tools derived from the
data had to be implemented by the physicians (often by the clinical lead).
DA2 sat outside of the organization’s businesses, but still operated as a cost center. DA2 reported
to Lovin and the strategy function, rather than information services. As a result, Dulin and his team
created a business plan for DA2 to show its return on investment (ROI) over the long term, and
prioritized projects of strategic importance to the organization. DA2 was composed of five groups:
“Applied Outcomes Research,” “Data Services,” “Client Services,” “Project Management,” and
“Advanced Analytics.” Compulsory reporting was one of DA 2’s core responsibilities: most of the staff
of 120, including 12 PhD-level analysts, worked on reports submitted to the government. Within DA 2,
over time, a 15-person team dedicated to serving the medical group performed predictive work. A
smaller research team worked to help measure the strength of the models and the interventions they
delivered. Another DA2 team studied cost analytics to measure the ROI of quality-increasing
investments. The businesses focused on identifying revenue opportunities, and DA2 helped to assess
what each opportunity would be worth. Patient data could be used to support investment decisions,
such as which surgical devices to purchase, since patient data contained information on the quality of
outcomes (which DA2 then combined with cost and device lifespan to assess ROI). One of Dulin’s
responsibilities was managing DA2’s internal customers, whose demand for analytics quickly
outstripped the team’s capacity. Many saw the potential for DA2 to become an additional revenue
stream by outsourcing its analytics services to third parties in the future.
Shortly after DA2 launched, the team received more than twice as many requests as it had capacity
to accept. DA2 created an advisory board for issues related to effectiveness, priority setting, and other
key focus areas. Then, CHS created a priority-setting process for developing predictive analytics:
first, DA2 conducted interviews with the clinical teams and an internal focus group (which was
incorporated into the proposals for each project); then DA2 determined each proposal’s alignment
with existing systems and CHS strategy, and balanced the resources the proposal required against
those needed for DA2 to conduct ongoing reporting and data warehouse management
responsibilities. Finally, it assessed the projects against a matrix of criteria, including size, patient
impact, mortality vs. quality-of-life improvements, speed of implementation, cost, and commercial
viability. Included in the process, DA2 would provide updates to the Clinical Integration Council
(CIC), the highest physician leadership team.
DA2 sought outside partners to improve the breadth and quality of its data. In 2013, CHS
partnered with four healthcare systems and IBM to form the Data Alliance Collaborative (DAC),
which focused on improving population health by creating scalable data models. The healthcare
partners contributed data to a communal warehouse, which contained data from over 100 hospitals
6
Page 169 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
preceded DA2. Much of DA2’s capacity was devoted to providing tools to support CHS-affiliated
hospitals in delivering best-in-class healthcare to patients, although, over time, DA2 also developed
analytical tools for evidence-based population health management, personalized patient care, and
predictive modeling.
Carolinas HealthCare System: Consumer Analytics
515-060
CHS used strategic partnerships to incorporate provider and payer data with consumer data into
its predictive algorithms. Dulin believed that such data would give DA2 additional insight into
communities and patient populations, and provide indications for early interventions for potential
problems beyond EMR data. The spending data, along with other inputs, were used to create a risk
score for admitted patients, which were then distributed to doctors and other healthcare providers to
reprioritize care delivery. This approach allowed the hospital system to focus limited resources on
high-risk patients to improve their outcomes and their health status.
DA2 and the Data Governance Committee
In 2012, CHS established a data governance committee, headed by Alicia Bowers, vice president of
corporate privacy, and Michael Trumbore, assistant vice president of advanced analytics (and
sponsored by Ray). The group included representatives from DA2, clinical and translational research,
information services, human resources, financial services, audit services, systems business, and the
office of the general counsel. Members of the CHS institutional review board (IRB) e also had seats on
the committee. Since CHS was a research organization, it followed the IRB standards to guide its
handling of data. For example, one of the IRB privacy standards dealt with creating a geographically
informed dataset and prevented disclosure of information if fewer than 50 people lived within a
single census tract.
The data governance group was formed to protect, manage, and determine accountability for the
data generated in the day-to-day operations of CHS. CHS recognized that the organization had a data
strategy, whether or not it was made explicit. With the formation of the data governance group, it
hoped to signal executive support for DA2 and facilitate engagement with the information services,
clinical, and business groups. The group met monthly to:

Create data governance policies.

Prioritize data governance initiatives (including pilot programs).

Define data governance policies, standards, processes, metrics, and principles.

Communicate the vision and activities of the group to the broad CHS organization.

Address the governance structure, data access, and data quality.
A key initiative of the advisory group was the appointment of “domain owners,” who were
responsible for upholding governance standards for the data and business processes in a given
domain. Domains referred to clusters of data that were organized around CHS businesses. For
example, the research domain might contain documentation of patient consent, IRB compliance, and
grant information, whereas the patient domain could contain EMR information and treatment plans.
The domain owners led multifunctional teams composed of process owners, data stewards, and
project managers, as well as representatives from information services and data governance who
were responsible for tracking performance against KPIs, data quality, and compliance with internal
data governance policies. The domain owners reported to the Strategic Governance Council, chaired
by Dulin, who ultimately reported to the Executive Governance Council, led by Lovin (see Exhibit 7
for an organization chart).
e The IRB was a committee formed to protect the welfare of human subjects involved in research, including maintaining the
subjects’ privacy. All IRBs were registered with the Office for Human Research Protections, a division of the U.S. Department
of Health and Human Services.
7
Page 170 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
and 1,600 non-acute care sites serving 28 million patients.14 IBM provided the data infrastructure,
which could incorporate clinical, claims, and financial data, to support the analytics.15
515-060
Carolinas HealthCare System: Consumer Analytics
By November 2014, DA2 had achieved three key results: it collected and handled vast amounts of
data efficiently; it created a data governance structure; and it helped shift the organization away from
an anecdotal culture to an evidence-based one. The year prior, DA2 had spent $5 million16 to create its
enterprise data warehouse (EDW). The EDW initially contained 10 terabytes of data derived from the
system’s 1.5 petabytes of data, which doubled in size by 2015. The data warehouse incorporated
clinical, billing, and claims data, which enabled DA2 to create models including hundreds of different
patient variables. In 2014, DA2 had a pipeline of nearly a dozen predictive risk models in
development and consistently more requests from the organization than it had capacity to accept.
Lovin said, “We went from having no DA2 to wanting to check their opinion on everything.” DA2
had made progress toward changing the culture of CHS to be evidence-based and data-driven;
however, its success meant DA2’s capacity was strained by demand for its analytics services. DA 2
began to provide the business lines with tools and education so that some analytics could be
performed independently and DA2 could reserve capacity for more complex questions.
Key DA2 Pilots
In 2014, the three ongoing strategic priorities for DA 2 were to predict health needs; continually
enhance patient outcomes; and drive transformative solutions to address community health issues.
By that time, DA2 had launched a number of successful pilots covering a variety of medical
conditions, geographies, and functional capabilities.
Mapping underserved communities In 2009, the Dickson Institute launched a project to
reduce unnecessary emergency department (ED) utilization in Charlotte by identifying areas
underserved by primary care facilities. The project leveraged descriptive data and clinical data from
local primary care and ED facilities in Charlotte to find the best variables to indicate poor access to
primary care. Over 367,000 clinical records were sourced from all 2007 patient visits to CHS primary
care and ED facilities. Data that did not contain the patient’s address, or from patients who lived
outside of the county, was excluded, which left a dataset of 187,000 ED visits and 50,000 primary care
visits, as well as patient insurance status, f to be used in the model. 17 For descriptive data, the team
used U.S. Census data at a census tract level. g
After mapping and testing multiple variables, the team selected five for use in the primary care
access model: population density, median household income, the uninsured/Medicaid population,
the incidence of ED utilization for primary care–treatable or preventable conditions, and the
proportion of the population that currently used primary care facilities.18 Equal weight was given to
each variable in the model, although the use of primary care facilities was given an inverse, but still
equal, weight (see Exhibit 8 for variables). The numeric value of each variable was calculated for
each census tract before being combined to create the single measure of need for primary care
facilities in that area. Census tracts with values greater than one standard deviation above the mean
were highlighted as high-need areas (see Exhibit 9 for map).
The geo-tagging technique was relatively inexpensive and could quickly identify community
candidates for additional primary care clinics. The U.S. Census data was free to the public, and, once
all data was geocoded, the process of building a new model with weighted attributes took only a few
f Patient insurance status contained five categories: Medicare, Medicaid, commercial, uninsured, and other.
g Census tracts were geographic regions defined by the U.S. Census Bureau that contained between 1,200 and 8,000 people.
Census tracts were typically slightly smaller than zip codes.
8
Page 171 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
DA2 in 2014
Carolinas HealthCare System: Consumer Analytics
515-060
hours. The team believed a similar approach could be used to anticipate localized future demand for
healthcare professionals, to develop interventions to improve access to primary care facilities, to
provide data to support policy decisions regarding healthcare initiatives, and to measure the impact
of interventions designed to improve healthcare access.
The Readmission Predictive Risk Model was launched by DA2 in
2013 to help the clinical teams in CHS hospitals identify high-risk patients. DA2 developed an
algorithm that calculated a “readmission risk score” for each admitted patient. The score was
modeled from data on the thousands of patients that had been discharged from one of the CHS
facilities in the prior two years. Out of 600 variables in the data, DA2 found 40 to be highly predictive
of readmission, including history of ED visits, sodium levels, language, and late-stage renal disease.19
Patients who were identified as high risk for readmission within 30 days of discharge received extra
focus from clinicians while still in the hospital. The model had an accuracy rate of 79% in predicting a
patient’s risk of readmission within 30 days of discharge. In addition, the model clustered patients
into one of five segments (see Exhibit 10), each of which possessed a unique set of guidelines for
transition planning (which were given to the discharge care manager). The discharge care manager
could then select appropriate interventions, such as scheduling follow-up visits to the patient’s home,
helping patients manage their medications, and connecting them with dietitians, trainers, and/or
coaches to provide appropriate follow-up care.
Advanced Illness Management CHS created an Advanced Illness Management Group
(AIM), which reported to Ray, to help patients with complex medical conditions avoid hospital stays.
Those patients carried a higher risk of hospital readmission, so the program offered access to a team
of experts who helped the patients better understand their health conditions, symptoms, medications,
and lifestyle choices in order to empower them to manage their own health and reduce unnecessary
ED visits and hospitalizations. Eligible patients had at least two chronic conditions, had visited the
ED or hospital more than twice over the last six months, took multiple medications to treat the same
condition, and were not actively involved in another care management or intensive health program.
The first AIM cohort included 25 patients who collectively visited the hospital 96 times (they had
visited the ED 41 times and were hospitalized 55 times) in the six months before the start of the
program in 2014. The multidisciplinary AIM team included licensed clinical social workers, nurse
practitioners, licensed practical nurses, and registered nurses. For each patient, the team assessed the
unmet educational, psychosocial, and resource needs, and created a care plan in conjunction with the
physician’s medical plan. The team was in frequent contact with patients to monitor changes in
health status, social circumstances, and/or psychosocial needs; to answer questions regarding
medications or medical jargon; to discuss care options before going to the ED; and to remind patients
of upcoming doctor’s appointments. In select cases, licensed practical nurses attended the
appointments with the patients. Patients in the first cohort experienced less pain, improved quality of
care, and increased satisfaction, while also incurring lower costs to the healthcare system. The
patients collectively visited the hospital only 33 times in the six months following the start of AIM.
Due to the initial success of the AIM program, CHS enrolled additional patients in cohorts two and
three.
Patient segmentation model In 2014, CHS reviewed the data of 2.2 million active patients
who received care within the system and collected 2,000 data points per patient, including clinical
data, medication compliance, education attainment, socioeconomic factors, and consumer spending
profiles (DA2 hoped to one day add genetic data and data generated by wearables). DA2 created a
segmentation model that grouped patients into one of seven distinct segments (e.g., “high risk of
cancer”), so that patients in each group could receive care and communications tailored to their
9
Page 172 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Reducing readmissions
515-060
Carolinas HealthCare System: Consumer Analytics
The benefits to CHS were twofold: segmentation helped clinicians to identify high-risk patients
quickly and help them change harmful behaviors; segmentation also helped CHS to estimate the
expected cost of providing care to each segment and influenced how CHS bid on new business
contracts with healthcare payers. Providers like CHS were often at a disadvantage when negotiating
with payers, since the payers frequently had more information on the patients than the providers,
who often lacked well-developed analytics capabilities. Since CHS acted as a payer for its own
employees, it was aware of the types and quality of data the payers possessed. Trumbore believed
that the CHS possessed better data than the average payer, as payers could only see what happened
(e.g., claims data), rather than the clinical treatment process. Over time, CHS would use the insights
gleaned from the segment analysis to optimize its care delivery model.
Key Challenges Ahead for DA2
As Dulin and his team prepared their strategic plan for the next three years, he pondered which of
the existing pilots might be extended to different yet related issues without requiring the design and
implementation of an entirely new model. The internal demand for DA2’s services could fill its
current capacity several times over. CHS identified many opportunities for DA2 to reduce waste
while also improving outcomes, but recognized the need to satisfy as many internal constituents as
possible. CHS leadership was interested in exploring external business opportunities that could
potentially turn DA2 into a source of profit for CHS, particularly given the mounting investments
DA2 had received for its technological infrastructure. However, high internal demands for DA2’s
services constrained Dulin’s ability to test DA2’s capabilities in the external marketplace. Another key
focus for DA2 was ensuring insights translated into action through the creation of user-friendly
reports, and ultimately, consistent care plans and sets of orders to follow for patients with a given
condition. Engaging with clinicians to ensure that data from the predictive models improved their
workflow would be a core focus of DA2 for years to come.
10
Page 173 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
specific needs. Within each segment, DA2 selected a subgroup from which to gather qualitative data
on segment-specific lifestyle and healthcare needs. The qualitative data in turn influenced the
recommendations for the communication plans for the sufferers of various disease groups within
each segment (a process managed by the AIM group, which partnered with care managers, social
workers, pharmacists, and rehabilitative and palliative care centers).
Carolinas HealthCare System: Consumer Analytics
CHS Summary Financials, 2013
Revenues
Tertiary & Acute Care Services
Continuing Care Services
Specialty Services
Physicians’ Services
Other Services
Non-Operating Activities
Total Revenues
Expenses
Wages, Salaries & Benefits
Materials, Supplies & Other
Depreciation & Amortization
Financing Costs
Funding for Facilities, Equipment & Programs
Total Expenses
Source:
Dollar Total (million)
Percentage of Total
$5,832
$293
$53
$1,487
$225
$469
$8,358
68%
4%
1%
18%
3%
6%
100%
$4,616
$2,615
$454
$125
$547
$8,358
55%
31%
5%
2%
7%
100%
Company documents.
Exhibit 2
CHS Charitable Expenditures, 2013
Charitable Expenditures
Cost of financial assistance to uninsured patients
Bad debt costs by patients who do not pay for services
Losses incurred by serving Medicare patients a
Losses incurred by serving Medicaid patients
Cost of community-building activities and other services
Cost of medical education, research, and cash and in-kind contributions to charities
Total value of uncompensated care and other community benefits
Source:
$, million
$324
$290
$563
$161
$56
$146
$1,540
Company documents.
a Medicare and Medicaid offered fixed compensation per recipient, which occasionally fell short of the actual cost of care;
hospitals were not permitted to refuse care to these patients.
11
Page 174 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Exhibit 1
515-060
515-060
Carolinas HealthCare System: Consumer Analytics
CHS Strategic Priorities, 2014
Strategic Priority
Details
Quality & Patient Experience
Design a customer relationship management tool; increase health literacy;
improve critical and diabetes care; improve clinical outcomes via collaboration
software.
Integrated System of Care
Deploy the CHS care management platform; manage population health; deploy
virtual care; transform service lines, continuing care, community health and point
of care.
Strategic Growth
Deliver competitive, consumer-facing retail services; commercialize existing CHS
services; develop payer/risk strategies with payers; deliver best-in-class specialty
care.
Transformative Operations
Improve processes to reduce patient wait times; share and implement best
practices across the organization; put the patient first in operational decisions;
leverage engaged workforce.
Source:
Company documents.
Exhibit 4
CHS Strategic Performance Measures, 2014
Performance Measures
Metrics
Quality & Patient Experience
Inpatient mortality; breast cancer screenings; physician satisfaction; patient
likelihood to recommend; patient safety score; diabetes treatment outcomes;
appropriate care score (for both ambulatory and acute care).
Integrated System of Care
Readmission rate; CHS/payer collaboration performance; progress against
integrated system of care goals; Medicare spend per beneficiary; CHS medical
plan performance.
Strategic Growth
Actively managed primary care patients; population share; commercial/managed
care population; use rates (versus industry benchmarks); evidence-based
screening volumes; commercialized products or services.
Transformative Operations
Average length of stay; emergency department transformation score; operating
cash flow margin; productivity improvement; revenue cycle improvement;
process enhancement product standardization savings and speed.
Teammate Engagement
Commitment indicator score (percentile ranking)
Community Benefit
Number of individuals screened for pre-diabetes; hours of community service.
Source:
Company documents.
12
Page 175 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Exhibit 3
Carolinas HealthCare System: Consumer Analytics
Source:
Medicare Advantage Payments as a Percentage of Traditional Medicare, 2006–2014
“Medicare at a Glance,” Kaiser Family Foundation, September 2, 2014,
http://www.cdc.gov/obesity/data/adult.html, accessed October 2014.
Exhibit 6
Personally Identifiable Data Protected under HIPAA, 2014
Protected Personal Data Types
Names
Addresses (including zip code)
Dates (birth, admission, discharge, death)
Telephone numbers
Fax numbers
E-mail addresses
Social security numbers
Medical record numbers
Health plan beneficiary numbers
Source:
Account numbers
Certificate/License numbers
Vehicle identifiers and serial numbers (including license plate)
Device identifiers and serial numbers
Web Universal Resource Locators (URLs)
Internet Protocol (IP) addresses
Biometric identifiers, including finger and voice prints
Full-face photographic images and any comparable images
Any other unique identifying number, characteristic, or code
Adapted from “HIPAA Background,” Office of Corporate Compliance, University of Chicago Medical Center,
February 2010, http://hipaa.bsd.uchicago.edu/background.html, accessed February 2015.
13
Page 176 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Exhibit 5
515-060
515-060
Carolinas HealthCare System: Consumer Analytics
Source:
Company documents.
Exhibit 8
Source:
Data Governance Committee Organizational Chart, 2014
Variable Definition Process for the Primary Care Access Model,a 2013
Dulin, Michael F., Thomas M. Ludden, Hazel Tapp, Heather A. Smith, Brisa Urquieta de Hernandez, Joshua
Blackwell, and Owen J. Furuseth. “Geographic information systems (GIS) demonstrating primary care needs for a
transitioning Hispanic community.” The Journal of the American Board of Family Medicine 23, no. 1 (2010): 109–120.
Reproduced by permission of the American Board of Family Medicine.
a ED referred to emergency departments; AHP referred to analytic hierarchical process.
14
Page 177 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Exhibit 7
Carolinas HealthCare System: Consumer Analytics
Source:
Map of High-Need Areas in Charlotte Produced by the Primary Care Access Model, 2013
Dulin, Michael F., Thomas M. Ludden, Hazel Tapp, Heather A. Smith, Brisa Urquieta de Hernandez, Joshua
Blackwell, and Owen J. Furuseth. “Geographic information systems (GIS) demonstrating primary care needs for a
transitioning Hispanic community.” The Journal of the American Board of Family Medicine 23, no. 1 (2010): 109–120.
Reproduced by permission of the American Board of Family Medicine.
Exhibit 10
Population Segments of the Readmissions Model, 2013
Segments
Low
Risk
Medium
Risk
Insured Healthy Adult
Medicaid Pediatric
Medicare Independent
Medicare with frequent visits
Middle age with frequent visits
Total
14.4%
4.1%
5.1%
0.8%
0.6%
25%
10.9%
2.5%
6.6%
2.7%
2.3%
25%
Source:
High
Risk
6.0%
1.2%
6.1%
5.6%
6.0%
25%
Very High
Risk
Total
4.2%
0.4%
5.1%
5.2%
10.3%
25%
35.5%
8.2%
22.9%
14.2%
19.1%
100%
Company documents.
15
Page 178 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Exhibit 9
515-060
515-060
Carolinas HealthCare System: Consumer Analytics
Endnotes
1 “Health Expenditure—OECD Health Statistics 2014,” OECD iLibrary, http://www.oecd.org/els/health-systems/health-
expenditure.htm, accessed February 2015.
http://kff.org/slideshow/u-s-health-care-resources-compared-to-other-countries-slideshow/, accessed February 2015.
3 “GME Funding: How to Fix the Doctor Shortage,” Association of American Medical Colleges,
https://www.aamc.org/advocacy/campaigns_and_coalitions/fixdocshortage/, accessed February 2015.
4 “GME Funding: How to Fix the Doctor Shortage,” Association of American Medical Colleges.
5 “HER Incentives & Certification,” HealthIT.gov, January 15, 2013, http://www.healthit.gov/providers-professionals/ehr-
incentive-programs, accessed February 2015.
6 OECD, “Health at a Glance 2013: OECD Indicators,” OECD Publishing, 2013, http://dx.doi.org/10.1787/health_glance-2013-
en, accessed February 2015.
7 Company documents.
8 Martha Hamilton, “Why walk-in health care is a fast-growing profit center for retail chains,” Washington Post, April 4, 2014,
http://www.washingtonpost.com/business/why-walk-in-health-care-is-a-fast-growing-profit-center-for-retailchains/2014/04/04/a05f7cf4-b9c2-11e3-96ae-f2c36d2b1245_story.html, accessed February 2015.
9 Hamilton, “Why walk-in health care is a fast-growing profit center for retail chains.”
10 “Summary of the HIPAA Privacy Rule,” U.S. Department of Health and Human Services, May 2003,
http://www.hhs.gov/ocr/privacy/hipaa/understanding/summary/privacysummary.pdf, accessed February 2015.
11 “HIPAA Background,” Office of Corporate Compliance, University of Chicago Medical Center, February 2010,
http://hipaa.bsd.uchicago.edu/background.html, accessed February 2015.
12 “Summary of the HIPAA Privacy Rule,” U.S. Department of Health and Human Services.
13 “HIPAA Background,” Office of Corporate Compliance, University of Chicago Medical Center.
14 Ken Terry, “Healthcare Collaborative, IBM Partner On Big Data Platform,” Information Week, June 18, 2013,
http://www.informationweek.com/healthcare/clinical-information-systems/healthcare-collaborative-ibm-partner-on-bigdata-platform/d/d-id/1110419?, accessed February 2014.
15 Terry, “Healthcare Collaborative, IBM Partner On Big Data Platform.”
16 Joe Carlson, “Carolinas centralizes data analytics to reduce readmissions and redesign care,” Modern Healthcare, December 9,
2013, via Factiva, accessed October 2014.
17 Michael F. Dulin, Thomas M. Ludden, Hazel Tapp, Heather A. Smith, Brisa Urquieta de Hernandez, Joshua Blackwell, and
Owen J. Furuseth, “Geographic information systems (GIS) demonstrating primary care needs for a transitioning Hispanic
community,” Journal of the American Board of Family Medicine 23, no. 1 (2010): 109–120.
18 Dulin et al., “Geographic information systems (GIS).”
19 Carlson, “Carolinas centralizes data analytics.”
16
Page 179 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
2 “U.S. Health Care Resources Compared to Other Countries Slideshow,” Kaiser Family Foundation,
9 -5 1 0 -0 5 7
REV: AUGUST 11, 2011
JOHN DEIGHTON
United Breaks Guitars
On July 8, 2009, United Airlines offered professional musician Dave Carroll $1,200 in cash and
$1,200 in flight vouchers to ‘make right’ a situation in which his guitar had been damaged at
Chicago’s O’Hare airport during transfer from one airplane to another, in full view of passengers
seated in the plane.
Carroll had spent 15 months seeking compensation, but the effort appeared to have reached an
impasse when a United Airlines customer service representative told him that the airline had
concluded that the damage was Carroll’s responsibility and that she considered the matter closed.
He replied that in that case he would be composing three songs about his experience and posting
them to the video-sharing site YouTube. The first song was posted on July 6, 2009. At the same time,
he wrote a blog entry detailing the ordeal and posted the link to the YouTube video on his Twitter
account. Within a week the video had been viewed three million times, and United Airlines had
reached out to Carroll to offer the compensation.
Online and offline media helped propagate the story. On July 22, 2009, The Times of London wrote
that “the gathering thunderclouds of bad PR caused United Airlines' stock price to suffer a mid-flight
stall, and it plunged by 10 per cent, costing shareholders $180 million.” 1 Hundreds of news reports
repeated the story of how a single poorly handled customer complaint had, thanks to the power of
social media, cost the company $180 million. News channel CNN urged viewers, “Anyone who’s
ever been frustrated with an airline needs to see this video.” 2 The first song, “United Breaks Guitars,”
began to be referred to as the complaint anthem of airline travelers, and Carroll was called the
Accidental Chief Marketing Officer of United Airlines. His catchy melody stuck in people’s minds.
Carroll observed, “It’s been said that in the old days (maybe only a decade ago) that people who
had a positive customer service experience would share that with 3 people. If they had a bad
experience, they would tell 14. . . . [A]s of today I have reached more than 6 million people on
YouTube with my story and, according to some estimates, some 100 million people if you total all
media references.”3
By October 2009, the video seemed to have receded from the public mind. Worldwide viewings
had slowed to 5,000 a day, and traffic to the websites of Dave Carroll and his band, Sons of Maxwell,
had declined from 150,000 unique visitors a month in July 2009 to 2,000 in October. And yet evidence
suggested that Carroll’s song had had a more persistent influence on perceptions of the United
Airlines brand. British blogger Peter Cochrane recounted an incident that took place in October 2009
on a shuttle bus from his hotel to New York’s JFK airport. “Barely awake,” wrote Cochrane, “I heard
the driver call for airlines and terminals. Someone piped up ’United’ and the immediate rejoinder
from the rest of the passengers was a chorus of the song ’United Breaks Guitars.’”4
________________________________________________________________________________________________________________
Professor John Deighton and Research Associate Leora Kornfeld prepared this case. HBS cases are developed solely as the basis for class
discussion. Cases are not intended to serve as endorsements, sources of primary data, or illustrations of effective or ineffective management.
Copyright © 2010, 2011 President and Fellows of Harvard College. To order copies or request permission to reproduce materials, call 1-800-5457685, write Harvard Business School Publishing, Boston, MA 02163, or go to www.hbsp.harvard.edu/educators. This publication may not be
digitized, photocopied, or otherwise reproduced, posted, or transmitted, without the permission of Harvard Business School.
Page 180 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
LEORA KORNFELD
510-057
United Breaks Guitars
United Airlines
With close to 50,000 employees and more than 3,300 flights per day, United Airlines was one of
the largest international airlines based in the U.S. 5 Since deregulation of the industry in 1978, airlines
had been unconstrained by pricing and scheduling mandates. Deregulation initially yielded profits
for the airlines, but profitability soon gave way to cycles of losses.
In common with all major airlines, United Airlines struggled with competitive fare slashing and
high fuel costs. In December 2002, United filed for Chapter 11 bankruptcy protection. Its losses for the
year totaled $3.21 billion. In 2003, it took further measures to trim operational costs, including a
workforce reduction of 20%, renegotiation of salaries, and outsourcing of maintenance services, and
its yearly loss decreased to $2.81 billion. 6 The layoffs and cuts continued, with flight attendants and
in-flight services bearing the brunt of the reduction. In 2006, United laid off 11% of its salaried
workers,7 and in 2008 the airline announced it would no longer be handing out complimentary
pretzels and biscuits to economy class passengers in North America.8 Shortly thereafter, the
American Customer Satisfaction Index Airline Rankings for 1996–2009 were released. United’s score
was the lowest, with the largest percentage drop in the 13 years of the survey (see Exhibit 1; United
Airlines’ stock price is shown in Exhibit 2).
The Incident
For more than 20 years, Dave Carroll had been making a modest living as a musician, performing
his brand of pop-folk music with his band, Sons of Maxwell. Carroll’s life as a working musician took
him back and forth across Canada most years, into several U.S. cities, and sometimes to international
destinations for music festivals. On March 31, 2008, Carroll and members of the band were flying
from their hometown of Halifax, Canada, for a week of shows in Omaha, Nebraska. During a
connection in Chicago, other passengers aboard the flight noticed some very rough handling of cargo,
and Carroll’s bandmates watched helplessly as Dave’s $3,500 Taylor guitar was mishandled by
United’s baggage handlers. Carroll shared his concerns with a flight attendant. He was told, “Don’t
talk to me. Talk to the lead agent outside.”9 Carroll complied and was informed by the employee at
the gate that he should take the issue up with the ground crew in Omaha. But the flight was delayed,
and Carroll saw no ground crew when he arrived in Omaha after midnight.
On his return to the airport in Omaha, he spoke with a United agent, who advised him that he
would need to start a claim at the originating airport, in Halifax. Once he was back in Halifax, Carroll
was given a phone number that he called a few times, eventually being rerouted to a call center in
India. Several calls later, Carroll was directed to United’s baggage offices in Chicago, where he was
told that he would need to bring the guitar to Chicago for inspection. When Carroll explained that he
was over 1,200 miles away from Chicago, the agent told him to go through United’s central baggage
center in New York, which eventually led Carroll back to the call center in India.
To Carroll’s delight, the customer service manager in India promised to get in touch with a United
representative in Chicago. He did so, and the representative reviewed Carroll’s file and made direct
2
Page 181 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Just what was the significance of this incident to a brand like United Airlines? Could its scale have
been anticipated or influenced? How, when, and how much could the brand have reacted? And
looking forward, what contingency planning would be appropriate? Was this kind of incident the
responsibility of brand management, customer service management, or media and public relations, or
did responsibility lie elsewhere in the organization?
510-057
e-mail contact with him. It was now seven months since the process began, but Carroll felt that at last
the matter was being addressed. When the email arrived, however, the representative said she was
sorry about what had happened to the guitar, but that standard airline policy held that claims be
made within 24 hours of damage (a precaution against fraudulent claims). She told him that his claim
was going to be denied. Carroll asked to speak with a supervisor, but was refused. His final request
was for $1,200 worth of United flight vouchers, the amount he had paid to repair his guitar. The
representative told him no, United considered the case closed, and there would be no further
communication on the matter. “At that moment,” wrote Carroll in his blog, “it occurred to me that I
had been fighting a losing battle. . . . The system is designed to frustrate affected customers into
giving up their claims and United is very good at it.” In his final exchange with the representative, he
told her that he would be writing three songs with video about United Airlines and sharing them on
YouTube. His goal, he said, was to get one million hits in one year. 10
Carroll wrote the first song, and with his friends at Curve Productions in Halifax, produced a
biting video to accompany it. The budget for the video was $150, with people donating time, props,
and locations for the shoot.11 The location that stood in for the O’Hare airport tarmac was the fire
station in Waverly, Nova Scotia, where Dave Carroll worked as a volunteer firefighter. On July 6,
2009, the video was posted to YouTube.
The Video Takes Off
Carroll’s friend, Ryan Moore, posted the video to YouTube at about 10 p.m. on Monday, July 6.
Throughout the night and the following day, a small team of friends used Twitter to introduce their
followers to the video. They also tweeted to those on Twitter who had themselves tweeted about bad
experiences with United Airlines, and to members of the media including Jay Leno, Jimmy Fallon,
and Perez Hilton. They posted the story to Digg and other social news sites to which people could
submit stories and vote them up and down.
At 1:49 p.m. on Tuesday, July 7, the video was picked up by Consumerist.com, a website
affiliated with Consumers Union, America’s leading not-for-profit consumer advocacy organization
and the publisher of Consumer Reports magazine. That evening at 7:02 p.m., the story made its first
mainstream news appearance in the travel section of the website of the Los Angeles Times. The
reporter had seen the video when it was e-mailed to a colleague by a friend. The story reported that
by that time the video had received 24,000 views and 461 comments, most of them maligning United
Airlines.
On Wednesday, July 8, HuffingtonPost.com and NBCChicago.com relayed the story to their
followers, and on that day there were 190,000 views on YouTube (see Exhibit 3). Mainstream media
such as CNN, the CBS Morning Show, and Associated Press began calling Carroll for interviews, and
on July 9 and 10 mainstream media mentions peaked at 150 and 155, respectively.
By Friday, July 10, YouTube views per day crested, with cumulative views reaching nearly 1.6
million. By this time “United Breaks Guitars” was YouTube’s number one rated music video of all
time, and number three in any category of video. 12 Beginning on July 23, a second surge in YouTube
traffic arose when the British news media picked up the story. By the end of July, the video had been
viewed 4.6 million times.
The popularity of “United Breaks Guitars” on YouTube spilled over to other online media. Traffic
to Dave Carroll’s website, www.davecarrollmusic.com, on which he sold his CDs, surged from a few
hundred unique visitors per week to more than 20,000 per week (see Exhibit 4). Song sales on iTunes
3
Page 182 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
United Breaks Guitars
510-057
United Breaks Guitars
On July 10, Bob Taylor of Taylor Guitars, the California-based manufacturer of high end guitars,
posted a two-minute video on YouTube to express his support for Carroll and to offer advice on
transporting guitars on planes. Taylor guitars were used by some of the top names in the music
industry, among them Prince, Taylor Swift, and Aerosmith. In the description section of his YouTube
video, Bob Taylor wrote, “Taylor has had an artist relationship with Dave for several years now. In
2006, our owners’ magazine, Wood & Steel, reviewed Dave’s CD, “Sunday Morning.” As we’ve had an
ongoing relationship, we have made the offer to replace and/or further repair his damaged 710ce . . .
we’ve offered Dave [as a Taylor artist] our support, a choice of a new guitar, and to re-repair the
damaged 710ce. Dave and his bandmate Julian made the trip to the Taylor factory in July and have
met many of our staff. We’ll also be running a story on Dave’s experiences in the fall issue of Wood &
Steel.” In the video, Bob Taylor expressed his concern for Dave Carroll’s situation and went on to say,
“If your guitar is broken and you had it in a hard shell case, there was clearly some negligence and
abuse there, because the case can protect the guitar from all kinds of damage.” 15
United Airlines’ Response
United Airlines had its own presence in both online and social media. Its website, like that of most
commercial airlines, gave customers the ability to browse fares and schedules, make bookings, check
their flyer miles balance, interact with customer service, file damaged baggage complaints, and view
the latest news from the airline (see Exhibit 5 for more information on united.com). In July 2009,
United maintained a presence on Twitter that had approximately 18,000 followers. United Airlines
used Twitter actively, tweeting two or three times a day with information useful to travelers, such as
Twitter-only airfare deals and system disruptions. All United employees were encouraged to monitor
social media for mentions of United Airlines. All they needed to do to monitor tweets involving
United Airlines was to subscribe to the service and search for mentions.
At noon on July 7, before any blogger or mainstream news medium had reported the story, and
with cumulative views of Carroll’s video under 20,000, a United Airlines staff member read the
following tweet from one of Carroll’s friends: “psssst . . . @UnitedAirlines breaks guitars! And they
don't even care!” Minutes later, Robin Urbanski of United Airlines’ media relations team in Chicago
called Rob Bradford, managing director of customer solutions at United, and told him, “We need to
call Carroll.” The call was made, but Carroll was not available to return the call until the next
morning. Urbanski then sent out United’s first tweet: “This has struck a chord w/us and we've
contacted him directly to make it right.”
United Airlines tracked the Twitter conversation throughout the day and onward, joining the
conversation many times and frequently tweeting the sentence, “This has struck a chord with us,” as
new people joined the conversation. United would often have to state that it had reached out to Dave
Carroll, as in this interchange:

At 1:02 p.m., from a member of the public: “Check the @unitedairlines account. They've apologized
and accepted responsibility. Cool stuff.”
4
Page 183 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
increased from one or two per day to hundreds per day. “United Breaks Guitars” became a Top 20
iTunes download in Canada and the number-one country music download in the United Kingdom
for July 2009.13 "My mother handles our mail-order,” said Carroll. “I went round there on Saturday
and her couch was piled high with CDs ready to be mailed out. It was the third ‘couch full’. . . . We
might have to hire someone to help her, or at least buy her a bigger sofa."14
510-057

At 2:28 p.m., from one of Carroll’s team: “word has it @UnitedAirlines isn't trying to make it right,
they are just tweeting is that way.”

At 2:31 p.m., from United Airlines: “The word you hear is wrong. We have called him and the person
who answered his phone scheduled a call for tomorrow morning.”
United’s Rob Bradford reached out to Carroll on July 8 to apologize for the situation and to ask if
United could use the video internally to help change its culture.16 He offered him $1,200 in cash, the
amount Carroll had spent on repairing his guitar, plus $1,200 in flight vouchers. Carroll declined the
offer and suggested that United give it to a customer of its choosing who had been affected in a
similar way. The airline chose instead to donate $3,000 to a music school.
Meanwhile, the tweeting continued. United did not respond to taunting from Carroll’s friends:





At 1:46 p.m., on July 7: “Why'd you guys have to go and break his guitar? http://bit.ly/rI2ef Stop being
a bully and fess up!”
At 3:40 p.m., on July 7: “Learn from the United Breaks Guitars song that its NOT ok to treat any
customer bad.”
At 9:50 p.m., on July 7: “You can say creatively that this has struck a chord with you but lets be real
how do you plan on changing?”
At 9:53 p.m., on July 7: “You realize that Dave Carroll is one of many people burned by your
‘Airline’ ~ how'd you plan to make things right 4them?”
At 9:55 p.m., on July 7: “And since I'm on a roll, shame on you for taking over a year to bother . . . too
much truth in your bad PR?”
On more than one occasion, however, United Airlines used Twitter to try to defuse the situation:




At 3:00 p.m., on July 8, from Ryan Moore: “i posted a video for a client of mine monday night and it's
like the biggest vid on youtube canada now. http://bit.ly.”
At 3:39 p.m., on July 8, United Airlines replied: “Love your client's video. Not all r as honest as he.
That is why policy asks for claims w/in 24 hours. No excuse; we're sorry.”
At 4:56 p.m., on July 8, from a Twitter member: “I love this song about @unitedAirlines Check it out!
http://bit.ly/8RDMI”
At 5:02 p.m.. on July 8, from United Airlines: “It is excellent and that is why we would like to use it
for training purposes so everyone receives better service from us.”
United Airlines continued to monitor references to the video and react to them:


At 6:00 p.m., on July 9, from a musician in Cleveland: “Funny how @UnitedAirlines denies someone's
claim until it's made public #UnitedBreaksGuitars—You COULD have made it right a year ago.”
At 7:59 p.m., on July 9, United Airlines replied: “Absolutely right, and 4 that (among other things), we
are v. sorry and are making it right. Plan 2 use video in training.”
Occasionally, United made comments about the incident that were not reactions to other Twitter
members’ tweets:


At 6:44 p.m., on July 10: “Wud like Dave 2 sing a happy tune—as asked we gave 3K to Thelonius
Monk Institute of Jazz 4 music education 4 kids.”
At 6:46 p.m., on July 10: “Can’t wait 2 make music w/Dave 2 improve service 4 all.”
For the rest of the week, the Twitter feed served as the channel by which United responded:
5
Page 184 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
United Breaks Guitars
United Breaks Guitars

At 9:34 a.m., on July 10, a Florida resident wrote: “United Breaks Guitars! LOL this is a funny vid!
I hope you guys buy him another.”
At 10:13 a.m., on July 10, United Airlines replied: “As Dave asked we donated 3K to charity and
selected the Thelonius Monk Institute of Jazz 4 music education 4 kids.”
At 7:31 p.m., on July 13, a Twitter member commented: “The sad thing is, I still don't think
@unitedairlines really gets what they did wrong.”
At 3:56 p.m., on July 13, another member commented: “It should have been fixed sooner & not have
happened in the 1st place.”
At 11:24 a.m., on July 14, the airline replied: “Should it regretfully happen to anyone, pls file a claim
w/in 24hrs at airport, online or phone.”
At 6:45 p.m., on July 14, United Airlines continued: “That was a mistake that we made, have
apologized for, have fixed, and most importantly, learned from too.”





United Airlines was selective in the media it used to discuss the incident. It responded to inquiries
from journalists about the incident, but did not address it on its website or its YouTube channel. It
posted a comment to Carroll’s band’s YouTube channel, but the message was deleted. By August
2009, United was responding to tweets with direct messages (private communications through the
Twitter medium) inviting a longer e-mail communication.
These emails came from Robin Urbanski in United’s media relations group and read:
Yes, these videos have struck a chord with all of us here.
In recent statements on YouTube, Mr. Carroll described our baggage service representative as a “great
employee who acted in the best interests of the company,” and I could not agree more with that.
He has made his point, I have since been in contact with him to fix, and I am now his BFF. 1
The second video is suggesting we do something that we’ve already done — and that is to provide our
agents with a better way to escalate and respond to special situations.
While his anecdotal experience is unfortunate, the fact is that 99.95 percent of our customers’ bags are
delivered on time and without incident, including instruments that belong to many Grammy award–
winning musicians.
As you know, in our business, how we conduct ourselves is important, and we do understand that
treating each other and our customers in a courteous and respectful manner is a vital part of running a
good airline.17
People who communicated with United Airlines’ customer relations department received the
following written response:
Thank you for contacting United Airlines Customer Relations. I appreciate the opportunity to respond
to your inquiry.
At United, we continually work to ensure the proper handling of your items when you fly with us, and
we transport thousands of checked bags each day without incident.
We have had discussions with our customer to make what happened right, and at his request, we
donated the money that would have gone toward a new guitar to the Thelonius Monk Institute of Jazz
1 Popular Internet abbreviation for ‘best friend forever’.
6
Page 185 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
510-057
United Breaks Guitars
510-057
that provides music education for kids with potential. The video provides us with a unique learning
opportunity that we plan to use for training purposes to ensure all customers receive better service from
us.
On September 14, Dave Carroll met with two senior vice presidents and a vice president as he was
passing through O’Hare airport in Chicago. They gave him a tour of O’Hare’s baggage-handling
facilities and explained the challenges of shaping internal culture in an organization where many
customer-facing employees spent most of their time traveling. They acknowledged that Carroll’s
claim should not have been denied and told him that customer service representatives were now
being trained to use discretion in applying rules like the one requiring notification of damage to
baggage within 24 hours.
The Aftermath
The path of Carroll’s life had been upended by the success of his song. As the media began to call,
he and his wife reached out to his wife’s father, Brent Sansom, an international management
consultant: “Dad, I think we need your help.” Sansom responded by relocating from Moncton, New
Brunswick, to Halifax, Nova Scotia, and set about handling the hundreds of e-mails and phone calls
that arrived every day from around the globe. “For the past four months we’ve been getting three to
four hours of sleep a night,“ Sansom said. “We are being offered new audiences and distribution
opportunities for Dave’s music. There are new business relationships with manufacturers like Taylor
Guitars and Calton Cases, and service providers such as Mariner Partners and RightNow
Technologies, which offered customer experience software. There are speaking and live performance
invitations, and companies want to commission songs and video recordings. Dave has done over 200
media interviews with everyone from the Wall Street Journal to Oprah Radio, Rolling Stone to the
Reader’s Digest. The story is resonating with many people.”
FlyersRights.org, a large nonprofit airline consumer organization, organized what it called a
stakeholder hearing on September 22 in the Rayburn House Office Building in Washington, D.C., on
airline passenger rights, where Senator Barbara Boxer of California spoke and Carroll performed an
acoustic version of “United Breaks Guitars.” For a group of Canadian broadcasters lobbying against a
proposed increase in cable fees, Carroll wrote “The Cable Song,” which was broadcast after the local
evening news across Canada for a week.
In October, Carroll traveled to Colorado to give the keynote address at the RightNow
Technologies User Summit. He took a United Airlines flight operated by a regional carrier, SkyWest,
whose baggage was being handled under contract to Air Canada. One of Carroll’s two items of
luggage was lost, providing him with fresh material for his speech. Reporting on the story, the New
York Times labeled him “the Everyman symbol of the aggrieved traveler.”18
Advertising Age columnist David Klein reflected on the meaning of the “United Breaks Guitars”
incident for marketing and branding in the age of social media, “To really incite the full range of
customer reaction to a brand, and by full range I mean everything from bitter rage at the low end to
fantastic appreciation at the high end, traditional advertising is not the way to do it. In these
postmodern times where every interaction with the customer is a marketing event, the real crunch
point comes when the customer meets your customer-service department. Seriously, who didn't
enjoy watching musician Dave Carroll's takedown of United Airlines for not only breaking his guitar
but then refusing to make things right by reimbursing him?”19
7
Page 186 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Your business is important to us, and I hope you will give us an opportunity to serve you in the future.
510-057
Exhibit 1
United Breaks Guitars
American Customer Satisfaction Index Airline Rankings, 1996–2009
Base96 97 98 99 00 01 02 03 04 05 06 07 08 09
line
Previous
First
Year
Year
%
%
Change Change
Southwest Airlines
78
76 76 74 72 70 70 74 75 73 74 74 76 79 81
2.5
3.8
All Others
NMa
74 70 62 67 63 64 72 74 73 74 74 75 75 77
2.7
4.1
Continental Airlines
67
66 64 66 64 62 67 68 68 67 70 67 69 62 68
9.7
1.5
Average Airline
72
69 67 65 63 63 61 66 67 66 66 65 63 62 64
3.2
-11.1
Delta Air Lines
77
67 69 65 68 66 61 66 67 67 65 64 59 60 64
6.7
-16.9
American Airlines
70
71 62 67 64 63 62 63 67 66 64 62 60 62 60
-3.2
-14.3
US Airways
72
66 68 65 61 62 60 63 64 62 57 62 61 54 59
9.3
-18.1
Northwest Airlines
69
67 64 63 53 62 56 65 64 64 64 61 61 57 57
0.0
-17.4
United Airlines
71
70 68 65 62 62 59 64 63 64 61 63 56 56 56
0.0
-21.1
Source: American Customer Satisfaction Index, http://www.theacsi.org/index.php?option=com_content&task=view&id=
147&Itemid=155&i=Airlines, accessed November 2009.
a Not measured.
8
Page 187 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Airlines
United Breaks Guitars
United Airlines Stock Price, 2009
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Exhibit 2
510-057
Source: Wall Street Journal Market Data Center, http://online.wsj.com/mdc/public/page/marketsdata.html,
accessed January 2010.
United Airlines Stock Price Compared with S&P Index, February to December 2009
Source: Wall Street Journal Market Data Center, http://online.wsj.com/mdc/public/page/marketsdata.html,
accessed January 2010.
9
Page 188 of 282
510-057
6-Jul
7-Jul
8-Jul
9-Jul
10-Jul
11-Jul
12-Jul
13-Jul
14-Jul
15-Jul
16-Jul
17-Jul
18-Jul
19-Jul
20-Jul
21-Jul
22-Jul
23-Jul
24-Jul
25-Jul
26-Jul
27-Jul
28-Jul
29-Jul
30-Jul
31-Jul
Media Activity
YouTube
Views
per Day
Blogs
per
Day
Tweets
per Day
Mainstream
Media
Mentions
0
25,000
190,000
500,000
910,000
650,000
240,000
210,000
190,000
180,000
120,000
90,000
80,000
60,000
70,000
95,000
50,000
180,000
190,000
120,000
100,000
80,000
90,000
70,000
70,000
50,000
0
50
180
420
140
80
90
140
45
45
0
10
20
20
40
10
0
30
60
40
45
50
45
60
80
50
20
50
200
860
900
500
250
300
280
560
220
230
80
80
60
20
30
350
200
150
140
160
140
100
110
80
0
0
40
150
155
50
20
30
60
10
15
5
5
0
5
0
40
40
30
5
5
0
0
0
5
0
Source: Compiled by casewriters.
10
Page 189 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Exhibit 3
United Breaks Guitars
United Breaks Guitars
510-057
Exhibit 3 (continued)
New and Traditional Media Activity per Day, July 2009
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Number of activities per day (logarithmic scale)
1000000
100000
10000
YouTube (views per day)
1000
Blogs per day
Tweets per day
Mainstream media mentions
100
10
1
4-Jul
6-Jul
8-Jul
10-Jul
12-Jul
14-Jul
16-Jul
18-Jul
20-Jul
Day
Source: Adapted from Media Miser website, http://www.mediamiser.com/resources/archive/090821_united.html,
accessed October 2009.
11
Page 190 of 282
510-057
Website Traffic at www.davecarrollmusic.com, May–August 2009
Source: Compete website, http://www.compete.com, accessed November 2009.
Exhibit 5
United Airlines Website Traffic, 2007–2009
Source: Compete website, http://www.compete.com, accessed November 2009.
12
Page 191 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Exhibit 4
United Breaks Guitars
United Breaks Guitars
510-057
Endnotes
1 Chris Ayres, “Revenge is best served cold—on YouTube,” The Times (U.K.), July 22, 2009, via Factiva.
3 Dave Carroll, “Statistical Insignificance,” AdWeek, November 25, 2009, http://www.adweek.com/aw/
content_display/community/columns/other-columns/e3ia7f0e1dcab3176840f88bf567f95ce7d, accessed
December 5, 2009.
4 Peter Cochrane’s Blog: “United breaks guitars?,” October 14, 2009, http://networks.silicon.com/
webwatch/0,39024667,39574741,00.htm, accessed October 23, 2009.
5 United Airlines website, http://www.united.com/pressreleases/0,7057,1,00.html, accessed October 22, 2009.
6CBC
News, “World airline woes: Carriers hit financial turbulence,” January
http://www.cbc.ca/news/background/aircanada/airlinewoes.html#a6, accessed October 26, 2009.
29,
2004,
7 “United Airlines Plans to Lay Off 11% of Its Salaried Workers,” New York Times, June 15, 2006,
http://www.nytimes.com/2006/06/15/business/15air.html, accessed October 27, 2009.
8 George Raine, “United Airlines to drop free snacks,” San Francisco Chronicle, August 20, 2008,
http://www.sfgate.com/cgi-bin/article.cgi?f=/c/a/2008/08/20/BUC812E8DR.DTL#ixzz0VAYnTkI8, accessed
October 26, 2009.
9 Dave Carroll blog posting, July 7, 2009, http://www.davecarrollmusic.com/story/united-breaks-guitars,
accessed October 23, 2009.
10 Ibid.
11 Linda Laban, “Dave Carroll Smashes YouTube Records with ’United Breaks Guitars,’” July 14, 2009,
http://www.spinner.com/2009/07/14/dave-carroll-breaks-youtube-records-with-united-breaks-guitars/#,
accessed October 26, 2009.
12 Ibid.
13 MktgCliks website, http://mktgcliks.blogspot.com/2009/07/united-breaks-guitars-united-airlines.html,
accessed October 28, 2009.
14 Laban, 2009.
15 “Taylor Guitars Responds to United Breaks Guitars,” YouTube video, http://www.youtube.com/watch?
v=n12WFZq2__0, accessed October 25, 2009.
16 Dave Carroll, interviewed by Beverly Thomspon, “Online broken guitar video gets airline’s attention,”
July 9, 2009, Canada AM, CTV, transcript via Factiva.
17 Brett Snyder, “United Aggressively Responds to ‘United Breaks Guitars Part 2,’ “ BNET Travel website,
http://industry.bnet.com/travel/10003236/united-aggressively-responds-to-united-breaks-guitars-part-2/,
accessed December 8, 2009.
18 Christine Neggroni, “With video, a traveler fights back,” New York Times, October 29, 2009, via Factiva,
accessed November 4, 2009.
19 David Klein, “Your most crucial moment comes when the customer calls,” Advertising Age, July 27, 2009,
via Factiva, accessed November 2, 2009.
13
Page 192 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
2 CNN, The Situation Room, July 8, 2009, http://transcripts.cnn.com/TRANSCRIPTS/0907/08/sitroom.
03.html, accessed November 2, 2009.
S
w
9B02E014
David Singer prepared this case under the supervision of Professor Darren Meister solely to provide material for class discussion.
The authors do not intend to illustrate either effective or ineffective handling of a managerial situation. The authors may have
disguised certain names and other identifying information to protect confidentiality.
Ivey Management Services prohibits any form of reproduction, storage or transmittal without its written permission. Reproduction of
this material is not covered under authorization by any reproduction rights organization. To order copies or request permission to
reproduce materials, contact Ivey Publishing, Ivey Management Services, c/o Richard Ivey School of Business, The University of
Western Ontario, London, Ontario, Canada, N6A 3K7; phone (519) 661-3208; fax (519) 661-3882; e-mail cases@ivey.uwo.ca.
Copyright © 2002, Ivey Management Services
Version: (A) 2009-12-01
Bruce Jones, visual effects (VFX) manager at Toybox, the visual effects division of Command Post and
Transfer, looked around his downtown Toronto office and recalled the post-mortem discussion about their
recent work on Panic Room. The discussion made it seem that the film’s title should have been seen as
prophetic. At the meeting, a lot of frustration had been expressed. Jones wanted to make sure that future
post-mortems were going to be easier.
Many of the complaints at the post-mortem talked about problems with the flow of information within
Toybox. Jones worried that Toybox’s information systems were not going to handle the company’s
anticipated growth. After the post-mortem, Jones had begun investigating solutions. NXN’s Alienbrain
software, (www.NXN-software.com) was one solution that intrigued him. Although the program was
designed initially for computer gaming companies, NXN was eager to include film postproduction
companies as clients. The internal information technology (IT) department was also proposing a solution.
Jones thought this was a great time to be implementing a solution. The workload was somewhat less due
to the tail-end effects of the threat of a writers’ strike in Hollywood, followed by the economic slowdown
after September 11, 2001. Before business started flooding in again, Jones wanted to set the stage for
success.
COMPANY OVERVIEW
Command Post and Transfer Corporation
In 2002, the Canadian production industry (including television, film and commercial production) was
approximately $5 billion in size and had experienced growth of nine per cent in the past year. Ontario was
responsible for 40 per cent of the production activity in Canada. Command Post and Transfer Corporation
was formed in 1986 by Michael Flanigan, Andy Sykes, Michael Ellis and Stephen Robinson to service the
growing commercial market for postproduction services in Metropolitan Toronto. Since its inception, the
company had grown into Canada’s largest full service postproduction company, with revenue for 2001 of
Page 193 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
TOYBOX: MANAGING DYNAMIC DIGITAL PROJECTS
Page 2
9B02E014
Over the last four years, Command Post had grown aggressively through acquisition strategy. Although
there was some overlap in capabilities in each of the acquired companies, Command Post was working
towards making each division a focused shop, each one offering specific postproduction services. As part
of this initiative, operations at Toybox had become quite independent of the rest of Command Post.
Toybox
Toybox was the visual effects division of Command Post. Toybox provided creative and technical services
such as 3D computer animation, 2D compositing, motion control, film transfer and editorial and highresolution pre-press. The Toybox team had more than 30 computer animators, 10 VFX compositors,
technicians, VFX supervisors and a full production staff.
Toybox categorized projects as either short or long format. Short format included commercial work and
music videos, while long format included feature films, television series and movies of the week. Short and
long format projects each had unique challenges and demands. For example, the required turn-around time
for commercial work was much shorter, usually ranging from a couple of hours to a couple of weeks
versus months for a long format project. Long format projects also differed from short format projects in
the amount of development that was required. Short format projects also rarely used 3D work, relying
mostly on 2D compositing effects.
Some of the company’s more recent film credits included Chicago, Panic Room, Undercover Brother,
Jason X, The Cell and Fight Club. A well-known commercial production undertaken by Toybox was the
‘Robaxacet’ Celtic dancing puppets.
Before the slowdown, VFX services provided by Toybox accounted for a significant portion of Command
Post’s revenue growth. Demand for these services had exceeded management’s expectations in recent
years.
Culture
Jones believed that creativity was the cornerstone of Toybox’s success. The Toybox office layout was
designed to nurture a supportive, informal and creative culture. Much of the workspace at Toybox was
based on an open-concept design. The 3D department was a large area with no walls or dividers between
individual workstations, thus encouraging collaboration amongst employees. Individual workstations were
cluttered with toys, action figures and comic books. The shelves were lined with movie memorabilia from
past projects. The dress code at Toybox was very casual. Employees dressed in anything more formal than
denim jeans and a T-shirt could be the subject of friendly ridicule. The tone for this casual environment
was set by the company’s founders.
Page 194 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
approximately $45 million. Command Post was an industry-leading provider of high-quality technical and
creative services, including visual special effects for film, television, video and digital audio services.
Clients included owners, producers and distributors of television programming, feature films, commercials
and other entertainment content in North America.
Page 3
9B02E014
Toybox was functionally organized. There were two overlapping lines of reporting: a creative line and
operational line. On the creative line, VFX supervisors and producers reported directly to Michael Ellis,
the vice-president of VFX and Digital Film. For operational issues, VFX supervisors, producers and
System Administration answered to the VFX manager, Jones (Exhibit 1). The creative group is
responsible for realizing the client’s vision for the shot. Operational responsibility is focused on ensuring
that resources were available for projects. Although this was only his second year at Toybox, Jones was no
stranger to the film production industry as his previous position was digital technology manager at Disney
Canada. Jones held a Bachelor of Arts in fine arts and an MBA.
THE LONG FORMAT PROCESS
VFX Supervisors/Producers
VFX Breakdown
The process for film postproduction work began when a prospective client submitted a script. A producer
and VFX supervisor team were assigned to the script to provide a visual effects “breakdown.” The
breakdown involved identifying sequences in the script that required a visual effect, recommending
solutions and estimating the time and resources required to complete the shot. The unit price/resource was
provided by the sales department. Based upon the estimated hours from the VFX breakdown, a price was
assigned to each VFX shot. The producer and VFX supervisor were not provided with cost data such as the
cost the company allocated for a specific resource. Eliminating cost responsibility from the VFX
supervisor/producer team was viewed as necessary by senior management to ensure that quality was the
priority. The VFX breakdown became the initial bid and, if clients approved, became the baseline
schedule for the project.
Time estimates for the VFX breakdown relied heavily on the experience of the VFX supervisor and
producer. Once a project had begun, there had never been attempts to determine the accuracy of the time
estimates of the VFX breakdown, and therefore accountability was not required.
Scheduling and Tracking
Scheduling at Toybox was reactive, dictated primarily by the arrival of inputs (filmed elements) provided
by clients. Clients did not always have or follow a schedule. This resulted in the building of flexible
processes at Toybox. Detailed working schedules for long format projects with long-term horizons were
useless as the future work demands were uncertain. The scheduling uncertainty meant that, some weeks, a
resource, such as an animator, would have nothing to do and during other weeks there would be more work
than they could handle.
When a bid was awarded to Toybox, Jones was responsible for working with the producers and VFX
supervisors to build project teams and schedule resources based on the VFX breakdown. As the schedule
for the project deliverables was uncertain, resources (including animators, compositors, technical staff or
equipment) assigned to a long format project were reserved for large blocks of time to be available when
required. This resulted in low capacity usage.
Page 195 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Organization
9B02E014
Scheduling and resource booking was centralized at Command Post using the Program 1 (PM1) software.
Work orders (Exhibit 2) were generated from PM1 and distributed on paper to individual employees. At
the end of each day, the animators and compositors were required to sign the work order and provide
details about the work accomplished. These work orders were collected by producers and given to the
accounting department. Accounting used the collected work orders for billing and determining allowable
revenue recognition for the month. This process proved useful for short format projects (commercials) that
had quicker turnover times, required less flexibility and used resources from different divisions of
Command Post. The scale and flexibility required by long format projects demanded that Toybox isolate
resources specifically for long format projects. As a result, PM1 proved less useful. In spite of its notorious
instability and frequent crashes, PM1 was still used for long format projects partly because, when capacity
allowed, long format resources could still be booked for short format projects. PM1 was resource
management software and consequently was not capable of providing many common project management
functions, such as task dependencies and critical path identification.
The VFX breakdown (Exhibit 3) was organized using an MS Excel template that had evolved over many
projects to include a substantial amount of information. As a result, the current spreadsheet template was
quite large, and its complexity meant that it was difficult to retrieve meaningful information regarding cost
and time variances. PM1 could not read MS Excel files. Information regarding scheduling had to be input
manually into PM1 using the MS Excel document as a guide.
Communication
The most important responsibility of the producer/VFX director was to ensure the realization of a client’s
vision. Due to the limited amount of information that was available on a PM1 work order, producers/VFX
supervisors had to rely on informal communication channels to transmit the client’s wishes internally.
Considerable time was spent communicating with 3D and 2D artists, partly through e-mail but more often
through face-to-face conversations. The time required to communicate through these informal channels
was significant and, as a result, VFX supervisors and producers were occasionally unavailable to address
the immediate concerns of all those under their supervision. Misunderstandings and work duplication
often resulted.
Animators and Compositors
Computer animators designed 3D elements to be used in a film. Maya (by Alias/Wavefront) and Houdini
(by Side Effects) were the primary 3D software packages used by computer animators. Animators worked
on Windows NT workstations. Compositors, using Inferno editing suites, blended live film provided by
clients with computer-generated images provided by the animators to produce the final digital output for
the film. Equipment costs in this area were significant. For example, the four Inferno editing suites cost
approximately $1 million each.
Animators and compositors were informed about their assignments through two channels: PM1 work
orders, which provided limited information about the specific daily tasks; and face-to-face conversations
with producers and VFX supervisors. Frustration with the process was common for animators and
compositors. Given the communication process, individuals could receive partial information without an
avenue to obtain complete information. This resulted in wasted time and resources. Compositors and
animators were often awaiting feedback; subsequent miscommunications often resulted in a duplication of
work. This created anxiety amongst the talent, as experience had taught them to expect an inevitable
Page 196 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Page 4
Page 5
9B02E014
“ramping up” of work near a project’s deadline, requiring significant overtime. This informal channel of
communication had other consequences, such as a lack of documentation of change requests and the
subsequent lack of accountability.
Requirements for long format projects were quite different from those of short format projects. The billing
schedule for a long format project was done by a per-cent-completion method (e.g., one-third upon signing
the agreement, one-third at an agreed upon date or stage of completion and one-third upon completion). In
order to determine allowable revenue recognition each month for a long format project, the accounting
department required producers to provide an estimated per-cent-completion on a per-shot basis. Producers
spent considerable time generating monthly per cent estimates, which were arguably arbitrary, leaving
them vulnerable to criticism and questioning from lenders.
A benefit of the PM1 system was its integration with the company’s accounting software package,
Dynamics. Integration of PM1 and Dynamics was achieved using third-party services. Inputs from PM1
generated work orders, such as the project’s ID, were easily transferred to Dynamics for invoicing and
bookkeeping purposes.
System Administration
The internal IT group at Toybox was called System Administration. Beyond day-to-day operational
concerns, Systems Administration was responsible for all IT- related activities including system acquisition
and implementation. The System Administrator’s chief operational responsibility was to ensure the
smooth functioning of the network (Exhibit 1). The main server at Toybox, Atlas, was a Silicon Graphics
UNIX server with 16 CPUs and five GB of RAM. The hard drive capacity of Atlas was approximately one
Terabyte.
In order to handle the amount of data that was required for long format projects, System Administration
added 12 Linux-based computers to the network last year. Each Linux machine had a 1.4 GHz processor
with 1.5 GB of RAM. All of the available CPUs on Atlas and the additional Linux computers comprised
the ‘Render Farm,’ a network of processors available to perform rendering jobs.
One particular area of concern for System Administration was the final stage for computer animators, i.e.,
rendering. When a computer animator finished working on the 3D elements for a particular shot, the file
was submitted to the Render Farm. The Render Farm took all of the available information (such as model,
texture, animation, lighting) from the 3D software package and generated a series of images.
Rendering was a critical and system-intensive job. Each rendered image represented a single frame of
film. Film runs at 24 frames per second and the amount of time the Render Farm required to generate a
single image was highly variable. A frame could take anywhere from a few seconds to a few hours to
render, depending on the complexity of the image. The average image would render in about five minutes.
A potential bottleneck existed with the Render Farm, as processors for rendering were a scarce resource
near deadlines. In order to organize the rendering process, the System Administration staff at Toybox
developed the render arbitrator program, R-Arb. R-Arb was the access point through which animators
submit jobs to the Render Farm. The program was essential as it distributed rendering jobs to available
Page 197 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Accounting
Page 6
9B02E014
As previously mentioned, computer animators worked on Windows NT workstations while the rest of the
network was entirely UNIX/Linux. This was the source of occasional problems with the Toybox network.
A long-term goal of System Administration was to transfer all of the computer animators’ workstations to
the Linux operating system. Conversion to Linux was delayed, partly because of the lack of available
software packages for Linux, but this was expected to change in the near future.
Post-mortem Meeting
After Toybox completed its work on Panic Room, Jones held a post-mortem meeting. While Toybox was
proud of its creative work for the film, Jones knew the process could have gone more smoothly. He was
concerned that future projects might suffer creatively if action was not taken. At the meeting, several
problems were voiced.
Jeff Morris, a computer animator, complained about having to work on the weekend. He had been given
clear, verbal confirmation that his work met client approval, only to be informed later that the client had
requested something significantly different. Consequently, Morris’s weekend was sacrificed to meet the
deadline. Morris assured Jones this was not a unique situation and that more accountability was required.
Andrea Kern, a Toybox producer, criticized the resource management software. She felt it hurt her
productivity by making it impossible to respond to approval requests quickly. Kern felt that better project
management tools would enable her to respond more quickly to approval requests. The accounting
department had chimed in too. Hannah Riley felt that the actual time spent on film work was tracked
inefficiently and inaccurately. She spent a lot of time getting accurate information that was never used in
billing for long format projects.
From this discussion, Jones felt that he had to take a hard look at some possible approaches to improve the
flow of information. He knew the solution would have to be reliable, cost-effective and would have to fit
the Toybox culture.
POSSIBLE SOLUTIONS
After some initial investigation, Jones felt that there were two possibilities that merited careful
consideration. The first, an off-the-shelf product, was NXN’s Alienbrain. The second was a proposal by
the System Administration group to develop an internal solution.
NXN — Alienbrain
From what Jones had read in trade magazines and from discussions with colleagues, Alienbrain appeared
promising as the program integrated project management with workflow tools. Alienbrain had been
Page 198 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
processors based on a two-tiered priority system. An animator submitting a file to the Render Farm could
enter the job as either high or low priority. With this system, if a render was submitted with a high priority,
the Render Farm would complete the entire high-priority job before it began work on any other file with a
low-priority setting. The combination of the lack of communication transparency and the fact that priority
setting worked on an honor system occasionally resulted in renders with critical deadlines being made to
wait until less critical work was completed. This was only the first version of R-Arb; an upgraded version
was already under development.
Page 7
9B02E014
originally designed for computer gaming companies (some of their more notable clients were Electronic
Arts and Sony Online Entertainment), but NXN was anxious to expand their client list to include film
postproduction studios. The software had many beneficial features but also presented some challenges.
Alienbrain used a client/server architecture. The Alienbrain was a Windows-based file manager for a
company’s main server; in Toybox’s case, this would be Atlas (see Exhibit 3). Off the shelf, Alienbrain
had several compatibility issues with the existing infrastructure. Although the software was fully
integrated with several 3D software packages including Maya, it was not integrated with Houdini, a less
widely used software package. Alienbrain was designed to be flexible and, according to NXN, could be
easily integrated with Houdini. However, no one at Toybox had Houdini programming skills. The System
Administration staff was confident that they could develop these skills themselves, but it would take time.
Compositors would be unable to access the Alienbrain network directly as the Inferno suites did not use the
Windows operating system. System Administration had significant experience programming for the
Inferno machines, but developing appropriate middleware could require significant time and resources.
When Jones mentioned Toybox’s concern, NXN suggested that Toybox might be interested in acquiring an
Inferno/Alienbrain integration program as Inferno suites were widely used in the postproduction industry.
Introducing Alienbrain would mean that Toybox would not need to use PM1 for project scheduling.
However, there was some concern that it would be more difficult for the accounting department to get the
information it would need. Information may have to be entered manually to the Dynamics software.
Version Control
The Version Control features of Alienbrain allowed users to check out a file from the server, edit the file
on a local computer and when finished, check the file back into the main server. Once a file had been
checked out by a user, further access was restricted, preventing accidental overwriting of another person’s
work. Each time a file was checked back into the server, Alienbrain compared it to the previous version,
determined the differences and saved only the changes made to the new version file. By saving only the
differences between the various versions, Alienbrain had the potential to save considerable hard drive
space as well as to track the history of any given file. This historical tracking could allow users to roll back
to any previous versions of a file.
Every time a file was checked in or out of the server, the user was able to attach comments and sketch
annotations onto the file representation. These comments and annotations were logged with each new
version. Users could modify the status of a file by using a color-coded menu that indicated whether the file
was a work-in-progress, awaiting feedback, awaiting modifications or awaiting final sign-off. Status was
also logged, providing a means of determining sources, external or internal, and duration of delays.
Communication
Alienbrain had an internal messenger system, similar to e-mail. It was not compatible with the company’s
internal e-mail system, as Toybox uses simple mail transfer protocol (SMTP) for e-mail and NXN uses a
proprietary protocol. It could allow workflow to be automated. For example, when an animator finished
Page 199 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Compatibility With Existing Systems
Page 8
9B02E014
working on a file, the VFX supervisor could be alerted that feedback was required. Similarly, after a VFX
supervisor annotated a file and changed its status, the animator could be alerted. To provide greater detail,
animators and VFX supervisors could include comments. These comments were then logged by the
system to provide a complete file history.
Using Alienbrain, a manager could assign resources and deadlines to projects. It was fully compatible with
MS Project. Updates made using either program could be immediately transferred to the other. The
awareness of the effects of production delays could help VFX supervisors/producers to reallocate resources
to meet the original project deadline and even out the workload. These tracking capabilities could aid in
assessing external/internal accountability. Greater accuracy of information could allow resources to be
booked more efficiently, reducing the need to book animators/compositors for large blocks of time. This
could mean higher billing utilization of resources and increased effective capacity. These functions
suggest that MS Project alone could provide all of functionality of the current VFX Breakdown MS Excel
model and more.
An additional feature of using Alienbrain with MS Project was that the responsibility of estimating per cent
completion for a shot could be transferred to the level at which work was executed, that is, to the
animators. Accounting could collect the data by accessing the MS Project file.
Improved Client Service
Alienbrain had the potential to improve the perceived quality of service at Toybox by providing a mediarich communication interface. In the current system, client delays on deliverables and feedback were
significant contributors to the end of project “ramping up.” Some of these external delays were attributed
to the busy schedules of the clients. Clients could remotely access the Alienbrain server and provide
feedback and would be able to access the same internal messenger features as the animators and VFX
supervisors. Their comments could also be logged, thereby making collaboration faster and easier. Much of
the work that required client feedback, however, occurred closer to the completion of a shot, at
compositing. Achieving the benefits of this remote access would require a significant change in the
compositor’s workflow.
Alienbrain Evaluation
NXN provided Toybox with evaluation copies of Alienbrain. The evaluation was to last for two months in
order to try the software out for an actual short format project environment. Initially there were problems
installing the evaluation copy of Alienbrain. Problems with third-party software used to regulate access to
the demo prevented using the software. After attempting to solve this problem for a couple of weeks,
NXN delivered a different version that removed this protection. However, difficulties persisted. After a
few more weeks, NXN sent in technical support from its home office in California to ensure that the
product would function correctly at Toybox. Even after this visit, there were still a couple of residual
glitches that were attributed to compatibility issues with the Toybox network. By the time the selected
short format project was supposed to begin, the demo software was still not fully functional.
Page 200 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Project Management
9B02E014
Jones felt that Alienbrain was a relatively inexpensive solution. The entire investment would be
approximately $120,000 for hardware and software licences. For Toybox, cost was not the primary issue.
Jones felt that the goal of a DPM system would be to enhance and smooth workflow in order to eventually
realize cost savings. However, these setbacks led to a negative “buzz” surrounding Alienbrain. As
Alienbrain software was not a primary responsibility of System Administrators, they would occasionally
gripe about the NXN software and the headaches it had already caused. Although very few people had
viewed the functioning software, animators and supervisors were feeling negative about the proposed
system.
An Internal Solution
Before the evaluation of Alienbrain began, System Administration had voiced their preference for a “home
grown” solution. They would be responsible for developing such a system. They cited several benefits,
including the ability to ensure compatibility with all segments of the workflow and maintaining the goal of
converting the animator’s workstations to Linux. Furthermore, close ties to a third-party software provider
was a concern to them. Experience had taught the System Administration staff that turnaround time for
problem solving was significantly faster if they built the system themselves. System Administration staff
thought that Alienbrain had more features than Toybox required. Designing a system in-house would offer
the opportunity to examine the needs of each potential user more closely and to develop a streamlined
solution.
System Administration proposed an iterative development strategy for an in-house solution. Using the RArb program as a starting point, they would expand its functionality to include a versioning and file status
functionality similar to those found in Alienbrain. Once the development of the tracking system had been
achieved, integration with the company’s e-mail system would be the next step. At the same time that
development was underway, producers and VFX supervisors could be introduced to MS Project, making it
an integral part of the workflow. As a final step, the file tracking system could be integrated with MS
Project. Each segment could be introduced to the workflow once it had been developed. This iterative
process would slow the pace of change, making it less abrupt to then “flip the switch on Alienbrain.”
Although this may increase the transition time, it might also have change management benefits. System
Administration thought it would go far in creating employee buy-in.
The three-person System Administration staff at Toybox was already heavily burdened ensuring the proper
functioning of the existing work pipeline. As a result, the human resources would be unavailable to devote
time to this type of project on a full-time basis. The proposed design strategy for the home-grown system
was given a generous timeframe of approximately eight months and would require hiring two to three more
employees. Contract programmers for this type of project could cost anywhere from $50,000 to $80,000
annually.
WHAT TO DO?
The fact that products such as Alienbrain were beginning to appear in the market indicated to Jones that
inefficient digital project management was a problem that other postproduction shops were facing. Jones
viewed the implementation of a new system as critically important for the company to remain competitive.
Movie scripts were beginning to pile up on his desk, and Jones knew that the time to implement change
was now. He wondered whether Alienbrain was the answer, or whether they should take the time and
resources to develop a tailor-made, in-house solution. Regardless of the decision, what would be the best
implementation strategy? How could the next post-mortem not be Panic Room II?
Page 201 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Page 9
Page 10
9B02E014
Exhibit 1
ORGANIZATIONAL CHART
Toybox T.O. — VFX Department
Michael Flanigan
President and CEO
Bill Varley
VP Engineering
Michael Ellis
VP VFX & Dig Film
VFX Sys Admin
Bruce Jones
VFX Manager
VFX Producers
Long Form
Page 202 of 282
SYS ADMIN
ASST.SYS.ADMIN
VFX & POST
Supervisors
VFX PRODUCER
VFX PRODUCER
VFX PRODUCER
VFX SUPERVISOR
VFX SUPERVISOR
DIG. POST PROD. SUPER
3D Animation
3D ANIMATORS
Digital Compositing
SEQUENCE SUPER
.
MAYA (4)
HOUDINI (2)
MATTE PAINTING
COMPOSITING SUPER
COMPOSITOR
LEAD ROTO ARTIST
VFX Prod. Support
PROD. ASSIST.
PROD. ASSIST.
Motion Control
& Digital Film
MOCO SUPER
DIG. FILM TECH.
DIG. FILM TECH.
SEQ. SUPER
SEQ. SUPER.
ART DIRECTOR
MATTE PAINTER
Technical Directors
& CG Development
MAYA TD
HOUDINI TD
CG DEVELOPER
VFX Sys Admin report to VP Engineering structurally, VFX Manager operationally
VFX Producers report to VP VFX & DF creatively, VFX Manager operationally and functionally
VFX Supervisors report to VP VFX & DF creatively, VFX Manager operationally and functionally
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Page 11
9B02E014
Exhibit 2
WORK ORDER
No. 6516/123756
Command Post Toybox
Date: Wednesday, Sept. 04, 2002
Transfer/CC
CLIENT
ADDRESS
SMEDI
Smith Editorial Inc.
180 Walnut St.
Toronto, Ont
M4T 2W2
PHONE / FAX
CONTACT
555-9538 / 555-8567
Fred Hubert / 555-9538x24
PROJECT
FOLDER
BOOKING
TITLE
PROMO
PROMO
HL
STATUS
Booking
Remarks
Normal
RESOURCES
START
END
DUR.
Colour Suite
01:00 p
05:00 p
04:00
Elaine Ford
Colourist
01:00 p
05:00 p
04:00
Molly Moyer
Coordinator
01:00 p
05:00 p
04:00
Kevin Camilleri
Assistant
01:00 p
05:00 p
04:00
Film to Tape
01:00 p
05:00 p
04:00
Digital Beta 525
01:00 p
05:00 p
04:00
LENGTH
COMMENTS
Colour 3
**C1-Digital 525**
VTR 6
QTY
STOCK
Bk’d by Hubert 8/08/02
BID AS PER ERIN ( per hour
+ hard costs)
PSA#1263
C.O.D.
35MM TO DBC (SUPPLIED STOCK)
NEG IS FROM TBA
# OF SHOTS TBA
LIST & CC WITH HUBERT?
ONLINE @ TBA
AGENCY – TBA
SUPERVISED BY – TBA
START
END
TOTAL
DOWN
_______________________________________
______________
_____________
________________________________
Signature of Operator
Archive
Audio
Fonts
Page 203 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Order taken by SYSADM
Page 12
9B02E014
Exhibit 3
VFX BREAKDOWN
ITEM
DESCRIPTION
Page 204 of 282
RESOURCE
DAYS
VFX 025
SC 69
Board #6
TECHNICAL REMARKS
CLUBHOUSE WALL – Platform shoes platform straight up
st
1 shot – ECU of shoes
LOCK OFF CAMERA
I
3D/CGI
VFX Comp
10
2
1
*One time cost to R&D CGI platform shoes
animate to suite scene
composite with plate of CU shoes
VFX 026
SC 69
Board #7
CLUBHOUSE WALL – Platform shoes platform up
2nd shot – POV above UB & SG rising up towards camera
towards top of wall
SOME CAMERA MOVE OK
3D/CGI
Design/Paint
VFX Comp
3
0
1
animate CGI platform shoes to suite scene
VFX 027
SC 69
Board #9
CLUBHOUSE WALL – Platform shoes platform straight up
3rd shot – POV through/behind guards as UB & SG rise up wall
SOME CAMERA MOVE OK
3D/CGI
Design/Paint
VFX Comp
3
0
2
animate CGI platform shoes to suite scene
VFX 028
SC 69
Board #10
CLUBHOUSE WALL –
4th shot: Camera POV from above, UB turning around
SOME CAMERA MOVE OK
3D/CGI
3D/CGI
VFX Comp
5
0
1.5
animate CGI platform shoes to suite scene
VFX 029
SC 69
Board #11
CLUBHOUSE WALL – shoes retract & expand as UB steps over
th
5 shot – side view of UB stepping over wall, shoes retract &
expand accordingly
LOCK OFF CAMERA
3D/CGI
Design/Paint
VFX Comp
3
0
1
animate CGI platform shoes to suite scene
VFX 030
SC 69
Board #13, 14
CLUBHOUSE WALL – Platform shoes platform down
6th shot – shoes retracting and expanding, as insert
shot as UB steps over top of wall
LOCK OFF CAMERA
3D/CGI
Design/Paint
VFX Comp
2
0
0.5
animate CGI platform shoes to suite scene
VFX 031
SC 69
Board #15
CLUBHOUSE WALL –
7TH shot: Camera POV from above, UB steps over wall
SOME CAMERA MOVE OK
3D/CGI
3D/CDI
VFX Comp
1.5
0
1.5
animate CGI platform shoes to suite scene
allowance for rig removal
allowance for rotoscoping & rig removal
allowance for rig removal
allowance for rig removal
composite with plate of CU shoes
allowance for rig removal
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
9B13E026
Douglas Scott wrote this case under the supervision of Joe Compeau solely to provide material for class discussion. The authors do
not intend to illustrate either effective or ineffective handling of a managerial situation. The authors may have disguised certain
names and other identifying information to protect confidentiality.
This publication may not be transmitted, photocopied, digitized or otherwise reproduced in any form or by any means without the
permission of the copyright holder. Reproduction of this material is not covered under authorization by any reproduction rights
organization. To order copies or request permission to reproduce materials, contact Ivey Publishing, Ivey Business School, Western
University, London, Ontario, Canada, N6G 0N1; (t) 519.661.3208; (e) cases@ivey.ca; www.iveycases.com.
Copyright © 2013, Richard Ivey School of Business Foundation
Version: 2013-09-17
As Lesley Stowe gazed out her office window at a rainy Vancouver evening, she couldn’t help but reflect
on the remarkable growth that her company had experienced over the past 10 years and its implications for
managing Information Systems and Information Technology (IS/IT). It was April 2012, and Lesley Stowe
Fine Foods (LSFF) was in the final stages of selecting an enterprise resource planning (ERP) system from
a shortlist of proposed solutions. Stowe began to review the notes she and her senior managers had
compiled over a long series of presentations and discussions with each of the candidates. Even though at
times the content felt overly technical, Stowe and her team were confident that implementing an ERP
system was an important step to take. She brewed a cup of her favourite loose leaf tea and began to think
things over once again.
COMPANY OVERVIEW
Parisian-trained chef Lesley Stowe founded Lesley Stowe Fine Foods in Vancouver, British Columbia in
1990 out of frustration with the lack of specialty foods in the city. LSFF’s combined offering of premium
catering services, cooking classes and specialty grocery products quickly became a hit with Vancouverites,
and the business enjoyed early success that continued through the 1990s. Recognizing the seasonal nature
of the catering industry, Stowe also negotiated contracts to provide desserts for many of Vancouver’s top
restaurants, a key component of the company’s stable profitability.
In the early 2000s, Stowe created a recipe for an artisan cracker to include in LSFF’s gift baskets. The
crackers — named Raincoast Crisps after Vancouver’s local ecosystem — were instantly successful, and
demand for the product resulted in several months of sustained stock-outs. Soon after, a difficult decision
was made to discontinue the company’s food service operations in order to focus on production of the
crisp.
A decade later, Raincoast Crisps were available in over 4,000 stores across North America and had been
featured on Oprah’s O List and in Martha Stewart’s Whole Living Magazine. In 2012, LSFF offered six
flavours of Raincoast Crisps, as well as a line of gluten-free oat crisps and a line of all-natural “power
cookies” aimed at competing with meal replacement bars (Exhibit 1 shows the product line for LSFF).
Page 205 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
LESLEY STOWE FINE FOODS: THE ERP DECISION
Page 2
9B13E026
Though Raincoast Crisps were widely available across Canada, LSFF’s distribution in the United States
was still limited. Given the company’s considerable domestic success, as well as strong preliminary results
across the border, it seemed that LSFF was positioned for exponential growth as it entered the U.S. market.
PRODUCT
Stowe developed the crisps as an accompaniment to other gourmet products that LSFF featured in its
signature gift baskets at the time. The first crisps, prepared in Stowe’s oven at home, were made from
brown bread dough mixed with a flavourful variety of seeds and nuts. The result was a cracker that had a
very distinct flavour and texture but that was neutral enough to complement a very broad range of
toppings, especially the premium cheeses, dips and antipasto for which they were designed. This recipe
was still used in the original flavoured crisps that remained bestsellers in 2012.
The immediate success of the original recipe soon led to the development of new flavours based around
more specific taste profiles. LSFF offered seven varieties: original, rosemary raisin pecan, salty date and
almond, cranberry hazelnut, fig and olive, cinnamon raisin and a seasonal winterfruit flavour. In response
to recent changes in consumer preferences, in 2012 the company released three flavours of gluten-free
crisps, which were made with oat flour.
2012 also marked the launch of LSFF’s first product outside of the cracker category: the Raincoast Cookie.
Stowe created the cookies after identifying the opportunity to make a better-tasting alternative for meal
replacement bars. The company offered three flavours: apricot, ginger and slivered almond; dried
cranberry and toasted hazelnut; and dark chocolate, tart cherry and pecan. The cookies were selling well
through local distribution channels, but due to time constraints, management had not yet been able to push
expansion beyond western Canada.
LSFF’s commitment to exceptional quality was a key consideration for all of its products. Despite huge
increases in production scale, the crisps were still mixed, baked and sliced in small batches, using only the
best ingredients. Product quality also represented a critical point of differentiation between LSFF products
and imitation brands that attempted to sell replica products at a lower price point.
MANAGEMENT AT LSFF
LSFF’s management team and organizational structure remained largely stable over the years following the
company’s move to dedicated production of the crisps. As the founder, owner and president, Stowe was the
key decision maker and was actively involved in the management of all business functions. In recent years,
her role in marketing and sales had become an increasing time commitment as expansion to new regions
required her to travel for trade shows, in-store demonstrations and distributor negotiations. Stowe had
worked with Vice-President (VP) Maggie Arro for over two decades and relied on her for day-to-day
management of the company. Other key decision-makers included VP Operations Ali Samei, Director of
Sales Susan McVee, Accountant Chris Bray and Production Manager Maybo Wu.
Page 206 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
In 2010, production volume necessitated the move from the company’s small facility in midtown
Vancouver to a large-scale manufacturing space located outside of the city. The new plant, combined with
prudent capital investment in new equipment, dramatically increased LSFF’s production capacity without
significant overhead increases or capital constraints.
9B13E026
In its early years, the company’s operations were small and processes were very simple. As a result, when
issues arose within specific business segments, they tended to have implications for the entire firm. In
response, Stowe and her managers were forced to work together to address each unique challenge as it
arose. This cross-functional approach became a key component of corporate culture, and Stowe believed
that it allowed the LSFF team to be exceptionally nimble as the company went through its sustained period
of high growth. But by 2012, continued growth was beginning to strain the capacity of LSFF management.
Communication between managers became increasingly difficult as each business segment came to be
larger and more complex. As well, the absence of lower-level staff often forced managers to complete
time-consuming tasks that would have otherwise been outside the scope of their responsibility.
LEGACY SYSTEMS AT LSFF
When LSFF began dedicated production of the crisps, a combination of low volume and a focus on local
distribution created a need for only the most basic of information systems. For the first few years,
management relied on manual data collection, using a combination of handwritten notes and ad hoc Excel
spreadsheets to manually track and compile information.
However, the company’s rapid growth made information increasingly complex as it scaled up volume,
expanded geographic distribution and added new flavours. By 2006, management was no longer able to
effectively record and process data informally and, as a result, adopted a software-as-a-service (SaaS)
enterprise resource planning (ERP) platform through a boutique software firm. This system was
implemented with two specific goals. The first was to support the collection and reporting of data related to
specific business functions, in particular sales order fulfillment, production, inventory management and
accounting. The second was to consolidate the data recorded in separate functions in order to provide
information to support firm-level management and strategic planning.
The SaaS ERP platform supported four key areas of LSFF operations: sales order fulfillment, production,
inventory management and accounts receivable. LSFF used the Simply Accounting Product for Accounts
Payable, a product called Ceridian, for payroll, and Salesforce.com for marketing and customer
relationship management (CRM). Unfortunately, from its inception, the SaaS ERP platform frustrated
management with a variety of problems. For some business segments, such as production and inventory
management, the system’s limited functionality necessitated continued use of manual information
recording and processing. For other segments, such as shipping, the system offered no functionality at all,
and manual data collection processes continued. These processes were time-consuming and almost always
carried out by managers. Compounding these clear issues was the problem of poor ongoing service from
the vendor. LSFF management felt that the vendor was slow and offered insufficient responses for basic
problems. Moreover, LSFF had no dedicated IT staff, and this lack of system support only drove managers
further towards manual processes when technical issues arose.
In many cases, inadequate data had direct and significant implications for profitability, as highlighted by
one incident encountered by VP of Operations Samei:
As I was compiling my monthly inventory spreadsheets for a report, I grew curious about certain
inventory items, and was shocked to see upon compilation that in the last 12 months we had
purchased over 100,000 pounds of sunflower seeds! In order to use [our current system], I had to
manually track purchases on an individual basis, and aggregated information was not available.
Within days of this discovery I negotiated a 20 per cent discount with our supplier, which wouldn’t
have been possible if I hadn’t happened to add things up on my own.
Page 207 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Page 3
Page 4
9B13E026
It got to the point where a simple monthly performance report was taking me eight to 10 hours to
manually compile and format. At that time I’d say 90 per cent of all management reports were still
made individually in Excel.
Years of frustration came to fruition when, in fall 2011, Arro formally approached Stowe about the need
for change. After a series of meetings, the company commissioned IDEACA, a boutique ERP consulting
firm, to do an operational audit of LSFF, create a request for proposal (RFP) and advise during the
selection process.
THE REQUEST FOR PROPOSAL
The RFP was issued on February 23, 2012. It included details of the expected requirements of the system
and the implementation process, as well as information about LSFF and the decision making process.
Exhibit 2 contains excerpts of the key elements of the RFP.
Of the eight RFP responses received, the LSFF team invited five candidates to pitch their proposals to the
company’s board. The results of these interviews were mixed; three candidates were eliminated quickly.
The team liked one proposal made by SAP; however, the proposal was contingent on an additional thirdparty add-on called BatchMaster. This software appeared to provide many unnecessary functions and was
prohibitively expensive. Another vendor pitched an Infor ERP system, and at first the software seemed to
be a great match for LSFF’s functionality needs. However, after seeing demonstrations of the software, the
team felt its bare bones appearance and complicated dashboard were not user friendly enough for their
needs. Two separate vendors proposed Microsoft Dynamics NAV systems, one of which seemed like a
poor fit and was eliminated immediately. The other, provided by a Vancouver-based vendor called the
Underwood Group, was very well-received. A fifth proposal made by another local vendor, Barnes and
Co., was also viewed as very strong.
Both of these proposals were made by experienced vendors offering solid software, and the team was faced
with the difficult task of deciding between the two.
Final Proposals
Barnes and Co. — Sage ERP X3
Founded in 1995, Barnes and Co. was a Vancouver-based technology consulting firm, which provided
Sage ERP systems to mostly small and medium-sized enterprises (SMEs). In recent years, the company
had some success with implementations in the food space, including two national distributors of organic
foods and one regional restaurant chain. The company listed projects for a hardware manufacturing firm
and a supplier of specialty auto parts as specific examples of Sage ERP X3 implementations.
The Sage ERP X3 system was targeted at mid-market businesses (50 to 5,000 employees) with
international requirements and strong growth in manufacturing, distribution and service industries. X3 was
Page 208 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Moreover, the system offered almost none of the promised functionality relating to data integration across
business segments. This was of critical importance, as any information to support decision making at the
firm level had to be collected from each segment and compiled by hand. Being involved in day-to-day
management made this especially frustrating for Arro:
Page 5
9B13E026
Barnes and Co.’s response was 67 pages in length and included a very detailed table indicating how the
Sage software met or exceeded the RFP requirements (Exhibit 3 shows an excerpt from this table). The
Sage software was described in detail with information about each of the modules that made up the product
that came from Sage’s product literature.1
The project methodology began with the sales hand-off and then proceeded to a project kick-off meeting.
This was followed by a project planning phase where the preliminary statement of work (to be developed
by LSFF and Barnes and Co.) would be converted to the formal project plan. The next phase involved the
configuration and implementation of the software at the hosting facility, based on a requirements
document. The implementation activities followed an iterative approach using a series of “conference room
pilots” where transactions were tested as if they were being processed in a live environment. After
configuration and implementation, the next phase was project closure, followed by account manager
follow-up.
The proposal included provisions for off-site hosting, which would be provided by the implementer
through a partnership with an affiliated data services firm. Costs for this option, based on an hourly billable
rate of $160 with no billable travel time, is included in Exhibit 4.
The Underwood Group — Microsoft Dynamics NAV
Founded in Vancouver in 1989, the Underwood Group provided implementation, development, project
management and support services on Microsoft Dynamics ERP and was the largest and oldest Microsoft
Dynamics Partner in western Canada. Underwood Group’s proposal highlighted extensive and specific
experience implementing Dynamics NAV at mid-size food processing and manufacturing companies. The
implementer offered strong references from a processor of specialty Asian fish, a Vancouver-based roaster
of fair-trade coffee beans and a processor of high quality fruit products, all of which had international
distribution.
Microsoft Dynamics NAV was Microsoft’s ERP solution, aimed to provide comprehensive solutions and
management functionality for SMEs. The software did not have a specific industry focus but was designed
to streamline financial processes and reporting, generate real-time data for production functions, improve
sales and customer relationship management and support international operations.2 Dynamics NAV was
also designed around considerations for growth, and the software facilitates the addition of new business
processes to the system post-implementation. The response from the Underwood group was 23 pages in
length and included at least some discussion of each of the required sections, though it did not follow the
template precisely. The proposal included options for LSFF to host the Microsoft Dynamics NAV software
in-house (on premises) with professional services for implementation provided by Underwood, hosted by a
third-party provider (hosted) and through a SaaS option. Details on the hosted and SaaS options were a bit
limited in the proposal, but additional information gained during the presentations clarified the missing
information.
1
2
http://na.sage.com/~/media/Company/ERP/White%20Papers/Sage-ERP-X3-Version-6, accessed September 16, 2013.
http://www.microsoft.com/en-ca/dynamics/erp-nav-overview.aspx, accessed September 16, 2013.
Page 209 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
developed by Sage to specifically address the needs of mid-size companies for whom large-scale ERP
solutions were thought to be unnecessarily complex and unwieldy. The goal of X3 was to offer the benefits
of ERP performance while focusing on simplicity and functionality. The system was also intended to be
scalable in recognition of the growth opportunities available to medium-sized businesses. Sage ERP X3
was used daily by over 170,000 users at more than 3,000 companies in 53 countries around the world.
Page 6
9B13E026
Underwood emphasized that it was a local partner who could serve LSFF well. It had a staff of 35. The
proposed implementation team members were relatively young and energetic but very experienced, with
each of the six members having at least eight years in the field. The proposal did not detail the expected
resource effort required from LSFF employees but estimated it based on past experience at between 166
and 249 person days (two to three times their own implementation services).
Underwood followed a five-phase project methodology running from analysis to design, construction, and
deployment, culminating in “go live.” The proposal provided a fairly detailed training and implementation
plan, using a phased roll-out with core features implemented first and production and capacity planning
introduced only in phase two. The response argued that such a phased approach would facilitate learning
and adaption to the new software and would also make it easier for the project team to manage the rollout.
The costs for this option, based on time and materials billing, are shown in Exhibit 5.
THE DECISION
Stowe finished her tea and prepared to leave for the day. She was more confident than ever that an ERP
system was a necessary step for her company. That being said, she was still unsure about which of the two
options to choose. She knew that the key to the decision was identifying which proposal best addressed the
needs of the company both now and in the future. The next step would be sitting down with her team to
figure out which one that was.
Page 210 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Three tools for supporting data migration were described in the proposal. The data migration process
seemed to involve extracting data from current systems, cleaning and harmonizing it to a standard format,
transforming it to the format required for NAV and then loading it into the new tool. Underwood also
offered some access to industry standard data (such as payment terms, country codes, etc.) that could be
automatically imported.
Original and Rosemary Raisin Pecan Crisps
Raincoast Crisps served with Goat’s cheese
Page 211 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Page 7
9B13E026
EXHIBIT 1: LSFF PRODUCTS
Page 8
9B13E026
Rosemary Raisin Oat Crisps
Two flavours of Raincoast Cookies in packaging
Source: Company records,
accessed September 16, 2013.
http://www.newboro.com/Kilborn's/Gourmet%20Foods/leslie_stowes_raincoast_crisps.jpg,
Page 212 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
EXHIBIT 1 (CONTINUED)
Page 9
9B13E026
EXHIBIT 2: KEY EXCERPTS FROM REQUEST FOR PROPOSAL
Request for Proposals
LSFF-RFP-001: ERP Implementation
Closing time:
Proposal must be received before 2:00 PM Pacific Time on March 16, 2012
1.0 INSTRUCTIONS TO PROPONENTS
1.1 Introduction
This Request for Proposal (RFP) outlines Leslie Stowe Fine Foods Ltd (LSFF) high-level requirements for an
Enterprise Resource Planning (ERP) system. Interested Proponents are invited to respond to this RFP by March 16,
2012. Short-listed Proponents will be invited to present a live system demonstration and present the essentials of their
proposal to the LSFF Evaluation Team during the days of April 2 – 4, 2012. LSFF is extremely value conscious in
making this acquisition and it is anticipated that the initiative must provide substantive benefits relative to the total
lifecycle costs.
1.2 RFP Terminology










Evaluation Team: The individuals who will evaluate the Proposals on behalf of LSFF.
LSFF: Lesley Stowe Fine Foods Ltd.
Optional: A requirement not considered essential, but for which preference may be given.
Proponent: A company or consortium that submits, or intends to submit, a Proposal in response to this RFP.
Proposal: The Proponent’s response to the RFP, which includes all the Proponent’s attachments and presentation
materials.
Request for Proposal (RFP): This solicitation for Goods and Services, including attached appendices.
Services: The contracted services as specified in the RFP, the Proposal, and any resulting Contract.
Should or Desirable: A requirement having a significant degree of importance to the objectives of the RFP.
Validation: The stage in the RFP process during which the LSFF may confirm, through testing, that a
Proponent’s proposed solution complies with the requirements of the RFP.
Vendor: A successful Proponent to this Request for Proposal who enters into a Contract with LSFF relating to
the subject matter of this RFP.
1.3 Confidentiality
The contents of this RFP and all information provided by LSFF are to be considered confidential information and
remain the property of LSFF. We remind all vendors of the Mutual Non-Disclosure required. All Proponents should
hold all information received in this document, including follow-up information and documents, in strict confidence
and not disclose this information to any third party except to its employees, lawyers and consultants on a need to
know basis.
Copies of the document may be reproduced by the Proponents, only to the extent that it is required to allow the
Proponent’s party and its employees to respond to this RFP. Upon the request of LSFF, the Proponent agrees to
destroy or return to LSFF all copies of this RFP.
Proposals submitted to LSFF will also be held in confidence and shall be considered the property of LSFF.
Page 213 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Issue date:
February 23, 2012
Page 10
9B13E026
EXHIBIT 2 (CONTINUED)
1.4 Participation
1.5 RFP Availability
This RFP is open to Proponents who have been invited to submit a Proposal. Proposals from other Proponents will be
considered as long as they are submitted as per this RFP.
1.6 Eligibility to Respond
Proposals will not be evaluated if the Proponent’s current or past corporate or other interests may, in LSFF’s sole
opinion, give rise to a conflict of interest in connection with this RFP.
1.7 Changes to the RFP
In the event there are modifications or additions to the RFP, all Proponents who have returned the Receipt
Confirmation Form (Appendix A: Receipt Confirmation Form) will be notified.
1.8 Closing Date for Proposal Submissions
The Closing Date and Time for this Request for Proposal is March 16, 2012 at 14:00 PST. Late proposals will not be
accepted and will be returned unopened to the Proponent.
If Proposals are sent by mail to the LSFF, the Proponent shall be responsible for actual delivery of the Proposal to
LSFF before the advertised Closing Date and Time. If mail is delayed for any reason beyond the date and time set for
the closing, delayed proposals will not be considered and will be returned unopened.
1.9 RFP Cancellation
LSFF has the right to cancel this RFP at any time at its sole discretion, without award or compensation to any party.
1.10 RFP Schedule of Events
Planning dates are subject to change at the discretion of the LSFF. Any changes will be communicated as described in
section 1.7 Changes to the RFP section.
Page 214 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Participation in the RFP is at the sole risk and expense of the Proponent. The Proponent understands and
acknowledges that LSFF makes no representation or guarantee as to the Proponent’s chances of success. LSFF shall
not be responsible for, nor shall LSFF reimburse the Proponent for any costs associated in responding to this RFP,
including, but not limited to, any costs relating to goods and services provided for the purpose of demonstration.
Page 11
9B13E026
EXHIBIT 2 (CONTINUED)
April 2 – 4, 2012
April 5 – 11, 2012
April 16, 2012
April 30, 2012
September 30, 2012
Event
RFP issued to Proponents
Proponents meeting conference call — open question and
answer session
Proponents Receipt Confirmation Form due
RFP closing date and time
RFP evaluation period
Notification to short-listed Proponents for the scheduling of
demonstrations
Short-listed Proponent presentation / system demonstration
Proponent customer site visits
Awarded / cancelled by (target date)
Intended project start date
Intended project go-live (if not sooner)
1.11 Proponent Questions
All communications should be directed in email to: , ERP Evaluation Project Manager
(Email: ) with the following subject line: “LSFF-RFP-001”.
LSFF will provide responses to all Proponents, unless the information is of a confidential nature. Any information
that may be received from LSFF personnel other than the ERP Evaluation Project Manager is not to be relied on and
should not be considered accurate.
1.12 Acknowledgement and Acceptance to Respond
Proponents shall complete and return the attached “Receipt Confirmation Form” (See Appendix A: Receipt
Confirmation Form) on or before March 2, 2012 before 14:00 PST as indicated in section 1.10 RFP Schedule of
Events.
1.13 Limitation of Damages
Although LSFF has devoted considerable efforts to ensuring the accuracy of the information in this RFP, this
information is supplied solely as a guideline for Proponents. The information is not guaranteed or warranted to be
accurate by LSFF, nor is it necessarily comprehensive or exhaustive. Nothing in this RFP is intended to relieve
Proponents from making investigations or forming their own opinions and conclusions with respect to the matters
addressed in this RFP.
1.14 Limitation of Liability
LSFF shall in no event be responsible or held liable for damages, including without limitation, liability for costs of
preparing the quotation, loss of profits or loss of property, however the same may be caused, including the negligent
acts or omissions of LSFF, and the Proponent hereby releases, indemnifies and agrees to hold LSFF harmless from
any liability arising from the bidder’s submission of a bid proposal in accordance with the bid documents.
Page 215 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Date
February 23, 2012
February 29, 2012
10:00 – 11:00 AM PST
March 2, 2012, 14:00 PST
March 16, 2012, 14:00 PST
March 19 – 23, 2012
March 26, 2012
Page 12
9B13E026
EXHIBIT 2 (CONTINUED)
1.15 Price Quotation
Proponents’ price quotation must be comprehensive and include all costs incidental to the successful delivery of the
goods and services that are the subject of this RFP and are detailed herein. Proponents are responsible to determine,
from careful examination, the methods, materials, labour, permits, insurance and time required for successful delivery
of the goods and services which are the subject of this RFP and are detailed herein, and must reflect the same in the
proposal and price quotation. Any costs not stated in the proposal will be borne by the Proponent, unless specifically
agreed to by LSFF in writing.
Proposals shall be considered firm for at least 60 days from the closing date of the RFP and will be used as the basis
for an agreement. LSFF reserves the right to incorporate all or part of your proposal into any resultant agreement. The
Proponent’s response shall be legally binding on the Proponent.
Prices quoted are to be firm for the duration of the proposed Agreement (with the exception of those prices which
bidder cannot control). Costs which the vendor does not have direct control of which would be passed on to LSFF
should be stated in the bid.
1.16 Recommendations / Alternative Proposal
We are value oriented and price sensitive. You are encouraged to be innovative in your approach and proposed
application solution.
LSFF will consider recommendations for alternative solutions and for best practices for managing and executing the
work. If you feel that there are areas within the scope of this RFP, where costs could be reduced or value could be
improved by performing the work differently, or using alternative components of your solution with changes to
requirements, please provide an alternate proposal in your proposal. The alternate must be clearly described as an
alternate proposal.
1.17 Proposal Format
Proponents must at minimum provide an electronic copy of the Proposal via email including a note of the Proponents
authorized signatory. In addition, the Proponent can also submit four (4) copies of the Proposal, dated and signed with
the signature of a person authorized to bind the Proponent.
1.18 RFP Evaluation Criteria
1. Functionality and Fit:
1.1. Evaluation of ERP system functionality against LSFF list of identified future needs.
1.2. An assessment of technical capabilities and fit of the Proponent’s ERP system against identified LSFF needs and
requirements.
2. Investment and Ongoing Cost:
The overall cost of the Proponent’s proposed solution, including acquisition costs, total cost of ownership and
vendor’s time and material rates, of note:
2.1. Investment cost (up-front) e.g., implementation services and software licensing
2.2. Ongoing costs e.g., subscription fees and hosting fees. Preference will be given to proposals that offer a low
ongoing cost.
Page 216 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
All prices quoted shall be in either Canadian or US currency. Taxes or other charges such as estimated travel
expenses, where applicable, must be indicated and shown separately.
Page 13
9B13E026
EXHIBIT 2 (CONTINUED)
3.1. Corporate vision, financial viability, as well as past and future focus on British Columbia.
3.2. Ability to provide local implementation and ongoing support resources – ideally based within a 250 km radius of
Richmond, B.C.
3.3. Proposed resources – business savvy, prior relevant ERP implementation experience.
3.4. Implementation approach and willingness to provide insights, best practices, and typical / pragmatic
implementation experience.
3.5. Account management, customer service as demonstrated through relevant references, interactions during the RFP
process and assessment questions.
4. Software Vendor:
4.1. Corporate vision, financial viability and stability.
4.2. Ability and commitment to meeting the requirements of the RFP and / or future product improvement plans to
address gaps.
4.3. ERP product vision and future investment plans, overall ERP roadmap, vendor’s commitment to the ERP space
and industry focus (preference will be given to software vendors proven in the food manufacturing sector).
4.4. Account management, customer service as demonstrated through references, interactions during the RFP process
and assessment questions.
Proposals meeting all of the mandatory criteria will be further assessed against desirable criteria as outlined below.
Criteria
Functionality and Fit
Cost — Investment, Ongoing
System Implementer and Approach
Software Vendor
Total
Weighting (%)
30
30
20
20
100
2.0 SUMMARY OF LSFF ERP REQUIREMENTS
2.1 LSFF Company Overview
LSFF is a family-run business and was started by Lesley Stowe in 1990. Operating within the consumer goods
market, the company has gained its competitive edge by manufacturing hand crafted high-quality crackers containing
high quality products. The product range consists of cookies and crisps which are marketed as snacks to consumers.
Recently, LSFF has experienced exponential growth and the company is currently preparing for future expansion. The
key to both sustain and facilitate this growth is a technology base which provides a sound foundation for the daily
operations by which the company operates. The current processes are supported by a combination of bespoke and
isolated systems whilst significant time is spent on coordinating manual processes due to end-users do not trusting the
systems.
LSFF requires an ERP solution which is user friendly and simple in functionality. Current operations within the
company are not significantly contributing to a competitive advantage and the expectation is that an out-of-box ERP
installation should be a feasible option.
Page 217 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
3. System Implementer:
Page 14
9B13E026
EXHIBIT 2 (CONTINUED)
2.2 General Information
2.2.1 Product Lines
Product Line 1 “Raincoast Crisps” has six flavors
Product Line 2 “Raincoast Cookies” has three flavors
2.2.2 Key Facts






LSFF plans on 10% year over year growth
Approximately 6,600 sales orders are placed per year
LSFF currently operates from one production location in Richmond B.C.
Production volumes are based on seasonal demand with peak periods being from October to December
The maximum number of orders received, historically, was 55 on a single day
Orders are currently shipped between Mondays and Thursdays
2.2.3 Distribution Channels
LSFF does not currently interact directly with end-consumers of the product, nor is there any intention to do so in the
future. The following channels are used to distribute the products in Canada and the USA:
 Distributors
 Wholesalers
2.2.4 Production Process





The production process takes approximately two days to complete and is described in more detail in Section 6.2
Production.
Finished goods inventory is typically kept on-site up to 2 weeks prior to shipping it to customers.
The production line is able to facilitate the production of one product, consisting of one flavour, at a single point
in time. The most efficient method to structure production is to have one flavour produced on a day.
Production is predominantly based on made-to-stock for Crisps.
Production of Cookies is akin to make-to-order but is handled in a make-to-stock manner i.e., the production
order may be sized for a larger quantity than the sales order.
2.3 Scope
The scope of the intended ERP implementation project is focused on the following business functions and processes:
 Order Fulfillment (Sales, Shipping, Customer Invoicing)
 Production (Sales Forecast, Production Scheduling, MRP)
 Materials Management (Purchasing, Inventory Management)
 Exception Handling (Customer Returns, Finish Goods Recall, Supplier Returns)
In addition, consideration will be given to Proponents who can provision a future implementation (part of the ERP or
via third–party) for the following:
 Basic Customer Relationship Management e.g., contacts and contact history
 Bar Coding e.g., Inventory / Shipping
 Point of Sale
 Basic Contact / Relationship Management (to replace SalesForce.com)
Approximately 10 people would make use of the ERP system equating to 2 – 5 concurrent user equivalents.
Page 218 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.


Page 15
9B13E026
EXHIBIT 2 (CONTINUED)
HR and Payroll are considered fully out of scope other than a basic element required to support labour time tracking
for production.
The current technology landscape consists of the following:
 Sales Order Fulfillment (MonkeyMedia)
 Production (MonkeyMedia, coupled with manual processes and spreadsheets)
 Inventory Management (MonkeyMedia, coupled with manual processes)
 Shipping (manual process)
 Accounting (MonkeyMedia – Accounts Receivable, Simply Accounting – Accounts Payable)
 Marketing / Relationship Management (SalesForce.com) – optional scope
 Payroll (Ceridian) – out of scope
A core objective of the new ERP project is to consolidate the various functions enabled by the existing systems in
place at LSFF into one integrated ERP system.
2.5 Future System Scalability and Integration
Considerations:
 A pivotal requirement is the ability of the selected ERP system to scale up to increased LSFF business demand,
new product lines, increasing business complexity, multiple manufacturing sites in Canada and the USA, etc.
 At this time, there are no requirements to enable system integration within LSFF or with external organizations.
That said, consideration will be given to Proponents who propose systems that have proven capability or
extensibility options to enable:
o System integration with SalesForce.com e.g. customer master data synchronization (if not having the
features to replace SalesForce.com)
o Sales notifications to customers (system to system)
o Courier integration
o Electronic data capture, such as barcode scanning
 LSFF currently does not have an internal IT department. LSFF will consider the following options for the new
ERP system:
1. Software as a Services (SaaS) – preference will be given to offerings were LSFF could in the future
elect to bring the software and system in-house
2. Off-site, hosted by a third-party
3. On-premise, hosted at LSFF
3.0 RESPONSE REQUIREMENTS
The following section outlines the detailed requirements which Proponents are requested to provide.
 Proponents are asked to format their responses in the sequence provided in this section.
 For your convenience, an electronic copy of this document can be made available, should you wish to insert your
responses directly into the template.
 If you are unable to meet any of the requirements outlined in this inquiry, you must clearly state in your response
that you are unable to meet the requirement.
Page 219 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
2.4 Current Technology Landscape
Page 16
9B13E026
EXHIBIT 2 (CONTINUED)
3.1 Title Page and Executive Summary
3.1.2 Provide a brief Executive Summary which summarizes the proposal and why your firm and application suite are
uniquely suited for LSFF.
3.2 Proponent Profile
3.2.1 Provide an overview of your company’s background, highlighting any specialty areas.
3.2.2 Describe your company’s strengths and distinguishing factors.
3.2.3 Describe your company’s relevant experience with ERP implementations, specifically for the food
manufacturing and distribution industry.
3.2.4 Document how long your firm has been in business (for both the software vendor and the system implementer,
when incorporated) and facts describing the firm’s financial viability (e.g., sales revenues, operating profit for at
least the past five years).
3.3 Requirements Matrix
Please refer to Appendix B: Requirements Matrix to view the Requirements Matrix which will be used for measuring
the degree of fit. Significant preference will be given to Proponents who provide most of the functionality “out of the
box” while befitting the pricing and investment effort for a small-to-medium sized food manufacturer.
Proponents are requested to complete this matrix directly in the provided Microsoft Excel format in order to allow
LSFF to score the products in an efficient manner. We request that the Proponent answer every question for
Complexity; one of:
Complexity
Out-of-the-Box
Configuration
Functional-Workaround
CustomizationDevelopment
Add-on (ISV)
Does not exist
Definition
This requirement is included in the base functionality of the ERP solution. There is
no need for Configuration, Customization, Software Development, nor is it an addon from an independent software vendor (ISV). This is the most ideal status.
The functionality of this requirement is included in the Out-of-box ERP solution but
it will require much or little configuration to meet its needs.
This requirement can be met with an alternative use of the ERP’s base functionality
in order to meet the requirement. No software development is required.
This requirement can be met but only with custom Software Development within or
integration with the ERP solution. It may also consist of a Functional-Workaround.
This requirement can be met with the purchase of the functionality from an
authorized Independent software vendor (ISV) in which is partially or fully
integrated into the suggested ERP.
This requirement exceeds the ERP’s functional/technical capabilities, it cannot be
met neither by Configuration, Customization, Software Development, nor is it an
add-on from an independent software vendor (ISV). This is the least ideal status.
Page 220 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
3.1.1 Provide a title page showing the date, RFP number, Proponent’s corporate name and address, telephone number,
facsimile number and contact person.
Page 17
9B13E026
EXHIBIT 2 (CONTINUED)
3.4 Technology
3.4.2 Describe your company’s policy for data ownership. Would LSFF be able to extract its data from your solution,
if required? If yes, explain how and any costs your company would impose on LSFF.
3.4.3 Outline what data migration utilities your ERP software supports. Make note of any accelerators your software
has that would help expedite data migration from some, or all of, the existing LSFF systems in place.
3.5 Planned ERP Implementation Approach, Method and Tools
3.5.1 Describe your company’s planned approach in order to meet the requirements of LSFF.
3.5.2 Based on your past experience, outline a training approach which is most suited for a project of this nature.
3.5.3 Describe the challenges and risks which you foresee for this project.
3.5.4 Describe your ERP implementation approach, guiding principles and how your firm ensures delivery of a
quality solution aligning to the business requirements and future growth considerations.
3.5.5 Outline your company’s approach to data migration and expectations on LSFF.
3.6 Macro Project Plan
3.6.1 Provide a Gantt chart based on the estimated timelines which shows the macro planned project activities and
clearly highlight the critical project milestones and significant dependencies.
3.7 References
3.7.1 Provide three (3) references which are specific to your organization’s previous food manufacturing
implementations and/or are relevant to this ERP implementation. Note that local references will receive
preference during the evaluation of this RFP.
Please include response reference name and contact details, project description and comments
3.7.2 Should your company be short-listed for the next round of review, please specify if an on-site visit to one of
your reference / existing customers would be a feasible option? Preference will be given to customers that are
located within 250 km of Richmond B.C.
3.8 System Implementer Resources
3.8.1 Provide resumes for the team which your company would like to propose for this project highlighting relevant
experience. Clearly highlight if sub-contractors, or resources from a company other than your own, would be
used. LSFF understands that the system implementer may have to substitute some of the proposed resources for
others of equal, or better, experience depending upon LSFF’s decision and contract award timing.
Page 221 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
3.4.1 Please outline if your proposed solution can be offered as a hosted solution (on premise, off premise) or
Software as a Service, or both.
Page 18
9B13E026
EXHIBIT 2 (CONTINUED)
3.8.2 If possible, please denote which resources you can commit to assuming mutual adherence to the LSFF schedule
of events as outlined in Section 1.10 RFP Schedule of Events.
3.8.4 Provide an estimate for the involvement you would require from LSFF’s resources. For increased clarity, please
make note of these requirements by major project stage (e.g., blueprint, build, test, implement and stabilize).
LSFF Employee
Sales Manager
Operations Supervisor
Executive
Shipper
Finance
Other LSFF staff
Total Estimated Effort
3.9 Fees – Implementation Services
3.9.1 Provide the total effort estimates and consulting fees for each of the proposed resources according to the
following table (please specify currency – CDN or USD):
Resource Name And Role
Total Estimated
Effort (Hours)
Hourly
Rate
Total Cost
(Plus
Taxes)
3.9.2 Please specify the invoicing / financial basis (e.g., time and materials, fixed price) and whether your company
can offer any risk-reward mechanisms.
3.9.3 Please outline anticipated travel expenses, including planned number of trips per proposed consultant role, if
applicable. Please note LSFF does not plan to cover any travel and commuting costs within the Metro
Vancouver, BC region.
Resource Name And Role
Estimated Trips
Consultant
Home
Base (City)
Travel
Expenses
3.10 Fees – Software (and Hosting)
3.10.1 Outline the detailed cost estimates for the ERP software and the pricing options (e.g., concurrent user licensing,
named user licensing, subscription fees, etc.). Provide details for initial investment costs as well as ongoing
service / assurance fees (e.g., annual maintenance fees). Please provide pricing for any, or all, of the following
options for your proposed ERP solution:
1. Software as a Service (Cloud)
2. Off-site hosted
3. On-premised hosted
Page 222 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
3.8.3 Please outline where the proposed consultants are based i.e., home office.
Page 19
9B13E026
EXHIBIT 2 (CONTINUED)
3.10.3 If providing a quote for on-premise hosted software, please also provide a cost estimate for what LSFF would
need to provision in hardware (assume a DEV – PROD landscape), including hardware specifications. LSFF
understands that this is a quote only and that detailed hardware sizing would need to be done to finalize the final
hardware requirements and investment cost.
3.11 Risks
3.11.1 Outline the top risks and challenges your foresee LSFF facing in this ERP implementation and how your
company has, or would plan to, overcome such challenges to ensure successful project delivery.
3.12 Change Management
3.12.1 Outline how your company will help to ensure end-user adoption and buy-in of the new ERP solution. Outline
what activities your company employs to ensure this success.
3.13 Assumptions
3.13.1 Please state clearly all assumptions. Any exceptions to the requirements outlined in this inquiry should be
explicitly stated in your response.
3.14 Proponent Feedback
3.14.1 (Optional) Provide any other feedback or suggestions based on your organization’s previous experience.
3.15 Presentation Requirements


Short-listed Proponents will be given the opportunity to showcase their products on-site at LSFF in Richmond
B.C. during a 2 - 3 hour presentation session to be scheduled one of the days between April 2 and 4, 2012.
Short-listed Proponents will be required to showcase a live demonstration of their proposed ERP system and to
answer questions. Please use the scenarios identified in Appendix C: High-level Future Processes as a guideline
for the high-level functionality demo (Note: PowerPoint slides can be used to support the presentations, but a live
system demonstration is a critical requirement to qualify for consideration. Proponents are encouraged to focus
on the ERP system fit to the LSFF business requirements and how the Proponent will approach the project to
ensure success).
Source: Company Records
Page 223 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
3.10.2 If providing a quote for off-site hosted software, please also provide a proposed price for the ongoing hosting
fees and any initial set-up fees. Please clearly identify if a third-party hosting company is providing this service
and how this company would invoice for its services (e.g., direct to LSFF, via the System Implementer or
Software Vendor).
Page 20
9B13E026
EXHIBIT 3: FUNCTIONALITY REPORT (EXCERPT FROM BARNES AND CO. RFP RESPONSE)
Page 224 of 282
Module
Number
Importance
Time
Entry
103
75%
Time
Entry
104
25%
Time
Entry
105
50%
Time
Entry
106
50%
Time
Entry
107
50%
Time
Entry
108
25%
Complexity
Comments
Production
— Does your system provide the ability for
operation related users to enter time against specific
production orders — and/or by
production order operation?
Budget control
Does your system provide the ability to
track timesheets against a budgeted
amount? If so, please elaborate.
Ease of use
Does your system provide efficient and
user friendly timesheet entry? If so,
please elaborate.
Requirement
Question
Out-of-thebox
Yes, at the operation level for job
cost reporting
Add-on
Sage Timesheets
Out-of-thebox
Payroll
time Does your system enable easy timesheet
detail interface
data coordination between a Payroll
provider (e.g. Ceridian and ADP) and
your system? If so, please explain what.
Reporting
Does your system provide the ability to
report on the hours worked by employee,
project, and/or element to create reports
in Excel such as % utilization, OT*
hours worked, YTD** hours by project,
etc.? Please elaborate.
Automated
Does your system have the ability to
reminders
for issue automated reminders for timesheet
timesheet entry
entry?
Out-of-thebox
Yes, it does have the ability to easily
enter time sheets. Clock in and
clock out can also be done with an
automated time clock
Yes, time sheet entries or auto
clock-in clock-out can be loaded to
a third-party payroll provider
Out-of-thebox
Add-on
Yes, as far as labour reporting in
production. Employees will clock
on and off jobs. Sage ERP X3 will
account for all their time during the
day including indirect labour, i.e.
meetings, breaks etc.
Sage Timesheets
*OT: overtime
**YTD: year to date
Source: Company Records
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Page 21
9B13E026
EXHIBIT 4: BARNES AND CO. FINANCIAL INFORMATION
SAGE ERP X3 LICENSES
Unit Price
1
5
0
0
0
0
$3,380.00
$3,450.00
$540.00
$1,620.00
$1,620.00
$4,360.00
Extended
Price
$3,380.00
$17,250.00
$0.00
$0.00
$0.00
$0.00
$20,630.00
MAINTENANCE AND SUPPORT
Annual Maintenance
Maintenance Fee
Qty
Unit Price
Extended
Price
1
$4,126.00
$4,126.00
Qty
Unit Price
Extended Price
10
$60.00
$600.00
Hosting Services
Named Users
Payment Options:
1. 50% upon contract, 50% upon delivery
2. EasyPay: $185 / month / users for 36 months after which customer owns software outright and
pays annual maintenance only 20%. Easy Pay price includes annual maintenance. 5 user
minimum.
IMPLEMENTATION ESTIMATE
Milestone / Activity
Project Management
Implementation Preparation
Installation, training and design,
migration
Conference Room Pilot
Go-Live Preparation
Go-Live / Production Support
Total: Implementation (Estimate)
Days
Rate
Amount
5
10
$1,200.00
$1,200.00
$6,000.00
$12,000.00
10
8
5
5
43
$1,200.00
$1,200.00
$1,200.00
$1,200.00
$12,000.00
$9,600.00
$6,000.00
$6,000.00
$51,600.00
data
Page 225 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Sage Standard ERP X3 License
Full Concurrent User
Inquiry / Business Intelligence Users
Operational Users
Warehouse / Handheld Device Users
Sage HRMS Payroll and HR
Total: Software
Qty
Page 22
9B13E026
EXHIBIT 5: UNDERWOOD GROUP FINANCIAL INFORMATION
IMPLEMENTATION SERVICES
Stage
Days
Price
12
25
21
17
8
83
$16,800.00
$35,000.00
$29,400.00
$23,800.00
$11,200.00
$116,200.00
Qty
Unit Price
Extended
Price
1
6
1
1
1
1
1
$5,350.00
$5,350.00
$1,080.00
$1,500.00
$1,080.00
$1,080.00
$675.00
$5,350.00
$32,100.00
$1,080.00
$1,500.00
$1,080.00
$1,080.00
$675.00
Analysis
Design
Construction
Deployment
Support & Sustainment
Total: Implementation
SOFTWARE AND HOSTING
Estimated Microsoft Licensing Costs by User
Software License
Advanced Mgmt. Foundation Pack (+1 user)
Advanced Mgmt. Users (+6 users)
Electronic Payments
AP Advanced Check
Bin Setup
BOM Version Management
Machine Centres
Total Microsoft Dynamics NAV User & Modules
$42,865
Promotional Discount*
Microsoft Support Plan Annual Maintenance (16%)
($23,750)
$6,858.40
Total: Software License
$25,973.40
Note: Provides First five users for Cdn$3,000.
Source: Company Records
Page 226 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Estimated Breakdown by Project Stage
Page 23
9B13E026
EXHIBIT 5 (CONTINUED)
The Underwood Group has collaborated with a local partner to provide a SaaS solution for LSFF:
 Hosted Solution
− 7 named users, 7 concurrent users @ 211.76 per user/month
− Monthly Total = $1,482.32 CAD
 NAV Licenses under Service Provider Licensing Agreement (SPLA)
− 7 Dynamics NAV BRL Advanced Management License @ $111.94 per license/month
− Monthly Total = $783.58 CAD
 Complete SaaS Solution Cost for 7 users per month
− Total cost = $2,265.90 CAD
Hosted Solution Option
Our hosting partner will provide the hosting only.
 Hosted Solution
− 7 named users, 7 concurrent users @ 211.76 per user/month
− Monthly Total = $1,482.32 CAD
 Microsoft License Cost
− Total Cost = $25,973.00
− Annual Maintenance = $6,858.00
On-Premise Solution Option
The Underwood Group will install the software on site.
 Installation Time
− 2 days @ $1400 day
− Total Cost = $2,800.00
 Microsoft License Cost
− Total Cost = $25,973.00
− Annual Maintenance = $6,858.00
Source: Company Records
Page 227 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Software as a Service (SaaS) Option
TRANSFORMING THE BUSINESS SERVICE PORTFOLIO AT GLOBAL
CONSULTANCY
Naqaash Pirani wrote this case under the supervision of Professor Ning Su solely to provide material for class discussion. The
authors do not intend to illustrate either effective or ineffective handling of a managerial situation. The authors may have disguised
certain names and other identifying information to protect confidentiality.
This publication may not be transmitted, photocopied, digitized or otherwise reproduced in any form or by any means without the
permission of the copyright holder. Reproduction of this material is not covered under authorization by any reproduction rights
organization. To order copies or request permission to reproduce materials, contact Ivey Publishing, Ivey Business School, Western
University, London, Ontario, Canada, N6G 0N1; (t) 519.661.3208; (e) cases@ivey.ca; www.iveycases.com.
Copyright © 2014, Richard Ivey School of Business Foundation
Version: 2014-01-17
The message was clear: find ways to improve efficiency across Global Consultancy Canada through
strategic sourcing. The chief executive officer (CEO) had left no room for ambiguity when he tasked John
Wong with this assignment. As the lead partner of the sourcing practice at the firm, Wong was more than
qualified to deliver on this. He had done it countless times for his clients and had always achieved
impressive results. But this time felt different. The more he analyzed the data, the less certain he felt. For
every opportunity to realize cost savings for the firm, there was an almost equally significant risk. The
CEO was expecting his report and recommendations at the next leadership update meeting, and Wong
was no closer to making a decision than he had been at the last. Coming up with a viable strategy for the
firm was going to require all of his 30 years of industry experience.
COMPANY BACKGROUND
Global Consultancy (GC) was a leader in the professional services industry, with over 160,000 employees
operating in 150 countries worldwide. The firm provided an array of services to clients that spanned the
domains of consulting, tax and accounting. GC generated revenues by charging clients an hourly rate for
the resources assigned to various projects and engagements. Its business model required the maximum
utilization of the company’s human resources. For this reason, GC’s client-facing staff were often
allocated to multiple projects with different clients. Additional incentives such as bonuses, awards and
variable pay structures were provided to compensate for the highly demanding nature of the job.
The client-facing lines of service at GC were supported by a department of internal employees that were
responsible for activities including marketing and communications, human resources (HR), finance and
information technology (IT) (see Exhibit 1). While external GC consultants were revenue generating,
these internal support employees were mostly financed by the activities of the client-facing lines of
business. To create value, they contributed to the success of the firm in other intangible ways.
Specifically, the marketing and communications department was responsible for positioning and brand
management, executive communications, design and research services. HR administered the firm’s
employee benefits, processed payrolls and pensions, organized recruitment and managed staffing based
Page 228 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
9B14E001
Page 2
9B14E001
on project needs. Finance handled payment for employees and contractors, reviewed and processed
expense reports from GC employees and managed billing and client payment. IT offered a service desk to
support software, network and Internet related issues and desktop support to help resolve computer and
hardware related issues. A description of specific activities of each of the four areas can be found in
Exhibit 2.
“Strategic sourcing” of business services can be broadly categorized into three types of arrangements:
business process outsourcing (BPO), knowledge process outsourcing (KPO) and IT outsourcing (ITO).
BPO involves turning over the management of a particular business process, such as accounting or
payroll, to a third party that specializes in that process. KPO is a subset of BPO that focuses on the
outsourcing of high-end knowledge-based services, such as market research. Lastly, ITO involves
transferring computer or Internet related work, such as programming, to other organizations that
specialize in IT services. The sourcing industry can be traced as far back as 1949, when Automatic Data
Processing, Inc. (ADP) began to offer clients payroll services. In 1989, sourcing made headlines and
gained international attention when Kodak Eastman Company outsourced its data centre operations to
IBM in a landmark $250 million, 10-year deal. 1 Over the next 15 years, many Fortune 500 companies
would follow suit, bringing the total value of outsourced IT services to an estimated $288 billion in
2013. 2
Strategic sourcing requires making informed decisions about the delivery of products and services within
an organization. Based on who delivers the service (i.e., in-house units or third-party vendors) and from
where the service is sourced (i.e., home country or offshore locations), strategic sourcing can be broadly
categorized into four major models: onshore outsourcing, offshore outsourcing, insourcing and captive
centre. Each sourcing model has its benefits and risks. For example, outsourcing may reduce cost,
improve efficiency and allow an organization to focus on its core competencies. On the other hand,
contracting and coordinating with a third-party vendor can lead to a number of risks. Similarly, offshoring
may also bring cost savings and access to a global talent pool, but the offshore location’s political,
economic, legal and technological environment, as well as privacy and security related issues, need to be
taken into account. For these reasons, strategic sourcing decisions require careful analysis.
In combination with the four main models, organizations may consider several staffing arrangements.
One option is staff augmentation, where resources are drawn from another company office or a third-party
vendor to assist in the delivery of a specific project. Staff augmentation provides the benefit of added
human resources without the costs of full-time employment, but this also means that when hired resources
return to their own organizations, they take their knowledge and experience with them. Another option is
a shared services model, where services that were previously delivered locally are centralized and
delivered by an internal or third-party provider. A shared services model provides the benefit of increased
standardization and scale through centralization but may reduce the responsiveness and adaptability of
services to local needs and concerns.
1
John K. Halvey, “The Top 24 IT Outsourcing Transactions,” HRO Today, 2003, www.hrotoday.com/content/532/top-24-itoutsourcing-transactions, accessed December 7, 2013.
2
Gartner, “Gartner Says Worldwide IT Outsourcing Market to Reach $288 Billion in 2013,” 2013, www.gartner.com/news
room/id/2550615, accessed December 7, 2013.
Page 229 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
STRATEGIC SOURCING
Page 3
9B14E001
Given the complexity of these decisions, GC provided consulting services to companies interested in
pursuing strategic sourcing. The practice had experienced high growth over the past decade, now boasting
a number of large clients in both the public and private sectors, including government agencies, utilities
companies, financial institutions and telecommunications companies. The scope of work encompassed all
aspects of the sourcing life cycle, from the readiness assessment that determined if certain functions were
appropriate for transformation, to request for proposal (RFP) services that helped in the identification and
selection of third-party vendors, and finally to transition support that helped clients redesign their
organization and business processes to ensure a smooth transition to a new model.
In a recent engagement with Canada’s National Post Office (NPO), for example, GC’s sourcing team
provided advisory services to facilitate the development and implementation of an RFP for call-centre
services. NPO was faced with an increasing volume of support and technical calls (over three million
each year) and was seeking to transform a large portion of its transactional call centre activities. GC
consultants first conducted a needs assessment to understand the organization’s current state. They then
used this information to draft an RFP and evaluation framework, as well as a service level agreement
(SLA) to be used when a vendor had been selected. Once the RFP had been administered and a contract
signed, the GC team stayed on to help with ongoing vendor management and stakeholder
communications. Further details about this engagement can be found in Exhibit 3.
Through such engagements, GC was able to develop a systematic sourcing methodology and gain
invaluable experience that helped develop a strong reputation in the marketplace and fuel further growth.
The Canadian Sourcing practice, whose lead partner was John Wong, now had over 40 employees
dedicated to providing consulting services. Wong had joined GC 15 years ago after a successful tenure as
a senior executive at a large utilities company. During that time, he oversaw sourcing transactions in the
millions of dollars and built a reputation as a hard-working and highly intelligent individual. When GC
decided to expand its service offering from accounting and tax to consulting services as well, he seemed
like a natural fit to lead the Sourcing practice.
GC’S OWN STRATEGIC SOURCING ENGAGEMENT
As a multinational corporation, GC had also adopted strategic sourcing in its own organization, but its
sourcing strategy differed across its global network of offices. The U.S. and Australian offices, for
example, had made strategic sourcing a high priority and thus had deep experience in the area as well as
future plans to expand their efforts. The Canadian office, on the other hand, was more cautious toward
outsourcing and offshoring. Compared to the U.S. and Australian offices, GC Canada generally
outsourced fewer of its own internal services. In fact, all of the work performed by its national Marketing
and Communications, HR, Finance and IT functions was done at regional offices by full-time GC
employees. Under the firm’s previous leadership, internal services like these were seen as critical
capabilities since they supported the client-facing employees in their day-to-day work.
However, when leadership changed and the firm came under the direction of a new CEO, this sentiment
shifted and efficiency became a guiding principle. The strategic sourcing project assigned to Wong would
be the first major initiative under this new leadership and direction. The CEO wanted Wong to include in
his scope each of the support functions currently being delivered within the firm. While the CEO
respected Wong’s experience and expertise, there was clearly pressure on him to reduce cost as much as
possible through strategic sourcing. This presented a significant challenge — Wong knew that
transforming an organization internally could be far more difficult than doing so for an external client.
Page 230 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
GC’S STRATEGIC SOURCING CONSULTING PRACTICE
Page 4
9B14E001
When he was tasked with identifying strategic sourcing opportunities for transformation, Wong
approached the project as he would any other client engagement. The first step was to collect as much
data as possible to understand the current state of the firm. He conducted an analysis of each of the four
service areas identified as being in scope by the CEO and compared them to industry standards in order to
benchmark the firm against its competitors. Where appropriate, he also sought to obtain information on
third-party vendors, both onshore and offshore, to better understand the marketplace.
Marketing and Communications
Wong began by interviewing front-line marketing and communications staff. Many interviewees
commented that their department faced long hours and unpredictable demand because they were often
seen as “order takers” and thus received a high proportion of ad hoc requests from client-facing lines of
business. Furthermore, the tasks and deliverables they had to complete often required multiple iterations
of review and revision, and bottlenecks were created while they waited for their clients to get back to
them. Wong also noted that one business unit in particular, Executive Communications, was required to
provide “high-touch support” for GC senior leadership and thus had to go “above and beyond” to ensure
that their clients were satisfied.
Wong conducted an external market scan to understand how other organizations delivered their marketing
and communication services. He found that many of them, especially large global organizations like GC,
outsourced their marketing functions to third-party agencies and public relations firms. These firms were
able to provide a vast portfolio of services that ranged from campaign management to content creation
and had the ability to target both online and offline audiences. Outsourcing to a third party could provide
GC opportunities to reap the benefits of many emerging areas in the marketing space such as social media
and analytics. Where GC internal employees had only begun to experiment with such tools, external
vendors had successfully implemented them and delivered impressive results for their clients.
Human Resources
Next, Wong sought to determine the potential for HR functions to be delivered using a better sourcing
model. In his analysis of HR services, he found that the level of standardization varied significantly across
the portfolio. For example, services such as payroll, benefits administration and pension administration
were fairly straightforward services that were not unique to GC. For this reason, GC had decided to use
third-party vendors to deliver the majority of these services. Benefits were delivered by Aon Hewitt and
payroll by Ceridian; pensions were managed by Sun Life. However, there were still 10 FTEs at GC who
were responsible for the fractional services associated with these portfolios, as well as ongoing vendor
management, with an estimated annual cost to the firm of between $100,000 and $150,000 each. For
example, Aon Hewitt managed the firm’s flexible benefits and health and safety programs, but disability
management and partner benefits were still delivered in-house. Similarly, the firm relied on Ceridian to
administer regular pay processes and annual events such as T4 forms, pay increases and bonuses, but GC
Page 231 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
For example, while client engagements often ended shortly after a contract had been signed and services
transitioned to a vendor, internal transformation would require ongoing monitoring and support. Many
problems could present themselves during the lifetime of a sourcing contract (e.g., fluctuating service
levels, escalating costs), and these would fall on Wong and his team for resolution. Furthermore, such
transformation often required the client to let go of full-time staff (FTEs) in order to realize the cost
savings identified by GC. In this case, those FTEs would not just be numbers on a page; they would be
people Wong knew very well, some of whom he had worked with since joining GC 15 years ago.
Page 5
9B14E001
Recruitment and Resource Management, on the other hand, relied on tools and processes that were either
developed in-house or were tailored to the specific objectives and strategies of GC. The Campus
Recruiting program, for example, was a key priority for the firm as it served as an entry point for almost
all of the new hires in the Accounting practice. Staff in this area group were needed to host information
sessions at universities across the country, schedule and facilitate interviews with candidates and issue
offer letters once a candidate had been selected. They were often the first point of contact for many
students and remained engaged with them throughout the application process. Resource Management was
also a highly critical area group as they managed the deployment of GC consultants to client
engagements. This required careful decision making that took into account client needs as well as
employee training and scheduled vacations. It was also a highly sensitive task as the allocation of GC
consultants was directly tied to their annual compensation. These services were currently being delivered
regionally by local GC offices rather than collectively at the national firm level.
Finance
GC’s accounting processes were very well-documented, likely due to the fact that they had been
aggregated under a shared services model two years earlier. In a shared services model, a firm merges
common services performed by multiple business units into a single service delivery organization that is
either located in-house or outsourced. In the case of GC, accounts payable functions — including
invoicing clients for work completed; reimbursing staff expenses such as parking, travel and
accommodations; generating monthly statements for the firm; and issuing corporate credit cards to
employees — were being delivered nationally by a core team in the GC Toronto office. Currently, the 25
full-time GC finance employees cost the firm roughly $37 an hour.
In evaluating the external finance market for sourcing opportunities, Wong identified offshore
outsourcing as a potential solution. Offshore resources were much less expensive than internal GC
finance employees, costing only $13 an hour. However, the costs of employee severance and transitioning
to a new vendor could run as high as $500,000. One vendor, Offshore Resource Solutions (ORS), was
currently being used by the GC U.S. and Australian offices for their finance functions. ORS had offices in
India and Uruguay; ORS India resources were roughly 15 per cent lower than those at ORS Uruguay.
ORS India also offered a “follow the sun” model where employees would work a night shift to ensure
adequate coverage.
Information Technology
Finally, Wong conducted an analysis of the services currently delivered by the GC IT Department. The
two main areas of interest to Wong were the IT service desk and desktop support teams. These functions
were the first point of contact for GC employees whenever they had an issue with their personal
computer, smartphone or GC-provided software. Wong captured current service levels being met by the
IT service desk and desktop support teams and compared them to standard packages offered by thirdparty vendors (see Exhibit 4). He also captured the cost to the firm for the resources currently staffed in
each function. The IT service desk currently employed 12 FTEs at an annual cost of $1.1 million; the
desktop support function was staffed by 54 FTEs at a cost of $5.6 million.
Page 232 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
employees were still being used to perform some manual cheque processing and administering records of
employment.
9B14E001
In speaking with GC client-facing employees, Wong found that they were very satisfied overall with the
level of service being provided by the IT Department. In fact, many of them commented that this was one
of the most responsive IT departments they had been supported by in their career to date. They
appreciated the willingness of IT resources to try and troubleshoot unique and complex problems that
sometimes were not caused by GC provided software or hardware solutions. They were also impressed
with the level of knowledge of IT personnel — the department had just recently obtained the IT
Infrastructure Library (ITIL) compliance certificate, confirming that the processes in place were based on
industry leading practices.
THE DECISION
Now that he had all of this information, Wong could see some potential for GC Canada to pursue strategic
sourcing. However, a more systematic analysis needed to be conducted in order to create a blueprint for
the company’s transformation. What were the key criteria for making sourcing decisions? What were the
specific benefits and risks of each sourcing model? What was the most suitable model for each of the four
business areas? To make a recommendation to the CEO at the next leadership team meeting, Wong would
need to answer these questions and articulate the rationale for his decisions in a highly compelling way.
Page 233 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Page 6
Page 7
9B14E001
EXHIBIT 1: GC ORGANIZATIONAL CHART
Tax
Accounting
Consulting
Internal Services
Business Process
Marketing and
Communications
Operations
Human
Resources
Risk Management
Information
Technology
Technology
Finance
Sourcing
Legend
= Client-facing line of business
= Internal GC department
Source: Company files.
Page 234 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
CEO
Page 8
9B14E001
EXHIBIT 2: KEY ACTIVITIES OF THE FOUR MAJOR INTERNAL BUSINESS SERVICES
Executive Communications: This area develops and has oversight for the execution of firm-wide
communication strategies for major internal and market-facing initiatives on behalf of the GC executive. It
also develops an annual firm communications strategy and plan and counsels the executive on protecting
and enhancing the firm’s brand and reputation.
Positioning and Brand Management: This group’s mandate is to maximize the impact of GC’s brand via
proactive consultation on internal and external marketing. It also monitors compliance with brand
guidelines and develops templates and programs to support the proper application of visual guidelines
and best practice in brand expression in internal and external communications.
Design Services: This team creates objective-focused materials, including newsletters, brochures,
proposals, invitations, conference materials and other print and presentation projects as required. It also
supports the development of proposal documents and sales support materials, develops interactive
multimedia marketing materials and consults on best practice integration with print materials.
Research Services: The Canadian research centres support the firm by conducting competitive
monitoring, reporting on relevant current events and supporting major projects for the leadership. The
group also supports the practice by producing a number of industry news briefs and supporting business
development.
Human Resources
Benefits: The benefits group administers the flexible benefits, disability management, partner benefits,
and health and safety programs for the national firm. It also administers a number of projects in addition
to managing the relationship with the benefits vendor (Aon Hewitt).
Payroll: The payroll group administers regular pay processes as well as special events — i.e., leaves,
hires and terminations — and regular annual events, such as T4 forms, pay increases and bonuses. The
payroll group also manages the payroll vendor (Ceridian).
Pensions: The pension group is responsible for administering member pension requests and processing
any required changes, responding to queries from employees/HR/vendors, producing annual statements
and filing information returns. It also has responsibility for ongoing pension governance and investment
manager searches and managing the vendor (Sun Life).
Recruitment: The recruitment work stream consists of entering job requisitions into a national database,
posting jobs and supporting the process of screening, interviewing and extending offers to potential
recruits. The recruiter is the main contact person for potential hires until the offer letter is signed.
Page 235 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Marketing and Communications
Page 9
9B14E001
EXHIBIT 2 (CONTINUED)
Finance
Accounts Payable: The accounts payable work stream consists of processing payments for internal
employees, rent, parking and independent contractors and running weekly, monthly and annual reports
for the firm.
Accounts Receivable: The accounts receivable work stream requires setting up new client engagements
in the online resource management system to support billing; processing payments, bank wires and cash
transfers from clients; maintaining the general ledger; and creating monthly and annual reports for the
firm.
Time and Expenses: The time and expenses work stream includes reviewing expense reports from GC
employees, approving and processing them and issuing corporate cards to employees.
Information Technology
Service Desk: The IT service desk provides phone support to GC employees and contractors who are
experiencing technical issues. Call topics range from Internet connectivity to questions about GC and
third-party software installation, connecting to printers and other network resources. There is also a Level
2 support team that deals with large-scale IT issues such as GC network downtimes or system failures.
Desktop Support: The desktop support team supports the personal computer and hardware resources
that are deployed across the firm. This includes the installation of new workstations and printers, setting
up personal computers for new hires at the firm, setting up and deploying smartphones to GC employees
and providing phone and in-person support for technical issues that can arise during these processes.
Source: Company files.
Page 236 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Resource Management: Resource management is responsible for developing schedules for staff based
on client needs, taking into consideration training and vacations. The process also involves staff coaching
and performance management. The team assigns managers to assignments based on calls from industry
partners.
Page 10
9B14E001
EXHIBIT 3: SAMPLE GC SOURCING CONSULTING ENGAGEMENT
Client Issue:
NPO sought to transform a major portion of the transactional contact centre activities
within the organization. Its customer relationship network had four call centres in Canada
and received more than three million support and technical calls each year. The large
majority of these calls were from the general public (consumers) who had shipping
inquiries or were looking for delivery status updates. The remaining calls were from
commercial or business customers requesting pick-ups, placing orders or requesting
delivery status updates.
Approach:
Working closely with NPO to understand its business needs, the GC team was able to
develop an RFP, draft service level agreements (SLAs), master services agreement
(MSA) and an evaluation model. Collectively, its work positioned NPO well to procure the
desired services in a fair, competitive and comprehensive manner. Based on GC’s
experience, the team was aware of key objectives that would be important to NPO during
this RFP including:
x Improved service delivery in terms of quality, efficiency and productivity.
x Building customer loyalty by reducing customer effort through first call resolution,
quality improvements and other operational improvements.
x Maintaining a high standard of service levels.
x Flexibility and scalability to accommodate NPO’s spikes in volumes.
x Contact centre, telephony and IT capabilities that supported industry standard
practice.
x Seamless and effective transition to the selected solution.
x Reducing total operating costs for the life of the agreement.
Result:
The RFP was successfully administered and a vendor was selected for the delivery of
contact centre services. GC stayed on after the engagement to ensure a smooth
transition to the new model by providing leading practices that NPO could use to develop
an internal vendor governance model and allocating a “stay-back team” to manage the
multiple vendor interface points. The GC team also provided guidance on stakeholder
and communicaiton management as well as on how to operationalize the vendor
management framework and manage the relationship. In the end, GC helped create an
internal centre of excellence for vendor management at NPO.
Source: Company files.
Page 237 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Engagement: National Post Office (NPO) — Call Centre Request for Proposal (RFP)
Page 11
9B14E001
EXHIBIT 4: IT SERVICE LEVELS — GC VS. INDUSTRY STANDARDS
Metric
Hours of operation
Average speed to answer
Abandonment rate
Cost ($/call or email)
GC Current State
12 hours/day, 5 days/week (24x7
via pager)
20 seconds
4%
$17
Industry Basic Support
9 hours/day, 5 days/week
GC Current State
30 minutes
2 hours
½ day
Best effort
Industry Basic Support
4 hours
1 business day
2 weeks
Not supported
$68
Onshore: $39 to $72*
Offshore: $30
90 seconds
10%
Onshore: $19 to $33*
Offshore: $10
Desktop Support
Metric
Time to respond
Time to repair
Install time (new PC)
Repair non-standard business
software
Cost ($/PC per month)
* A range of costs is provided from low to high, assuming similar services, volumes and levels provided by
third-party Canadian vendors.
Source: Company files.
Page 238 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
IT Service Desk
IN1896
Challenges in Commercial Deployment of AI:
Insights from the Rise and Fall of IBM Watson’s AI Medical
System
Source: tungnguyen0905, Pixabay.com
02/2023­6753
This case study was written by Dr Lisa Simone Duke, case writer, under the direction of Quy Huy, The
Solvay Chaired Professor of Technological Innovation and Professor of Strategy at INSEAD, Timo
Vuori, Associate Professor at Aalto University, and Tero Ojanpera, Professor of Practice of Aalto
University and CEO/ co­founder of Silo.AI. The authors thank Riikka Markkula at Aalto University for
research assistance. The case is intended to be used as a basis for class discussion rather than to
illustrate either effective or ineffective handling of an administrative situation.
To access INSEAD teaching materials, go to https://publishing.insead.edu/
Copyright © 2023 INSEAD
COPIES MAY NOT BE MADE WITHOUT PERMISSION. NO PART OF THIS PUBLICATION MAY BE COPIED, STORED, TRANSMITTED, TRANSLATED,
REPRODUCED OR DISTRIBUTED IN ANY FORM OR MEDIUM WHATSOEVER WITHOUT THE PERMISSION OF THE COPYRIGHT OWNER.
Page 239 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Case Study
Yet success ultimately proved elusive. Watson was helpful in combatting a number of chronic
diseases, but in the fight against cancer – where thousands of variations and genetic mutations
resulted in different outcomes – its achievements were less than optimal. After years of
development and heavy investment, rumours surfaced in 2021 that IBM was looking to sell Watson
Health.
IBM in Brief
Legally incorporated as the Computing­Tabulating­Recording Company in New York in 1911, IBM
initially made measuring equipment such as commercial scales, industrial time recorders and
tabulators. Changing its name to International Business Machines in 1924, it expanded into
electric typewriters and office machines. After launching its first computer in 1951, within five years
IBM held 85% of the computer market. The IBM Personal Computer was invented in 1981. From
the 1960s to the 1980s its offering developed to include hardware, software and service
agreements. It became America’s most admired corporation.
But as competition intensified in the 90s, the company increasingly looked like a dinosaur. Losses
amounted to $16 billion between 1991 and 1993. Thereafter, under Lou Gerstner Jr, it became
leaner and focused on high­margin opportunities and a global service model that integrated
technologies beyond IBM’s own for its clients.
In 1997, IBM’s Deep Blue AI beat world chess champion Garry Kasparov. With its strong
commitment to R&D, IBM employees would earn five Nobel prizes, four Turing Awards, five
National Medals of Technology, and five National Medals of Science. Henceforth it continued to
develop its AI capabilities. 2
IBM Watson
In 2006, an IBM lab scientist, David Ferrucci, suggested to that to advance research on natural
language processing and automated question answering IBM should develop a computer to play
the Q&A game show Jeopardy!. Despite initial skepticism, he was given the go­ahead to build a
room­sized computer that stored thousands of Wikipedia entries, electronic books and digitized
reference works. 3 The computer was named after IBM’s founder, Thomas J. Watson.
1
Hsu, Jeremy. “IBM Invests $1 Billion to Grow Watson Supercomputer’s Struggling Business.” Spectrum.ieee.org,
January 9, 2014, https://spectrum.ieee.org/ibm­invests­1­billion­to­grow­watson­supercomputer­struggling­business,
accessed November 2021.
2
See https://www.ibm.com/cloud/learn/what­is­artificial­intelligence.
Lohr, Steve. “What Ever Happened to IBM’s Watson?” The New York Times, July 17, 2021,
https://www.nytimes.com/2021/07/16/technology/what­happened­ibm­watson.html, accessed November 2021.
3
Copyright © INSEAD
1
Page 240 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
IBM Watson’s artificial intelligence (AI) was more than just a gimmick. Winning the US gameshow
Jeopardy! against human champions in 2011 was proof that the technology was not simply an
experiment for the future. Watson AI would soon become a central pillar of IBM’s strategy, with
healthcare the most promising area of focus. In 2014, IBM announced it was investing more than
$1 billion and launching a $100 million venture capital fund to encourage apps built on the
technology. 1
After a few years of training, Watson was ready to play. In February 2011, it won against two of
the most successful quiz champions, working through 200 million pages of structured and
unstructured content. 5 Thereafter, IBM executives were approached by customers about
commercial applications for Watson but Ferrucci insisted it was designed to play Jeopardy! and
not for commercial purposes. That was not what his colleagues wanted to hear – “It wasn’t the
marketing message,” said Ferrucci, 6 who left IBM in 2012.
Healthcare – AI to Conquer Disease
Healthcare was a promising market for Watson, worth an estimated 10% of global GDP. 7 Given
the vast amount of research data on genetics, treatments and procedures published on a daily
basis, medical staff were overloaded with information and hard pressed to keep up to date. It was
estimated that healthcare professionals made accurate treatment decisions only 50% of the time. 8
Six elements made up the diagnostic process: the patient’s story, data acquisition, accurate
problem representation, hypothesis generation, selection of the illness script, and the diagnosis.9
Hence a potential application of AI was to process research data to support medical staff as they
formed their diagnoses and formulated treatment plans by keeping them abreast of the latest
findings.
On February 17, 2011, IBM announced that Watson would focus on healthcare, the cost of which
was rising in many countries. A company press release quoted Dr John E. Kelly III, SVP and
Director of IBM Research, saying: 10
4
5
6
7
8
9
10
Strickland, Eliza. “How IBM Watson Overpromised and Underdelivered on AI Health Care.”, Spectrum.ieee.org, April 2,
2019, https://spectrum.ieee.org/how­ibm­watson­overpromised­and­underdelivered­on­ai­health­care, accessed
November 2021.
Jackson, Joab. “IBM Watson Vanquishes Human Jeopardy Foes.” PCWorld.com, Feburary 16, 2011,
https://www.pcworld.com/article/494961/ibm_watson_vanquishes_human_jeopardy_foes.html, accessed November
2021.
Lohr, Steve. “What Ever Happened to IBM’s Watson?” The New York Times, July 17, 2021,
https://www.nytimes.com/2021/07/16/technology/what­happened­ibm­watson.html, accessed November 2021.
Chesbrough, Henry. “IBM Watson and the Value of Open.” Forbes, February 12, 2020,
https://www.forbes.com/sites/henrychesbrough/2020/02/12/ibm­watson­and­the­value­of­open/?sh=11bedf4a2cfb,
accessed November 2021.
Upbin, Bruce. “IBM’s Watson Gets Its First Piece of Business in Healthcare.” Forbes, February 8, 2013,
https://www.forbes.com/sites/bruceupbin/2013/02/08/ibms­watson­gets­its­first­piece­of­business­in­
healthcare/?sh=747c8a805402, accessed November 2021.
Bowen, Judith. “Educational Strategies to Promote Clinical Diagnostic Reasoning.” The New England Journal of
Medicine, November 23, 2006, https://www.nejm.org/doi/full/10.1056/nejmra054782, accessed November 2021.
“IBM to Collaborate with Nuance to Apply IBM’s ‘Watson’ Analytics Technology to Healthcare.” PR Newswire, February
17, 2011, https://www.prnewswire.com/news­releases/ibm­to­collaborate­with­nuance­to­apply­ibms­watson­analytics­
technology­to­healthcare­116395589.html , accessed November 2021.
Copyright © INSEAD
2
Page 241 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
To play Jeopardy! Watson examined phrase structure and grammar to work out what the quiz
question was asking and to understand jokes, slang, metaphors and other sophisticated elements
of natural language to arrive at the “most appropriate answer”. The team experimented with
several machine learning algorithms and trained it using thousands of Jeopardy! clues and
answers labelled correct/incorrect. 4 The AI looked for patterns to work out how to get from a clue
to a correct answer.
Unlike a search engine, Watson would take an input query from a healthcare professional in
natural language, understand the question, produce possible answers and evidence, analyze
them, compute the level of confidence, then deliver the response, evidence and confidence to the
professional to support their decision making. 11
To frame the business case for medical AI, IBM started several projects targeted at different
players in the healthcare system – physicians, admin staff, insurers, and patients 12 – all linked by
a common goal: providing decision support using AI to analyze big data sets. The company
anticipated launching its first commercial offering in 18 to 24 months.
Partnering With Experts
IBM joined forces with Columbia University Medical School to identify critical issues in medical
practice that Watson could help with, and with the University of Maryland School of Medicine to
identify the best way to interact with practitioners. 13 Watson had already been fed medical journals
and textbooks as part of its training for Jeopardy! and by May 2011 was “as good as the smartest
second­year med student,” according to Dr Eliot Siegel, a senior radiologist at the University of
Maryland. The next step was to feed Watson anonymized patient records to match diagnostics
with procedures, treatments and outcomes. Siegel believed that pilot testing with doctors was
three to five years away and that a diagnostic tool could be ready in eight to ten years. 14
IBM also formed alliances with the Memorial Sloan Kettering Cancer Center, the Mayo Clinic,
Cleveland Clinic, CVS Heath, and Johnson & Johnson. In March 2012, working with Memorial
Sloan Kettering, Watson was fed tens of thousands of patient records and histories. By 2013 it
had analyzed 605,000 pieces of medical evidence, 2 million pages of text, 25,000 training cases,
and had benefitted from 14,700 clinician hours to adjust its decision accuracy.
In October 2013, IBM CEO Virginia Rometty set a target for Watson to generate $10 billion in
revenues within the decade and projected annual revenues of $1 billion by 2018. 15 That same
11
12
13
14
15
Kohn, Martin. “IBM Watson in Healthcare.» Slideshare, May 6, 2013,
https://www.slideshare.net/AndersQuitzauIbm/ibm­watson­in­healthcare, accessed November 2021.
Strickland, Eliza. “How IBM Watson Overpromised and Underdelivered on AI Health Care.”, Spectrum.ieee.org, April 2,
2019, https://spectrum.ieee.org/how­ibm­watson­overpromised­and­underdelivered­on­ai­health­care, accessed
November 2021.
“IBM to Collaborate with Nuance to Apply IBM’s ‘Watson’ Analytics Technology to Healthcare.” PR Newswire, February
17, 2011, https://www.prnewswire.com/news­releases/ibm­to­collaborate­with­nuance­to­apply­ibms­watson­analytics­
technology­to­healthcare­116395589.html, accessed November 2021.
Upbin, Bruce. “IBM’s Watson Now a Second­Year Med Student.” Forbes, May 25, 2011,
https://www.forbes.com/sites/bruceupbin/2011/05/25/ibms­watson­now­a­second­year­med­
student/?sh=57aa4056300a, accessed November 2021.
Hsu, Jeremy. “IBM Invests $1 Billion to Grow Watson Supercomputer’s Struggling Business.” Spectrum.ieee.org,
January 9, 2014, https://spectrum.ieee.org/ibm­invests­1­billion­to­grow­watson­supercomputer­struggling­business,
accessed November 2021.
Copyright © INSEAD
3
Page 242 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
We can transform the way that healthcare professionals accomplish everyday tasks
by enabling them to work smarter and more efficiently. This initiative demonstrates
how we plan to apply Watson’s capabilities to new areas such as healthcare.
In lab conditions Watson repeatedly impressed. It was able to make diagnoses of rare diseases
that most doctors would have difficulty recognizing. Its superior performance was based on a huge
database of every rare disease and an unbiased information­processing approach. Given a
description of symptoms, it could match them in seconds using the knowledge it had accumulated
to make a diagnosis and treatment recommendations. A Business Insider reporter wrote: 17
As IBM scientists continue to train Watson to apply its vast stores of knowledge to
actual medical decision-making, it’s likely just a matter of time before its diagnostic
performance surpasses that of even the sharpest doctors.
MIT’s Andrew McAfee concurred: “I’m convinced that if it’s not already the world’s best
diagnostician, it will be soon.” 18 At MD Anderson, a doctor­in­training told The Washington Post,
“Even if you work all night, it would be impossible to be able to put this much information together
like that.” 19 Wellpoint, the exclusive reseller of Watson, was reported as claiming the system was
“significantly better than human doctors at diagnosing lung cancer”. 20 Wellpoint sold Watson as a
tool to recommend the best cancer treatments to doctors around the world. It even recommended
new approaches to cancer care, according to one report, for a per­patient fee of between $200
and $1,000, depending on the number of products a hospital purchased. 21
One commentator wrote:
There are reasons for enthusiasm. Computers continue to get cheaper even as they
get more powerful, making it easier than ever to crunch vast amounts of data in an
instant. Also sensors, smartphones, and other tech devices are all over the place,
feeding more and more information into computers that are learning more and more
about us.
Investors began pouring money into AI start­ups. Apple, Facebook, Microsoft and Google began
acquiring start­ups and researching AI. An estimated $8.5 billion was invested in AI in 2015. 22
There was a bidding war for software engineers with AI skills, who were treated like “star athletes”.
16
17
18
19
20
21
22
Herper, Matthew. “MD Anderson Benches IBM Watson In Setback for Artificial Intelligence In Medicine.” Forbes,
February 19, 2017, https://www.forbes.com/sites/matthewherper/2017/02/19/md­anderson­benches­ibm­watson­in­
setback­for­artificial­intelligence­in­medicine/?sh=6603d1c33774, accessed November 2021.
Friedman, Lauren F. “IBM’s Watson Supercomputer May Soon Be The Best Doctor in the World.” Yahoo! Finance, April
22, 2014, https://finance.yahoo.com/news/ibms­watson­supercomputer­may­soon­141433414.html, accessed
November 2021.
Friedman, Lauren F. “IBM’s Watson Supercomputer May Soon Be The Best Doctor in the World.” Yahoo! Finance, April
22, 2014, https://finance.yahoo.com/news/ibms­watson­supercomputer­may­soon­141433414.html, accessed
November 2021.
Herper, Matthew. “MD Anderson Benches IBM Watson In Setback for Artificial Intelligence In Medicine.” Forbes,
February 19, 2017, https://www.forbes.com/sites/matthewherper/2017/02/19/md­anderson­benches­ibm­watson­in­
setback­for­artificial­intelligence­in­medicine/?sh=6603d1c33774, accessed November 2021.
Friedman, Lauren, F. “IBM’s Watson Supercomputer May Soon Be The Best Doctor in the World.” Yahoo! Finance,
April 22, 2014, https://finance.yahoo.com/news/ibms­watson­supercomputer­may­soon­141433414.html, accessed
November 2021.
Ross, Casey & Swetlitz, Ike. “Watson is smart, but cancer is still smarter.” The Boston Globe, September 10, 2017.
Lohr, Steve. “Fulfilling Watson’s Promise.” The New York Times, February 29, 2016.
Copyright © INSEAD
4
Page 243 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
month, IBM announced that the MD Anderson Cancer Center at the University of Texas was “using
the IBM Watson cognitive computing system for its mission to eradicate cancer”. 16
However, computer scientist, entrepreneur and author Jerry Kaplan observed: “Expectations are
way ahead of reality.” 23
Having created the IBM Watson unit dedicated to developing and commercializing cloud­based
cognitive computing technologies in 2014, Watson Health was formally launched in April 2015 –
the Watson Health cloud would drive timely, evidence­based decisions to “provide a secure and
open platform for physicians, researchers, insurers and companies focused on health and
wellness solutions.” 24 The main focus was oncology, where IBM hoped to deploy Watson’s
cognitive abilities to turn big data into personalized cancer treatments. Watson partnered with
more than a dozen cancer centers in 2015 to achieve this goal.
To add further muscle to the technology, IBM acquired two healthtech companies, Explorys (cloud­
based solutions that identified patterns in diseases, treatments and outcomes) and Phytel (patient
communications) for an estimated $560 million. 25 Clients of Explorys included 360 hospitals and
317,000 providers who oversaw $69 billion in care. 26 IBM paid $1 billion for Merge Healthcare
(specialized in analysis of complex healthcare data) in 2015, and $2.6 billion for Truven (analysis
of mammograms and MRI scans) in 2016. 27
In 2016, it launched IBM Watson Genomics with Quest Diagnostics, a company specialized in
genomic sequencing and oncology diagnostics. Watson Genomics, supported by Memorial Sloan
Kettering, would “give precision medicine – and oncology – a boost by combining cognitive
computing with genomic tumor sequencing”. Laboratories would sequence and analyze a tumor’s
genomic makeup and Watson would compare its mutations against relevant literature to identify
therapies. 28
IBM CEO Rometty told a 2017 health IT conference that AI could usher in a ‘golden age’ in
medicine, saying: “It is real, it’s mainstream, it’s here, and it can change almost everything about
healthcare.” 29 She described an inflection point where adding AI to businesses would produce “an
23
24
25
26
27
28
29
Lohr, Steve. “Fulfilling Watson’s Promise.” The New York Times, February 29, 2016.
“IBM and Partners to Transform Personal Health with Watson and Open Cloud.” PR Newswire, April13, 2015,
https://www.prnewswire.com/news­releases/ibm­and­partners­to­transform­personal­health­with­watson­and­open­
cloud­300065025.html accessed November 2021.
Global Research. “International Business Machines Corp. Potential Divestiture Could Help Further Streamline the
Business and Improve Focus.” UBS, February 19, 2021.
“IBM Acquires Explorys to Accelerate Cognitive Insights for Health and Wellness.” Cision PR Newswire, April 13, 2015,
https://www.prnewswire.com/news­releases/ibm­acquires­explorys­to­accelerate­cognitive­insights­for­health­and­
wellness­300065024.html, accessed November 2021.
Cooper, Laura & Lombardo, Cara. “IBM explores sale of Watson Health.” The Australian, February 20, 2021.
Monegain, Bernie. “IBM Watson, Quest Diagnostics, Memorial Sloan Kettering Cancer Center, MIT, Harvard combine
forces for massive oncology, precision medicine initiative.” Healthcare IT News, October 17, 2016,
https://www.healthcareitnews.com/news/ibm­watson­quest­diagnostics­memorial­sloan­kettering­cancer­center­mit­
harvard­combine­forces, accessed November 2021.
Strickland, Eliza. “How IBM Watson Overpromised and Underdelivered on AI Health Care.”, Spectrum.ieee.org, April 2,
2019, https://spectrum.ieee.org/how­ibm­watson­overpromised­and­underdelivered­on­ai­health­care, accessed
November 2021.
Copyright © INSEAD
5
Page 244 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Launch of Watson Health
exponential curve”– a phenomenon that might one day be called “Watson’s Law”. 30 She called on
health IT leaders to “play offence” on AI in five priority areas: 31
2) Transparency: Knowing who trained and what data is used is essential to have
confidence in insights. This is an industry full of professionals who want to understand
‘How did that answer come up?’
3) Domain-specific: For health, it must be trained by physicians.
4) Cloud-based: It will need to be cloud-based to be ubiquitous, built for scaling and
security.
5) Open: It has to be an open platform. This is an industry all about innovation.
From Promise to Reality
Yet despite the investment, partnerships and technological progress, it proved difficult to build
Watson in a way that added value to actual patient care for the community of practicing doctors
and medical centers. While Watson could learn how to scan clinical studies and determine basic
outcomes, it proved impossible to teach it how to read them in the way a doctor could. 32 One
doctor said Watson’s thinking was based on statistics about main outcomes – “but doctors don’t
work that way” – suggesting that the information doctors extracted from medical journals to change
their care might not be the “major point of the study”. 33 The Medical Futurist magazine would later
remark: “Setting up a diagnosis and treating a patient are not linear processes. It requires creativity
and problem-solving skills… Patients and their lifestyles vary to the degree that people differ.
Diseases have the same features.” 34
At MD Anderson, Dr Courtney DiNardo, Assistant Professor of Leukemia, defended the use of AI,
saying: “It’s not just a Google search or analysis of published research literature.” She predicted
that the Oncology Expert Advisor’s (OEA) capabilities in merging a patient’s ‘clinical story’ with the
research literature to produce individualized treatment options would create more humanizing
work opportunities for doctors. The doctors would still make the decision: 35
30
31
32
33
34
35
Strickland, Eliza. “Layoffs at Watson Health Reveal IBM’s Problem with AI.” IEEE Spectrum, June 25, 2018.
Hoeksma, Jon. “IBM’s CEO says AI will unlock a golden era for health.” Digitalhealth.net, February 20, 2017,
https://www.digitalhealth.net/2017/02/ibms­ceo­says­ai­will­unlock­golden­era­for­health/, accessed November 2021.
Strickland, Eliza. “How IBM Watson Overpromised and Underdelivered on AI Health Care.”, Spectrum.ieee.org, April 2,
2019, https://spectrum.ieee.org/how­ibm­watson­overpromised­and­underdelivered­on­ai­health­care, accessed
November 2021.
Strickland, Eliza. “How IBM Watson Overpromised and Underdelivered on AI Health Care.”, Spectrum.ieee.org, April 2,
2019, https://spectrum.ieee.org/how­ibm­watson­overpromised­and­underdelivered­on­ai­health­care, accessed
November 2021.
“5 Reasons Why Artificial Intelligence Won’t Replace Physicians.” The Medical Futurist, March 2, 2021,
https://medicalfuturist.com/5­reasons­artificial­intelligence­wont­replace­physicians/, accessed November 2021.
The Moon Shots Program.” The Future is Here Smithsonian, Smithsonian.com, May 2015,
https://issuu.com/toantritue/docs/toantritue_­_smithsonian_magazine_­, accessed November 2021.
Copyright © INSEAD
6
Page 245 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
1) Service range: Those who are successful must apply a range of cognitive services.
As early as 2014, delays had been reported in the project with MD Anderson to make a Watson
version capable of recommending leukemia treatments. 36 Interpreting doctor’s notes proved
challenging for the OEA “because documentation in medical records often contains private
acronyms, sentence fragments and grammatical errors with unintended or ambiguous meanings”
that could only be understood by the doctor. To overcome these issues, the OEA retained links to
the original content, but the information was still required as input for the Center’s electronic health
records system. Either the engineers needed to develop OEA’s algorithms to correctly analyze
the notes, or doctors would need to spend more time making their notes intelligible. 37 It was later
reported that doctors became frustrated with using the OEA, which took time away from patient
care. Moreover, when MD Anderson changed its electronic health records system, Watson could
not access patient data. 38
Three years later, MD Anderson withdrew from the partnership with IBM because of poor
performance. The OEA product had been tested but not commercialized. MD Anderson had paid
for the entire project, which was described as ‘unusual’ as companies typically paid research
centers. 39 In total, MD Anderson had invested more than $62 million dollars (as well as staff time)
in the failed project. 40 An internal audit report in 2016 stated: 41
Staff told us that the plan to pilot Leukemia OEA internally was suspended mid-project,
with lung cancer chosen instead because project leaders thought that area would
provide greater opportunity for a timely completion. Medical oncology staff also told us
that internal pilot testing of Lung OEA achieved an accuracy of prediction near 90
percent but advised that significant updating is needed before OEA can be tested
further. We were told that OEA must be integrated to the current medical records
system, that drug protocol and clinical trial data must be updated before internal pilot
36
37
38
39
40
41
Hsu, Jeremy. “IBM Invests $1 Billion to Grow Watson Supercomputer’s Struggling Business.” Spectrum.ieee.org,
January 9, 2014, https://spectrum.ieee.org/ibm­invests­1­billion­to­grow­watson­supercomputer­struggling­business,
accessed November 2021.
Greenstein, Shane, Martin, Mel & Agaian, Sarkis. “IBM Watson at MD Anderson Cancer Center.” Harvard Business
School, 9­621­022, February 18, 2021.
Lohr, Steve. “What Ever Happened to IBM’s Watson?” The New York Times, July 17, 2021,
https://www.nytimes.com/2021/07/16/technology/what­happened­ibm­watson.html, accessed November 2021.
Herper, Matthew. “MD Anderson Benches IBM Watson In Setback For Artificial Intelligence In Medicine.” Forbes,
February 19, 2017, https://www.forbes.com/sites/matthewherper/2017/02/19/md­anderson­benches­ibm­watson­in­
setback­for­artificial­intelligence­in­medicine/?sh=6603d1c33774, accessed November 2021.
Chesbrough, Henry. “IBM Watson And The Value of Open.” Forbes, February 12, 2020,
https://www.forbes.com/sites/henrychesbrough/2020/02/12/ibm­watson­and­the­value­of­open/?sh=11bedf4a2cfb,
accessed November 2021.
“Special Review of Procurement Procedures Related to the MD Anderson Cancer Center Oncology Expert Advisor
Project.” The University of Texas System Administration, November 2016, accessed November 2021.
Copyright © INSEAD
7
Page 246 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
If there are two equivalent therapies and one is given as a daily, the other as an
injection less frequently, the choice is up to the doctor and patient. OEA can’t answer
that question.
testing can resume, and that testing cannot be conducted with a network partner until
OEA is successfully piloted within MD Anderson.
Thereafter, IBM focused on selling the product it had developed with Memorial Sloan Kettering. It
was successful outside of the US, particularly in Asia and India, but in the US anecdotal reports
indicated that oncologists preferred to rely on their own judgement and said Watson suggested
only standard treatments they already knew. 43 By mid­2018, Watson for Oncology was in use in
approximately 230 hospitals worldwide. 44
The Watson for Genomics product developed with the University of North Carolina and Yale
University, among others, proved successful. Launched in 2016, the tool was used by genetic labs
to generate reports for oncologists. Within minutes of taking a patient’s genetic mutations, it could
produce a report of all relevant drugs and clinical trials – a feat that IBM engineer Vanessa
Michelini said would “enable[s] the labs to scale”. 45
Yet that success was not repeated elsewhere. In 2018, internal IBM documents shown to
journalists pointed to serious problems, including erroneous and unsafe advice provided in
Watson’s Oncology product, noting that it ­ 46
…often spit out erroneous cancer treatment advice, and that company medical
specialists and customers identified ‘multiple examples of unsafe and incorrect
treatment recommendations’ as IBM was promoting the product to hospitals and
physicians around the world.
The leaked documents indicated that Watson was undermined by the “inadequacy of the training
cases” – synthetic cases compiled by doctors and IBM engineers to train Watson with clinical
scenarios and hypothetical (rather than actual) patient records. This meant its recommendations
were derived from the doctor’s preferences rather than as a result of machine learning from actual
cases. 47 Back in 2014, historic patient cases had been used, but IBM had determined that
42
43
44
45
46
47
Herper, Matthew. “MD Anderson Benches IBM Watson In Setback For Artificial Intelligence In Medicine.” Forbes,
February 19, 2017, https://www.forbes.com/sites/matthewherper/2017/02/19/md­anderson­benches­ibm­watson­in­
setback­for­artificial­intelligence­in­medicine/?sh=6603d1c33774, accessed November 2021.
Ross, Casey & Swetlitz, Ike. “IBM pitched its Watson supercomputer as a revolution in cancer care. It’s nowhere close.”
Stat news, September 5, 2017, https://www.statnews.com/2017/09/05/watson­ibm­cancer/, accessed November 2021.
Ross, Casey & Swetlitz, Ike. “Documents raise alarm over Watson’s diagnostic flaws.” The Boston Globe, July 30,
2018, https://www.bostonglobe.com/business/2018/07/29/ibm­documents­raise­alarm­over­watson­diagnostic­
shortcomings/DB1xvAgLjjZDjrCRTWXDHK/story.html, accessed November 2021.
Strickland, Eliza. “How IBM Watson Overpromised and Underdelivered on AI Health Care.”, Spectrum.ieee.org, April 2,
2019, https://spectrum.ieee.org/how­ibm­watson­overpromised­and­underdelivered­on­ai­health­care, accessed
November 2021.
Ross, Casey & Swetlitz, Ike. “IBM’s Watson supercomputer recommended ‘unsafe and incorrect’ cancer treatments,
internal documents show.” Stat news, July 25, 2018, https://www.statnews.com/2018/07/25/ibm­watson­recommended­
unsafe­incorrect­treatments/, accessed November 2021.
Ross, Casey & Swetlitz, Ike. “IBM’s Watson supercomputer recommended ‘unsafe and incorrect’ cancer treatments,
internal documents show.” Stat news, July 25, 2018, https://www.statnews.com/2018/07/25/ibm­watson­recommended­
unsafe­incorrect­treatments/, accessed November 2021.
Copyright © INSEAD
8
Page 247 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
An IBM spokesperson defended the Oncology Expert Advisor, insisting its recommendations were
accurate and agreed with experts 90% of the time: “The OEA R&D project was a success, and
likely could have been deployed had MD Anderson chosen to take it forward.” 42
At Memorial Sloan Kettering, where a team of 24 doctors supported the training efforts, lead trainer
Dr Mark Kris said of Watson: “It’s been a struggle to update, I’ll be honest.” It took doctors and
engineers six years to train Watson in seven types of cancer, as well as keeping it up to date.
Citing an example of a change in treatment guidelines for every metastatic lung cancer patient
worldwide within one week following a conference presentation that suggested testing for a
particular gene, he acknowledged: 49 “Changing the system of cognitive computing doesn’t turn
around on a dime like that. You have to put in the (research) literature. You have to put in cases.”
Leaked documents later revealed that rather than thousands of cases being used as learning
input, the numbers ranged from 635 cases of lung cancer to 106 for ovarian cancer. 50 One article
suggested that IBM had ­ 51
…not publicly acknowledged the shortcomings of the software… To the contrary, top
company executives told customers and others that Watson for Oncology’s advice for
physicians is based on data from real patients, and that it had won nearly universal
praise from doctors around the world… John Kelly, the senior vice president for IBM’s
cognitive solutions division, said that Watson ‘has ingested all of the Memorial Sloan
data, historic patients and results’, and [claimed] at another event that …Watson for
Oncology is ‘going fabulously’.
The documents shown to journalists also stated ­
Studies conducted by IBM on the software, whose findings were touted as evidence
of the system’s usefulness, were designed to generate favorable results. 52
Dr Kris later said:
I believe in analytics, I believe it can uncover things… But when it comes to cancer, it
really doesn’t work. 53
Another commentator said cancer was “a textbook example of a ‘big problem’ that AI could solve
but it may also explain why some of the doctors who have worked with Watson seem less than
48
49
50
51
52
53
Ross, Casey & Swetlitz, Ike. “Documents raise alarm over Watson’s diagnostic flaws.” The Boston Globe, July 30,
2018, https://www.bostonglobe.com/business/2018/07/29/ibm­documents­raise­alarm­over­watson­diagnostic­
shortcomings/DB1xvAgLjjZDjrCRTWXDHK/story.html, accessed November 2021.
Ross, Casey & Swetlitz, Ike. “Watson is smart, but cancer is still smarter.” The Boston Globe, September 10, 2017.
Ross, Casey & Swetlitz, Ike. “Documents raise alarm over Watson’s diagnostic flaws.” The Boston Globe, July 30,
2018, https://www.bostonglobe.com/business/2018/07/29/ibm­documents­raise­alarm­over­watson­diagnostic­
shortcomings/DB1xvAgLjjZDjrCRTWXDHK/story.html, accessed November 2021.
Ross, Casey & Swetlitz, Ike. “Documents raise alarm over Watson’s diagnostic flaws.” The Boston Globe, July 30,
2018, https://www.bostonglobe.com/business/2018/07/29/ibm­documents­raise­alarm­over­watson­diagnostic­
shortcomings/DB1xvAgLjjZDjrCRTWXDHK/story.html, accessed November 2021.
Ross, Casey & Swetlitz, Ike. “Documents raise alarm over Watson’s diagnostic flaws.” The Boston Globe, July 30,
2018, https://www.bostonglobe.com/business/2018/07/29/ibm­documents­raise­alarm­over­watson­diagnostic­
shortcomings/DB1xvAgLjjZDjrCRTWXDHK/story.html, accessed November 2021.
Strickland, Eliza. “How IBM Watson Overpromised and Underdelivered on AI Health Care.”, Spectrum.ieee.org, April 2,
2019, https://spectrum.ieee.org/how­ibm­watson­overpromised­and­underdelivered­on­ai­health­care, accessed
November 2021.
Copyright © INSEAD
9
Page 248 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
synthetic cases representing patient cohorts were better suited to the development of Watson for
Oncology. 48
With regards to cancer, we’re talking about a constellation of thousands of diseases,
even if the focus is on one type of cancer. What we call ‘breast cancer’, for example,
can be caused by many different underlying genetic mutations and shouldn’t really be
lumped together under one heading. AI can work well when there is uniformity and
large data sets around a simple correlation or association. By having many data points
around a single question, neural networks can ‘learn’. With cancer, we’re breaking
several of these principles.
Former chief medical scientist at IBM Research, Dr Martin Kohn, had originally recommended a
narrower focus for Watson, such as predicting if a patient would have an adverse reaction to a
specific drug, but “was told I didn’t understand.” 56
Mounting Problems
Warning signs began to appear. From mid­2018, IBM Watson Health started to reduce headcount.
An ex­employee described how ­ 57
We were losing a lot of clients and we hadn’t merged the [data] assets that could have
created a really impressive market. Things just sort of spiraled out of control from
there. Clients were fed up. We couldn’t meet their needs anymore.
Engineers who had been ‘let go’ told reporters: “IBM Watson has great AI. It’s like having great
shoes but not knowing how to walk – they have to figure out how to use it.” An ex­Phytel engineer
suggested that IBM’s goal of new products based on a combination of Phytel and Explorys 58 had
floundered because IBM’s managers responsible for coming up with products, “couldn’t decide on
a roadmap… We pivoted so many times.” The people managing the offering had no technical
background – they came up with products that were impossible to produce. 59 Former GM of
54
55
56
57
58
59
Hruska, Joel. “IBM Watson Recommends Unsafe Cancer Treatments.” ExtremeTech.com, July 30, 2018,
https://www.extremetech.com/extreme/274453­ibm­watson­recommends­unsafe­cancer­treatments, accessed
November 2021.
Taulli, Tom. “IBM Watson: Why is Healthcare AI so Tough?” Forbes, February 27, 2021,
https://www.forbes.com/sites/tomtaulli/2021/02/27/ibm­watson­why­is­healthcare­ai­so­tough/?sh=598632c35375,
accessed November 2021
Lohr, Steve. “What Ever Happened to IBM’s Watson?” The New York Times, July 17, 2021,
https://www.nytimes.com/2021/07/16/technology/what­happened­ibm­watson.html, accessed November 2021.
Ross, Casey * Swetlitz, Ike. “When it comes to health data, Watson hasn’t been much help.” The Boston Globe, June
18, 2018.
Phytel offered disease management and preventive care protocols that identify and notify non­compliant patients of
needed healthcare actions. It developed and maintained prescription, lab and clinical data for health and disease
management (https://www.ibm.com/watson­health/about/phytel). Explorys was a healthcare intelligence cloud company
that had built one of the largest clinical therapeutic data sets in the world, which its algorithms analyzed and supported
hospitals and healthcare systems to deliver better care and quality of life (https://www.ibm.com/products/explorys­ehr­
data­analysis­tools).
Friedman, Lauren F. “IBM’s Watson Supercomputer May Soon Be The Best Doctor in the World.” Yahoo! Finance, April
22, 2014, https://finance.yahoo.com/news/ibms­watson­supercomputer­may­soon­141433414.html, accessed
November 2021.
Copyright © INSEAD
10
Page 249 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
enamored of it”. 54 Dr Nirav R. Shah, Chief Medical Officer at Sharecare, highlighted the complexity
involved: 55
Lack of a realistic timeframe was an issue identified by Jeremy Howard, CEO of radiology imaging
AI startup Enlitic. He believed AI could transform healthcare, but “That’s a 25­year project.” 61 Oren
Etzioni, CEO of the Allen Institute for AI and former computer science professor, made the
following analogy: 62
IBM Watson is the Donald Trump of the AI industry – outlandish claims that aren’t
backed up by credible data… Everyone – journalists included – know[s] that the
emperor has no clothes, but most are reluctant to say so.
Professor Robert J. Marks III, Director of the Bradley Center for Natural and Artificial Intelligence,
pointed to inflated expectations: 63
I’ve also heard that they mismanaged it. I know as an engineer, that you get a good
result and you think the world’s going to beat a path to your door and use your
invention. And boy that doesn’t work! You need to interface, and you need to get down
and dirty with the people that are actually applying it.
Another commentator concurred: 64
IBM brought the big ideas but putting them into practice was an afterthought.
Predictably, their top-down approach was met with resistance. Doctors spend roughly
a decade of their lives learning how to be a doctor and then they continue to learn
throughout their careers. It’s only natural for them to be skeptical that an AI system
would know their patients better than they do.
I should clarify here that clinical AI is not designed to replace the judgment of clinicians,
but rather to augment it with new information they may not have been aware of or had
at their disposal. Despite this, misconceptions about AI’s role in healthcare persist and
the idea that AI replaces doctors is another source of resistance.
Likewise, the MIT Technology Review said that in the case of the OEA project, IBM and MD
Anderson “both overinflated expectations for the technology”.
60
61
62
63
64
Lohr, Steve. “What Ever Happened to IBM’s Watson?” The New York Times, July 17, 2021,
https://www.nytimes.com/2021/07/16/technology/what­happened­ibm­watson.html, accessed November 2021.
Steve Lohr. “The Promise of Artificial Intelligence Unfolds in Small Steps.” The New York Times, Febuary 28, 2016,
https://www.nytimes.com/2016/02/29/technology/the­promise­of­artificial­intelligence­unfolds­in­small­steps.html,
accessed November 2016.
Snapp, Shaun. “How IBM is Distracting from the Watson Failure to Sell More AI.” Brightwork Research, June 13, 2019,
https://www.brightworkresearch.com/how­ibm­is­distracting­from­the­watson­failure­to­sell­more­ai/, accessed
November 2021.
“Why was IBM Watson a Flop in Medicine?” MindMattersNews Podcast, August 9, 2019,
https://mindmatters.ai/2019/08/why­was­ibm­watson­a­flop­in­medicine/, accessed November 2021.
Frownfelter, John. “Why IBM Watson Health could never live up to the promises.” MedCityNews.com, April 8, 2021,
https://medcitynews.com/2021/04/why­ibm­watson­health­could­never­live­up­to­the­promises/?rf=1, accessed
November 2021.
Copyright © INSEAD
11
Page 250 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Watson, Manoj Saxena, admitted: “The challenges turned out to be far more difficult and time­
consuming than anticipated.” 60
We often call out overly optimistic news coverage of drugs and devices. But
information technology is another health arena where uncritical media narratives can
cause harm by raising false hopes and allowing costly and unproven investments to
proceed without scrutiny.
Robert Wachter, chair of the department of medicine at the University of California, summed it up:
IBM came in with marketing first, product second, and got everybody excited… then
the rubber hit the road. This is an incredibly hard set of problems, and IBM by being
first out has demonstrated that for everyone else. 67
What Next for IBM Watson Health?
Between 2011 and 2017 IBM had announced nearly 50 partnerships to develop AI products for
medicine (see Exhibit 1). Many had not produced commercial applications. Even Watson for
Genomics in partnership with the University of North Carolina was discontinued at the end of
2020.68 While Watson Health had started many ambitious projects, none had been successful.
An observer commented: 69
The projects that IBM announced that first day did not yield commercial products. In
the eight years since, IBM has trumpeted many more high-profile efforts to develop
65
66
67
68
69
Freedman, David. H. “A Reality Check for IBM’s AI Ambitions.” MIT Technology Review, June 27, 2017,
https://www.technologyreview.com/2017/06/27/4462/a­reality­check­for­ibms­ai­ambitions/, accessed November 2021.
Snapp, Shaun. “How IBM is Distracting from the Watson Failure to Sell More AI.” Brightwork Research, June 13, 2019,
https://www.brightworkresearch.com/how­ibm­is­distracting­from­the­watson­failure­to­sell­more­ai/, accessed
November 2021.
Strickland, Eliza. “How IBM Watson Overpromised and Underdelivered on AI Health Care.”, Spectrum.ieee.org, April 2,
2019, https://spectrum.ieee.org/how­ibm­watson­overpromised­and­underdelivered­on­ai­health­care, accessed
November 2021.
Lohr, Steve. “What Ever Happened to IBM’s Watson?” The New York Times, July 17, 2021,
https://www.nytimes.com/2021/07/16/technology/what­happened­ibm­watson.html, accessed November 2021.
Strickland, Eliza. “How IBM Watson Overpromised and Underdelivered on AI Health Care.”, Spectrum.ieee.org, April 2,
2019, https://spectrum.ieee.org/how­ibm­watson­overpromised­and­underdelivered­on­ai­health­care, accessed
November 2021.
Copyright © INSEAD
12
Page 251 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Those expectations had been amplified by the media. IBM had claimed in 2013 that “a new era of
computing had emerged” and gave Forbes the impression that Watson “now tackles clinical trials”
and would be used with patients in just a matter of months. In 2015, The Washington Post quoted
an IBM manager describing how Watson was busy establishing a “collective intelligence model
between machine and man”. The Post reported that the computer system was “training alongside
doctors to do what they can’t”. 65 Another commentator warned about the media hype: 66
AI-powered medical technology – many of which have fizzled, and a few of which have
failed spectacularly.
Watson health actually has some great technologies. AI was able to recognize pictures
of cancer at a higher rate than even some of the top doctors. But that was in the lab.
The business executives had minimum references and experiences in introducing a
business model powered by AI, so they mainly depended upon engineers to show
them what can be accomplished and went to market without a fully integrated
business-AI strategy.
Ultimately it was about the people, not the product:
The company’s top management, current and former IBM insiders noted, was
dominated until recently by executives with backgrounds in services and sales rather
than technology product experts. Product people, they say, might have better
understood that Watson had been custom-built for a quiz show…a powerful but limited
technology.
In January 2020, the board announced that Rometty would step down as CEO. Arvind Krishna
would be taking over from April 2020. He was a technologist who had played a significant role in
developing IBM’s AI, cloud, quantum computing and blockchain. 72
By early 2021 IBM Watson Health was generating $1 billion in annual revenues, but was not yet
profitable. 73 While IBM did not publish separate accounts, it was estimated that it had invested
over $4 billion in acquiring companies with medical data, billing records and patient diagnostic
images. 74
In February 2021 it was rumoured that IBM was paring back the unit and was considering selling
Watson Health. Rumours persisted throughout the year but at the end of 2021 the unit was still on
IBM’s website, mentioning AI, blockchain and data and analytics solutions, with its commitment to
“support our client’s digital transformations through a combination of technology solutions and
experienced consulting”. However, it was a far cry from “revolutionizing healthcare” through AI
and the suite of game­changing products promised in 2011.
70
71
72
73
74
Tan, Alan. “What did IBM Watson do wrong in Bsuiness AI Strategy.” Medium, July 1, 2019,
https://medium.datadriveninvestor.com/what­did­ibm­watson­do­wrong­in­business­ai­strategy­7392fd8f1a48, accessed
November 2021.
Tan, Alan. “What did IBM Watson do wrong in Bsuiness AI Strategy.” Medium, July 1, 2019,
https://medium.datadriveninvestor.com/what­did­ibm­watson­do­wrong­in­business­ai­strategy­7392fd8f1a48, accessed
November 2021.
Vanian, Jonathan. “IBM CEO Ginny Rometty to step down.” Fortune, January 31. 2020,
https://fortune.com/2020/01/30/ibm­ceo­ginni­rometty­to­retire/, accessed November 2021.
Cooper, Laura & Lombardo, Cara. “IBM explores sale of Watson Health.” The Australian, February 20, 2021.
Lohr, Steve. “What Ever Happened to IBM’s Watson?” The New York Times, July 17, 2021,
https://www.nytimes.com/2021/07/16/technology/what­happened­ibm­watson.html, accessed November 2021.
Copyright © INSEAD
13
Page 252 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Another said in 2019 that AI worked best in cases with high value (repetition) but that healthcare
was a high­stakes field 70 that needed an integrated strategy: 71
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
With Watson Health clearly facing a very different future, how could IBM’s deployment of AI have
gone so wrong?
Copyright © INSEAD
14
Page 253 of 282
Date
Partner
Project
Outcome by 2019
Nuance Communications
Diagnostic tool and clinical­
decision support tools
No tools in use
Sept.
WellPoint (now Anthem)
Clinical­decision support tools
No tools in use
2012
Mar.
Memorial Sloan Kettering
Cancer Center
Clinical­decision support tool for
cancer
Watson for Oncology
Oct.
Cleveland Clinic
2013
MD Anderson Cancer
Center
Clinical­decision support tool for
cancer
No tool in use
New York Genome Center
Genomic­analysis tool for brain
cancer
No tool in use
June
GenieMD
Consumer app for personalized
medical advice
No app available
Sept.
Mayo Clinic
Clinical­trial matching tool
Watson for Clinical Trial
Matching
2011
Feb.
Oct.
2014
Mar.
2015
April
Johnson & Johnson
Training tool for medical students;
clinical­decision support tool
Consumer app for pre­ and post­
operation coaching; consumer
app for managing chronic
conditions
Consumer app for personalized
diabetes management
No tools in use
No apps available
Sugar.IQ app
April
Medtronic
May
Epic
Clinical­decision support tool
No tool in use
May
University of North Carolina,
others
Genomic­analysis tool for cancer
Watson for Genomics
July
CVS Health
Sept.
Teva Pharmaceuticals
Sept.
Boston Children's Hospital
Dec.
Nutrino
Care­management tool for chronic
conditions
Drug­development tool; consumer
app for managing chronic
conditions
Clinical­decision support tool for
rare pediatric diseases
Consumer app for personalized
nutrition
No tool in use
No tool in use; no app available
No tool in use
No app available
advice during pregnancy
Dec.
2016
Jan.
Novo Nordisk
Consumer app for diabetes
management
No app available
Under Armour
Consumer app for personalized
athletic coaching
No app available
Feb.
American Heart Association
April
American Cancer Society
Consumer app for workplace
health
Consumer app for personalized
guidance during cancer treatment
Copyright © INSEAD
No app available
No app available
15
Page 254 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Exhibit 1
IBM AI for Medicine Partnerships
American Diabetes
Association
Consumer app for personalized
diabetes management
No app available
Oct.
Quest Diagnostics
Genomic­analysis tool for cancer
Watson for Genomics from
Quest Diagnostics
Nov.
Celgene Corp.
Drug­safety analysis tool
No tool in use
MAP Health Management
Relapse­prediction tool for
substance abuse
No tool in use
2017
May
Source: Strickland, Eliza, April 2, 2019, https://spectrum.ieee.org/how­ibm­watson­overpromised­and­underdelivered­on­ai­
health­care,
Copyright © INSEAD
16
Page 255 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
June
9B14E005
Brad Evans wrote this case using publicly available information under the supervision of Professor Derrick Neufeld and Professor
Ning Su solely to provide material for class discussion. The authors do not intend to illustrate either effective or ineffective handling
of a managerial situation. The authors may have disguised certain names and other identifying information to protect confidentiality.
This publication may not be transmitted, photocopied, digitized or otherwise reproduced in any form or by any means without the
permission of the copyright holder. Reproduction of this material is not covered under authorization by any reproduction rights
organization. To order copies or request permission to reproduce materials, contact Ivey Publishing, Ivey Business School, Western
University, London, Ontario, Canada, N6G 0N1; (t) 519.661.3208; (e) cases@ivey.ca; www.iveycases.com.
Copyright © 2014, Richard Ivey School of Business Foundation
Version: 2014-03-18
“Nobody’s madder than me about the fact that the website isn’t working as well as it should, which means
it’s gonna get fixed.” 2
U.S. President Barack Obama on Healthcare.gov
In October 30, 2013, Health and Human Services 3 Secretary Kathleen Sebelius had just endured several
hours of grilling by the House Energy and Commerce Committee. The fledgling $100+ million
Healthcare.gov website project that had been entrusted to CGI — an IT solutions supplier headquartered
in Montreal, Canada — was crashing and burning. Sebelius was the highest-ranking official subject to
congressional oversight and was there to represent her boss, the president. As the former governor of
Kansas, she was certainly no stranger to political grandstanding or the pressures that come with operating
in the spotlight of public office. But this was different. “Hold me accountable for the debacle. I am
responsible,” 4 Sebelius told the bipartisan committee, while at the same time displaying brave confidence
that a “gonna get fixed” plan was being executed. The reality was that there were still far more questions
than answers. In short, admitting that there were issues and taking responsibility was the easy part.
Getting the project on track and out of the headlines would not be so straightforward.
THE U.S. HEALTH CARE SYSTEM
Health care in the United States is subject to extensive government regulation. Under the system, the
federal government cedes primary responsibility to the states as per the McCarran-Ferguson Act of 1945. 5
1
This case has been written on the basis of published sources only. Consequently, the interpretation and perspectives
presented in this case are not necessarily those of the U.S. Government or any of its employees.
2
Obama on HealthCare.gov, ‘It’s Gonna Get Fixed’ video, mark 19:55, The Washington Post, October 21, 2013, accessed
March 2, 2014.
3
Health and Human Services (HHS) is a cabinet-level department of the U.S. federal government responsible for protecting
the health and providing essential human services for all Americans. The department employs more than 76,000 people with
an annual budget approaching $1 trillion.
4
S. Condon, “Sebelius: ‘Hold me accountable for the debacle’ of HealthCare.gov”, CBS News,
www.cbsnews.com/news/sebelius-hold-me-accountable-for-the-debacle-of-healthcaregov/, October 30, 2013, accessed
March 2, 2014.
5
The McCarran-Ferguson Act of 1945 gives states the authority to regulate the “business of insurance” without interference
from federal regulation, unless federal law specifically dictates otherwise. The act provides that the “business of insurance,
Page 256 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
THE OBAMACARE WEBSITE 1
Page 2
9B14E005
Due to the highly specialized nature of the health care industry, with its many guidelines and restrictions,
there were only about 35 private health insurance companies in the United States, not including Medicare
providers. 6 However, in aggregate the U.S. health care system was comprised of a vast number of players,
resulting in a complex framework with massive administrative overhead. A study by Harvard Medical
School and the Canadian Institute for Health Information determined that some 31 per cent of U.S. health
care dollars, or more than $1,000 per person per year, went to health care administrative costs — nearly
double the administrative overhead in Canada on a percentage basis. 7 The study concluded that a large
sum might be saved if administrative costs could be trimmed by implementing a nationalized single-payer
system such as those adopted in Canada and the United Kingdom.
Health care had become something of a political hot potato. A 2008 poll conducted by the Harvard School
of Public Health and Harris Interactive found that Americans were divided in their views of the U.S.
health care system and that there were significant differences by political affiliation. In one recent poll,
only 45 per cent of U.S. respondents replied that the U.S. health care system was better than that of other
countries, while 39 per cent said that other countries’ systems were better. 8 The belief that the U.S.
system was superior was highest among Republicans (68 per cent), lower among independents (40 per
cent) and lowest for Democrats (32 per cent). Among the Democratic respondents, over half (56 per cent)
said they would be more likely to support a presidential candidate who advocated making the U.S. system
more like that of other countries (this figure was only 37 per cent for independents and 19 per cent for
Republicans).
A 2013 Bloomberg study examined the health care systems of 48 countries and/or regions and ranked
system efficiency using three criteria: life expectancy (weighted 60 per cent), relative per capita cost of
health care (30 per cent), and absolute per capita cost of health care (10 per cent) 9 (see Exhibit 1). Hong
Kong placed first, followed by Singapore, Japan, Israel and Spain. Canada was seventeenth. The United
States ranked forty-sixth, with only Serbia and Brazil scoring lower. Health care costs per capita for the
United States came in at $8,608, which represents 17.2 per cent as a percentage of gross domestic product
(GDP) per capita. By comparison, the Hong Kong figures were $1,409 and 3.8 per cent while Canada’s
were $5,630 and 10.8 per cent. These data were not necessarily reflective of the best health care in the
world but rather provided a measure of overall quality as a function of cost.
In the United States, health care facilities were largely owned and operated by private sector businesses,
while the government primarily provided health insurance for public sector employees. That said, a 2004
report from the Organization for Economic Co-operation and Development (OECD) noted: “With the
exception of Mexico, Turkey, and the U.S., all OECD countries had achieved universal or near-universal
and every person engaged therein, shall be subject to the laws of the several States which relate to the regulation or
taxation of such business.”
6
Insurance Providers, “How Many Health Insurance Companies Are There in the United States?,”
www.insuranceproviders.com/how-many-health-insurance-companies-are-there-in-the-united-states, accessed January 16,
2014.
7
S. Woolhandler et al., “Costs of Health Administration in the U.S. and Canada,” New England Journal of Medicine 349,
September 21, 2003, pp. 768–775, www.nejm.org/doi/full/10.1056/NEJMsa022033, accessed February 6, 2014.
8
Harvard School of Public Health and Harris Interactive, “Most Republicans Think the U.S. Health Care System is the Best
in the World. Democrats Disagree,” www.hsph.harvard.edu/news/press-releases/republicans-democrats-disagree-us-healthcare-system/, March 20, 2008, accessed December 21, 2013.
9
Bloomberg Visual Data, “Most Efficient Health Care: Countries,” August 19, 2013, www.bloomberg.com/visual-data/bestand-worst/most-efficient-health-care-countries, accessed December 20, 2013.
Page 257 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
The federal Health and Human Services (HHS) department oversees the various federal agencies involved
in health care, whereas individual state governments maintain their own local health departments.
Page 3
9B14E005
(at least 98.4 per cent insured) coverage of their populations by 1990.” 10 In terms of universal coverage,
statistics reported in the 2010 U.S. census were troubling 11 (see Exhibits 2 and 3).
HEALTH CARE REFORM
Few people realized it, but universal health care was an important agenda item 60 years ago. President
Harry Truman attempted, but failed, to convince Congress to pass laws ensuring universal coverage.
What he got instead was the Hill-Burton Act, 14 the federal government’s first venture into building
hospitals. The bill was groundbreaking and, over six decades, had delivered billions of dollars for
equipment and completed projects in thousands of communities across the United States.
In 1965, President Lyndon Johnson enacted legislation that introduced Medicare and Medicaid 15 to cover
both hospital and general medical insurance for senior citizens along with partial federal funding of
programs for the poor (with Medicaid managed and co-financed by the individual states). More recently,
the Health Security Act was a health care reform package proposed by the administration of President Bill
Clinton in 1993. However, opposition to the plan was heavy from the beginning, and, under constant
attack by the pharmaceutical and health insurance industries, the proposed act was dropped by September
1994.
After his inauguration, President Obama announced in February 2009 that he intended to work with
Congress to build a plan for health care reform. By July, committees within Congress had approved a
number of bills. Meanwhile, the Senate Finance Committee began a series of meetings to develop a health
care reform bill. The president and congressional leaders recognized that progressive plans, such as a
single-payer or Medicare-for-All act, would struggle in the Senate. By deliberately drawing on bipartisan
ideas, they hoped to increase the chances of garnering the necessary votes for passage.
Democrats rallied behind an “individual mandate” proposal (i.e., a requirement that each citizen either
purchase coverage from a private company, or get it through an employer or obtain it as a government
benefit). But Republicans began to oppose the mandate and threatened to filibuster any bills that
10
E. Docteur and H. Oxley, “Health-system Reform: Lessons from Experience in The OECD Health Project Towards HighPerforming Health Systems Policy,” The OECD Health Project, October 19, 2004, p. 74.
11
C. DeNavas-Walt et al., “Income, Poverty and Health Insurance Coverage in the United States: 2010,” U.S. Census
Bureau, September 13, 2011, www.census.gov/prod/2011pubs/p60-239.pdf, accessed December 22, 2013.
12
S. Woolhandler et al., “Despite Slight Drop in Uninsured, Last Year’s Figure Points to 48,000 Preventable Deaths,”
Physicians for a National Health Program, September 12, 2012, www.pnhp.org/news/2012/september/despite-slight-drop-inuninsured-last-year’s-figure-points-to-48000-preventable-, accessed December 22, 2013.
13
CBS News, “Medical Debt Huge Bankruptcy Culprit,” June 5, 2009, www.cbsnews.com/news/medical-debt-hugebankruptcy-culprit, accessed December 22, 2013.
14
The Hill-Burton Act was named for its two sponsors in the Senate, Lister Hill, a Democrat from Alabama, and Harold
Burton, a Republican from Ohio. They believed that existing hospitals had been neglected during the years of the
Depression and World War II and that there was a desperate need for clinics and small hospitals, particularly in poor rural
areas.
15
Medicaid is the U.S. health program for families and individuals with low income and resources. Medicare is a U.S. federal
government social insurance program that guarantees access to health insurance for certain Americans and legal residents
aged 65 and older, younger people with disabilities, people with end stage renal disease and persons with amyotrophic
lateral sclerosis.
Page 258 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Recent evidence demonstrated that lack of health insurance resulted in some 45,000 to 48,000
unnecessary deaths per year in the United States. 12 Moreover, a study published in the American Journal
of Medicine reported that getting sick was a factor in 62 per cent of personal bankruptcies, an increase
from just 8 per cent in 1981. 13
Page 4
9B14E005
Despite significant ongoing political wrangling, the House of Representatives passed the Affordable
Health Care for America Act in a 220 to 215 vote on November 7, 2009 and forwarded it to the Senate for
passage. The bill then passed on December 24, 2009, with all Democrats and two independents voting for
it, and all Republicans voting against. On March 23, 2010, the Patient Protection and Affordable Care Act
(PPACA), commonly called the Affordable Care Act (ACA), and otherwise known in colloquial terms as
“Obamacare,” became law.
AFFORDABLE CARE ACT (ACA)
The ACA was made up of a combination of measures aimed at controlling health care costs as well as
expanding coverage through both public and private insurance. There were two primary mechanisms for
increasing insurance coverage: (1) expanding Medicaid eligibility to include nearly all U.S. citizens under
65 with family incomes up to 133 per cent of the federal poverty level ($30,675 for a family of four in
2012) 16 and (2) creating state-based insurance exchanges where individuals and small business could buy
health insurance plans. The Congressional Budget Office (CBO) originally estimated that the legislation
would reduce the number of uninsured residents by 32 million, leaving 23 million uninsured residents in
2019 after the bill's provisions had all taken effect. With the elderly covered by Medicare, the CBO
estimate projected that the law would raise the proportion of insured non-elderly citizens from 83 per cent
to 94 per cent. 17
The ACA established state-based health insurance exchanges. The exchanges were to be regulated online
marketplaces, administered by either the federal or state governments, through which individuals and
small businesses could purchase private insurance plans. The exchanges were to be accessible via the
website Healthcare.gov where providers could offer for sale private regulated insurance plans. Consumers
would surf the website or ring a call centre, compare the plans on offer, fill out a form to the government
that would be used to determine their eligibility for subsidies and then purchase the insurance of their
choice from the options available during designated open enrollment periods. 18
The insurance exchanges were designed with the intention of creating a regulated market for private
insurance that addressed shortcomings and failures in the current system such as medical bankruptcies,
the high number of uninsured, coverage limits, unaffordability and inflation. Only approved plans
meeting certain standards were permitted for sale on the exchanges, and insurers could not deny insurance
to applicants with pre-existing conditions.
16
American
Public
Health
Association
(APHA),
“Medicaid
Expansion,”
www.apha.org/advocacy/Health+Reform/ACAbasics/medicaid.htm, accessed December 23. 2013.
17
C. Schoen et al., “Insured But Not Protected: How Many Adults Are Underinsured?,” Health Affairs, 2005,
http://content.healthaffairs.org/content/early/2005/06/14/hlthaff.w5.289.full.pdf, accessed December 23, 2013.
18
Exchanges were to be fully operational by September 2013 in time for the initial open enrollment period starting October
1, 2013 and for coverage starting on January 1, 2014.
Page 259 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
contained it. Senate minority leader Mitch McConnell, who led the Republican congressional strategy in
responding to the bill, did not support it and worked to maintain party discipline and prevent defections.
Republican senators, including those who had supported previous bills with a similar mandate, began to
describe the mandate as “unconstitutional.”
Page 5
9B14E005
The ACA project as a whole was overseen by the Centers for Medicare & Medicaid Services (CMS),
which reported to the HHS. Numerous additional government departments were tasked with awarding
contracts to manage, support or service the implementation, including the Centers for Disease Control and
Prevention (CDC) and the Internal Revenue Service (IRS). According to the government website, “CMS
is developing new programs and tools as a result of the Affordable Care Act to help you deliver better
care. We are your partner in ensuring that millions of Americans are part of a better health care system.” 19
Although CMS had no experience running a project this big, a government agency had to be in charge,
and CMS was “it.” CMS also monitored and coordinated the development of all the state exchanges.
In late 2011, the Obama administration via HHS and CMS gave a website design and development
contract valued at $93.7 million 20 to CGI Federal, a wholly owned subsidiary of the Canadian firm CGI
Group (the initials stand for “Conseillers en Gestion et Informatique,” which roughly translates to
“Information Systems and Management Consultants”). CGI was founded in Quebec City in 1976 by two
26-year-olds named Serge Godin and Andre Imbeau. 21 The company was headquartered in Montreal,
Canada, and had a market capitalization of $8.9 billion on annual revenues of $4.8 billion. In 2013, CGI
acquired U.K. rival Logica, in the process doubling its employee base and becoming Canada's largest
technology firm. CGI Group employed 72,000 personnel in 400 offices worldwide, including 11,000 in
the United States. CGI’s business model relied on embedding personnel deeply within the client business;
as one company profile read, “The ultimate aim is to establish relations so intimate with the client that
decoupling becomes almost impossible.” 22
CGI’s winning bid stretched back to 2007 when it was one of 16 companies to receive certification on a
$4 billion “indefinite delivery, indefinite quantity” contract for upgrading the Medicare and Medicaid
systems. Government-wide acquisition contracts (GWACs) were designed to allow agencies to efficiently
issue task orders to pre-vetted companies without requiring full procurement processes. However, CGI
Federal was a relative newcomer on the U.S. government information technology (IT) contracting scene.
It bought U.S. contractor American Management Systems in 2004 but only started ramping up U.S.
business after 2008; this process was accelerated in 2010 with its $1.1 billion acquisition of U.S.-based
military IT contractor Stanley, Inc. CGI was the twenty-ninth largest federal IT contractor in the United
States, with about $950 million in contracts in 2012 (by comparison, Lockheed Martin was the largest IT
contractor with $14.9 billion in contracts).
Experience in similar types of projects was very important in winning federal contracts, and CGI could
demonstrate success in delivering complex projects on time and on budget. For example, in 2009, the
White House’s Recovery Board retained CGI Federal to adapt a well-functioning system it had previously
built for the U.S. Environmental Protection Agency into FederalReporting.gov (another complex, publicfacing, high-volume site that would handle all contracts granted under federal stimulus legislation). CGI
completed the adaptation in just six weeks, for much less money, and won accolades for flexibility and
reliability. Not every project went so well, however; in fact, after three years of missed deadlines on
Ontario’s health care medical registry for diabetes sufferers, officials cancelled a $46.2 million contract
with CGI. 23
19
CMS website, www.cms.gov/about-cms/aca/affordable-care-act-in-action-at-cms.html, accessed February 10, 2014.
All currency in US funds unless specified otherwise.
21
CGI website, www.cgi.com/en, accessed December 24, 2013.
22
L. Depillis, “Meet CGI Federal, The Company Behind the Botched Launch of HealthCare.gov,” The Washington Post,
October 16, 2013, www.washingtonpost.com/blogs/wonkblog/wp/2013/10/16/meet-cgi-federal-the-company-behind-thebotched-launch-of-healthcare-gov, accessed December 28, 2013.
23
Ibid.
20
Page 260 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
THE HEALTHCARE.GOV PROJECT
9B14E005
In addition to CGI, multi-million dollar contracts were awarded to a long list of companies. These
included deals with Serco ($114 million to support the federally facilitated marketplace and state-based
marketplaces for the eligibility support tasks under the ACA), Quality Software Services ($68 million to
develop the federal data services hub), McKinsey ($14 million across three project management and
analysis contracts) and many others. A June 2013 report from the Government Accountability Office
listed 55 multi-million dollar contracts (see Exhibit 4). Like CGI, most of these deals were done under
existing contracts, without an open bidding process. Notably, the contractor list does not appear to include
any companies with specialized experience building large-scale web services.
Under the contract, CGI would design, construct and implement an online system for the 36 states that
elected to participate in the federal marketplace (the remaining states had decided to build their own
exchanges). CGI was responsible for “putting all of the pieces together,” making the QSSI-developed
back-end data hub interface seamlessly with the front-end user interface and identity management
software and ensuring the system remained adaptable. As CGI Senior Vice-president Cheryl Campbell
put it in a September hearing on the progress of the site, CGI’s role involved “designing an IT solution
that is adaptable and modular to accommodate the implementation of additional functional requirements
and services.” 24
More specifically, the system CGI was charged with delivering was meant to be a fabulous “one-stop
shop” for consumers to buy health insurance. It must first welcome and validate the user; create a secure
online identity for them; collect and store relevant personal information; cross-check income and
immigration status; determine whether the applicant qualified for a subsidy; connect them with the
appropriate marketplace; advise them on insurance options; enable them to purchase insurance; and,
finally, communicate the result to the pertinent government agencies and insurers (see Exhibit 5). 25 In
short, the final product had to be easy enough to navigate individuals through a complex array of different
insurance offerings, secure enough to hold sensitive private data and powerful enough to withstand peak
traffic in the hundreds of thousands if not millions of concurrent users.
TECHNOLOGY INFRASTRUCTURE
Designing and implementing a health insurance exchange was a massively complex undertaking. It
involved creating an entirely new health insurance marketplace replete with new policies, procedures and
practices spanning regulatory, financial, programmatic, contractual, actuarial and administrative domains.
Enterprise architectures would have to be designed, with new hardware technology, applications,
databases, telecommunications, work flows, reporting capabilities and middleware to link the new
systems to legacy systems across multiple federal and state agencies and programs as well as participating
health insurers’ systems. It involved developing a seamless, consumer-friendly, real-time Web portal to
screen consumers for Medicaid and Children’s Health Insurance Program (CHIP) eligibility, receive
applications for complex new federal premium and cost sharing subsidies, verify compliance with
individual and parental mandates, facilitate comparison shopping among multiple qualified health plans
(QHPs) with multiple government-regulated products and benefit designs, enable collection of employers’
subsidies (if any) and allow for enrollment of self and family members with a QHP.
The magnitude of the resulting system-related complexity was the subject of political ridicule, as
exemplified in the unreadable (and indecipherable) graphic shown in Exhibit 6. Partisan jeering aside, the
24
Ibid.
T. McCarthy, “Contractors Defend Work on Obamacare Website at Congressional Hearing,” The Guardian, October 24,
2013, www.theguardian.com/world/2013/oct/24/obamacare-website-testify-congress-live, accessed January 4, 2014.
25
Page 261 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Page 6
9B14E005
challenges, and opportunity, of building such a large-scale enterprise system to meet the ACA
requirements were significant. With unprecedented funding opportunities for states, and the potential to
modernize existing Medicaid/CHIP systems, the benefits were obvious. On the other hand, the
infrastructure update work was enormous and would have to be completed in full compliance with ACA
expectations. A major update or complete replacement of legacy IT systems would be required by most
states. New interfaces to link individual eligibility and enrollment data among Medicaid, CHIP and the
states’ exchanges were also necessary. 26 In many cases, wholesale system replacements would be
required.
THE FAILED LAUNCH
As the project timeline progressed, several parties both within the Obama Administration and outside it
began to raise red flags. Key deadlines were being missed, and the “go-live” date was being jeopardized.
For example, a report from the Government Accountability Office stated:
CMS’s timelines for the remaining key activities provide a road map for completion; however,
factors such as the still-evolving scope of CMS’s required activities in each state and the many
activities yet to be completed — some close to the start of enrollment — could suggest the
potential for future challenges. And while missed interim deadlines may not affect
implementation, additional missed deadlines could do so. 27
A more pointed warning came via internal emails between CMS officials. Henry Chao, deputy chief
information officer and Healthcare.gov’s chief project manager, described struggles with contractors, staff
shortages and software problems 11 weeks before the October go-live date. In a July 16 email sent to CGI
Federal, Chao described the agency’s low confidence level in the project work, citing constant struggles
with releases, changed delivery dates and poor quality assurance on software. Said Chao: “I just need to
feel more confident they are not going to crash the plane at take-off.” The very next day, however, Chao
assured a House Oversight and Government Reform subcommittee that HealthCare.gov would be ready
on time, and on July 20 he distributed an email urging CMS staff to make good on his pledge to
lawmakers: “I would like you [to] put yourself in my shoes standing before Congress, which in essence is
standing before the American public, and know that you speak the tongue of not necessarily just past
truths but the truth that you will make happen.” 28
Unfortunately, the concerns Chao raised in July were confirmed when the Healthcare.gov website went
live on October 1, 2013. As expected, the new system was inundated with traffic on its first day. While a
small proportion of users experienced success using the site and were able to purchase new health
insurance, the majority complained of stalled pages, dead ends, error messages, dropped accounts, loading
delays and misinformation.
26
D. Bachrach et al., “Medicaid’s Role in the Health Benefits Exchange: A Road Map for States,” National Academy for
State Health Policy, March 2011, www.nashp.org/sites/default/files/maxenroll%20Bachrach%20033011.pdf, accessed
January 17, 2014.
27
U.S. Government Accountability Office, “Patient Protection and Affordable Care Act, Status of Federal and State Efforts to
Establish Health Insurance Exchanges for Small Businesses,” Report to the Chairman, Committee on Small Business,
House of Representatives, June 2013, www.gao.gov/assets/660/655285.pdf, accessed February 10, 2014.
28
D. Morgan, “Henry Chao, Healthcare.gov Official, Feared Website ‘Crash’ In July,” Reuters, November 15, 2013,
www.huffingtonpost.com/2013/11/15/henry-chao_n_4280291.html, accessed February 10, 2014.
Page 262 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Page 7
Page 8
9B14E005
During the following weeks and months, many questions were asked by politicians and the public alike.
Had CGI failed to correctly forecast performance and throughput requirements, resulting in an undersized
infrastructure? Had scheduling pressures led to insufficient testing prior to launch? (HHS had allegedly
taken years to issue final project specifications, and some observers suggested this prevented CGI from
getting started until the spring of 2013.) Or had pre-existing “silos” between the various project players
(e.g., contractors and departments) resulted in some kind of communication breakdown? 30
Other experts argued that the fault could be traced to fundamental, systemic government inefficiency
issues — e.g., arduous procurement processes that caused agencies to lock in contractors for longer
periods. While such an approach might provide an agency with certain short-term benefits, it might also
give preference to insiders, exclude “hungry” outside companies and dull competitiveness. As a result,
innovative programming frameworks and development methods might take a long time to reach the
government, since a provider that had already secured a 10-year contract might have less incentive to
innovate. Likewise, some observers suggested that government contractors would always find a way to
ensure the work they produced was complex enough to consume the available budget. 31
Scalability consultants commented that the government’s mistakes were not so different from errors
observed in many private companies when technical requirements were miscalculated. The big difference
with the ACA project was the highly charged partisan context surrounding it. Politicians who were trying
to salvage HealthCare.gov and the programs it represented were competing for political attention with
those who wanted them to fail. 32
As a result of the problematic Healthcare.gov launch on October 1, congressional committee hearings
with representation from both Democratic and Republican party members were called, and GCI
executives and senior administration officials were required to testify. The committee heard the following
testimony from Campbell: 33
CMS serves the important role of systems integrator or “quarterback” on this project and is the
ultimate responsible party for the end-to-end performance of the overall federal exchange . . .
Basically, it’s the government’s fault. We just build the damn thing. If they didn’t tell us to build
the right thing, or test it properly, well, it’s their fault. Also, someone else we won’t name is
really at fault. . . . Unfortunately, in systems this complex with so many concurrent users, it is not
unusual to discover problems that need to be addressed once the software goes into a live
production environment. This is true regardless of the level of formal end-to-end performance
testing — no amount of testing within reasonable time limits can adequately replicate a live
29
A. Jeffries, “Only Six People Managed to Enroll in Health Insurance on Healthcare.gov's First Day,” The Verge, November
1,
2013,
www.theverge.com/2013/11/1/5054302/only-six-people-managed-to-enroll-in-health-insurance-on-healthcare,
accessed February 10, 2014.
30
Healthcare.gov, “Why Projects Fail,” October 22, 2013, http://calleam.com/WTPF/?p=6061, accessed January 20, 2014.
31
C. Weaver et al., “Software Design Defects Cripple Healthcare Website,” The Wall Street Journal, October 6, 2013,
http://online.wsj.com/news/articles/SB10001424052702304441404579119740283413018, accessed December 28, 2013.
32
“How
Would
You
Fix
Healtcare.gov?”
Information
Week,
December
4,
2013,
www.informationweek.com/healthcare/leadership/how-would-you-fix-healthcaregov/d/d-id/1112896, accessed January 20,
2014.
33
“Contractors Who Built Healthcare.gov Website Blame Each Other for All the Problems,” techdirt,
www.techdirt.com/articles/20131023/18053424992/contractors-who-built-healthcaregov-website-blame-each-other-allproblems.shtml, accessed February 3, 2014.
Page 263 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
A House Oversight Committee document indicated that just six people were able to successfully enroll in
health insurance through the website during the first 24 hours, though 4.7 million people had visited the
site in this time. Only 242 people were able to enroll on the second day. 29
Page 9
9B14E005
environment of this nature. . . . Another contractor was awarded the contract for the data services
hub portion of the federal exchange.
It is relevant to note that the EIDM tool is only one piece of the federal marketplace’s registration
and access management system, which involves multiple vendors and pieces of technology.
While the EIDM plays an important role in the registration system, tools developed by other
vendors handle critical functions such as the user interface, the e-mail that is sent to the user to
confirm registration, the link that the user clicks on to activate the account and the web page the
user lands on. All these tools must work together seamlessly to ensure smooth registration . . .
WHAT TO DO?
One thing was certain: the website launch challenges would continue to feature in the daily news until
Sebelius could demonstrate to congressional leaders, political reporters, website users and her boss that a
“gonna get fixed” plan had in fact been developed and was successfully being executed. From the
perspective of an outsider looking in, the situation appeared chaotic. Everyone had an opinion to share,
and all forms of news media were adding to the hype, with some recommending a restart and others
arguing that the site could be fixed with a little more time. Taking a step back from the political rhetoric
and focusing on the core items, the next move was not clear. Wasn’t this supposed to be a “simple”
website design and implementation project?
Page 264 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Campbell’s reference to “another contractor” was to QSSI/Optum, which built the data services hub as
well as the enterprise identity management (EIDM) functions. QSSI/Optum’s executive vice-president,
Andy Slavitt, testified that the data service hub worked perfectly at all times, though he did admit to
EIDM having issues. In turn, Slavitt pointed to other vendors who were to blame:
Page 10
9B14E005
Source: K.A. Davidson, “The Most Efficient Health Care Systems in the World,” The Huffington Post, August 29, 2013,
www.huffingtonpost.com/2013/08/29/most-efficient-healthcare_n_3825477.html, accessed February 6, 2014.
EXHIBIT 2: U.S. CENSUS DATA CONCERNING HEALTH INSURANCE
Population (millions)
People with health insurance (some overlapping coverage)
… Employer, spouse’s employer, or parent (181.5)
… Government programs (85.1)
… Purchased individually (27.2)
People without any form of health insurance [see Exhibit 3]
People covered by Medicaid
People covered by Medicare
Children under age 18, living in poverty, without health insurance
Immigrants (legal and illegal) without health insurance
306.1
256.2
49.9
48.6
44.3
11.5
12.4
Source: C. DeNavas-Walt et al., “Income, Poverty and Health Insurance Coverage in the United States: 2010,” U.S. Census
Bureau, September 13, 2011, www.census.gov/prod/2011pubs/p60-239.pdf, accessed December 22, 2013.
Page 265 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
EXHIBIT 1: WORLD HEALTH CARE SYSTEM EFFICIENCY
Page 11
9B14E005
Source: C. DeNavas-Walt, B.D. Proctor and J.C. Smith, “Income, Poverty, and Health Insurance Coverage in the United
States: 2010,” U.S. Census Bureau: Current Population Reports, U.S. Government Printing Office, Washington, DC,
www.census.gov/prod/2011pubs/p60-239.pdf, September 13, 2011, p. 23, accessed March 2, 2014.
Page 266 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
EXHIBIT 3: UNINSURED AMERICANS
Page 12
9B14E005
Company
ABT Associates
Accenture
A. Reddix & Associates
BearingPoint
Booz Allen Hamilton
CDM Group
CGI Federal
Client/Server Software Solutions (CSSS.net)
Computer Sciences Corp.
Computing Solutions
David-James LLC
Deloitte Consulting
Genova Technologies
George Washington University
H.S.I. Network
Humanitas Inc.
ICP Systems LLC
Information Systems Consulting Group Inc.
International Business Machines
Intertribal Council of Arizona
IQ Solutions Inc.
JSI Research and Training Institute
KAT Video Productions
Macro International Corp.
Maximus Federal Services Inc.
McKinsey & Company
Mitre Corp
National Opinion Research Center
Northrop Grumman
Porter Novelli Public Strategies
PricewaterhouseCoopers
Quality Software Services Inc (United Health Group)
Rand Corporation
Research Triangle Institute
Science Applications International Corp.
Sentel Corp
Serco
Social and Scientific System Inc.
Soft-Con Enterprises
Summit Consulting
Thomson Reuters Healthcare Inc. (now Truven
Health)
Unicom Logistics
University of California (SF)
Urban Institute
Vangent (General Dynamics)
Verizon Business Network Services
Weber Shandwick
Westcott, John
TOTAL
Contract Award
$971,322
$2,136,176
$0
$251,427
$2,668,754
$0
$93,700,000
$3,880,000
$4,024,384
$7,802,076
$7,283,208
$12,921,094
$1,261,259
$51,274
$137,340
$33,837
$499,952
$6,270,789
$4,999,999
$97,500
$520,000
$15,500
$106,182
$2,584,665
$43,163,074
$13,767,707
$2,234,783
$297,889
$1,666,812
$11,670,603
$284,150
$68,339,812
$1,044,531
$404,255
$1,772,132
$5,487,434
$114,307,266
$293,280
$0
$1,090,753
$0
$6,270,789
$12,000
$1,988,575
$28,237,831
$1,193,916
$3,477,364
$24,599
$459,246,293
Government Agency
HRSA
IRS & CMS
HHS
CDC
IRS & CMS
Office of Asst. Sec. for Health Except Centers
CMS
IRS
Office of Asst. Sec. for Health Except Centers
CMS
CMS
IRS & CMS
CMS
CMS
CMS
HRSA
CMS
IRS
IRS
Indian Health Service, HHS
Office of Asst. Sec. for Health Except Centers
Office of Asst. Sec. for Health Except Centers
CMS
CDC
CMS
CMS
IRS
Office of Asst. Sec. for Health Except Centers
IRS
CMS
CMS
IRS
Office of Asst. Sec. for Health Except Centers
Office of Asst. Sec. for Health Except Centers
IRS
IRS
CMS
CDC
IRS
CMS
IRS
Office of Asst. Sec. for Health Except Centers
Office of Asst. Sec. for Health Except Centers
CMS
IRS
Office of Asst. Sec. for Health Except Centers
HRSA
Source: B. Allison, “Good Enough for Government Work? The Contractors Building Obamacare,” Sunlight Foundation,
October 9, 2013, http://sunlightfoundation.com/blog/2013/10/09/aca-contractors/, accessed January 30, 2014.
Page 267 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
EXHIBIT 4: ACA CONTRACTORS
Page 13
9B14E005
Source:
J.
Woo,
“The
Biggest
Problem
with
the
ACA
Federal
Exchange,”
Value
www.valuepenguin.com/2013/10/biggest-problem-federal-exchange, October 2013, accessed March 8, 2014.
Page 268 of 282
Penguin,
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
EXHIBIT 5: WEBSITE ARCHITECTURE
Page 269 of 282
EXHIBIT 6: HEALTH CARE SYSTEM COMPLEXITY
9B14E005
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Source: D. Leeper, “Obamacare’s Fatal Flaws: Complexity and Central Planning”, Tea Party Tribune, www.teapartytribune.com/2012/07/06/obamacares-fatal-flaws-complexityand-central-planning/, July 6, 2012, accessed March 2, 2014.
Page 14
9 -6 0 1 -1 1 4
REV: FEBRUARY 28, 2018
The iPremier Company (A):
Distributed Denial of Service Attack
January 12, 2018, 4:31 AM
Somewhere a phone was chirping. Bob Turley, CIO of the iPremier Company, turned beneath the
bed sheets, wishing the sound would go away. Lifting his head, he tried to make sense of his
surroundings. Where was he?
The Westin in Times Square. New York City. That’s right. He was there to meet with Wall Street
analysts. He’d gotten in late. By the time his head had hit the pillow it was nearly 1:30 AM. Now the
digital display on the nearby clock made no sense. Who would be calling at this hour? Why would
the hotel operator put a call through?
He reached for the phone at his bedside and held it to his ear. Nothing. The chirping was coming
from his mobile. Staggering out of bed, he located the noisy phone and opened the call.
“This is Bob Turley.”
“Mr. Turley?” There was panic in the voice. “I’m sorry to wake you, Joanne told me to call you.”
“Who is this?”
“It’s Leon. Ledbetter. I’m in Ops. We met last week. I’m new. I mean, I was new, last month.”
“Why are you calling me at 4:30 in the morning, Leon?”
“I’m really sorry about that Mr. Turley, but Joanne said—“
“No, Leon, I mean tell me what’s wrong.”
“It’s our website, sir. It’s locked up. I’ve tried accessing it from three different computers and
nothing’s happening. Our customers can’t access it either; the help desk is getting calls.”
“What’s causing it?”
Professor Robert D. Austin, Dr. Larry Leibrock (Chief Technology Officer, McCombs School of Business, University of Texas at Austin), and Alan
Murray (Chief Scientist, Novell Service Provider Network) prepared this case. This revised version was prepared by HBS Emeritus Professor
Richard L. Nolan, Professor Robert D. Austin (Ivey Business School), and Professor Michael Parent (Beedie School of Business, Simon Fraser
University). HBS cases are developed solely as the basis for class discussion. Cases are not intended to serve as endorsements, sources of primary
data, or illustrations of effective or ineffective management. The situation described in this case is based on real accounts of denial of service attacks
directed against several companies during 2000 and 2001. Company names, product/service offerings, and the names of all individuals in the case
are fictional, however. Any resemblance to actual companies, offerings, or individuals is accidental.
Copyright © 2001, 2002, 2003, 2005, 2007, 2018 President and Fellows of Harvard College. To order copies or request permission to reproduce
materials, call 1-800-545-7685, write Harvard Business School Publishing, Boston, MA 02163, or go to www.hbsp.harvard.edu. This publication
may not be digitized, photocopied, or otherwise reproduced, posted, or transmitted, without the permission of Harvard Business School.
Page 270 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
ROBERT D. AUSTIN
601-114
The iPremier Company (A): Distributed Denial of Service Attack
“Joanne thinks—if we could only—well, someone might have hacked us. Someone else might be
controlling our site. Support has been getting these e-mails—we thought it was just the web server, but
I can’t access anything over there. Joanne is on her way to the data center. She said to call you. These
weird e-mails, they’re coming in about one per second.”
“They say ‘ha.’”
“Ha?”
“Yes, sir. Each one of them has one word in the subject line, ‘ha.’ It’s like ‘ha, ha, ha, ha.’ Coming
from an anonymous source. That’s why we’re thinking—.”
“When you say they might have hacked us—could they be stealing customer information? Credit
cards?”
“Well, I guess no firewall 1—Joanne says—actually we’re using a firewall service we purchase from
the hosting company, so—.”
“Can you call someone over there? We pay for monitoring 24/7, don’t we?”
“Joanne is calling them. I’m pretty sure. Is there anything you want me to do?”
“Have we set our emergency procedures in motion?
“Joanne says we have a binder, but I can’t find it. I don’t think I’ve ever seen it. I’m new—“
“Yes, I got that. Does Joanne have her cell?”
“Yes sir, she’s on her way to the data center. I just talked to her.”
“Call me back if anything else happens.”
“Yes sir.”
Turley stood up, realizing only then that he had been sitting on the floor. His eyes were bleary but
adrenaline was now pumping in his bloodstream. Steadying himself against a chair, he felt a wave of
nausea. This was no way to wake up.
He made his way to the bathroom and splashed water on his face. This trip to New York was an
important assignment for someone who had been with the company such a short time. It demonstrated
the confidence CEO Jack Samuelson had in him as the new CIO. For a moment, Turley savored the
memory of the meeting in which Samuelson had told him he would be the one to go to New York. As
that memory passed another emerged, this one from an earlier session with the CEO. Samuelson was
worried that the company might eventually suffer from “a deficit in operating procedures.” “Make it
one of your top priorities,” he had said. “We need to run things professionally. I’ve hired you to take
us to the next level.”
1 A “firewall” is a combination hardware/software platform that is designed to protect a local network and the computers that
reside on it against unauthorized access.
2
Page 271 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
“What do the e-mails say?”
The iPremier Company (A): Distributed Denial of Service Attack
601-114
Looking himself over in the mirror, seeing his hair tussled and face wet, Turley lodged a protest
with no one in particular: “I’ve barely been here three months!”
Founded in 1996 by two students at Swarthmore College, the iPremier Company had evolved into
a web-based commerce success story. From its humble beginnings, it had risen to become one of the
top two retail businesses selling luxury, rare, and vintage goods on the web. Based in Seattle,
Washington, the firm had grown and held off incursions into its space from a number of well-funded
challengers. For the fiscal year 2017, profits were $20.1 million on sales of $320 million. Sales had grown
at more than 20% annually for the last three years, and profits, though thin, had an overall favorable
trend.
Immediately following its IPO in late 1998, the company’s stock price had nearly tripled. It had
continued up from there amid the euphoria of the 1999 markets, eventually tripling again. A follow-on
offering had left the company in a strong cash position. During the NASDAQ bloodbath of 2000, the
stock had fallen dramatically but had eventually stabilized and even climbed again, although not to
pre-2000 levels. In the decade plus since, the company had held its own and consolidated its leading
market position, enjoying better-than-average returns by streamlining and focusing its business to
achieve profitability.
Most of the company’s products were priced at a few hundred dollars, but there were a small
number of items priced in the thousands and tens of thousands of dollars. Customers paid for items
using their credit cards. The company had flexible return policies, which were intended to allow
customers to thoroughly examine products before deciding whether to keep them. The iPremier
customer base was high-end—so much so that credit limits on charge cards were rarely an issue, even
for the highest-priced products. Trust was critical to this relationship. Customers had to believe and
trust that the goods sold by iPremier were genuine. Otherwise, they could easily purchase the same
sorts of goods from a number of other websites, including iPremier’s fiercest competitor, MarketTop.
iPremier’s competitive advantage lay not necessarily in its array of goods, but more in its responsive
and attractive website, order fulfillment, and after-sales service. iPremier led its industry segment in
the quality of the “user experience” and constantly innovated to provide the best, and most seamless
service. As a result, the company had over one million regular customers in its database, and another
few hundred thousand casual buyers.
Management and Culture
The management team at iPremier was a mix of talented younger people who had been with the
company for a long time, and more experienced managers who had been gradually hired as the firm
grew. Recruitment had focused on well-educated technical and business professionals with reputations
for high performance. Getting hired into a senior management position required excelling in an intense
series of three-on-one interviews. The CEO interviewed every prospective manager at the director level
and above. The reward, for those who made the grade, was base compensation above the average of
managers at similar firms, and variable compensation, mainly in the form of stock options, that could
be a significant multiple of the base. All employees were subject to quarterly performance reviews that
were tied directly to their compensation. Unsuccessful managers did not last long. Most managers at
iPremier described the environment as “intense.”
Throughout the company, there was a strong commitment to doing “whatever it takes” to get
projects done on schedule and on budget, especially when it came to system features that would benefit
3
Page 272 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
The iPremier Company
601-114
The iPremier Company (A): Distributed Denial of Service Attack
customers. The software development team was proud of its record of consistently launching new
features and programs a few months ahead of MarketTop. Senior managers understood that their
compensation and prospects with the company depended on executing to plan. They pursued “the
numbers” with obsessive zeal.
The company had historically outsourced management of its technical architecture and had a longstanding relationship with Qdata, a company that hosted most of iPremier’s computer equipment and
databases, and provided connectivity to the Internet. Qdata was an early entrant into the Internet
hosting business, but it had been battered by the contraction of the Internet bubble and lost any
prospect of market leadership. Its data center was physically proximate to the corporate offices of
iPremier; some felt there was little else to recommend it. The company had not been quick to invest in
advanced technology and had had trouble retaining staff.
The iPremier Company had a long-standing initiative aimed at eventually moving its computing to
an internal facility, but several factors had kept this from happening. First, and most significant,
iPremier had been very busy growing, protecting its profits, and delivering new features to benefit
customers; hence the move to a better facility had never quite made it to the top of the priority list.
Second, the cost of more modern facilities was considerably higher—two to three times as expensive
on a per-square-foot basis. Third, there was a perception that a move might risk service interruption to
customers. Finally, one of the founders of iPremier felt a personal commitment to the owners of Qdata
because they had been willing to renegotiate their contract at a particularly difficult time in iPremier’s
very early days.
4:39 AM
Sitting at the hotel room desk, Turley began scrolling through the phonebook on his phone. Before
he could find the number for Joanne Ripley—his technical operations team leader—she called him.
“Well, Joanne. How are you this morning?”
A cautious laugh came from the other end of the call. “About the same as you, I’m guessing.
I assume Leon reached you.”
“He did, but he doesn’t know anything. What’s going on?”
“I don’t know much either, yet. I’m in the car, on my way to the data center. I ought to be there in
five minutes.”
“How long after that until we are back up and running?”
“That depends on what’s wrong. I’ll try restarting the web server as soon as I get there, but if
someone has penetrated our databases and stolen customer data, getting the server running will be the
least of our worries. Did Leon tell you about the e-mails?”
“The ‘ha, ha’ e-mails? Yeah. Makes it sound like something deliberate.”
“I’d have to agree.”
“No chance it’s a simple DDoS attack?”
4
Page 273 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Technical Architecture
The iPremier Company (A): Distributed Denial of Service Attack
601-114
“I doubt it’s a simple DDoS attack; we’ve got software to deal with those.”
“Not soon enough. They’re coming through an anonymizer that’s probably in Europe or Asia. If
we’re lucky we’ll find out sometime in the next decade who sent them. Then we’ll discover they’re
originating from some laptop or smart thermometer in Podunk, Idaho, and their owner has no idea
they’ve been compromised by hackers.”
“What are the chances they’re stealing credit cards? I know we don’t keep credit card numbers on
our database, but they could be stealing other sensitive information, right?”
Ripley paused before answering: “There’s really no way of knowing.”
“Should we pull the plug? Physically disconnect the communications lines?”
“If we start pulling cables out of the wall it may take us a while to put things back together.”
“Joanne, don’t we have emergency procedures for times like this? I don’t think I’ve seen it but it
comes up when people mention our business continuity plan (BCP).”
“We’ve got a BCP binder,” said Ripley. “I’ve got a copy with me. Keep it in my car. There’s one at
the office too, and we store it electronically on our shared drive. But to be honest, well—it’s out of date,
and we don’t really train people with it because of that. Lots of people on the call lists don’t work here
anymore. I don’t think we can trust the phone numbers and I know some of the technology has changed
since it was written. We’ve talked about practicing incident response but we’ve never made time for
it.”
“A Disaster Recovery Plan (DRP)? An Incident Response Plan (IRP)?” Turley was incredulous. It
boggled his mind, and created more than a little career-anxiety that he hadn’t thought to check these
since his arrival. He’d assumed that, as a publicly-listed company, iPremier had to have such plans.
But now was not the time, he decided, to grill Ripley about it. So he changed the subject: “What’s
the plan when you reach the data center?”
“Let me restart the web server and see what happens. Maybe we can get out of this without too
much customer impact.”
Turley thought about it for a moment. “Okay. But if you see something that makes you think
customer records or other information are being stolen, I want to know that immediately. We may have
to take drastic action.”
“Understood. I’ll call you back as soon as I know anything.”
“Good. One more thing: Who else knows this is going on?”
“I haven’t called anyone else. Leon might have. I’ll call him and call you right back.”
“Thanks.”
Turley disconnected. Just as he did so, his phone rang again.
5
Page 274 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
“Can we track the e-mails?”
601-114
The iPremier Company (A): Distributed Denial of Service Attack
“Damn.” It was Warren Spangler, VP of business development. Turley recalled vaguely that
Warren and Leon’s father were college buddies or something. Ledbetter had almost certainly called
Spangler.
“Hi, Warren,” said Turley.
“Something’s definitely going on, but we’re not sure what yet. We’re trying to minimize customer
impact. Fortunately for us, it’s the middle of the night.”
“Wow. So is it just a technical problem or is somebody actually doing it to us?”
Turley was eager to call the chief technology officer (CTO), so he didn’t really have time for this
discussion. But he didn’t want to be abrupt. He was still getting to know his colleagues.
“We don’t know. Look, I’ve got to—“
“Leon said something about e-mails—“
“Yes, there are suspicious e-mails coming in so it could be someone doing it.”
“Oh, man. I bet the stock takes a hit tomorrow. Just when I was going to exercise some options.
Shouldn’t we call the police?”
“Sure, why don’t you see what you can do there, that’d be a big help. Look, I’ve got to—“
“Seattle police? Do we know where the e-mails are coming from? Maybe we should call the FBI?
No. Wait. If we call the police, the press might hear about this from them. Whoa. Then our stock
would really take a hit.”
“I’ve really got to go, Warren.”
“Sure thing. I’ll start thinking about PR. We got you covered here, bro. Keep the faith.”
“Will do, Warren. Thanks.”
Turley ended that call and began searching through his cell phone’s memory to find the number for
Tim Mandel, one of iPremier’s co-Founders and now the company’s CTO. He and Mandel had already
cemented a great working relationship. Turley wanted his opinion. Just as Turley was about to initiate
the call, though, another call came in from Ripley.
Turley answered the phone and said: “Leon called Spangler, I know. Anything else?”
“Ah, no. That’s it for now. Bye.”
Turley dialed Mandel. At first the call switched over to voicemail, but he retried immediately. This
time Mandel answered sleepily. It took five full minutes to wake Mandel and tell him what was
happening.
“So what do you think, should we just pull the plug?” Turley asked.
“I wouldn’t. You might lose some logging data that would help us figure out what happened.
“I’m not sure knowing exactly what’s happening is the most important thing to me right now.”
6
Page 275 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
“Hi, Bob. I hear we’ve got some kind of incident going on. What’s the story?”
The iPremier Company (A): Distributed Denial of Service Attack
601-114
Turley heard a thumping sound, as if Mandel had fallen getting out of bed; his phone clattered as
it impacted something, the floor perhaps. A scant moment later, Mandel came back on the line and
continued: “Come to think of it, Bob, preserving the logs is irrelevant because I’m pretty sure detailed
logging is not enabled. Detailed logging adds a performance penalty of about 20%. Someone
somewhere at some point decided that unacceptably impacts the customer experience.”
“So we aren’t going to have evidence of what happened anyway.”
“There’ll be some, but not as much as we, or the FBI, will want.”
Another call was coming in.
“Hold on, Tim.” Turley kicked the phone over to the waiting call. It was Peter Stewart, the
company’s legal counsel. What was he doing awake?
“This is Turley.”
“Hey, Bob, it’s Pete. Pull the plug, Bob. Shut off the power, pull the cords out of their sockets, go
dark, kill it…everything. We can’t risk having PII (Personally Identifiable Information) stolen.”
“Spangler call you?”
“Huh? No, Jack. Samuelson. He called three minutes ago, said hackers had control of our web site
and were stealing information. Told me in no uncertain terms to call you and ‘provide a legal
perspective.’ That’s exactly what he said: ‘provide a legal perspective.’”
So the CEO was awake. The result, no doubt, of Spangler’s “helping” from that end. Stewart
continued to speak legalese at him for what seemed like an eternity. By this time, Turley was incapable
of paying attention to him.
“Thanks for your thoughts, Pete. I’ve got to go, I’ve got Tim on the other line.”
“Okay. For the record, though, I say pull the plug. I’ll let Jack know you and I spoke, and will write
a memo to file reflecting this conversation and my advice.”
“Thanks, Pete,” said Turley, acerbically.
Turley switched back over to the call with Mandel.
“Spangler’s got bloody everybody awake, including Jack. I recommend you get dressed and head
into the office, my friend.”
“Is Joanne on this?”
“Yes, she’s at Qdata by now.” Turley’s phone rang. “Got a call coming in from her now.”
He switched the phone.
“What’s up Joanne?”
7
Page 276 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
“I suggest you change your mind about that. If you don’t know what happened this time, it can
happen again. And, if you don’t know what happened, you won’t know what, if anything, you need
to disclose publicly. We might also need to preserve evidence of what has happened. A DDoS attack is
a federal crime, and we might eventually need to involve the FBI.”
601-114
The iPremier Company (A): Distributed Denial of Service Attack
“I’m in New York, Joanne. I’ve got no Qdata contact information with me. But let me see what I
can do.”
“Okay. I’ll keep working it from this end. The security guard doesn’t look too fierce. I think I could
take him.”
“Do what you can.”
Turley hung up. He noticed that Mandel had disconnected also. For a moment, Turley sat back in
the chair, not sure what to do next.
5:27 AM
The phone rang again, and Turley could see from Caller ID that it was the call he had been dreading:
Jack Samuelson, the CEO.
“Hi Jack.”
“Bob. Exciting morning?”
“More than I like it.”
“Are we working a plan?”
“Yes, sir. Not everything is going according to plan, but we are working a plan.”
“Bob, the stock is probably going to be impacted and we’ll have to put a solid PR face on this, but
that’s not your concern right now. You focus on getting us back up and running. Understand?”
“I do.”
Samuelson hung up abruptly.
That had gone better than Turley had feared. He avoided the temptation to analyze Samuelson’s
every word for clues to his innermost thoughts. Instead, he called Ripley.
“Hi, Bob,” she said, sounding mildly cheerful. “They let me in. I’m sitting in front of the console
right now. It looks like a SYN flood from multiple sites directed at the router that runs our firewall
service. So it is a DDoS attack. By the way, this is not a proper firewall, Bob; we need to work on
something better.” (Exhibit 1 explains the different types of Denial of Service attacks).
“Fine, but what can we do right now?”
2 The “Network Operations Center” is the control room from which production computer operations and networks are
monitored and operated.
8
Page 277 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
“They won’t let me into the NOC2,” she said angrily. There’s no one here who knows anything
about the network monitoring and that’s what I need to use to see the traffic coming into our site. The
Qdata guy who can do it is vacationing in Aruba. I tried rebooting the web server, but we’ve still got
a problem. My current theory is an attack directed at our firewall, but to be sure I’ve got to see the
packets coming in, and the firewall is their equipment. You got an escalation contact to get these dudes
off their butts?”
The iPremier Company (A): Distributed Denial of Service Attack
601-114
“Well, looks like the attack is coming from about 3000 sites. If the guys here will let me, I’m going
to start shutting down traffic from those sites. I’ll have to set the phone down for a minute.”
There was a pause of a couple of minutes. Turley heard some muffled conversation in the
background, rapid keyboard clicks, then several epithets. Ripley came back on the line.
“You’re going to have to translate that one for me, Ripley.”
“Every time we shut down traffic from one address, the zombie we’ve shut off automatically
triggers attacks from two other sites. I’ll try it a few more times, but right now it looks like that’s just
going to make things worse. My guess is the hackers are using a ‘bot net of enslaved machines.”
“If it’s a DDoS, they haven’t hacked us, right? It means it’s not an intrusion. They haven’t gained
entry to our system. So customer data are safe. Can we say that?” Turley was especially worried in
light of recent, gigantic data breaches, and the ensuing class-action lawsuits they provoked (see Exhibit
2).
“There’s nothing that makes a DDoS attack and an intrusion mutually exclusive, Bob.”
Turley knew this, but had hoped otherwise in a moment of wishful weakness. Hearing Ripley
remind him of the facts strengthened a growing, nauseating storm in his stomach. “I’ll let you get back
to it. Call me with regular updates."
Turley hung up and thought about whether to call Samuelson and what to tell him. He could say
that it was a DDoS attack. He could say that the attack, by itself, was not evidence that customer
information was at risk. But Turley wanted to think some more before he went on record.
Before he could do anything else, his cell phone rang again. It was Ripley.
“It stopped,” she said excitedly. “The attack is over.”
“What did you do?”
“Nothing. It just stopped. The attack just stopped at 5:46 AM.”
“So—what now?”
“The website is running. A customer who visits our site now wouldn’t know anything had ever
been wrong. We can resume business as usual.”
“Business as usual?”
“I’d recommend that we shut down, or at least disconnect from the public Internet, and give
everything a proper going-over. In the longer run, we’ll need to conduct a thorough forensic audit to
ensure nothing else bad has happened. I’ve been thinking about how they targeted the firewall, and I
don’t think it sounds like script kiddies. With your approval, I’d like to reach out to some cybersecurity
consultants and get them in ASAP.”
9
Page 278 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
“Damn it, Bob, they’re spawning zombies. It’s Dawn of the Dead out there.”
601-114
The iPremier Company (A): Distributed Denial of Service Attack
What to Recommend
Shutting down to conduct a thorough forensic audit seemed like a prudent course, but it was
unclear how long that would take. In such time, iPremier’s customers could flee to a competitor.
iPremier would have to explain a shutdown, and, for legal reasons, they’d probably have to admit that
such a precaution was motivated by concerns about a data breach.
Shutting down, then, could freak out customers, sink the stock, even kill the company. And,
Stewart’s memo notwithstanding, there were plenty of good arguments to keep the business up and
running. iPremier had done nothing to provoke the attack, and there was – as of this moment anyway
– no actual evidence of an intrusion or breach. Turley knew that DDoS attacks were a daily occurrence,
and that the bigger players like Amazon, Apple, Google, Yahoo and Microsoft were being attacked
constantly. It was usually just a cost of doing business in this space, not a material event.
But – Ripley had observed that this seemed like a particularly sophisticated DDoS attack. And he
realized, now, that the company had been lax in deploying and protecting its systems – the very thing
he’d been hired to do. His new job honeymoon was over – he’d have to obtain the resources to secure
company operations.
On the one hand, there was no evidence-based reason to shut down. Indeed, doing so could be
considered an irresponsible overreaction. After all, iPremier’s corporate officers were first and foremost
responsible to iPremier shareholders.
On the other hand, if customer data had been stolen, and if the site was insecure, keeping the site
running could lead to more mischief by hackers – and kill the company in a different way, by exposing
it to reputation damage, liability, and lawsuits.
Turley knew that the company’s senior management team would be conflicted and angry about all
this, and that he would have to make a recommendation soon. He guessed that Peter Stewart (Legal)
and, probably, Joanne Ripley (Tech Ops) would want to shut the site down for an indefinite period,
until such time as they would be reasonably assured that no data had been taken, or no ticking time
bomb had been planted within iPremier systems. But he could easily imagine that some members of
the senior management team would object to a shutdown; they would argue that until demonstrated
otherwise, it should be assumed that nothing bad had happened beyond what they already knew –
that iPremier had ‘dodged a bullet’.
Samuelson and the Board of Directors would be looking to Turley for guidance, and his
recommendation would have a profound impact on the future of the company.
10
Page 279 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Post attack, Turley realized that he immediately faced a new decision: Whether to recommend
shutting down – or, at least, disconnecting from the Internet as a precaution – while they figured out
what had happened. Doing either would shut down normal business operations.
The iPremier Company (A): Distributed Denial of Service Attack
Exhibit 1
601-114
Denial of Service Attacks Explained
Each such “conversation” with a web server begins with a sequence of “handshake” interactions.
The initiating computer first sends a “SYNCHRONIZE” or “SYN.” The contacted web server responds
with a “SYNCHRONIZE-ACKNOWLEDGE” or “SYN-ACK.” The initiating computer then completes
the handshake with an “ACKNOWLEDGE” or “ACK.” A “SYN flood” is an attack on a web server
intended to make it think a very large number of “conversations” are being initiated in rapid
succession. Because each interaction looks like real traffic to the website, the web server expends
resources dealing with each one. By flooding the site, an attacker can effectively paralyze the web server
by trying to start too many conversations with it. This is the essence of a Denial of Service, or DoS
attack.
In its simplest form, an attacker uses a single computer to send many requests in rapid succession.
Because these types of attacks (single source) can be more easily traced, they are seldom used, except
for the most unskilled of script kiddies.
More sophisticated hackers engage in Distributed denial-of-service (DDoS) attacks, such as
described in the case, where many requests come in rapid succession from many computers. This might
include the use of “Botnets”, large clusters of Internet-enabled devices such as cellphones, computers,
and even smart devices like thermometers, that have been infected with malware, allowing hackers to
control these devices – sometimes called “zombies” – remotely. In one recent example, the Mirai Botnet
was used to attack Internet service company Dyn in October 2016.
11
Page 280 of 282
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
Keying-in a website’s URL (Uniform Resource Locator), or web address on a web browser or search
engine’s input line begins a conversation with the web server that will eventually return, or send the
requestor the web page requested.
601-114
The iPremier Company (A): Distributed Denial of Service Attack
Exhibit 1
Denial of Service Attacks Explained (continued)
Source:
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
The most sophisticated denial-of-service attacks target other parts of source codes used by
companies’ websites. In one advanced case, many computers send requests to many other computers.
However, by using IP (Internet Protocol) spoofing, the source address of these many other computers
is set to the target’s address, causing replies to go to the target address and flood it. This is called a
Distributed Reflected denial-of-service, or DrDoS, attack.
Casewriters.
12
Page 281 of 282
The iPremier Company (A): Distributed Denial of Service Attack
Exhibit 2
601-114
A History of Large Data Breaches
Yahoo2
1000
Yahoo
500
MySpace
For use only in the course Leveraging Information Technology at Ivey Business School from 12/15/2024 to 3/5/2025.
Use outside these parameters is a copyright violation.
427
eBay
145
Equifax
143
LinkedIn
117
VK
101
AOL
92
Sony PSN
77
Dropbox
69
Tumblr
65
0
0
200
400
600
800
1000
1200
Millions of Records Compromised
Source:
Casewriters.
13
Page 282 of 282