.
i
^Hiey
HD28
.M414
rvo-
34I(»
COMPUTER-BASED DATA AND
ORGANIZATIONAL LEARNING:
THE IMPORTANCE OF
MANAGERS' STORIES
David K. Goldstein
April
CISR
Sloan
WP
WP
1992
No. 234
No. 3426-92
Center for Information Systems Research
Massachusetts
Institute of
Sloan School of
Technology
Management
77 Massachusetts Avenue
Cambridge, Massachusetts, 02139
COMPUTER-BASED DATA AND
ORGANIZATIONAL LEARNING:
THE IMPORTANCE OF
MANAGERS' STORIES
David K. Goldstein
April
CISR
Sloan
WP
WP
1992
No. 234
No. 3426-92
®1992 D.K. Goldstein
Center for Information Systems Research
Sloan School of Management
Massachusetts Institute of Technology
TSlT. UBRAHIhb
JUL 1 6 1992
R£CfclVtU
J
.:i>mputcr-based data and oraaniz-aliotial Icamini;
COMPUTER-BASED DATA AND ORGANIZATIONAL LEARNING:
THE IMPORTANCE OF MANAGERS' STORIES
While many organizations are investing large amounts of money to provide computerbased data to their managers, little is known about how. or even whether, managers use these
data to learn about the business environment.
This issue
is
explored by examining
how
managers use supermarket scanner data to leam about changes in the
Managers' stories play a central role in the four step process used by
one product management organization as it learns from analyzing computer-based data. First,
findings that contradict one or
a manager examines the data and looks for unexpected results
grocery
product
marketing environment.
—
more of her
stories about the
marketing environment.
If
carries out a relatively unstructured, multi-stage process to
result.
a surprise
make
is
found, the manager
sense out of the unexpected
This process can be viewed as a dialogue between the result and a
manager's disposal (including analyses of computer-based data).
set
of tools
Next, the manager
at the
tells the
and superiors, developing a common understanding.
manager creates an official story, that is used to 'sell' new marketing approaches
people outside the product manager organization
the sales force and supermarket buyers.
story to share her insights with peers
Finally, the
to
—
Keywords: Organizational Learning, Decision Support, Marketing Information. Managerial
Learning, Stories, Storytelling
-2-
.ompuier-based data and organizational Icamina
Introduction
Making sense of
the business environment
has increased
on
their
Victor, 1991).
importance
its
becoming a more important organizational
and technological changes increase environmental complexity and
task, as rapid organizational
dynamism (Boynton and
is
— the
operations captured
by
This task
computer.
their
is
facilitated
Organizations
by the same phenomenon
now have
own, and sometimes
access to detailed data
customers',
their
processing systems.
These data can play an increasingly significant role
learning, by helping
managers both
to better
in
transaction
organizational
These
understand the effect of their actions.
managers can then react more rapidly and accurately
to
changes
that
customer needs or
in
competitor actions.
While many organizations are investing large amounts of money
based data to their managers,
little is
known about how,
data to learn about the business environment.
computer-based data
only
when
investigate
in organizations'
individual
their
how managers
learn
The goal of
learn about the
(Simon,
this
paper
the
managers share
My
insights derive primarily
to explore the role of
Nonaka,
1991;
environment and organize
their insights both within
is
Since organizations learn
then explore the role of computer-based data in managerial learning.
how
provide computer-
or even whether, managers use these
understanding of their world.
members
to
1991),
their
I
will
knowledge.
Finally,
I
will
first
I
will
examine
and outside the organization.
from two sources
— from
studies of cognition in practice
carried out by anthropologists and organizational researchers and from interviews with grocery
product managers.
from study
in situ
Managerial understanding of the environment
a problem that can benefit
— through the examination of how managers use computer-based data as part
of their day-to-day work.
artificial settings
is
Since cognition in practice differs significantly from cognition
(Brown and Duguid. 1991; Lave, 1988),
who have
insights provided
in
by andiropologists
studied problem solving in the workplace and by organizational
(e.g., Orr,
1990)
researchers
who have observed
managerial cognition (e.g., Isenberg, 1986)
in practice will
help us understand the role of computer-based data in managerial learning.
The examination of
product managers
six
relatively small (sales
manufacturer.
research
is
supplemented by interviews carried out with
grocery product manufacturers.
in
from interviews with
this
Most of
the data for this paper
product managers and two information support personnel
come
at Butler',
a
under $2(X) million), independently run subsidiary of a major grocery
These interviews were conducted as part of a larger study of the use ot
computer-based data by product managers
'Company name
at five
grocery manufacturers (Goldstein and Cho,
as well as details about specific products are disgxused.
computer-based data and oreamzational leamine
1991;
Goldstein.
1990;
Goldstein and Zack.
managers used a new data source
what they learned from
scanner data for the
— purchased
The
their analyses.
first
1989).
focused on
TTiey
how
product
the
data collected from supermarket scanners
visit
— and
took place six months after Butler purchased
time and three months after
company-wide state-of-the-brand reviews (presentations
it
began using the data as part of
to senior
management on
the status of
each Butler product).
Product managers are responsible for developing marketing programs (e.g., consumer
promotions) for a particular product line (e.g.,
prices, plan marketing activities,
introduce
Computer-based data play a key role
(McCann, 1986).
Duncan Hines Brownie Mixes).
new product
varieties,
They
set
and monitor competition.
product managers' understanding of their environment
in
These managers integrate information produced by
on
their f~irm
sales,
shipments, and inventory, with purchased data on their and competitors' performance.
The data
available to product
managers greatly increased when
data
became
fifty
U.S. regions on each category (e.g., baking mixes)
available.
These data included
(e.g.,
Butler purchased, for
retail
volume
sales
price,
percentage of merchandise sold
its
in
supermarket scanner
product managers, monthly data for each of
in
in
which the company sold products.
dollars and units, and promotion activity
supermarkets with an end-aisle display or with a large
These data were for each item
or small ad in a supermarket circular).
in
(including both Butler and competitor products) with a unique Universal Product
Duncan Hines Deluxe Brownie Mix).
megabytes of data.
By analyzing scanner
price,
They could
data, product
in
(e.g.,
in
the
marketing
4Ps of the marketing mix
— on sales for their and their competitors'
The Role of Stories
products.
Managerial Learning
Researchers have proposed several frameworks to describe
understanding of their world.
Code
managers could increase the accuracy
better understand the impact of the
promotion, place, and product
category
Butler product managers could access hundreds of
and timeliness of the feedback they receive about causal relationships
environment.
the
how
people organize their
People have scripts (Schank and Abelson, 1977)
that contain
event sequences for frequently occurring actions, for example ordering a meal in a restaurant.
They have schema (Minsky, 1975)
based on
its
In addition,
attributes, for
that hierarchically structure
example the
characteristics of birds
knowledge about a given topic
and of a specific type of
people have stories to organize their knowledge about experience (Schank, 1991).
Stories are a type of narrative that contain settings, characters, a plot
or conflict
bird.
and
its
resolution
(Bower,
agglomeration of several incidents.
1976).
They can descnbe one
Stories also have a moral
-4-
— with a problem
—a
rule,
incident
lesson,
or
an
theme or
computer-based data and oreanizatiotial ieamina
inference
general
—
that
When
understand current events (Martin, 1982).
story (and
from
has derived
person
the
moral
This
story.
used
is
to
a person hears a story, he tnes to match that
moral) to one he already knows.
its
the
he tlnds a match, he believes that he
If
understands the story (Schank, 1991).
There
ample evidence
is
They use
stories.
For example,
people maintain some of their work-related knowledge as
stories to understand their
senior
comparing them
that
world both within and outside the organization.
managers develop a plausible
to stories of specific previous events (Isabella,
it
knows about previous
to stories she
events
by
1990; Isenberg,
1986).
In
announcement of a new product, an
assessing the reasoning behind a competitor's early
executive relates
of current
understanding
She
early product announcements.
might remember a story about a comparable announcement by another company one year
The
earlier.
was
lesson of that story
announcement was a market
that the early
signal.
Therefore,
she would determine that the current announcement served the same purpose.
Similarly,
photocopier technicians recall stories of previous repairs as they diagnose and
correct copier malfunctions (Orr,
Midwives
in the
Yucatan
recall
Stories are also used by workers in other cultures.
1990).
and
retell stories
of births to determine
how
handle difficult
to
deliveries (Jordan, 1989).
my
In
changes
in the
interviews with product managers,
4Ps of the marketing mix on
they recalled stories about the impact of
sales for their
For
and competitors' products.
example, a manager might recall a story about a $10 per case wholesale price reduction, whose
setting
was
the Phoenix region last September.
The
story's characters
would be consumers,
supermarket buyers, competitors, and members of the organization's sales force.
would be a description of
the rationale behind the price reduction (was
competitor's action?) and the reaction to
communicate the change and
it
be
a
that
$10 price drop
(how did
in
it
Phoenix
response to a
by buyers (how much did they purchase
and did they pass the savings onto consumers?),
reduction?), and by consumers
in
plot
by the sales force (how aggressively did they
rationale to buyers?),
its
it
Its
by competitors (did they
affect retail sales?).
should
only
be
The moral of
used
to
match
the story
increase
the
might
purchases
by
supermarkets, since they do not pass the savings onto consumers.
As a second example, consider
company
this story told
that recently introduced a line that directly
[The competitive
line]
bad
by a Butler product manager about a
competed with her product:
better distribution.
This
was
their Gist refrigerated
product and they paid top dollar to get into the dairy case, but
having problems.
Philadelphia.
now
they're
They are losing distribution in the West Coast and
They have two strong and two weak items
in their line,
while
in
we
computer-based daia and oraamzational leaminc
have three strong and one weaA'
volume per store
distribution. (Sales
In this story,
at the
1)
which the items are carried./
in
The
is
not doing well with
Other characters
competitor.
The
was
setting
plot described the
to stock the four items in its
the competitor
— the
the various regions.
time of the product introduction.
convince supermarkets
lessons:
m
new product
company's attempts
The
line.
new product and
this
the U.S. grocery
is
all
had two
vulnerable;
2)
of the competitor's items and stock more of Butler's items.
Another key characteristic of the above story
general
—
is
Not only do we know
detail.
its
suf)ermarkets to get them to stock
first
story
to
swap out some
therefore. Butler's saJes force should attempt to convince the supermarkets to
or
of
also have better sales per point
main character was a company
the
included Butler and the supermarkets
market
We
item.
its
new
product,
dairy product and they were committed to
story's impact (Martin. 1982).
detailed stories
When compared
— and
we
making
also
it
managers'
know why
a success.
—
it
stories
money
competitor paid
the
that
product
was
the
in
the
to
company's
This detail increases the
to abstract information,
people both can recall
more accurately (Reyes. Thompson and Bower, 1980) and
are
more
likely to
them (Kahneman and Tversky, 1973; Borgida and Nisbett. 1977).
act u{X)n
Managers' Use of Computer-Based Data to Confirm and Modify Stories
Stories play an important role in
how
a product
manager changes
his understanding of
When
the marketing environment through his analysis of computer-based data.
manager
studies computer-based data, he
product and
data and
its
sales
supermarket chain raised the
try to
match
in
Seattle
retail price
this data to stories
story about the last time the
attempting to relate
Consider a product manager
environment.
finds that both
is
were down by
raised
its
understands the data,
surprised by the data, he
would
if
15% and
the city's
price, or he
make sense of them
—
that
largest
The product manager would
The manager might
recall
a
might remember the impact of
The product manager would
they confirmed one of his stories.
try to
his
the latest scanner
about previous price changes.
same chain
knows about
who examines
of his product by 10%.
price increases by similar chains in Seattle or in other cities.
that he
to stories he
it
a product
at the
If the
say
manager were
same time modifying one of
his stories.
Analysis of computer-based data plays a key role
making process.
It
in
the product
managers'
sense
provides the raw material they use to confirm or modify their stories.
Product managers at Butler used scanner data to learn about the impact of changes
in the
4Ps
of the marketing environment on key outcome measures, such as sales and market share, for
Butler products and
its
competitors.
oorapulfir-based data and orcamzauonai learning
Product managers noted that analyzing computer-based data helped them monitor the
Tliey tracked sales, distribution, price, and promotions for their product and
environment^.
They often focused
key competition.
attention
their
on new products or new
Product managers studied the country as a whole and important markets
markets that accounted for a large proportion of sales or new or
They analyzed data
competitor.
One product
superiors.
The data
to evaluate their
varieties.
— usually
either
markets for Butler or a
test
performance and
its
to anticipate questions
from
manager-^ discussed the arrival of monthly market share data:
hand-delivered to the president [of the subsidiary], the executive vice
is
of marketing, and the product managers.
on a Tuesday. People wait tor the mail person. It's pretty
arrives at 10
president, the director
It
AM.
intense.
The managers tend to look at market share tor the total U.S. and tor a tew key
oHen get notes from them asking, 'why did this happen? what's
It's my job to know that our Den\'er numbers are off. because our
competitor ran a coupon. By 2 P.M. that day. a report is on the desk of the
markets.
I
going on?'
division president at corporate.
It
contains market share data plus trends tor
the year.
If the
manager
just tracked trends but did
These data contirmed one
understood the data and learned nothing new.
product and
its
environment (see Table
no further analysis, he would say
For example, one product manager closely
1).
analysis of the previous month's scanner data,
In discussing this
variety.
all varieties
new
two of
his stories
it
was
'tracking'
—
line.
were confirmed.
new
increased 11.1% and sales of the
variety, he noted that
he
his stories about the
followed the introduction of a new, and relatively spicy, variety of his product
found that sales for
that
variety
In his
First,
he
were up 11%.
selling as well as the average
Hence, these new data confirmed a portion of his story about the new variety.
Second, he noted that the new variety was selling best
aspect of his story about regional differences in
in the
consumer
This contirmed one
Southwest.
tastes
—
that
Southwestemers prefer
spicier foods.
Frequently an event occurs that
is
surprising to a manager.
from predictions" and hence "trigger[s] a need for explanation"
manager has decided
to 'notice'
from monitoring, which
^Rockart and
De Long
is
the event (Starbuck
The event
(Lxjuis,
1980,
and Milliken, 1988).
characterized as a more automatic or scripted
is
"discrepant
p.
241).
The
She must switch
mode of
(1988) also note that computer-based data play an important role
in
cognitive
monitoring the
environment.
^This manager was not at Butler, but was at one of the company's that participated
study.
-7-
m
the first phase of this
;omputer-based data and organizational learning
TABLE
contirm storv about..
1:
USING COMPUTER-BASED DATA TO CONHRiM
MANAGERS' STORIES
compuler-baacd dala and orcamzatiunal Icamine
among
these
product
elements on
Tlie
sales.
following
example
illustrates
the
process
followed by one product manager as she analyzed computer-based data and modified her
stories
about the impact on sales of promotions.
One
Butler product
was not as
manager was surprised
effective this
that her winter holiday
promotion
year as the year before. Her hunch
marketed under the supermarket's own labelJ
was that the growth
in private label products [products
She attempted to
over the past year reduced the impact of her promotions.
of
her
product tor the
data
on
sales
scanner
hunch
comparing
by
verify this
promotion period
other regions.
had strong private
in regions that
label sales with data tor
She found that overall her promotion was
less elective in regions
with strong private label sales.
While carrying out her analysis, the manager noticed that there were large
differences in the effectiveness of her promotions among those regions with
prompted her to investigate further. She looked
within those regions comparing regions in which her promotion was effective
with those in which it was not. The manager found that when the supermarkets
with private labels promoted their products during the weeks right before the
Thanksgiving and Christmas holidays their promotions were very effective at
strong private label sales.
This
generating increased sales, otherwise they were not.
The product manager modiSed her story about the impact of promotions on the
sales of her product, adding that selecting the appropriate week for promotions
was critical. She explained this phenomenon by noting that her product was
purchased to be eaten with holiday meals and that it was a relatively expensive,
impulse purchase.
Hence the promotions were most effective when timed to
bring the product to the purchasers' attention during the ^veek before they were
most
likely to
consume
The above example
unstructured one.
It
it.
illustrates
that
the
story
modification
can be viewed as a dialogue between the
set
process
a
is
relatively
of tools (including various
types of analysis of computer-based data) at the manager's disposal and her knowledge of the
environment and of how
(1966)'*.
In the
to use the tools.
above example,
scanner data that she performed to
the differences
was
not.
between regions
in
in the
manager's tools were the specific analyses of
test
her hunch about private label products and to examine
which her promotion was successful and those
the
marketing mix on
environment was organized as
sales.
She also possessed
how the bricoleur (handyman) enters into a dialogue
how to use them when attemptmg to repair something.
^Levi-Strauss descnbes
and his knowledge about
process referred to as bricolage by Levi-Strauss
the product
The manager' s knowledge about
impact of changes
A
in
which
it
stories about the
stories,
gained from
with the matenals available to htm
computer-based data and oreanizational leamine
TABLE
type of story
modification
2:
USING COMPUTER-BASED DATA TO iMODIFV
MANAGERS' STORIES
computer-based data and organizational leamine
between tools and knowledge (Levi-Strauss. 1962) and as a
as a dialogue
dialectical process in
which the gap "between resolution characteristics and information and procedural
is
closed (Lave,
1988,
Levi-Strauss and Lave emphasize that both the use of tools
145).
p.
possibilities"
and the definition of the problem affect each other; each changes as the dialogue proceeds.
In the
label
example, the product manager mitially defined the problem as one of private
growth and hypothesized
that the difference in the effectiveness
could be explained by the increase
in
The
private label sales.
of her holiday promotion
she used to
tool
test
her
hypothesis was an analysis of scanner data in which she compared the impact of her promotion
with strong private labels with
in regions
its
impact
supported her hypothesis.
As her dialogue proceeded,
one of promotion timing.
She tested a new hypothesis
when
effective
or Christmas
the
manager redefined
— supermarkets'
they promoted their private label product the
— using a new
tool.
While she
still
Her
in other regions.
relied
the
problem as
promotions were more
week before
on scanner
analysis partly
data,
either Thanksgiving
now
she focused her
analysis on the impact of promotion timing on sales.
A
second example
illustrates the use
outcome occurred.
specific
It
of multiple sources of information to learn
shows how talking with peers and market research can be
combined with analysis of computer-based
data.
This example also supports the view that the
story modification process can be characterized as a dialogue
between a
set
of tools and a
problem.
Another product manager at Butler did not
light
— were poor
in the
South region.
know why
sales
of a new
—
variety
His Srst step toward creating a story to
explain this event was to ask the regional sales manager.
many of this
guessed that Southerners did not consume as
The sales manager
variety as people in
manager
then entered into a dialogue.
The Srst tool that he used was an analysis of
scanner data, comparing sales of Butler and competitor products of this variety
The manager found that Southerners
in the South and in other regions.
consumed about as many of the competitors light variety as people in other
other areas.
With this serving as his tentative hypothesis, the product
'
regions.
The product manager then redeSned the problem as one of taste. He came up
with a hunch of his own
Butler's light variety does not taste as good as its
competition to Southern consumers. The manager tested this hunch using a new
—
tool.
He
6rm to conduct blind taste tests.
between Butler and its competitors in ffa vor.
hired a market research
showed no
differences
why a
The product manager finally accepted a
third
hypothesis
— that
These
the
tests
South
regional sales force was doing a poor job convincing supermarkets to stock
Butler light variety.
This hypothesis was supported by the scanner data, since
11 -
computer-based data and oreanizational leamine
He
they showed that sales of light were relatively poor in the South.
this to
be
of
result
this
sensemaking process,
manager modiGed
the
his stories
the effectiveness
manager
to
of the South regional sales three at introducing new
example, the product manager
good as
was not
its
is
— and
the story
consistent
with
modification process
(1986)
—
—a
used another tool
Isenberg
varieties.
on discussions with
light
vanety
about
the regional sales
in
the South.
his tool to test the tentative hypothesis.
developed a new one
verified, he
competition
manager ended
which
first relied
develop a hypothesis about poor sales of the
manager used analysis of scanner data as
hyf>othesis
Ma
and about
regional differences in product sales, about the taste of his product,
In this
believed
because he believed there was no plausible alternative.
true,
this
that Butler light variety did not taste as
—
market research study
when he developed
who
When
The
notes
that
to test
it.
The
a plausible explanation,
managers
strive
for
believable
explanations rather than optimal solutions.
As
in the first
example, the manager used a
problem and a plausible solution.
During
set
of tools to close the gap between the
gap closing process,
this
the
problem
itself
reformulated three times and the tools used to solve the problem were changed twice.
was
This
provides further evidence of a dialogue in which both the problem and the tools use to solve
affect each other.
They both are modified
Finally, a third
experiment
as
a
tool
experiments on product
example
during
taste
illustrates
story
it
as the process progresses.
how
a product
modification.
manager used a
While
managers
naturally occurring
routinely
conduct
or packaging using market research techniques, they also can use
computer-based data to examine the impact of a natural experiment.
Consider the following
example:
A
manager wanted to understand what level of price
Would sales increase
reduction was needed to make his promotion effective.
79C
99C
price? The manager
promotion price as with a
about as much with a
went back through scanner data. He noted that when there had been both 79C
and 99c prices accompanied by features (ads in the supermarket circular) and
end-aisle product displays in the same regions at different times, sales were
similar.
He summarized the lesson that he learned, stating, "We learned that
We'd
price is not what drives sales. It's getting the feature and the display.
move just as many cases at 99C with feature and a display as at 79C.
third Butler product
When
product managers analyze scanner data they increase the accuracy and timeliness
of feedback they receive about causal relationships
in the
marketing environment.
necessary condition to facilitate organizational experiments (Huber.
1991).
This
is
a
Each marketing
action taken by a product manager, or by a colleague or competitor, can be viewed as a
naturally occurring experiment,
from which the product manager can gather valuable
12-
insights.
omanuauonal Icamins
compuler-base<l data and
There was one type of
Butler product managers.
tool missing
from the story modification process followed by
They performed almost no
sense of computer-based data.
statistical
analyses to help them
the
make
managers had the training (many had M.B.A.
Tlie product
many of
degrees) and the computer software needed to perform these analyses.
Further,
surprises they encountered could be studied using statistical techniques.
For example, product
managers could have used regression
to
the
determine which of several causal variables (price
They
reductions, coupons, ads in supermarket circulars) were significantly related to sales.
could also perform analyses of variance to identify significant differences
sales
in
among
regions.
Only one manager
tried to
perform a
/ wanted to find out what drives sales
regression, but
it
didn
bottom three markets.
't
tell
me
of my product.
The simple
rich,
described his attempt:
I first tried
Then I decided to compare
much.
running a
my
top
and
of scanner measures. I found that
they were on price reduction more
I looked at the full set
had much deeper pace cuts,
had more AB features [larger ads
the top three
otien. they
He
statistical analysis.
analysis performed by the product
in
supermarket circulars].
manager was an
easier
way
detailed story about the factors that affected sales of his product.
him
for
The
to build a
story that the
product manager developed was also more easily communicated to his superiors and to the
sales force as will be discussed in the next section.
Sharing Stories Based on Analysis of Computer-Based Data
Managers share
the insights they have gained
Sharing insights
telling stories.
is critical
exchanging information and developing a
is
managers who
common
interpret their
community memory
—a
set
environment by
understanding (Daft and Weick, 1984).
also critical to organizations, permitting multiple interpretations of the
organizational learning (March, SprouU, and
a
to
from analysis of computer-based data by
Tamuz, 1991) and
same event and richer
facilitating the
of stories about the organization and
It
development of
environment (Brown and
its
Duguid, 1991).
People
in organizations like to tell
and
listen to stories.
and managers identify themselves as competent practitioners.
Telling stories helps workers
As
organizations have a desire to share their achievements with others.
managers
social
like to
dramas of
story tellers,
As
listeners,
people
in
workers and
hear stories; they have a "deep and abiding interest in the characters and
their
world" (Orr, 1990,
p. 75).
-13-
computer-based data and oreamzational leamine
Telling Stories to Others
Members
of the Organization
Product managers share stories that are based on their analyses of computer-based data.
They
tell
might be
each other,
stories to
in
to their
managers, and
to sales
managers.
This story telling
response to a specific inquiry from a superior or a sales manager,
review of product performance presented to senior marketing management, or
settings
in a
formal
in less
formal
— such as over lunch or coffee.
At Butler, many
presentations to
stories
based on analyses of computer-based data were shared during
The
management.
vice
of marketing
president
strongly
encouraged the
product managers to use the newly acquired scanner data as part of their state-of-the-brand
The managers developed
review.
environment and told them
from
detailed
They supported
at the presentation.
their analyses, presented in charts
about their product and
stories
and graphs.
companng
with charts
this story
supermarket circular ads
size
in the three regions
marketing
these stories with
For example, the manager
the story about factors that affected sales of his product discussed at the
supported
its
numbers
who developed
end of the
last section
and number of price cuts and number of large
with the best and worst sales.
Another product manager was singled out by her peers and superior for the quality of
She
her state-of-the-brand presentation.
steeply over the last
all
two years.
told a story
She stated
about
that there
why
sales of her product declined
were too many
varieties of the product,
but a few were doing very poorly, and only one was selling well.
also observed that while her product
was a fad food and experienced rapid
introduced about four years ago, the product category was
price
She also noted
sensitive.
competition;
her
that
had higher per store
it
The product manager
sales.
now
sales
growth when
aging and had become more
product
was performing well
Finally,
she pointed out that there were large
relative
to
its
seasonal variations in sales and in the effectiveness of promotions.
Based on
this story, the
included dropping
eliminated
all
product manager developed a detailed marketing strategy.
many product
varieties
promotions from the
first
and focusing on the best
quarter
when
sales
were low.
product manager attracted management support for her brand.
We got
She
selling one.
She also
Most importandy,
stated:
corporate commitment back to the business, because we could explain a
We could tell them a story. We could tell them what
we don't know. Before [we analyzed the scanner data/ we didn't even know
what we knew and what we didn
We got commitment to grow the business
despite seeing a 20-30% sales decline over the last 2 years, because the
presentation was impressive.
Before they were looking the other way when I
great deal more to them.
—
't.
walked in the hallway.
Now they're letting
-14-
us spend
This
money behind
the business.
the
computer-based data and Drganizaiionai leanuna
There was a second major benefit of the state-of-the brand analyses.
managers and
their superiors
found commonalities among the products.
elaborated on patterns present in organizational events (Boje,
The product
They found and
The product managers
1991).
developed a joint understanding about the impact of price, promotions, competition, and
marketing mix on
One product manager described
sales.
the story that had been developed
about promotions of Butler products through analyses of computer-based data by several of her
colleagues:
^o"-
We 're
The trade
a small brand.
[supermarkets} will only give us so
promotions per quarter and most things need
to
be supported quarterly.
many
Since
an impulse-driven, trade- intensive category, we're very dependent on
promotions. Frankly, we ha ve more than our lair share based on our size.
vve're
We
ha ve deep enough pockets to support all of our
can't support 14 categories, we can support at most 6.
learned that we don
We
product categones.
Theretbre.
we
will
be doing joint promotions of several products.
Another product manager added
We found
that
C
't
to the
promotion
story:
ads [the smallest ads in supermarket circulars] do next to
Two C ads cost the same as one A ad [the largest size ad], but don 't
bring the same results. End aisle displays with A ads are 50% more effective
than A ads alone. Further, joint consumer and trade are much more effective
nothing.
than separate ones.
This learning led to a change
in
Product managers identified those
promotion strategy.
products that could be linked together with a joint promotion (through further analysis of
computer-based data).
participate in
'C
ads.
Brokers* were instructed to strive for larger 'A'
They were
also instructed to
make
trade promotions (price reductions to
supermarkets) contingent on getting both end aisle displays and
promotions
coupons for consumers
to coincide with
ads and not to
that
A
ads
appear
in
in circulars
and
to time
free-standing inserts in
Sunday newspapers.
Official Stories
Told
to People
The product managers
Outside the Organization
also developed 'offical stories' (Schank,
marketing environment to people outside of the organization
(Butler's customers).
These
stories
were not used
The
they were used to enact change (Boje, 1991).
stories
developed for management or peers.
its
— brokers and supermarket buyers
make sense of
official stories
They were
obtained from analyses of computer-based data.
^Butler sold
to
1990) to describe the
the environment.
were often versions of the
also supported by charts and graphs
Since brokers met with supermarket buyers
product through independent regional brokers, not a dedicated sales force.
15
Rather,
compuier-hased data and organi7^iional leanuns
primarily to convince them to participate
stories
official
told
to
and
brokers
in
a promotion or to discuss shelf space (facings), the
supermarket buyers involved explainmg
changes
in
promotion strategy, oblainmg facmgs for a new or existing product, or defending shelf space
See Table 3 for a summary of these
from a competitor's product.
A
marketing manager
another company descnbed these stories as essential to
at
He viewed
based marketing.'
stories.
storytelling, supported
'fact-
by appropriate computer-based data, as
an alternative to reducing price or other incentives to get a supermarket to stock a product or
participate in a promotion.
His view supports the idea that having more information
is
symbol of organizational competence (Feldman and March, 1981) and. therefore, makes
manufacturer 'look
The
better' to the
They were
story about joint promotions described
also given the
We'll
the
tell
new
above was packaged for
'party line' about joint promotions
numbers (based on scanner data
manager described a
Butler product
the
supermarket buyer.
The brokers were given
buyers.
a
telling
and no
the
to
C
ads.
One
analysis) to supp>ort the story.
part of this story:
we asked you to execute 6 separate
was too much. Our promotion [for one product]
the trade [buyers] that last year
in October and it
was executed in only 22% of all supermarkets. We'll try to convince the trade
that one big promotion [involving several product categories] is the way to go.
We prefor one
We'll show them that C ads did nothing for [another product].
promotions
joint promotion with
These
stories
an
A ad for [each
of two
sets
of related products}.
were sometimes biased, since they were used
to put Butler's products in
as favorable a light and competitor products in as unfavorable a light as possible.
The
stories
about competitor products always focused on problems (e.g., a new product introduction was
not successful,
or the competitor was getting out of a certain product line).
information was puqxjsely omitted
product manager
who compared
when
it
his best
did not support the official story.
and worst markets found
supported the official story about Butler promotions.
were used
regions.
relatively
He
noted,
more frequently
"We
left that
He
For example,
most of
the
his insights
did note, however, that smaller
C
ads
with high sales of his product than in low sales
in regions
off the chart, because
-
that
Sometimes
16-
it
didn't add to our argument."
computer-based data and orgamzaiional leamins;
TABLE
3:
OFTICIAL STORIES TOLD TO BROKERS
SUPERMARKET BUYERS
tuld
to..
OR
oompuier-hased data and organizational learning
Discussion
The Butler product management organization followed a four-step process
about their environment from the analysis of computer-based data.
and look for unexpected
the data
results
— findings
that contradict
If
unstructured, multi-stage process to
make sense out of
is
be viewed as a dialogue between the result and a
insights with peers
and superiors, developing a
used to
one or more of her
the unexpected result.
of tools
set
Next, the manager
(including analyses of computer-based data).
common
at
new marketing approaches
product manager organization
— the sales force and supermarket buyers.
the product
manager's disposal
the
understanding.
is
While
stories
This process can
tells the story
creates an official story, that
'sell'
manager exammes
First, a
found, the manager carries out a relatively
about the marketing environment.
a surprise
in learning
to share her
Finally, the
manager
to people outside the
managers made extensive use of large amounts of computer-based
data as they attempted to understand changes in their environment, they used almost no formal
statistical
methods
in their analyses.
designed to assist people
in
This
is
somewhat
surprising in that statistical methods are
summarizing and drawing inferences from large data
sets.
The
Butler product managers preferred to develop rich, detailed stories which could be easily told
to
and understood by
supporting role.
modification.
their peers, superiors,
The managers used
They
and customers.
In these stories, the data played a
the data to confirm the stories or as a tool in story
also used the data to
make
the stories
more credible
— charts
and graphs
reinforced the message of the stories.
Another example of the process of creating
complex
statistics,
York Times.
can be found
in the financial
report that
by
from large data
sets,
without using
pages of The Wall Street Journal and The
New
Reporters create stories that explain the changes in stock, bond and currency
prices that occurred the previous day.
A
stories
Consider the following
showed only a small increase
story:
in inflation
was ultimately ignored
credit market participants yesterday as traders sold Treasury securities
moved higher.
and
of notes and bonds rose sharply immediately
its Producer Price Index rose by a modest
two-tenths of a percent last month, and only one-tenth of a percent when volatile
energy and food prices were excluded. But the gains, which amounted to as
much as ^ point on some securities, could not be sustained, as sellers emerged
Selling of short- and intermediate-term notes was
at the higher price levels.
particulariy hea vy. Rather than focus on the good inQation news, memories of
stronger January and February retail sales figures reported on Thursday
continued to linger. .And preliminary results of a University of Michigan survey
interest rates
Prices
a Her the Government announced that
reinforced that economic recovery
is
under way. [Kenneth G. Gilpin. The
York Times. March 14. 1992].
18
New
compuurbased
The
bond
reporter created a rich story to explain the change in the
statistical
data and orgamzauonal learning
He
marlcet.
did not use
analyses to correlate producer prices or retail sales and bond prices.
created a plausible explanation and supported
it
with both several sets of data and
Rather, he
text.
This study has provided insights into one aspect of organizational learning
organizations
make
organizational learning process.
who
and Tamuz (1991)
studied
Those authors noted
of data.
confronted with single
They then share
of
out
sense
amounts
large
how
— and
on
perspective
a
the
organizations learn from single events and small amounts
most when the event
that organizations learn
critical events,
theme of creating and sharing
data
findings are strikingly similar to those of March, Sproull,
Its
this history to
of
— how
is
a surprise.
When
organizations create a detailed history to explain
it.
The
allow for multiple interpretations and richer learning.
stories runs through both studies.
who
There are also imp>ortant lessons for researchers
psychology researchers have compared the impact of
study
detailed
rich,
Cognitive
stories.
with abstract
stories
numerical information on decision making and have found that people were more influenced
by
stories
(Kahneman and Tversky,
managers used data
Borgida and Nisbett,
1973;
to support their stories to influence their superiors
than studying the differences between abstract data and stories,
examine the degree
to
which adding supporting data
Butler product
1977).
and colleagues.
Rather
could be more beneficial to
it
to stories increases their credibility
and
impact.
The study
also highlights the importance of
when managers must understand changes
in
making sense out of computer-based data
the
business
While
environment.
many
researchers and practitioners have discussed the benefits of organizational learning (e.g.,
Senge, 1990; Stata, 1989), the role of computer-based data in organizational learning has been
all
but
ignored
by
information
systems,
management), and organizational behavior
functional
researchers''.
focused on the value of more sophisticated tools,
area
(e.g.
marketing
operations
Information systems researchers have
such as expert systems,
learning and in storing organizational knowledge (e.g., Sviokla,
1990).
in
facilitating
In a similar vein,
functional area researchers have tried to use computer-based data to develop
to explain underlying business processes (e.g.,
or
complex models
Kalwani, Yim, Rinne, and Sugita, 1990). Most
organizational behavior researchers seem to recognize the value of stories and storytelling in
organizational learning,
transmitting
but they have focused their attention on the role of stories
norms and values
Much more
research
computer-based data
^Zuboff (1988)
is
is
(e.g., Martin,
Feldman. Hatch, and
needed on organizational learning
in the organizational learning process.
a notable exception.
-19
We
for
Sitkin, 1983).
in general
and on the role of
need to better understand
how
computer-based data and orsanizational learning
organizations interpret their environment (Ruber, 1991) and the role of computer-based data in
The techniques employed by cognitive anthropologists
this process.
how managers
and how they share
others.
Research
is
information technology investment
is
little
known about
organizational learning,
is
exists
that
linked to organizational performance (Harris and Katz.
Through
the causality of this link.
we can
also needed to quantify the impact
While some evidence
of computer-based data on organizational performance.
1991),
1991.
analyze computer-based data, what they learn from these analyses,
knowledge with
this
Hutchins.
Using these techniques, we
Orr. 1990. Lave. 1988) can play an important role in this process.
can observe
(e.g.,
study
how an
investment
in
the intervening process of
information technology might
improve organizational performance.
Finally,
managers need guidance
specifically through the analysis of
that organizations
in
designing organizations that encourage learning,
computer-based data.
Some
researchers have suggested
can promote learning by fostenng communities of practice (Brown and
Duguid. 1991) and by building overlap into responsibilities and rotating work assignments
(Nonaka. 1991).
from
Little is
the analysis of
Butler
is
known, however, about how
computer-based data.
one mechanism
to
in place,
meet changes
At a
customers.
the
state-of-the brand analyses
move
too slowly to
environment.
must constantly adapt
to survive in
an environment
by rapid product changes and the diverse demands of a wide variety of
Toward
that end.
they must gather, interpret, and disseminate knowledge about
Analysis of computer-based data can play a
changing market conditions and customer needs.
key role in managing these organizational 'knowledge assets' (Boynton and Victor,
The more
1991).
detailed data that are collected about the environment and factors that influence
organizational performance, the
into the
At the
With too many formal
organizations might over-analyze data and hence
in the business
employed by
Others must be found.
danger of 'analysis-paralysis.'
strategic level, organizations
characterized
develop and share insights gained
encourage organizational learning.
same time, we must consider
mechanisms
The formal
to
changes
more important
that organizations
This paper has provided
initial
it
will
might make and
insights
into
how
computer-based data to improve their performance.
-20-
be to use these data
to
provide insights
to evaluate the
impact of these changes.
organizations
might go about mining
compmer-based
data and
orgamzauonal learning
References
Boje, David
A
(1991). "The Storytelling Organization:
J.
Study of Story Performance
an
in
Office-Supply Firm," .Administrative Science Quarterly, 36. 106-126.
Borgida,
E.
and R.E. Nisbett (1977), "The Differential Impact of Abstract Vs. Concrete
7, 3. 258-271.
Information on Decisions." Journal of Applied Social Psychology,
Bower, G.H. (1976), "Expenments on Story Understanding and Recall," Quarterly Journal of
Experimental Psychology, 28, 511-534.
Boynton. Andrew C. and Bart Victor (1991)
Stable Organization," Califyrnia
,
"Beyond
Managing
Flexibility:
Management Review,
34,
1,
the
Dynamically
53-66.
Brown, John Seely and Paul Duguid (1991), "Organizational Learning and Communities of
Toward a Unified View of Working, Learning, and Innovation."
Practice:
Organization Science,
Daft,
2, 1,
40-57.
R.L. and K.E. Weick (1984),
Systems,"
elSawy,
Omar
"Toward a Model of Organizations
Academy of Management Review,
(1985),
Quarterly, 9,
1,
"Personal
9,
Systems
Information
as Interpretation
284-295.
For
Strategic
MIS
Scanning,"
53-60.
Feldman, Martha
S. and J.G. March, (1981) "Information in Organizations as Signal and
Symbol," Administrative Science Quarterly, 26, 171-186.
Goldstein, David K. (1990), "Information Support for Sales and Marketing:
a Small Grocery Manufacturer," Information
and Management,
19,
A
Case Study
at
257-268.
David K. and Namjae Cho (1991), "Environment, Strategy, and the Use Of
Computer-Based Data: Case Studies in Product Management," Marketing Science
Goldstein,
Institute
Report 91-132.
Goldstein, David K. and Michael H.
Zack (1989), "The Impact of Marketing Information
Supply on Product Managers:
An
OfBce Technology and People,
4, 4,
Organizational Information Processing Perspective,"
313-336.
Harris, Sidney H. and J.L. Katz, "Organizational Performance
and Information Technology
Investment Intensity in the Insurance Industry," Organization Science, 2, 3, 263-295.
Huber, George
P.
(1991),
"Organizational Learning:
Literatures," Organization Science, 2,
Hutchins,
1,
The Contributing Processes and
88-115.
Edwin (1991), "Organizing Work by Adaptation," Organization
39.
21
the
Science, 2,
1,
14-
-•iimputer based
Lynn A. (1990), "Evolving
Isabella.
Interpretations as a
dau and
Change Unfolds;
organizaiional Ieamin2
How Managers
Construe Key Organ izalional Events," .Academy of Management JournaL 33,
Isenberg, Daniel
J.
(1986),
Managerial Action,"
in
Brigdet
(1989),
7-41.
and Process of Understanding: Implications for
Sims. Henry P., Jr.. Dennis A. Gioia and Associates (Eds.). The
"Tlie Structure
Thinking Organization, San Francisco,
Jordan.
1,
CA: Jossey-Bass
"Cosmopolitical Obstetncs:
Traditional Midwives." Social Science
Some
and Medicine,
Kahneman, D. and A. Tversky (1973), "On
Publishers.
Insights
from the Training of
28. 9, 925-944.
the Psychology of Prediction,"
Psychological
/?eweiv, 80, 237-251.
Rmne, and Yoshu Sugita (1990), "A Price
Expectations Model of Customer Brand Choice," Journal of Marketing Research, 27,
Kalwani. Manohar U., Chi Kin Yim, Heikki
J.
251-262.
Lave, Jean.
(
1988), Cognition in Practice, Cambridge, England:
Cambndge
University Press.
Levi-Strauss, Claude (1966), The Savage Mind, Chicago: Chicago University Press.
Louis, Meryl Reis and Robert
Mind
to
Sutton (1990), "Switching Cognitive Gears:
I.
Active Thinking,"
Human
Relations, 44,
From
Habits of
1.
Meryl Reis (1980), "Surprise and Sense-Making: What Newcomers Experience
Louis,
in
Entering Unfamiliar Settings," .Administrative Science Quarterly, 25, 226-251.
March, James G., Lee
S. Sproull,
and Michal
or Fewer." Organization Science, 2,
1,
Tamuz
(1991), "Learning from Samples of
One
1-13.
Martin, (1982), "Stories and Scripts in Organizational Settings," in Hastori', Albert H. and
Alice
M.
Joanne,
Martin.
Isen (eds.
Martha
),
J.
(1986),
Homedale,
IL:
Feldman,
S.
Uniqueness Paradox
438-453.
McCann,
Cognitive Social Psychology,
in
New
York, Elsevier/North-Holland.
Mary Jo Hatch, and Sim
B.
Sitkin
(1983),
The
Organizational Stories." .Administrative Science Quarterly, 28,
The Marketing Workbench: Using Computers for Better Performance,
Dow-Jones
Irwin.
Minsky, Marvin (1975), "A Framework for Representing Knowledge,"
(Ed.), The Psychology of Computer Vision, New York: McGraw-Hill.
in
P.H.
Winston
Nonaka. Ikujiro (1991), "The Knowledge Creating Orgamzation." Har\ard Business Review,
69, 6, 96-104.
computer-based data and on;anizational learning
"Shanng Knowledge, Celebrating
Orr, Julian (.1990),
Memory
in
a Service Culture,
Remembering: Memory
"
in
in Society,
Identity:
War
Stories and
D.S. Middleton and D. Edwards (Eds.)
Beverly Hills,
CA: Sage
Community
,
Collective
Publications.
W.C. Thompson, and G.H. Bower
(1980), "Judgmental Biases Resulting from
Arguments,"
Journal
of Personality and Social Psychology,
Differing Availabilities of
Reyes, R.M.,
39, 2-12.
Rockart, John F. and David
Dow-Jones
W. De Long
(1988), Executive Support Systems, Homedale, IL:
Irwin.
Schank, Roger C. (1990), Tell
Me
A New Look at
a Story:
Real and ArtiGcial Memory,
New
York: Charles Scribner's Sons.
Schank, Roger C. and Robert Abelson (1977), Scripts. Plans, and Knowledge, Hillsdale, NJ:
Erlbaum.
Senge,
M.,
Peter
(,1990),
New
Organization.
The Fifth Discipline:
York:Doubleday,
The Art and Practice of the Learning
.
Simon, Herbert A. (1991), "Bounded Rationality and Organizational Learning," Organization
Science, 2,
Stata,
1,
125-134.
~ The Key
Ray, (1989) "Organizational Learning
Management Review, 31, 63-74.
Starbuck,
W.H. and
to
Management Innovation," Sloan
What They Notice
The Executive Effect: Concepts and
F.J. Milliken (1988), "Executives' Perceptual Filters:
How
They Make Sense," in D. Hambrick (Ed.),
Methods for Studying Top Managers, Greenwich, CT: JAI
and
Sviokla. John
J.
(1990),
"An Examination of
CaseofXCON," MIS
Press, 35-66.
the Impact of Expert
Systems on the Firm: The
Quarterly, 14, 126-140.
Zuboff, Shoshana (1988), In the
Age of the Smart Machine, New York:
-23-
Basic Press.
10
nif]
Date Due
-r
9v
Lib-26-67
MIT
3
TOflO
LIBRARIES
DD7nMDS