Fall 2006 Davison/Lin
PLANNING: Things to Know
BEFORE You Start…
CSE 197/BIS 197: Search Engine Strategies 6-1
SEM Proposal
How to sell it?
Strategic Planning
How to do it?
Measure SEM performance
How good is my search?
Site Analysis
How good is my site?
Goal Analysis
Why SEM?
Module II Overview
−
−
−
Conversion Rate
Counting dollars
And more…
CSE 197/BIS 197: Search Engine Strategies 6-2
Tools that help your web analytical needs
How to measure web site performance based on your
goals
Counting visitors
How to diagnose a web site’s well being using multiple
performance metrics [notice not just HOW, but HOW WELL!!]
In this section, we discuss:
Fall 2006 Davison/Lin
If you can’t measure, you can’t manage
By using cookie, servers store the state of your machine
Often session = visit in SEM context
−
session-id-time
1159167600l
amazon.com/
Technically, hard to accurately measure
−
CSE 197/BIS 197: Search Engine Strategies 6-3
Session: theoretically, one visit paid by the customer to your web
site
−
disk. The pieces of information are stored as name-value pairs
Cookie: a piece of text that a Web server can store on a user's hard
Fall 2006 Davison/Lin
Metric 1: # of Visits / Visitors
CSE 197/BIS 197: Search Engine Strategies 6-4
Both were actively used to measure site performance around 2000
Number of unique visitors:
the number of unique machine IDs that visited the server
Number of visits:
the number of unique sessions as counted by the server
Fall 2006 Davison/Lin
Metric 1: # of Visits / Visitors
How unique is “unique”?
−
CSE 197/BIS 197: Search Engine Strategies 6-5
Financial analysts have become increasingly skeptical of nonfinancial metrics [Gupta et al. 2004]
Issue with session
−
Issues with visits / unique visitor
Fall 2006 Davison/Lin
Metric 1: # of Visits / Visitors
Heuristics to solve the problem (but distributed environment!)
Single pixel tracking
−
−
CSE 197/BIS 197: Search Engine Strategies 6-6
Stickiness = average duration / page view
Simple log for each requested file (but file != page)
−
Textbook (as well as Brian ☺) explains in detail how page
views are counted
Visit duration / page views
Fall 2006 Davison/Lin
Metric 2: Stickiness - Related
Implicates higher likelihood to purchase
−
CSE 197/BIS 197: Search Engine Strategies 6-7
Stickiness is capable of explaining the share price of
Internet firms [Demers and Lev 2001]
Page views offer some explanatory power but do not
appear affecting firms’ net incomes [Trueman et al 2000]
Reflects high loyalty
−
Conventional wisdom suggests stickiness to be a valuable
metrics
Fall 2006 Davison/Lin
Metric 2: Stickiness - Related
Page views significant for search goods
−
CSE 197/BIS 197: Search Engine Strategies 6-8
Duration significant for experience goods
−
Positive relationship between stickiness and conversion [Lin et al,
2006]
Positive relationship between stickiness and purchase [Wu et al.
2005]
Fall 2006 Davison/Lin
Metric 2: Stickiness - Related
330
Number of websites in shopping category that offer direct sales services
CSE 197/BIS 197: Search Engine Strategies 6-9
601
Total number of websites that offer direct sales services
1,392,713
Total number of websites visited
46,942
174,990
Total number of online purchases made in shopping websites
Total number of websites belonging to shopping category
342,706
213,356,003
Total number of online purchases made
Total number of website visits
100,000
Or they could buy service from companies such as comScore to
get client-side monitoring capability
Companies could invest in technology and build stronger server
side monitoring programs (limitations exist)
Total number of participating households
−
−
Server – side solution vs. Client – side solution
Fall 2006 Davison/Lin
Solving Measurement Issues
Poisson
Poisson with variation
BG/NBD (Fader et al 1004)
Dynamic NBD (Moe and Fader 2004b)
Fall 2006 Davison/Lin
Dynamic Model (Allenby et al 1999)
Inter-purchase time: Gamma
Inter-purchase time: Poisson
Poisson
Pareto/NBD (Schmittlein 1987)
**
Poisson
NBD (Gupta and Morrison 1997)
Joint Model (Boatwright, et al)
Purchase Rate Assumption
Models
*
None
***
None
None
Geometric
Exponential
None
None
None
None
Beta
Gamma
None
Death Events
Heterogeneity
Death Rate Assumption in death rate
CSE 197/BIS 197: Search Engine Strategies 6-10
Inverse Generalized Gamma
Gamma
Gamma
Gamma
Gamma
Gamma
Heterogeneity in purchase rate
Purchase Behavior
Metric 3: Stochastic Models
Mostly academic research, no industry adoption yet
CSE 197/BIS 197: Search Engine Strategies 6-11
Hard to implement
Fall 2006 Davison/Lin
Enables individualized prediction
Metric 3: Stochastic Models
−
CSE 197/BIS 197: Search Engine Strategies 6-12
Average conversion is at 5% and decreasing [Moe 2004]
Conversion = number of actions / number of visitors
Fall 2006 Davison/Lin
Metric 4: Conversion
Average conversion is at 5% and decreasing [Moe 2004]
CSE 197/BIS 197: Search Engine Strategies 6-13
Depending on the goal of the web site (remember last chapter?),
the meaning of “action” might differ
A relatively well-accepted metric for measuring web site
performance
−
Conversion = number of actions / number of visitors
Fall 2006 Davison/Lin
Metric 4: Conversion
Use web analysis program to find number of visits
Purchase / visits
−
−
CSE 197/BIS 197: Search Engine Strategies 6-14
Use e-commerce system to find number of transactions
−
Simplest to calculate
Fall 2006 Davison/Lin
Revenue Generation
Metric 4: Conversion
CSE 197/BIS 197: Search Engine Strategies 6-15
You can also continue to track these visitors and capture their
purchase events as well
That means connect the form with lead management system
−
−
Count each visitor who fills in web contact form as an action
−
For lead generation with the goal of acquire new
customer information
Fall 2006 Davison/Lin
Lead Generation
Metric 4: Conversion
“Call Me” button
Questionnaires at the offline locations
−
−
CSE 197/BIS 197: Search Engine Strategies 6-16
Special phone number, special coupon, etc
Need innovative methods for the offline channel to identify traffic
re-directed from online
−
−
For lead generation with the goal of offline sales:
Fall 2006 Davison/Lin
Lead Generation
Metric 4: Conversion
−
?
CSE 197/BIS 197: Search Engine Strategies 6-17
Measuring the lead is only the first step…
For lead generation with the goal of offline sales:
Fall 2006 Davison/Lin
Lead Generation
Metric 4: Conversion
−
−
CSE 197/BIS 197: Search Engine Strategies 6-18
Then implement possible mechanism to capture conversion
behaviour
First, define what is a conversion based on the goal of your
campaign
Abstract and hard to measure
Fall 2006 Davison/Lin
Brand Image
Metric 4: Conversion
For example, 1,000 visitors, $ 1 per visitor, 20 end up purchasing, then cost per
purchase = ?
Advertising Cost / Total Completed Actions
CSE 197/BIS 197: Search Engine Strategies 6-19
For example, last month you spent $1,000 on advertising to generate 2,000 visitors
and 20 bought at an average of $100 per sale with a gross profit margin of 90.
Average Action Value = revenue / action
−
−
Average Action Value x Gross Profit as % of Sales
−
completed action
Value of a Buyer: the average gross profit you earn from a
−
−
completed action
Cost per Action: the advertising cost you pay for one
Fall 2006 Davison/Lin
Metric 5: Monetary Measurements
−
CSE 197/BIS 197: Search Engine Strategies 6-20
number of visitors to the site,
unique/return visitors
traffic referrers
search engine referrers
search keywords used
page views
visit paths
average number of page views per visitor
entry and exit pages
Basic Level:
Tools that help analyzing the web site performance
Fall 2006 Davison/Lin
Web Analysis Tools
CSE 197/BIS 197: Search Engine Strategies 6-21
Conversion stats.
Dividing the website into logical categories and monitoring
each separately
Bounce rates- the percentage of visitors who leave the
website within the first x seconds of the visit.
Advanced Level:
Example
−
Tools that help analyzing the web site performance
Fall 2006 Davison/Lin
Web Analysis Tools