Merck Moving Beyond Passwords: FIDO Paris Seminar.pptx
Learn How to Do a Conjoint Analysis Project In 1 Hour
1. Learn How to Do a Conjoint
Analysis
Project In 1 hour
2. •Started in 2002 in Seattle, WA
•#172 on Inc. 500 Fastest Growing Private Companies
•#12 on Puget Sound Journal's Top 100 in Washington
•Over 6K+ clients and growing!
•QuestionPro, SurveyAnalytics, IdeaScale, MicroPoll
Esther LaVielle – Chief Education Director
Who is Survey Analytics?
3. Andrew Jeavons has been active in the survey and market research business for over 25 years around the world.
He has worked in Europe, the USA and APAC. He currently lives in Cincinnati, Ohio, USA.
After studying neuropsychology at Birkbeck College in London UK, he then worked in the medical statistics
department of the Institute of Neurology in London UK.
Andrew worked as a software developer for Quantime. He was one of the founders of the software company E-
Tabs, and a founder of a software consulting company now called Cobalt Sky. His areas of interest include
statistics, text analytics and visualization, neuropsychology, writing and speaking.
In the last ten years he has worked for survey software companies in a marketing, sales and strategic
development capacity. He has also written numerous articles for ESOMAR publications and a range of
international conferences.
He is currently Western area convenor for the New MR 2010 conference (www.newmr.org)
Who is Andrew Jeavons?
4. AGENDA:
1.What is Conjoint, Why you need to use it, Core Concepts
2. How to Put together a Conjoint Analysis Question
Wizard based interface to create Conjoint Tasks based on simply entering
Features (Attributes) and Levels for each of the features.
3. Adding Conjoint Design Parameters
Tweak your design but choosing the number of tasks, number of profiles per task as well as "Not-
Applicable" option.
4. Preview Survey
5. Review Utility Calculation & Relative Importance
6. Market Segmentation Tool
Filter the data based on criteria and then run Relative Importance calculations
7. Best Practice & Tips / Q&A
5. Why Choose Survey Analytics
over
Sawtooth Software?
Flexible monthly pricing available
Most User-friendly Conjoint Tool
6. What is Conjoint Analysis ?
Conjoint analysis is a method developed over the past 50 years by market researchers and statisticians to
predict the kinds of decisions consumers will make about products by using questions in a survey.
The central idea is that for any purchase decision consumers evaluate or “trade off” the different
characteristics of a product and decide what is more important to them. For instance , it may be that the
container size is the most important factor, or it may be environmental friendliness of the product and the
price. Obviously for different products there are a whole range of possible characteristics or “attributes” that
consumers may take into account.
Conjoint analysis is away of presenting a set of possible products to consumers via a survey and ask them to
make a choice about which one they would pick. A set of attribute for a product (perhaps color, size, price)
are chosen and then a set of “levels” of the attributes are selects. For instance we could have 3 colors of a
product, red, green and blue, then maybe 3 sizes, 4, 8 and 12 oz, then 3 prices, $10, $20 and $30 . This
would give 3 x 3 x 3 possible product combinations.
A set of alternative “products” based on the attributes you have defined are presented to
respondents who make choices as to which product they would purchase in real life. It is important to
note that there are a lot of variations of conjoint techniques. SA uses a conjoint technique which we
feel best simulates the purchase process of consumers.
7. Why use Conjoint Analysis ?
Conjoint analysis is used to help evaluation new products, or variations of products,
against an existing range of products or a marketplace.
It is very expensive to develop a new product and then put it out into the marketplace
with no guarantee of success.
Conjoint analysis allows market researchers to simulate the decisions consumers
would make in the market place.
This means a company can get an idea about how a new product with be received in
the marketplace much more easily than if they had to really develop and market the
product. It is also used to see what effect changes in price of existing products may
have on the sales of the product.
With Survey Analytics conjoint analysis system you can get feedback on new
products or variations of an existing product very quickly and at a low cost.
8. 1) Attributes/Feature: Define the attributes of the products for your market. These
are the properties of your product.
Seattle Tourism Study: vs. Hours, Time of Day, Tour Type
2) Levels: The different properties of the attributes. Define at least two levels for
each of the attributes.
Seattle Tourism Study:
Hours - 3 levels
Time of day - 4 levels
Tour Type: 5 levels
3) Utility or Part Worth functions: These are what are produced by the conjoint
analysis. These can then be used to determine how important an attribute is to the
purchase or choice process and in “market simulations”.
4) Relative importance: how important an attribute is in the purchasing/choice
decision ?
Core Concepts
9. (1)Survey Analytics uses Multinomial Logistic
Regression for part worth calculations.
Used in calculating utility values for each level
(2) Survey Analytics use an Orthogonal Profile
Generation
Any set of attributes will have a minimal set of profiles
that can be generated to form a balanced design.
Have greater confidence in the results you receive!
Analysis
10. Set up a Discrete Choice Conjoint Analysis Study:
Add Instructions and Features
11. Set up a Discrete Choice
Conjoint Analysis Study:
Set Attributes /
features for each
Level
12. Set up a Discrete Choice
Conjoint Analysis Study:
Prohibited Pairs- You can create as many "Pairs" as you want and
the Engine will never display two levels that have been marked as
"Prohibited" in the same concept (as a product) for the user to choose
13. Set up a Discrete Choice
Conjoint Analysis Study:
Set Concept Simulator- This can be used to determine what
choices will be presented to the respondents when your survey
is actually deployed.
Click on the Simulate Concept
Choices button.
14. Set up a Discrete Choice
Conjoint Analysis Study:
Prohibited Pairs- Determine Levels not to be paired
together
Example: Weird Seattle tour can’t be 4-6 hours
Note: too many prohibitions are
not recommended - may skew results.
15. Set up a Discrete Choice
Conjoint Analysis Study:
Preview Using the Preview Option
24. Market Segmentation Simulator
Gives you the ability to "predict" the market share of new products and concepts that may
not exist today.
Ability to measure the "Gain" or "Loss" in market share based on changes to existing
products in the given market.
Important steps in Conjoint Simulation:
1- Identifying and describing the different products or concepts that you want to investigate.
We call these "Profiles".
Example one of the profiles could be: Tour Type: Weird, Hours: 1-2 , Time of Day: Evening
2- Find out all the existing products that are available in that market segment and simulate
the market share of the products to establish a baseline.
3-Try out new services and ideas and see how the market share shifts based on new
products and configurations.
26. Setting Up Simulator:
1) Click on Online tools >>Name Simulator Profile>>change
profiles
2)Click to see results . . . . .
27. Results: Simulator Output Defined
The market simulator uses utility values to project the probability of
choice and hence the market share
28. Now that we know
how to use this . .
What can we ask
and find out with the
Market Segmentation Simulator?
29. Market Segmentation Simulator
In our second example we have a 20%/80% split of market share when
we just vary the DVD or DVD with Podcast feature. What happens if we
change the guide to family guide from separate adult and children’s
guides ?
Answer: The market share for the Family guide with DVD option goes
down to 10% from 20%
30. Market Segmentation Simulator
In our first example what happens if have a tour of 1-2 hours as opposed to 4-6
hours in the afternoon for “Weird Seattle” ?
Answer: We find that the 1-2 hour tour would attract about 75% of the market
share.
31. Market Segmentation Simulator
Using our second example, what happens if we take the most preferred
profile and change the family exercise calendar from 60 days to 30 days
? What effect does the 90 day calendar have ?
Answer: We get a 60% market share for the 60 day calendar vs 40% for
the 30 day calendar. A 90 day calendar vs a 60 day calendar has 30%
market share for the 90 day calendar and 70% for the 60 day calendar.
60 days 60%
30 days 40%
60 days 70%
90 days 30%
33. You must use qualitative research first!
What are the top attributes?
What range?
What language?
A focus group or surveys with open-ended
questions will help define your top attributes
needed for your study
Best Practices & Tips
34. You need some numbers to get good statistics
Best Practices & Tips
Sample size is a question that comes up very frequently. Richard Johnson , one of the inventors of conjoint analysis,
has presented the following rule of thumb for sample size in choice based conjoint:
(nta/C) > 500
Where n = the number of respondents, t= the number of tasks, a=the number of alternatives per task , C= the largest
number of level for any one attribute.
So if you have 50 respondents, 3 tasks per respondent, 2 alternatives per task and the maximum number of levels on
an attribute is 3 you get:
(50 x 3 x 2 x 3) = 900
The general opinion now seems to be that 500 may be too small a number, 1000 is a better value.
Generally speaking sample sizes tend to be around 200 – 1200 respondents, admittedly a wide range. It does seem
that the value of 300 comes up most often for a single group of subjects.
35. Keep the options clear and simple as possible
No more than 10-12 trade-off exercises (5-7 standard)
No more than 5-6 attributes
Keep the ranges simple
You can ask more intimate questions of current customers than
potential customers, but don’t let that stop you from trying!
Follow general good online survey techniques
Test your survey
Make it clear responses keep strictly confidential
Keep survey results to 15-20 minutes
Provide incentives
Best Practices & Tips
36. Pricing for SurveyAnalytics Conjoint Analysis
SurveyAnalytics offers pricing on a monthly or yearly basis
Unlimited Conjoint Projects + Dedicated Account Manager +
additional Survey & Analytics Tools
Cost per Month: $500/Mo per user limited time discount
Yearly License : $6,000/yr per user
Billed via credit card only
37. Conclusion and Question & Answer Session
Esther LaVielle
SurveyAnalytics
http://www.surveyanalytics.com
esther.rmah@surveyanalytics.com
Andrew Jeavons
apj@andrewjeavons.com