Method category: Evaluative market experiment
How to Use This in GLIDR
Conjoint Analysis is a quantitative survey method that will help you figure out which particular combination of features or attributes are most important to your customers. There are various software tools for performing this analysis, linked at the end of this article.
In GLIDR, you can use this technique in an experiment to determine product features and attributes. First, make an assumption (idea) for each feature in the test. Then, connect those ideas to an Experiment with a hypothesis about your expected results. Run the Experiment and add the Conjoint Analysis as Evidence (you can attach the survey itself and any raw data to the Evidence). In the Analyze phase, figure out what you learned from this Experiment and update your project accordingly.
Learn more about each of those aspects of GLIDR:
Broken Conjoint Analysis
Article excerpted from The Real Startup Book
A complex survey method where customers choose between product offerings that have different attributes such as price, screen size, or weight. Statistical analysis is then used to reveal the relative value of each attribute and predict the value of each possible combination of features.
Does a particular new feature have any value in the eyes of the customer?
Which product attributes are more/less important to the customer?
What dollar value can we be assigned with each feature?
Which features should we build next?
Time Commitment and Resources
1-2 hours offline for B2C or 24 hours online to gather responses. For B2B, participantrecruitment times can vary widely. Analysis of the data can be very rapid with off the shelf
software for analyzing less than 10 attributes. For analyzing dozens of factors the expertiseand software required can take several weeks and human analysis.
Create a list of the top 3-5 product attributes you want to rank based on deep consumer understanding gained from prior experience and research.
No hard & fast rule but a good guideline is: Sample Size = [ (Total # of levels for allattributes) - (Number of attributes + 1) ] x 10 -- ref
Use software program to mix attributes into new product offerings
Show participants selected product offerings side by side and have them choose which one they would prefer (see example below from Pragmatic Marketing article)
Use statistical analysis software (or consultant / service) to analyze the results and compute a relative value ranking for each stand-alone attribute
Output formula revealing the “ideal” product based upon optimized mix of attributes.
Conjoint analysis can be complex and, depending on the tools and exact statistical method employed, the results from the analysis be be extremely difficult to understand as raw statistical data:
Other methods of displaying the results can be more straightforward, but lack details:
What matters most is:
The relative importance of each attribute compared with the others
The formula that allows you to predict the relative preferences of any kind of mix
Elasticity of demand for pricing simulations (presuming that price was an attribute that was tested)
Conjoint analysis is most often used in existing markets where the product attributes are generally known by the customer. When brand new attributes are introduced, customers may not initially understand them and therefore may not be able to accurately include the potential value of those attributes in their choices, producing a false negative of sorts.
For this reason, generative market research methods are generally used before conjointanalysis to ensure that the attributes being tested are the correct one.
The method also tends to be extensive and requires a high level of expertise to design. Earlystage innovation projects therefore rarely use this method.
Confirmation Bias: If administering the surveys face-to-face, overly enthusiastic entrepreneurs will sometimes over explain, correct, or nonverbally prompt the participant with the desired answer.
Invalid Target Audience: Key to the success of conjoint analysis is knowing your audience well enough to be able to create products with useful mix of attributes to begin with. Conjoint analysis works best for products and services that rely more on logical comparison and less on emotion or impulse.
Homogenous Market: Boiling market segments down to a series of equations and values has the drawback of treating every member of that segment identically. This is can often be reasonable within the specific price range studied but outside of these ranges, the mix of desirable attributes can change quite dramatically as customers enter and leave the market.
Indication of Price Sensitivity NOT Exact Pricing. Conjoint analysis is great for providing an indication about which variables and ranges influence pricing but a separate pricing and elasticity of demand analysis should be performed separately to really nail this down.
Got a tip? Add a tweetable quote by emailing us: firstname.lastname@example.org
MIT Sloan Courseware - Note on Conjoint Analysis by John R. Huaser
Chris Chapman - 9 Things Clients Get Wrong About Conjoint Analysis
Brett Jarvis - Conjoint Analysis 101
Sawtooth Software - Interpreting the Results of Conjoint Analysis
Got a case study? Add a link by emailing us: email@example.com
Conjoint Software Resources
Conjoint Survey Design Tool , Harvard 2014 (Free)
XLSTAT-Conjoint ($50 Student, $275 private)
Conjoint Analysis in Excel (Free)
Choosing By Advantage ($30/mo, similar to CA)
Survey Gizmo ($95/mo, free 7-day trial)
1000minds.com ($20,000 for enterprise, free for students)
Got a reference? Add a link by emailing us: firstname.lastname@example.org