Tips For Making Conjoint Useful ¹

By Paul Richard McCullough

Download This Article

Conjoint Analysis is the single most powerful technique available to the marketing researcher. How else, with a single study, can you answer all of these questions, as well as many others:

• Which product features should I include in the final product?
• Which product features should I emphasize in marketing communications?
• What price will generate the most revenue? The most profit?
• Who should be my primary target segment? Secondary segment?
• How should my messages differ by segment?
• What will first year sales be?
• How will my sales be affected by competitor changes?

But like all powerful tools, conjoint analysis needs to be handled properly to get the most out of it. Here are some tips to make your next conjoint study more actionable:

• Use Lots of Attributes: Use enough attributes to make your choice tasks realistic. The world is complicated. That usually means you’ll need a lot of attributes in your conjoint study if you want your conjoint tasks to look anything at all like your marketplace. Don’t be afraid of a large number of attributes. There are a variety of good ways to design conjoint studies today that can accommodate a large number of attributes.

• Build Complex Experimental Designs: Design your study to reflect the complexities of the market. Having a large number of attributes usually isn’t enough to mimic the real world. Nested variables, such as separate price attributes for each brand, are often powerfully informative. Multiple product categories combined in the same conjoint exercise are almost essential in categories with extensive substitutability. Don’t let the conjoint design dictate what the market looks like. Let the market dictate what the design looks like.

• Use Pictures: Use visuals to express product attributes whenever possible. Visuals, such as photos of packages, communicate quickly and effectively large numbers of attributes. Visuals can also communicate more profoundly. Think of the word “Porsche” and compare that experience with seeing a metallic blue 911 gleaming in the sunlight. But maybe that’s just me.

• Do Multiple HB Runs: Hierarchical Bayes is commonly used to estimate disaggregate choice utilities. But HB tends to flatten responses across subgroups. If you have subgroups that you know you want to look at separately and you suspect they should have different utilities, estimate their utilities in separate HB runs. It’s a little more work but the results make more sense. That’s useful when you’re presenting to non-geeks, like the Sales VP.

Conjoint analysis is the most powerful technique available. But like driving that Porsche on a residential street, we rarely use its full potential.

¹ A version of this Newsletter was sent to our subscription list in March,2010.

MACRO CONSULTING
Intelligent Decisions