asking respondents to make choices among a set of products is more realistic than rating each product, or attribute of a product, individually.
Tech blogs are raving about yesterday’s release of Microsoft’s Surface Book. Some are calling it a ‘revolutionary’ product in a category that’s been lacking innovation. But product development isn’t easy and there is no guarantee that this product will succeed in the marketplace. Stakeholders are likely asking themselves - Did we incorporate the right combination of features to generate substantial interest? Did we price the product so that it’s both competitive and profitable?
If I were in Microsoft’s shoes, I’d probably solicit the feedback of consumers to help me answer these questions before going to market. And if I were a betting woman, I’d bet Microsoft did just that.
While there are many marketing research techniques that can aid in the product development cycle – both qualitative and quantitative – there is usually always a ‘product pricing’ phase. In this phase, one typically asks:
- How much are people willing to pay for this product?
- What do we have to offer to take share away from our competitors?
One solution, is choice-based conjoint analysis, or CBC. In a CBC exercise, a respondent is forced to trade-off different product scenarios, uncovering the attributes (brand, price, etc.) that drive them to their decision. Why is this ideal? Because it’s hard for consumers to articulate what they want in a product. Take the example below, how would you respond?
Are you willing to pay more for a rear camera, stylus pen, and removable keyboard? Or are you an Apple loyalist? Maybe you are budget conscious? Or maybe you don’t like any of these options!
While this is a hypothetical scenario you can see that asking respondents to make choices among a set of products is more realistic than rating each product, or attribute of a product, individually. In choice-based conjoint analysis, respondents complete a series of questions, typically 10 to 30, similar to the example above. The questions are designed carefully, using experimental design principals of independence and balance etc., so that based on the respondent’s choices, we can statistically deduce which features are most desired and which have the most impact on choice. The result of the analysis is a full set of preference scores, or part-worth utilities, for each attribute level in the study. These scores allow companies to gauge the price sensitivity and elasticity of their products. The scores can also be taken into a market simulator to test a variety of what-if scenarios, estimate shares of choice and even understand substitution (including cannibalization) effects, offering valuable advice on how to price your product when you go to market.
There is so much more to applying conjoint analysis than is explained here... Interesting concept? Check out an “old school” technical paper that sampled respondents by sending discs in the mail and before individual estimation was even possible in choice-based conjoint!