At the beginning of my research career I grew accustomed to clients asking us for proposals using a methodology that they had pre-selected. In many cases, the client would send us the specs of the entire job, (this many completes, that length of survey) and just ask us for pricing. While this is certainly an efficient way for a client to compare bids across vendors, it didn’t allow for any discussion as to the appropriateness of the method being proposed.
Today most research clients are looking for their research suppliers to be more actively involved in formulating the research plan. That said, we are often asked to bid on a “conjoint study.” Our clients who’ve commissioned conjoint work in the past are usually knowledgeable about when a conjoint is appropriate, but sometimes there is a better method out there. And sometimes the product simply isn’t at the right place in the development “chain” to warrant conjoint.
Conjoint, for the uninitiated, is a useful research tool in product development. It is a choice-based method that allows participants to make choices between different products based on the product’s make-up. Each product comprises various features and levels within those features. What keeps respondents from choosing only products made up of the “best” features and levels is some type of constraint – usually price.
We look to conjoint to help determine an optimal or ideal product scenario, to help price a product given its features, or to suggest whether a client could charge a premium or require a discount. It has a wide range of uses, but it isn’t always a good fit:
- When the features haven’t been defined yet. One problem product developers face is having to “operationalize” something that the market hasn’t seen yet. You need to be able to describe a feature, what its benefits are, and its associated levels in layman’s terms. We can’t recommend conjoint if the features are still amorphous.
- When there are a multitude of features with many levels or complex relationships between the features. The respondent needs to be able to absorb and understand the make-up of the products in order to choose between them. If the product is so complex that it requires varying levels of a lot of different features, it’s probably too taxing for the respondents (and may tax the design and resulting analysis as well). Conjoint could be the answer – but the task may need to be broken up into pieces.
- When there are a limited number of features with few levels. In this case, Conjoint may be overkill. A simple monadic concept test or price laddering exercise may suffice.
- When pricing is important, but you have absolutely no idea what the price will be. Conjoint works best when the product’s price levels range from slightly below how you want to price it to slightly above how you want to price it. If your range is huge, respondents will gravitate toward the lower priced product scenarios and you won’t get much data on the higher end. It may also confuse respondents that similar products would be available at such large price differences.