Conjoint analysis captures how buyers trade off features and prices, so estimating Willingness to Pay (WTP) seems like a natural byproduct of the analysis. However, traditional methods for estimating WTP tend to overstate it and are unrealistic because they don't consider relevant competition or that the buyer can walk away (choose the "None alternative").
Conjoint Analysis Strengths
Since the inception of conjoint analysis in the 1970s, managers have naturally wanted to use this valuable tradeoff technique to estimate how much buyers are willing to pay (WTP) for enhanced product features. This seemed like a natural request, since when price is included as an attribute in conjoint analysis studies we observe how respondents trade off features vs. price in their product choices. Conjoint analysis mimics the buying process and is much better for WTP estimation than methods that naively ask people to state how much they are willing to pay for features.
Using established approaches like CBC (Choice-Based Conjoint, also known as DCM) leverages the most trusted methodology in the industry for getting at WTP. Proper execution is key. If we interview respondents who are in the market to buy, who are informed, and we clean the data to remove speeders and random responders, then we are doing the right kinds of things to set us up for success.
Weaker WTP Approaches
Despite the strength of conjoint analysis, not all WTP calculations using conjoint data are created equal. Two common approaches, the algebraic approach and the two-product 50/50 simulation approach, have been used for decades and tend to overstate WTP. They overstate it largely because they fail to consider competition: respondents might obtain identical or substitute features elsewhere in the marketplace. An example involving the TV show Gilligan’s Island illustrates the issue. The rich Mr. Howell would be willing to pay a million dollars or more to a boat that offered passage back to freedom. But, if a second boat shows up offering passage for $500, Mr. Howell will choose that option. A more managerially useful conception of WTP should account for the competitive realities in the marketplace, otherwise we risk overstating WTP.
With WTP, we estimate the monetary amount that buyers are willing to pay for a product enhancement compared to a reference (base) level of an attribute. For example, the reference level might be “low quality” and we estimate the WTP for “high quality” relative to low quality.
Better WTP Approaches Considering Competition
About 20 years ago, we at Sawtooth Software recommended a market simulation approach to estimating WTP involving a client’s offering vs. a rich set of competitive offerings (including the possibility of the “None” alternative). First, we record the share of preference for the client’s offering in that competitive context. Next, we enhance that offering (which increases its share) and find the increase in price that drives that share back to the original share prior to making the enhancement. That difference in price that equalizes share is taken as WTP. This approach focuses on respondents on the cusp of choice, rather than averaging WTP across all respondents. If you have a good sense for your product and your client’s product positioning, then you can estimate WTP using that realistic and relevant competitive context.
Sampling Of Scenarios (SOS)
With Sampling Of Scenarios (SOS), you don’t have to define a specific base case competitive scenario for your product offering and competitors. Sometimes we don’t know quite what our offering will be, let alone our competitors. With SOS, we can draw 100s of randomly designed simulation scenarios (random attribute specifications for your product and competitive products). For each randomly drawn simulation scenario, we compute WTP in the way we described above and we summarize the WTP by taking the median result across these random scenario draws. This approach estimates a generalized WTP for each feature relative to all possible competitive positioning and reactions.
We’ve compared the traditional methods for WTP (algebraic and two-product approaches) to our generalized SOS approach to WTP for nine commercial CBC datasets we had on hand. On average, the SOS approach leads to WTP values about 20% lower than the algebraic approach. For a couple datasets, the WTP is cut about in half with the SOS approach.
An additional strength of SOS is the extension in which we specify that we have (say) a patent on a feature and that competitors cannot take on that feature. Furthermore, we can assume that our offering carries our brand name and competitors cannot use our brand name. We find that the WTP for a feature depends on whether we assume exclusive rights to that feature and the strength of our brand equity.