Welcome to the Listening Post!

Newton, Miller and Smith's Extension to Van Westendorp

Tweet

 

One reader of my post Revenue Index Curves: Maximizing Gross Sales through Pricing Research asked, “But how do I generate a revenue index curve using Van Westendorp PSM?” (Van Westendorp PSM is a method of estimating price sensitivity.)

The answer is to use the Newton-Miller-Smith extension to Van Westendorp. At the 1993 AMA ART Forum, Jeff Miller (now president of Burke, Inc.), Dennis Newton and Paul Smith presented their adaption of the Van Westendorp question battery. One of Van Westendorp’s strengths is that its question battery asks about price points that are so low that customers are unwilling to buy products at those prices, but Newton, Miller and Smith thought the way Van Westendorp analyzed the survey results lacked reliability and validity. Therefore, they expanded the question battery with two additional questions:

• At <price just named as “cheap”>, what is the likelihood that you would actually buy the product?  Not at all likely, Slightly likely, Moderately likely, Very likely, Completely likely
• At <price just named as “expensive”>, what is the likelihood that you would actually buy the product?  Not at all likely, Slightly likely, Moderately likely, Very likely, Completely likely

They didn’t ask likelihood for the prices named as “too cheap” and “too expensive”, assuming that there was 0% purchase likelihood at those price points.

Based on their experience in tracking particularly product categories over time, they mapped answers to different probabilities – so a respondent who said “Slightly likely” might be assigned a 10% probability of purchasing, for instance. Absent such historical data, you can use 0% (Not at all likely), 25% (Slightly likely), 50% (Moderately likely), 75% (Very likely) and 100% (Completely likely). This is not to say that someone who tells you in a survey that they are completely likely to purchase a product will actually purchase a product, but for our purposes we are looking for relative revenue index curves and the actual values don’t matter so long as they are equidistant from one another.

You can then use these probabilities to prepare a revenue index curve.

Van Westendorp is just one method of using surveys for pricing research. See also:

• Willingness to Pay
• Monadic Price Testing
• Sequential Monadic Testing 
• Conjoint Analysis [by Michaela Mora of Relevant Insights]