Derived vs Stated Importance
PMG generally favors using derived importance rather than stated importance for two reasons. First, customers will be hesitant to rate any attribute as unimportant since there is a perception that in so doing, the vendor will reduce efforts on that attribute or redirect resources to other areas. Customers fear the possibility of losing what they currently have, so the tendency is for customers to rate every attribute high on the scale. When we have asked customers to rate importance (stated) we often see all of the attribute average importance ratings fall between 6 and 7 on a 7-point scale. The bunching of the averages at the top of the scale makes it impossible to isolate the attributes that are truly the important drivers. On the other hand, using derived importance typically yields greater variance between attributes, providing a much clearer picture of the most important attributes.
The second reason we like to use derived importance by not asking importance directly, the survey length is dramatically reduced – as is the potential for respondent fatigue. By using derived importance, the overall number of survey questions is reduced by the number of attributes, as participants only need to rate performance for each attribute.
It is also interesting to look at stated versus derived importance as an analytic tool. Results can be mapped in a four-cell matrix with stated importance on one axis and derived importance on the other axis. The attributes of interest are those that are off the diagonal of the matrix. Attributes with high stated importance and low derived importance are false signals, those where customers are saying something is important but the relationship to the outcome variable is tenuous. Attributes with low stated importance and high derived importance are hidden levers, those that customers do not talk about, but are in fact strong drivers of the outcome variableis because by
Although the temptation is “to just ask” the importance of attributes, we would strongly recommend deriving attribute importance. It creates clearer, more distinct results, and it shortens the overall length of the survey.