When you have worked with survey data for over thirty years – gulp – you take for granted things that other people do not. For instance, unlike many, I tend to assume that consideration and stated purchase intent questions do anticipate consumer behavior. However, I do not assume that an individual’s stated intentions will guarantee they buy the brand they state.
I decided to write this post after I heard yet another person decry the ability of surveys to anticipate consumer behavior. The speaker, as with many marketers, evinced far more confidence in behavioral trend than the survey data. Uh oh, I thought, here we go again. Yet another person assuming that one single measure of attitude is going to anticipate behavior, or rather, finding that it does not.
Now, let us be clear, there is academic evidence to suggest that measures of stated intention are not as accurate at predicting behavior as we might like, lending credence to the general belief. But again, is the general assumption that one data point is likely to anticipate behavior valid in and of itself? Honestly, I think it is naïve to assume that there is a direct link between stated intentions and behavior when there are just so many ‘out of context’ factors that could influence people’s actual behavior.
For a start, it depends on what product category you are dealing with. Purchasing of consumer packaged goods purchases is innately variable. They are typically a repertoire purchase, with people seeking to satisfy different needs and occasions, buying for others, and dealing with the inherent vagaries introduced by out-of-stocks, in-store promotions and a desire to try something different for a change.
That said, even stated intentions of packaged goods purchase prove reasonably accurate over time. “Beneath the surface the hidden world of individual buying behavior” by Kyle Findlay, Constantin Michael and Jan Hofmeyr (available with WARC subscription) reports an analysis which compares stated purchase intention, e.g. buy most often, against actual purchasing behavior from panel and loyalty records and finds little relationship over three months but a far stronger relationship over 12 months. Given the chaotic influences on any individual packaged good purchase, this is hardly surprising. It takes time for people’s intentions to play out.
Similarly, while the predictive power of individual stated intentions may be relatively weak, when aggregated up their predictive capability is far stronger, as the influence of chaos tends to cancel out. Predictions can be further improved by taking into account factors known to influence the probability of purchase, like brand size (a surrogate for physical availability) and relative price (people are less likely to follow through on their intention to buy if a brand proves more expensive than its competition).
The practice of marketing is one of influencing the probability that someone will buy a specific brand (you cannot guarantee it). So why do so many people seem to assume that probabilities do not apply to stated intentions, or that one survey measure is likely to prove an adequate predictor of behavior? Please share your thoughts.