What People Say Versus What People Do


The problem with a lot of interview based shopper research is that the responses are not verified. The answers are simply taken at face value. But by digging a little deeper, it’s usually revealed that the respondent isn’t accurately saying what they will do.

History is full of bad marketing decisions caused by only listening to what people say they will do:

  • New Coke: Based on favorable taste test reviews in focus groups, Coke went ahead with this new formulation and quickly discovered it flopped in market.
  • The Ford Edsel: A complete flop based on market poll data from car owners.
  • Personal Stereos: When asked, people said they would never give up their home stereo in favor of earbuds yet today most of us listen to music using earbuds.
  • Bank Machines: when first introduced people said they wouldn’t use them and would rather get cash from a teller in a bank.

There is much disagreement about how accurate questionnaires are. However, one area that most researchers agree upon is that behavior based approaches are a more accurate reflection of what will occur in market.

So, why do people say one thing and do another? It could simply be that the respondent believes the answer they are giving is correct. Or it could have to do with the fact that what is being asked about happened too long ago to accurately recall (small purchases that happened more than a month ago). Also, a respondent may not be able to find an answer that accurately reflects their beliefs (multiple choice) or the interview may take longer than expected and so they rush through the answers. All these issues can be avoided with improved questions,  shorter surveys and a behavioral research approach.

Here are 3 reasons why respondents may give an answer that doesn’t reflect their real behavior:

  1. They have an intention- to lose weight, to save more money or exercise more. However, these intentions don’t always translate into behavior. When asked they are telling you their intention.
  2. Respondents may be overly enthusiastic about a product tested in isolation of the retail environment. They may inflate their likelihood of buying a product in absence of the product being tested in the retail environment.
  3. Respondents give socially desirable answers. The best example of this is voting polls. Survey estimates of voting are always higher than actual voting turnout because there is a social expectation that you vote. So, when asked, people say they vote even though they don’t.

So how do marketers get to the real answers to improve the quality of their research.

  1. Start by measuring behavior. See what people react to in a real life-sized environment. Don’t ask any questions during this phase just measure their behavior.
  2. Use technology to measure the subconscious drivers of behavior, so instead of asking people about their behavior you can measure what they are responding to positively or negatively. We always say that behavior trumps attitudes in research.
  3. When you do ask questions, ensure appropriate quality control measures are in place to filter out any respondents who are not paying attention. Simple to do and a minor investment to get better results!
  4. Ideally a more in-depth one-on-one interview approach allows for the interviewer to probe more and to pick-up on the nuances of what a respondent is trying to communicate.

Contact us if you would like to learn more about our behavioral approaches to market research.

Anne Stephenson
Partner, Explorer Research

anne stephenson

Marketing Research Best Practices