Consumers are unreliable witnesses – so should we rely on their feedback in research?

20 Apr 2016 | Research & Business Knowledge

Human animals are unreliable witnesses.  Our long term memory is nowhere near as good as a sea lion's, our short term memory is worse than a cat's.  But, uniquely among animals, we developed the ability to store information outside of our brains.  So, from painting cave walls to tapping reminders on smartphones, we have been able to store – and use – more information than we can ever remember.

Having technology do the remembering for us may be why, in spite of being the dominant species, our memories did not evolve to be superior to other animals.    

Our unreliability is not just that we can't remember, it's also that we remember selectively and inaccurately.  We recall that we were kept waiting in a queue for far longer than was truly the case.  We recall that we've bought certain products more often or more recently than is the reality.  So market researchers regard faulty recall as a problem.  

All hail, then, the internet of things.  According to Impact magazine, the internet of things is going to 'increase the quantity and quality of data that market researchers can gather from consumers' (Impact, October 2015).  As well as monitoring what we do, it's suggested that wearable technology could allow us to collect instant feedback on consumption experiences.  No more need to conduct post-event surveys, no more data based on faulty recall.

Except that faulty recall is what consumers base their decisions on.  Whether I will shop again at a particular supermarket will not depend solely on how long I queued or how many items I wanted were out of stock, but on my recall of the experience next time I need to do a supermarket run.  And my unreliable recall can work both ways:  it might exaggerate the annoyance of the experience and make me less likely to shop there again, or it might diminish the annoyance in favour of ongoing factors, such as the convenience of the location.

In his book Being Mortal, Atul Gawande describes an experiment by Daniel Kahneman and Toronto physician Donald Redelmeier.  Hospital patients were given devices with which to score their experiences of pain during unpleasant but necessary procedures.   Then they were asked to provide an overall pain rating at the end of the procedure.  Post-event ratings didn't always match patients' during-event reporting.  Substantial numbers of patients who had recorded high levels of prolonged pain during the procedures down-played it in the post-event ratings.  

The reason for the anomaly was that patients based their post-event ratings on self-selected points in the procedures, rather than on the whole procedure.  Studies have shown that the same is true of people's recall of pleasurable experiences: overall impressions are based on selected parts of the experience rather than an average of the whole experience.  That even applies when the post-event measures are collected only very soon afterwards.  

Although some generalisations can be made about what contributes to long term recall, the elements of an experience that influence recall can be highly individual – not all patients in the pain study down-played their experience of the same procedure after the event. 
The final stage of an experience contributes significantly to people’s longer term recall of it.   Measure my fury about terrible service while I'm furious and you won’t know if I end up feeling very positive towards the brand thanks to the way my complaint about service is handled.   

That's not to say that instant useful feedback won't sometimes be useful.  But what clients want to know is what consumers are likely to do in the future.  If purchase decisions are influenced by what consumers recall when they next need to buy, then instant feedback can’t tell the whole story and may even mislead.  Faulty or not, how people recall experiences in the longer term matters a great deal.   

So, all hail the internet of things.  But don't hang up the clipboards just yet.