Facial expression analysis to detect emotions – a Diet Coke case study

John Habershon
Momentum Research, United Kingdom


Non-verbal communication is something which is part of our business as qualitative researchers, but conventionally an implicit part. We take as given that the moderator, for example, will pick up non-verbal signals from the respondents, using them to gauge the dynamics of the group. We regularly scan the table to detect if anyone is looking uncomfortable or appears to disagree with what other respondents are saying. In the debrief a client would expect to hear, not simply a report of what was said, but observations and insights on how it was said – body language, tone of voice, facial expressions – and what was not said.

But this level of engagement with non-verbal information is neither explicit nor systematic. Moderators are assumed to be sensitive to non-verbal clues, to have a high degree of emotional intelligence. If we say a product is greeted with suspicion, or it is more attractive than respondents want to let on, then clients will tend to believe us, without evidence as such. In a sense the role of a qualitative researcher is of an emotional conduit between customers and clients. We are hired for our ability to understand consumers and to an extent this ability can't be monitored or measured – the proof is in the quality of our analysis and conclusions.