Why Facial Recognition Measurement Has Put A Frown On My Face
Having recently discussed the multitude of unhelpful ways in which we now measure engagement, I'm less than thrilled to discover that no less a body than the BBC is now experimenting with facial recognition software in order to measure emotional responses to content marketing campaigns.
Apparently, the BBC used software from CrowdEmotion to measure the 'conscious and subconscious emotional responses' of a research group of some 5,000 people. Conscious responses were measured, one presumes, by eye tracking and propensity to click, while unconscious responses were measured by the facial recognition software, which recorded a choice of six possible emotions: sadness, happiness, puzzlement, fear, rejection and surprise.
Leaving aside the fact that I have seldom been made to feel rejected by content marketing ("I'm sorry Kev. It's not me. It's you. And I want my CDs back."), I must admit that I struggle to see the point of measuring emotional responses, for a number of reasons.
Turn That Frown Upside Down, And It Still Won't Tell You Much
If I'm honest, my initial reaction to the BBC's use of facial recognition was pretty positive. After all, when I was privileged to witness a trial of brainwave analysis at King's College Hospital, courtesy of a company called NeuroSense, I was absolutely blown away.
They can seemingly overcome all of the limitations of traditional market research by analysing brain activity to establish the veracity of subject responses and thus bridge the 'truth gap' between what people say and what they really think. However, taking advantage of their impressive methodology requires what we in the trade call: 'a pretty significant budget.'
So if facial recognition software could democratise the opportunity to get at the hidden truth, and bring it within the reach of every brand, we might well be interested. However, I do have to question the value of knowing that 90% of respondents frowned at our latest campaign, while 9% looked happy and a further 1% looked as if their significant other had just run off with the milkman.
After all, like many people, despite being warm and charming in the extreme, I am somewhat prone to the first world condition known acronymically as RBF - or Resting Bitch Face - particularly when I am deep in thought; so what might a researcher really read into a welter of frowns? Confusion? Unhappiness? Deep interest? Or simply an unhelpful cocktail of all the above…
Recognising The Strengths - And Limitations - Of Facial Recognition
Plainly, if you produce comedy pilots for the BBC, research based on facial recognition techniques could be invaluable. After all, if 90% of the audience looked happy 90% of the time during the test screening of a new comedy, it's probably time to tie the cast and writers down and get straight into production.
That's because, at the risk of stating what John Cleese likes to call "the bleedin' obvious", the job of comedy is to make you happy, which is a pretty easy emotion to gauge.
Unfortunately, the job of marketing is to sell; and measuring the range of emotions people go through during the decision-making cycle is not only a great deal more complex, it would demand that measurement took place over a sustained period of time, and would need to be allied to many other measurements, not least of which is the propensity to stop smiling and simply make a purchase.
Can Market Research Really Stay In Your Face All Day Long?
Some research organisations do indeed follow their 'panels' around: using 'People Meters' and 'Return Path Data' to take note of not simply how many TV programmes they consume 'live' and how many through time-shift viewing on other platforms, but also how many radio commercials and pieces of music they hear, albeit subconsciously, during their daily round.
Perhaps, one day, facial recognition software may also do this, enabling us all to enjoy a more fully-rounded single customer view. In fact, I'm given to understand that CrowdEmotion is already working on facial recognition software that will serve next page content based on the emotion currently registering on one's face, as opposed to simple eye tracking.
However, impressive though that is, it seems that while facial recognition software is clearly a useful security tool for spotting when terror suspects walk into airports for example, it is still very much in its infancy when it comes to decoding the real meaning of facial expressions.
In fact, the last I heard, everyone from experienced customs officers to skilled police interviewers and body language experts read pretty much everything but faces in order to spot dishonesty, tension and alarm in their subjects. They look at the way you sit, the degree of eye contact you make, hand gestures and all sorts of other non-facial and non-verbal 'tells' in order to decide what you're thinking.
So, as long as I continue to know so many people who radiate grumpiness when they're simply thoughtful, not to mention smiling broadly through a brilliant hand of poker when they're holding anything but, I'll probably have to reserve judgment on the ultimate usefulness of facial recognition as a content marketing research tool.
Article first published here on 03/02/16