Emotient - Finding Predictive Capabilities Through Facial Recognition

Emotional testing, neuroscience and community based measurement are expanding the boundaries of traditional research by offering new, insightful tools to predict audience and consumer behavior. A particularly interesting example is the work being done by Emotient which is a San Diego based research company that has been operating in the emotional measurement space for about three years. Their position is that emotions drive spending and that by focusing on facial recognition and registering an emotion via a facial expression, it is possible to predict the strength of any creative such as an advertising message or even a political candidate.

There are several companies using biometrics to help marketers ascertain the non-verbal and unstated impact of their content. Tools like Galvanic Skin Response (GSR) which measures changes in the skin and Functional Magnetic Resonance Imaging (FMRI) which measures brain activity through the blood flow, as touted as good methods to measure subconscious consumer response. But according to Marian Bartlett, Ph.D., a Co-Founder and Lead Scientist of Emotient, GSR’s are noisy and slow while FMRIs are slow and cumbersome.

Emotient’s IP has the ability to capture hundreds of facial expressions at one time in a non-intrusive manner which, they say, offers the capability to amass big data by individuals’ gender and age. There are nine basic emotions, each offering an insight into the desirability (or lack thereof) for the messaging and creative, and twenty more facial muscle movements that can be captured. And, with technology, facial expressions can be captured anywhere and, once opted-in, passively, not requiring a visit to a laboratory and donning a gel cap or wearing a measurement vest. Imagine gathering data via a tablet or smartphone or in a car with a sensor pointed at the driver.

The result is an ability to scale and to create a range of baselines based on creative or product type, gender and age, for example. Joshua Susskind, Emotient’s Co-Founder, Lead Deep Learning Scientist, explained that via their methodology, P&G’s Tide detergent was able to more accurately predict intent to purchase across three different types of detergent fragrances using facial recognition while the survey the client fielded at the same time, could not.

One of the more interesting applications of Emotient’s methodology was in predicting the winner of the first Republican debate on Fox News. According to the measurement of audience reactions through the debate, Emotient projected that Donald Trump would be, by far, the winner. This was based on a preponderance of facial expressions of “Joy” for Trump while the other contenders elicited expressions of “Anger” for Scott Walker and “Disgust” for Chris Christie and Rand Paul.

This certainly makes for some compelling arguments for the expanded use of facial recognition to help move the media needle. Such methodology might become a tool for creatives that can be applied to advertising messages so that the ads can be placed in program content with the greatest emotional match. It can also be used to fine tune a message so that it better resonates with a desired target audience and can be placed in compatible content. A fast food company whose commercial angers women may want to tweak their message for a more engaging and positive response from them or place their ad in more male dominant content. Interestingly, not all negative emotions are bad for messaging, according to Susskind. An ad that elicits “Disgust” may be perfect for detergent but not for food.

But the real question is, if actual cause and effect can be measured through the scalability of emotional reactions, do we need to rely on our current, historical data-centric stats that are less nuanced and predictive? Isn’t it time to integrate these new neuroscience-based measurement capabilities and approaches into our current measurement toolkit?

My opinion is that, as an industry, media has been historically wedded to delivery metrics such as ratings and GRPs. These measurements have been embedded in the system since inception. While new technology and neuroscience applications are certainly exciting and can produce impactful results, I am not sure they will ever replace standard metrics as the next currency. These new solutions will more than likely remain as a consultative, supportive measurement tools. Our industry moves slowly and is loath to change from those data-driven metrics that have become the bedrock of our business. But, who knows? I could be wrong.

No comments:

Post a Comment