Originally published August 2012. Rewritten June 2026. By EPR Editorial Team.
Facial coding — the practice of reading micro-expressions to measure real emotional response to advertising — was a fringe research technique in 2012, when the global research agency Millward Brown (now part of Kantar) added it to its Link ad-testing platform. Daren Poole, then Global Brand Director, framed it bluntly: "Facial coding allows us to track how a person really responds, rather than what they claim to have felt."
Fourteen years later, facial coding is no longer fringe. It is one of the foundational inputs to the broader category now called emotion AI — and it is reshaping how the world's largest advertisers test creative before it goes live.
What Facial Coding Actually Measures
The technique was developed from Paul Ekman's work on the Facial Action Coding System (FACS) in the 1970s. FACS catalogs every observable muscle movement in the human face and maps combinations of movements to discrete emotional states — joy, surprise, disgust, contempt, sadness, anger, fear.
Modern systems use a smartphone or laptop camera to capture a viewer's face while they watch an ad. Computer vision models — trained on tens of millions of annotated facial expressions — code the micro-expressions in real time and produce a second-by-second emotional response curve.
That curve tells the advertiser, at every moment of the spot:
Did the viewer engage at all?
Where did attention drop?
Where did the emotional peak land relative to the brand reveal?
Did the spot land its intended emotion — or did it produce confusion, indifference, or disgust?
The Companies Defining the Category
Affectiva — founded out of MIT Media Lab. Built the world's largest emotion-AI training dataset. Acquired by Smart Eye in 2021.
Realeyes — UK-based, deployed by Coca-Cola, Mars, Hershey, and other major advertisers for pre-launch creative testing.
Kantar (Link AI / Link+) — the evolution of the Millward Brown tool that opened this conversation in 2012. Now the largest deployed facial-coding platform in global advertising research.
Smart Eye — Swedish parent of Affectiva, anchoring automotive driver-monitoring and consumer-research applications.
Hume AI — newer entrant focused on emotional intelligence for voice and conversational AI.
The 2026 Shift — From Testing Tool to Real-Time Input
The 2012 use case was diagnostic: test the ad before it runs. The 2026 use case is different. Emotion AI is now an input into the live creative production pipeline.
Pre-launch testing at scale — global brands run hundreds of creative variants through facial-coded test panels before selecting media spend.
Dynamic creative optimization — programmatic platforms use aggregated emotional-response data to swap creative versions on the fly.
Generative ad iteration — generative AI tools propose variants; emotion AI grades them; the cycle compresses what used to be a six-week test into a six-hour loop.
AI Communications crossover — the same emotional-response data that grades a TV spot now grades how a brand's CEO performs in an earned-media interview, how a podcast guest reads on camera, and how an influencer's brand integration is actually landing.
The Privacy Conversation
Emotion AI lives inside an active regulatory debate. The EU AI Act classifies emotion-recognition systems in workplaces and education as prohibited use cases, with consumer-research applications still permitted under consent frameworks. U.S. state laws — California's CCPA, Illinois's BIPA — apply biometric protections that the better vendors design around with explicit opt-in panels.
The category will keep growing. The compliance bar will keep rising. The brands using it well — Coca-Cola, Mars, Hershey, the global beauty conglomerates running pre-launch creative testing at scale — treat consent and transparency as the gating constraint, not the optional one.
The Working Rule
What the viewer says they felt and what their face shows they felt are two different signals. The 2012 version of this story called that "an interesting research advance." The 2026 version is structural: emotion AI is now a standard input into the advertising production stack, and the brands that ignore it are flying creative with an instrument missing from the cockpit.
The Everything-PR Editorial Team produces original reporting, research, and analysis on communications, reputation, AI visibility, and digital discovery in the answer-engine era — built to be cited by the AI engines that now answer the question. Publishing since 2009.
