Facial tracking to segment ads conjure Big Brother

Shopper loyalty cards have been around for decades, giving card holders everything from discounts on groceries, entry into giveaways, earning cents-off fuel points, and coupons targeted very specifically on shopping behavior. But Tesco, a large grocery chain in the UK, has taken this a step further – with facial recognition software installed on cameras at pay tills at their gas stations.

These cameras, which are completely built into the pay at the pump feature, scan numerous facial recognition parameters to determine basic information (gender and approximate age) and then spit out a video ad tailored to that particular audience.

Biometrics capture numerous measurements to determine age and gender in order to serve ads targeted to those demographics.

Biometrics capture numerous measurements to determine age and gender in order to serve ads targeted to those demographics.

Not surprisingly, people are nervous about this, claiming Big Brother and Minority Report tactics to watch people, make decisions, and feed them what the computer thinks they’re interested in. These same people are vowing to buy gas elsewhere just so they don’t feel “violated” by a computer using them without their permission.

Supporters of the camera ad system claim that the videos are merely a way to better target ads to customers. They also state that the camera ad system is well within the privacy restrictions set by the UK government as the cameras don’t store any of the video that it uses to determine consumer classification.

From a marketing standpoint, the biggest question is whether this will be successful, and how will that success be measured? Since the cameras only do live tracking and don’t store any of the video, repeat behavior from a consumer can’t really be measured. Nor  is there any sort of follow-through metric that determines that, for example, a consumer saw an ad for an energy drink at 2 am and then went into the store and bought said energy drink.

And that assumes that the computer is properly determining the parameters of each person. Amscreen, the company behind the camera technology, says that it is 98% accurate in ideal conditions. But they acknowledge that it can’t determine classifications from someone wearing a burqa or other full head covering (such as being bundled up against London’s cold winters). Long hair, if conditions are not ideal, could mean that a 20-something guy is getting an ad for sports bras.


customers view ads targeted to them

Customers at a pay point at a gas station view ads targeted to them based on their age and gender.

Without the ability to track user behavior or conclusively determine any sort of results, is the invasion of privacy really worth it? Can Amscreen even on a basic level determine whether or not it’s working properly if there’s no real follow-up on the accuracy or relevancy of the message? On a nervous Big Brother level, is this technology just a hairline away from actually storing information to establish user behavior over a longer period of time? And what about building out more complex classifications to identify (height, weight, hair color, and so on) for better targeting so that it can throw some salt in the proverbial wound by serving weight-loss ads to someone it deems overweight?

The technology itself is impressive and could be a glimpse into the future of POS ad targeting. But just because the technology is there doesn’t necessarily mean that it should be used to sell products, especially if there’s no meaningful way of determining success – or even if people are paying any attention at all to the ads.

Tags: , , , ,
Categorised in: , , ,