How L'Oreal Knows What You Look Like

Photograph by Simon Dawson/Bloomberg

L’Oreal knows what hair color you want. Well technically GumGum, a Santa Monica (Calif.)-based technology and advertising company, knows the hair color contained in images you browse and sells that information to L’Oreal. The cosmetics company, in turn, uses the information to spread ads for its line of Ombre hair color products across more than 1,000 news and entertainment websites, including digital properties owned by Tribune Co., the New York Times Co., and TMZ.

Think of it as the online equivalent of sales personnel who stand outside hair salons and lure women inside with promises of free highlights. “We’ve gotten really good with this ‘image detection,’ stuff,” says Ophir Tanz, founder and chief executive officer of GumGum, which has been around since 2008. “We can do things like nudity detection, eye color recognition, hair color detection.” How is this possible? Tanz says GumGum doesn’t scour Facebook photos or anything like that. Instead, the company tracks and uses your Web history to create a digital picture of what you probably want to look like.

“Let’s say you’re browsing articles about hair care,” Tanz explains. “If you’re reading something about curly hair, we can see that.” Gumgum can then pair that data nugget with all the other crumbs of personal information you leave around the Internet—where you live, your gender, your income level, and how often you visit the websites GumGum tracks. Then it uses all that information to decide which ads to place in front of you.

If you’re a young woman with dark hair, you might see the Ombre ads. If you’re an older man who keeps searching “how to stop going bald,” then—well, L’Oreal will leave you alone.

This is an important shift for online advertising, which until now has taken the approach best described as: “Did you recently buy a pair of shoes? Well then, maybe you want ads for more shoes!” Tanz thinks his hyper-targeted ads may help the media industry, which right now is grasping at everything from native advertising to cat-video click-bait to help monetize digital news in the face of declining print sales. If online ads can be directed to the right customers, the ads will become more effective and therefore worth more—letting websites charge higher prices for them. “I think what we’re seeing today is a very, very basic version of what online advertising will be like in the future,” says Tanz.

Some of these techniques can leave you feeling that your privacy has been invaded. And as targeted ads become ever more sophisticated, people are increasingly startled by the results.

So are GumGum’s hair color ads creepy or helpful? Probably a bit of both. “Look, you’re going to be advertised to, no matter what,” says Tanz. “So ultimately, you might as well have advertising that you’re actually interested in—and that’s hopefully less annoying.”

    Before it's here, it's on the Bloomberg Terminal.