Every breath you take, every move you make, Coca-Cola is segmenting you into personas.

“We’re totally committed to behavior-based marketing,” says Greg Chambers, global group director of digital innovation for Coca-Cola (and featured speaker at VB’s upcoming MobileBeat 2017 next week). “Marketers today focus so much on privacy-based information—like your name, where you were born, your ethnicity. But those details don’t matter as much anymore.”

Consider their recent digital signage pilot with grocery chain Albertsons in which Google-integrated endcaps served up Coca-Cola ads based on data gathered from nearby smartphones. The setup can’t dig up any specific personal identifiable information, but gathers substantial info from DoubleClick ad data that tracks customers’ browsing habits.

In addition to ballparking table stakes data like gender, age range, and income bracket, the endcaps also identify affinity categories, dietary preferences, and past online shopping — then serves up an ad designed to appeal to that profile.

“Our goal is to use machine learning to figure out which path-to-purchase a consumer is on based off any scrap of information we can get,” Chambers says. “And eliminate one more step, one more choice, one more click from that process for the busy professional in this day and age.”

Chambers also emphasizes the utility of observational data for segmentation, pointing at how just the speed a shopper is walking at through the store can yield a wealth of essential information about what they want and need — is this someone who wants to learn about products, or are they a deal seeker who just wants a coupon and they’re done?

The company uses on-premise beacons that measure how quickly or slowly you’re trolling down the aisle, allowing them to identify, for instance, the customer who is on a fill-up trip moving leisurely up and down every aisle versus the hurried shopper rushing in after work to find something to put on the table for dinner.

“No name, no PII, no nothing—just the rate of speed,” Chambers says. “Just looking at those small things that have nothing to do with privacy, you can easily — and with a pretty high level of confidence — determine what kind of shopping trip that person is on and then tailor the marketing to that kind of shopping trip.”

Or consider an entirely different setting. In a theme park filled with hot and thirsty people, consumers identify themselves by which rides they choose to go on, from the gentle kiddie rides to the all-ages rides to the high-speed roller coasters that appeal to the adrenaline junkies. It’s all valuable persona data, Chambers says, and could even be described as common sense observation.

“Think about any aspect you can learn just by sitting on a bench, looking at people, and making general assumptions about what they’re doing on their journey. We’ve started to figure out ways that we can pick that kind of information up and feed into machine learning models,” Chambers explains. “Every aspect that can be measured, and can be measured in a way that is not intrusive to anybody’s privacy, can be used in that factor.”

The company’s top objective, Chambers says, is using all this data to create great experiences for both their own customers (those buying Coke products for resale), and the end consumer. AI has been a game changer in that regard.

“We use some form of AI in a lot of stuff that we do, and AI is a tool that can be used to create amazing experiences for people,” he says. “So all the tools that we’re building with AI are all about making that experience more intelligent, more seamless, less friction-based, more natural.”

Their strategy is about answering one ultimate question, he adds.

“How do we get to a place in the future where you can think that you’re thirsty, and a product just appears in your hand? We’ll stop innovating when we get to that point!”