Cedric Chereau, Managing Director and Co-Founder of EagleAI, explains why treating loyalty program members as groups instead of individuals wastes marketing budget, and how his team borrowed algorithms built for Netflix to fix it.
Two 80-year-old billionaires who both live in London look identical on paper. Same age bracket, same city, same income tier. One of them is King Charles. The other is Mick Jagger. If you built a marketing segment around “wealthy octogenarian Londoners” and sent them the same offer, you’d be wrong about both of them. That’s the example Cedric Chereau, Managing Director and Co-Founder of EagleAI, used to explain why grocery retail personalization built on segments is fundamentally broken, and why he spent the last decade building something built on individuals instead.
Chereau spent 15 years as a consultant helping retailers across Europe and North America get value out of loyalty program data, then left to build something more scalable. Ten years ago, that became EagleAI, now part of Eagle Eye, working with grocery retailers and CPG brands to recreate the kind of personal relationship a small local grocer used to have with regulars, at the scale of millions of customers.
Why Does Customer Segmentation Waste Marketing Budget?
Segments feel efficient. Build one offer for your most loyal customers, another for less-engaged ones, and you’ve covered your base. Chereau’s argument is that this approach quietly bleeds money in both directions.
“If you’re looking at averages by design, you’re treating people the same way within a segment. And the customers can be completely different in that segment.”
— Cedric Chereau, Managing Director and Co-Founder, EagleAI
Say your top-spending segment averages $100 a visit, so you build a $10-off-$100 offer for them. Some of those customers already spend $300 or $500 regardless of any offer, so that discount is pure margin loss. Others in that same segment can only manage $50, so the offer is unreachable and useless to them.
“For the customers who spend $300, it’s money wasted because they would have spent that money anyway. And for the customers who spend $50, it’s completely useless because spending $100 is way too difficult for them.”
— Cedric Chereau, Managing Director and Co-Founder, EagleAI
EagleAI’s approach, which the industry has started calling “segment of one,” treats every loyalty member as their own category. Instead of rewarding a segment average, it rewards individual positive behavior: spend $60 instead of your usual $50, and you get recognized for that specific increase, not a group-wide discount most people didn’t need.
What Can AI Actually Learn From What’s in Your Fridge?
The data source here is deceptively simple: transaction history from a loyalty card. No cookies, no third-party data, just what you actually bought, when, and how often. Chereau’s read on that data source is blunt.
“You are what you eat, and I think that’s a very good thing. If I know as a retailer what you bought last month, I understand better what your affinities are, what you like, what you dislike.”
— Cedric Chereau, Managing Director and Co-Founder, EagleAI
He’s also skeptical of the data everyone assumes matters most: demographics and location. When retailers ask whether EagleAI factors in age, income, or ZIP code, his answer is consistently no.
“What is much more interesting is to look at what people are actually putting in their fridge. If you look at this and this only, you have much more granular, much more valuable insights than if you’re looking at personal or location data. It’s more misleading than anything else.”
— Cedric Chereau, Managing Director and Co-Founder, EagleAI
The same skepticism applies to just asking customers what they like. Self-reported preferences are biased, not because people lie, but because they round themselves up.
“They might say, ‘I’m buying a lot of organic product because health is very important,’ but in reality, what is in your fridge? You’re making choices as a customer always. Having access to that kind of very granular data will give much more insight than if you ask customers what their real affinities are. They will always be biased.”
— Cedric Chereau, Managing Director and Co-Founder, EagleAI
How Does Real-Time Personalization Actually Work in a Grocery Store?
EagleAI runs a suite of purpose-built algorithms, and Chereau gave them names simple enough to explain on a podcast. The “people pleaser” identifies the specific product that makes each customer unique, not milk or eggs, which everyone buys, but the one item, like a specific brand of Parmesan cheese, that a shopper will make a special trip for if it’s not in the fridge. The “influencer” flags products similar customers buy that you don’t, a signal you might already want them elsewhere. And a third model predicts natural spend, then tests how much headroom exists to nudge it upward with the right reward.
But the more important shift, Chereau argued, isn’t real-time processing, it’s real timing.
“More than real time, the right timing in those interactions is what matters. For some customers, the right message will come two days before, when they’re planning for their next visit and building their list. For others, it has to happen the moment they’re finalizing their transaction.”
— Cedric Chereau, Managing Director and Co-Founder, EagleAI
That distinction, between when data is technically available and when a customer is actually receptive to it, is easy to overlook and expensive to get wrong.
The Bigger Picture
What struck me most in this conversation is how directly it validates something we talk about constantly on identity resolution: averages hide the individual, and the individual is where the value actually is. Chereau built his first algorithms in the grocery world by adapting open-source recommendation research from Netflix, of all places, proof that the underlying problem (predict what this specific person wants next, not what their group wants on average) is the same whether you’re recommending a movie or a jar of pickles.
Retailers are increasingly turning loyalty programs into ecosystems, tying grocery spend to gas station discounts or streaming perks, precisely because more connected data points mean better individual predictions. The retailers who win that game won’t be the ones with the biggest segments. They’ll be the ones who stopped using segments and started treating every customer like the individual they actually are.
Find Cedric Chereau on LinkedIn, or learn more about EagleAI at eagleeye.com/eagleai.
To build a real, individual-level view of who your customers actually are instead of relying on averages, visit bdex.com and click “Talk to an Expert.”
Watch the full episode: Why First-Party Data Beats Silos and Drives Actionable Marketing on YouTube
This article was adapted from an episode of Deconstructing Data, BDEX’s weekly podcast on data-driven marketing. Tune in live every Thursday at 4:15 PM Eastern on LinkedIn.
About BDEX: For companies that need clean identity data to power their products, BDEX offers unmatched quality and execution. Visit bdex.com and click “Talk to an Expert” to get started.
