Uber — the company that once made waiting for a Corolla feel like a metaphor for life itself — has done something absurdly bold. It took the one thing Silicon Valley usually ignores — boredom — and turned it into a measurable asset.

The result is Custom AU, a joint invention with Adelaide (the brilliant data workshop run by Marc Guldimann) and Kantar, the research giant of brand lift and ad effect. Together, they built a model that doesn’t just count how many pixels you looked at, but how long you actually cared.

And here’s the part that gets me: I want to believe in it. Against my better judgment, I want to believe this one’s real.

Uber didn’t invent attention. It cornered it.
While every social platform fights for half-second bursts of chaos-scrolling, Uber looked at the quiet moments — waiting for the driver, staring at the map, zoning out on the way home — and said: That’s it. That’s the ad space.

It’s the economics of stillness.

A moment that used to be worthless is now premium media inventory because Uber can prove you were there, looking.

Adelaide’s predictive model maps how likely you were to notice; Kantar’s surveys show whether it actually landed; Uber’s ride data stitches it together. The machine learns, retrains, and hums quietly in the background — feeding on patience.

And it works because it’s honest. The rider isn’t pretending to multitask or faking interest. They’re just sitting there, captive and curious, existing in a measurable pause.

Here’s where I ramble a little. Because I’ve seen the “next big thing” in measurement more times than I can count — from viewability to verification to whatever nonsense “engagement quality” was supposed to be. Each one promised enlightenment, and each one dissolved into spreadsheet dust.

Then along comes Marc Guldimann, and somehow he talks about attention like it’s the missing variable between art and math.

He doesn’t hype it; he dissects it. He says things like, “Why are we still talking about viewability?” not as a burn, but as an existential question about how far advertising has drifted from reality.

He built Adelaide on the idea that media quality should be predicted by context, not guessed at through vanity stats. And when Uber plugged that system into its own universe of first-party ride data, the whole thing clicked in a way that feels — I don’t know — earned.

For once, the math doesn’t feel like an illusion. It feels like a map.

Unlike almost everyone else in adtech, Uber controls everything: the screen, the context, the timing, even your location.

It knows how long you’re there, what you’re doing, and whether you’re going to the airport or Taco Bell.

In theory, your wife knows you’re going to the “Gentleman’s Club.”

That means its ads don’t live in the chaos of the open web — they live in an ecosystem that’s physically tied to your attention. There’s no “maybe they saw it.” Uber knows if they did.

In the strange hierarchy of ad environments, that makes Uber something like the Tesla of media: closed, sleek, slightly cultish — but undeniably effective.

And when that data runs through Guldimann’s AU model and Kantar’s brand validation, the result is more than a report. It’s a feedback loop between behavior and belief.

Yes, it’s still proprietary.

Yes, advertisers can’t see the algorithm.

But if you’ve been in this industry long enough, you know that transparency is the great myth — everyone’s “open” until you ask for the dataset.

So maybe it’s fine. Maybe a closed-loop metric built on real-world data is better than another round of synthetic visibility charts. Maybe Uber’s “black box” is less of a mystery and more of a mirror — showing us how fragmented, fragile, and desperate the rest of the ecosystem looks.

I’m not naïve. I know this could all turn into another beautifully packaged illusion — an expensive proof deck to sell higher CPMs. But I can’t shake the feeling that something here actually fits.

For all its corporate gloss, “Custom AU” feels like the rare case where the industry built something it can actually defend. It ties predictive attention to brand lift, grounding every claim in something observable.

It’s not perfection, but it’s progress — a metric that treats human focus like a natural resource instead of a clickstream.

And Marc… yeah, maybe he’s the reason I haven’t given up on this entire side of the business. He talks about attention the way scientists used to talk about gravity: invisible, inevitable, worth studying.

Every time I read one of his interviews, I find myself thinking maybe — just maybe — there’s still magic in this mess of media math. Maybe we can measure something real without ruining it.

And I’ll take that feeling, even if it’s fleeting. Because in an industry built on illusion, wanting to believe might be the most honest data point we have left.

Uber has decided that just tracking where you’re going isn’t enough. Now it wants to predict what you’re looking at. Partnering with Adelaide and Kantar, Uber has built a hybrid model called Custom AU — a predictive measurement engine that merges data science, psychology, and brand alchemy. It’s part machine learning, part marketing theater, and it might just become the playbook for every platform trying to prove that its ads actually work.

Custom AU doesn’t just count impressions; it forecasts engagement. It promises advertisers something beyond “views” — proof of attention that translates into brand impact. By merging Adelaide’s predictive modeling, Kantar’s brand lift validation, and Uber’s first-party ride data, the company has built an algorithm that thinks it can predict when your brain lights up.

Adelaide’s Attention Unit (AU) works like a weather model for human focus. Instead of forecasting rain, it forecasts how likely you are to notice an ad — and whether that attention will matter.

Here’s what the model chews on:

Then Uber adds its secret sauce: first-party rider data.
Ride duration, app interactions, time of day, even behavioral cues — these contextual details turn Adelaide’s general model into an environment-specific prediction engine. It’s not just tracking if the ad was seen; it’s estimating how focused you were when you saw it.

Finally, Kantar’s brand lift surveys provide the feedback loop. The model is continuously trained on post-exposure outcomes — unaided awareness, brand favorability, consideration. It learns what combinations of attention signals correlate with positive brand movement. Think of it as a self-updating metric that studies human behavior and then adjusts its own expectations.

The result is a closed, adaptive feedback loop:

It’s behavioral machine learning meets brand science, fine-tuned for Uber’s captive ecosystem.

Uber has something most Retail Media Networks (RMNs) can only dream of: end-to-end visibility.

That’s a rare trifecta in modern advertising — exposure, engagement, and outcome all contained within a single, trackable ecosystem. It eliminates the attribution noise that plagues open-web campaigns, where half the data disappears into the void of privacy walls and click fraud.

Uber doesn’t have to infer attention; it can observe and model it directly. That makes Custom AU not just a brag sheet for higher CPMs but a credible proof mechanism in a market starving for trust.

It’s a simple but brutal advantage: when you own both the audience and the data, you don’t need to argue — you just show the numbers.

Of course, like any data-driven marvel, Uber’s new metric comes with its fine print. The same closed-loop control that makes it powerful also makes it opaque.

It’s the same paradox haunting all predictive systems: the more precise the model, the less portable it becomes.

And while Uber promises integrity, advertisers are still being asked to trust a black box wearing a white lab coat.

This isn’t about one metric. It’s about who controls the definition of “attention.”

Uber’s model blends data precision with marketing psychology to create something the industry’s been chasing for a decade: a quantifiable, outcome-linked measure of ad quality. It’s not perfect — it’s proprietary, self-referential, and borderline cultish in its belief that it can quantify human focus — but it’s undeniably the future of platform-native measurement.

It wasn’t just dial-up.

It was chaos with a hint of destiny. DSL was the luxury flex, Flash banners were witchcraft, and half the industry thought “streaming” was what happened to their morning coffee.

That’s when I launched ADBUMb — back when the internet still squeaked and no one could agree whether “digital media” was a fad or a cult.

ADBUMB became ADOTAS, and suddenly the misfits were right.

We called it all — video, gaming, programmatic before it had a name, attention metrics before they became the new religion. Then I sold it. The suits came in. Everything got focus-grouped to death. ADOTAS died the way most revolutions do: quietly, with a press release.

Now it’s ADOTAT+, and the pulse is back.

If you’ve made it here, you already know what this is. You’ve been around long enough to recognize when the industry starts repeating its own mistakes in higher resolution. You want the truth, not the polished keynote version.

ADOTAT+ gives you that:

The free edition gives you the headlines.
ADOTAT+ gives you the story under the story.

If you were here when ADBUMB called the internet’s future, you know what happens next.
Don’t watch it from the sidelines again.

👉 Upgrade to ADOTAT+ — before everyone else pretends they saw it coming.