In her latest column, Shirley Marschall takes a look at the rise of AI as a targeting tool, and intelligence-grade analytics...

Maybe data isn’t oil but honey? Rare, wild, and super sticky forest honey, used by platforms to lure in advertisers. Promising this magic honey is only available right there, on that specific platform. Telling advertisers this is where all targeting problems will be solved. Magic honey! And isn’t that how [enter any walled garden] hold advertisers by the bank accounts? It’s not just attention. It’s data.

And AI? That’s the oil (yes, we’re back to oil) in this data–targeting fire. The accelerant that could set the whole thing ablaze.

From demographics to "Oh, you shouldn’t have"

Remember the "good old days," when targeting meant picking a demographic? Age, location, maybe income bracket, and a creative you hoped would land. Cold, precise, a little military-sounding and not exactly consumer-friendly.

So we softened (read: rebranded) it: personalisation! Friendlier, less scandal-y, promising relevance.

Then came hyper-personalisation, layering real-time behaviour, context, and AI triggers to adapt the ad and the moment. A digital ad just for me? Shiny eyes. Excitement. Blushing. "Oh, you shouldn’t have!" …said no one ever.

And now? We’re skipping straight to trialling military-grade AI to decide who sees which ad.

Yes, you heard right and yes, it’s a true WTF moment!

Consumers still stuck on cookies

Don’t expect protests. Most consumers are still processing the cookie massacre, shaking off the last retargeting ad that followed them around for weeks. Annoyed by consent banners, usually clicking yes, occasionally wondering why and if it even matters since “all companies” already have all of their data anyway.

Advertisers and publishers alike are still part mourning cookies, part cursing Google for their Privacy Sandbox U-turn, while trying to enrich, collect, and monetise their first-party data.

And then there are some tech and data companies that clearly didn’t get the memo about privacy. Or, more likely, they read it, shrugged, and went straight back to "whatever works."

Case in point: Stagwell meets Palantir

Stagwell is piloting Palantir (yes, the one best known for defence and intelligence work) to see if military-grade pattern recognition can supercharge targeting.

The trial combines:

• BERA: brand insights from 50,000+ weekly consumer interviews
• NRG: rich media behaviour data
• Stage: media planning and buying platform

Structured and unstructured data, fed into Palantir’s AI, aiming for new targeting efficiencies. One client. Controlled trial. "Drug to market" approach.

Technically fascinating. Reputationally… it’s complicated.

The quiet return of fingerprinting

That’s not the only escalation. Fingerprinting, as in identifying users by unique device/browser traits, is back, but not just the version Google quietly re-enabled earlier this year under "less prescriptive data activation."

This time AI is in the mix, and fingerprinting becomes fingerprinting-on-steroids: harder to detect, seamlessly embedding into cohorts and clusters that make targeting decisions with minimal human oversight.

Nice. And totally unproblematic, especially for consumers. Or… maybe not?

When data turns against the customer

The ad industry has long sold a simple bargain: free content in exchange for data. But what if "free" was always fiction? 

The data, the honey, the ads targeting, are the polite use case at best. It’s no secret that the same data signals can serve political influence, law enforcement, predictive policing, or intelligence work.

And what happens when the data-for-value equation (or illusion) collapses for good? When that data is weaponised against the customer? If it’s used to hike prices through dynamic pricing, reduce service quality compared to others, or predict and influence behaviour with intelligence-grade tools… 

…then the value exchange stops looking like reciprocity and starts looking a lot like exploitation.

So yes, consumers will trade data if the deal is clear. Right now? There is no clear deal. The line between "consumer marketing" and "data intelligence" is getting harder to see.

The blind spot

Public debate is still stuck on cookies, while privacy laws lag behind AI funnels, silent fingerprinting, and cross-context tracking that doesn’t need a single cookie to function.

By the time people realise they’ve moved from banner ads to behavioural profiling with intelligence-grade analytics, the infrastructure will be firmly in place.

It’s a slippery slope and each step (sort of) feels incremental until you zoom out and realise we’re halfway down the slope, feet off the ground.

Closing thought

Advertisers will always chase the honey. Platforms will always guard it. And as long as the hive stays hidden, consumers won’t swarm.

But if the "free" story was always fiction, and the honey can be used for anything, then the value exchange isn’t just fragile… it was never real in the first place.

Palantir’s trial, fingerprinting’s return, and AI-driven cohorts aren’t separate trends. They’re symptoms of the same disease: the relentless pursuit of unique, high-value data and the growing opacity in how it’s gathered and used.

Anyone else craving honey?