Shirley Marschall explores vibe targeting, and why it's becoming the industry's latest black box.

Oh AI, you really are the gift that keeps giving…

You gave us vibe coding a.k.a. “going with your gut, no formal training, chaos energy, trust-the-vibe” coding. Or startup-speak for: “I don’t know what I’m doing, but I’m shipping anyway.” Definitely one of the bigger gift boxes so far, punching AI holes straight through the digital ozone.

From prompt to code, it was only a matter of time until this trend reached ad tech. And here we are: vibe targeting or prompt-to-target has arrived.

AI is already creeping into every part of the ad tech ecosystem, from legit use cases to sketchy “growth hacks”. But targeting? That’s a particularly sensitive corner, both from a results and privacy perspective. And it’s been questionable for a looooong time.

The eternal obsession with targeting

Digital advertising has always been obsessed with one thing: getting sharper, smarter, more precise with targeting.

We moved from broad demographics to behavioral segments, then from behavioral to hyper-personalised experiences powered by real-time data, AI-driven creative, and dynamic delivery.

By the time we reached peak personalisation, ads were following people from app to inbox to smart fridge. Consumers got spooked. Regulators raised eyebrows. Marketers started saying things like “privacy-safe personalisation” with a straight face.

TL;DR: Each step gave us more granularity. More signals. More control. And a whole lot more creepiness.

But at least – at the very least – we still had some idea what was going on. We knew how segments were built. We knew who we were targeting and (mostly) why.

Vibe Targeting: Feels safer… but isn’t

If vibe coding, as Andrej Karpathy put it, means: “You fully give in to the vibes, embrace exponentials, and forget that the code even exists…”

…then what does that mean for vibe targeting?

“Vibe targeting” is exactly what it sounds like and also, nothing like it sounds. Vibe targeting sounds soft, emotional, and yes, human even. It promises to match the tone, tempo, and mood of the moment. Not just what they’re doing, but what vibe they’re in.

Sounds… kind of lovely, right?

Vibe targeting isn’t about empathy though, it’s about automation.

Instead of building segments, marketers are now told to set a goal, describe a feeling, or write a prompt and let the system take care of the rest.

The platform chooses the audience. The AI scores the creatives. Bidding strategies run themselves. And no one really knows what’s happening behind the curtain.

That’s not creepy anymore. That’s full on clueless. And weirdly, that’s the selling point.

The black box era of targeting

Brian O’Kelley described a version of this on his Substack… a clean, seductive model of prompt-driven automation:

“Why not have prompts be the targeting?”

“You write the prompt. YOU produce the vibes.”

Sounds a bit like Zuckerberg’s “let the machine optimise everything” mantra, just with a softer filter. A “Zuck Lite”, if you will.

Sure, it’s efficient. It’s convenient. Nobody loves building segments.

But now we’re not targeting, we’re vibe targeting, we „forget that the code even exists“…so we’d basically be targeting, without knowing:

• Where the data comes from
• How segments are created
• What exclusions are applied

Any flashing red lights yet? Because when you prompt the machine and walk away…who’s responsible for what happens next?

The disappearance of accountability

In the era of behavioral targeting and hyper-personalisation, someone, somewhere, still built the segment.

In vibe targeting? No one can explain how the audience was defined.

Marketers become prompt writers. And the system decides everything else.

So when a campaign backfires – when it’s getting weird, biased, exploitative, or just plain off-brand – there’s always a convenient fallback: “We didn’t choose that audience. The algorithm did.”

And that’s the best part about vibe targeting: Unlimited. Deniability.

No more worrying about brand safety slip-ups. No more questions about dodgy data brokers. No more segment audits or targeting justifications.

The vibe feels good. The campaign “performs.” But no one can explain how or why.

We’ve moved from precision to opacity. From creepy to clueless. From strategy… to vibes.

And while marketers might love the vibe…regulators probably won’t. Because if you can’t explain how something works (and you run it anyway) that’s not a feature, it’s a bug. 

But let’s not ruin the good vibe, shall we?!

Some more good and bad vibes

The good: So far, these discussions mostly happen inside the ad tech echo chamber, meaning the real-world impact is still limited… for now…(don’t hold your breath)

The bad: We’re not just optimising toward losing the last bit of control, we’re once again chasing a flawed premise of targeting. But that’s nothing new. Targeting is basically a Gen Z’er by now… and under the banner of “performance”, everything is allowed, even stacking new abstractions on top of old flaws.

Yes, AI will shift ad tech. That’s not in question.

But it doesn’t have to be a shift from bad to worse (does it?).

It doesn’t have to mean throwing more money at black box monopolies or blindly following “ad tech godfathers” (does it?).