AI isn’t coming for search. It already owns the lease, changed the locks, and left you standing outside with your SEO playbook from 2013.

Search used to be prime real estate — the Fifth Avenue of the internet, where being on the first page meant status, traffic, and cash. You didn’t just rank; you arrived. Now? That street’s empty, and in its place is a concierge service run by your friendly neighborhood algorithm. It doesn’t wait for you to ask questions. It anticipates them. It greets you at the door with your credit card history, Spotify preferences, and a disturbingly accurate guess of what you’re about to Google.

Simon Poulton said it bluntly: search itself was a construct defined by the UI. In plain English — the only reason we ever yelled single words like “yoga pants” into a text box is because that’s all the interface allowed. The UI trained us to think like robots. Now that robots can think like us, that entire ritual is over.

The keyword era is dead, and we’re the ones still showing up to the shivah with Google Ads reports and sad CTR graphs.

Today, intent is expressed in conversation. Not in staccato nouns. Not in metadata hacks. People don’t search by typing — they ask, describe, rant, and wonder. The future of discovery is dialogue. Which means the job of marketers isn’t about gaming syntax anymore. It’s about teaching machines how to understand meaning.

If that sounds terrifying, it’s because it is. But it’s also the most exciting creative reset since banner ads stopped being ironic.

Poulton calls it the “agentic ecosystem” — a new world where digital agents don’t just fetch information; they negotiate, compare, and transact on your behalf.

There will be good agents and bad ones. Helpful agents and hustlers. “Agents that wanna sell you things, and agents that wanna help you,” as Simon put it. In other words, the internet is about to become one massive dating market for AIs — and your brand’s reputation is what shows up on the shidduch résumé.

Think of it like this: your new audience isn’t the consumer — it’s their digital proxy. Your success depends on how clean, structured, and honest your data looks when that agent goes browsing for options. Forget the human first impression. You now need machine hygiene.

Are your feeds structured?
Is your content accurate?
Can your brand be understood by an algorithm, not just about one?

Because if your data looks like a pile of loose socks, no agent’s introducing you to anyone.

The industry’s still worshipping clicks like they mean something. They don’t. They haven’t for a long time. Poulton dropped the hammer: “There isn’t necessarily bad measurement — there’s just bad inference.”

That’s the line that should be written above every marketing dashboard in the world.

AI search doesn’t care about engagement bait, thumb-stopping colors, or A/B-tested emojis. It cares about understanding. It cares about truth signals.

CTR was the old god — dumb, binary, transactional. The new metric is comprehension: how well your ecosystem communicates its intent, provenance, and utility to the network of agents reading, ranking, and recommending.

The old job was seduction.
The new job is education.

You’re not trying to lure humans anymore — you’re teaching machines to trust you.

Here’s the part marketers keep missing: people aren’t searching less. They’re just outsourcing the act of searching.

“The user isn’t not searching,” Simon said, “it’s just changing.” Exactly. Users still want answers, but now they expect an AI to fetch, filter, and decide before they ever see the options.

That means every future search will begin and end with an agent. Whether it’s Gemini, ChatGPT, or some yet-to-be-named assistant living inside your phone, they’ll do the questioning — and they’ll decide who answers.

And that’s why optimizing for humans is no longer enough. You must optimize for understanding. The battle for attention has become a battle for representation.

If you’re not in the data the agents trust, you don’t exist.

AI search won’t reward cleverness; it rewards clarity. It won’t favor noise; it favors structure. You can’t hack your way into the algorithmic inner circle. You can only earn your spot through consistency, transparency, and precision.

Poulton calls it a “crossing the chasm moment.” He’s right. We’re waiting for the Google 2.0 of this era — the moment when the AI networks gain the critical mass of trust that makes them indispensable. When that happens, your brand’s ability to be read, reasoned about, and recommended will define your survival.

So stop thinking of “ranking” as visibility. Start thinking of it as machine fluency. Your job now is to make your content, products, and policies legible to the intermediaries doing the thinking for everyone else.

Search used to be a tool. Now it’s an actor. It’s got a personality, preferences, and yes — biases. If your marketing still assumes it’s neutral, you’re already invisible.

Let’s stop pretending: the ad industry treats data like religion. Everyone swears they “believe in measurement,” but no one reads the fine print of the scripture. Models are holy, spreadsheets are prophets, and we still manage to get every single parsha wrong.

Simon Poulton isn’t having it. In our conversation, he dismantled the cult of data worship one polite sentence at a time — starting with his devastating line: “There isn’t necessarily bad measurement, there’s just bad inference.” That’s it. That’s the gematria truth we’ve all been avoiding. The problem isn’t the math; it’s the people interpreting the math like fortune tellers with a dashboard fetish.

Ah, the CDP — the ultimate expensive relationship you didn’t need but got talked into because someone at a conference said it “future-proofs your data stack.”

Simon rolled his eyes at this mythology: when the AI apocalypse comes, no one’s going to care that you bought a Customer Data Platform you barely configured. “If it’s truly sentient,” he joked, “I’d offer it compassion — not a CDP. It would be offensive to offer it a CDP.”

That’s the most honest take on martech I’ve heard in years. Half the industry is waving around overpriced “solutions” like a bouquet on a first date, desperate to look sophisticated while quietly hoping the system doesn’t ghost them.

Simon was just warming up when we got to the Church of MMM. He described it perfectly: “We get told this is the thing that will mitigate opinions… but it’s so loaded with opinion you cannot separate opinion from analytics.”

Exactly. Marketing Mix Modeling isn’t the problem; it’s the blind faith that turns it into pseudoscience. Teams demand “FDA-level precision” from data that barely qualifies as anecdotal. They torture the numbers until they confess what the CMO wants to hear, and then act shocked when the business doesn’t move.

Simon added, “There isn’t necessarily bad measurement… it’s the interpretability factor.” In other words, we’ve built cathedrals to precision while completely ignoring the part where we’re just guessing, loudly, in PowerPoint.

By this point in the interview, we’d both stopped pretending to be polite. Simon dropped what might be the line of the season: “If your creative looks like a woman with a dog, don’t test it against the same photo with a cat — test it against a woman on a roller coaster eating granola bars.”

Why? Because algorithms don’t reward timid tweaks. They reward step-change. Meta’s own engineers, he said, design learning systems that crave meaningful variation — not the timid pixel-shifts that brand teams mistake for bravery.

In other words, stop dressing your creative in a beige cardigan and calling it performance marketing. It’s not cozy; it’s boring. Turn up the temperature or the algorithm will put you in the digital nursing home with every other risk-averse campaign that once tested “headline A vs. headline B.”

Eighteen months. That’s how long the average CMO lasts before being “moved on” — which in corporate speak means “gently escorted out before the next board meeting.”

Simon didn’t mince words: “The longevity of CMOs is always one of the shortest C-level positions… you can’t think much beyond three months at a time.” That’s not a career; that’s a reality show.

He painted the picture perfectly — the CFO breathing down one shoulder asking, “Can we cut spend next quarter?” while the CEO is on the other side demanding growth. No wonder marketers cling to short-term KPIs like a security blanket. They’re not optimizing campaigns; they’re optimizing survival.

And when the data doesn’t tell the right story? The flashlight turns into a weapon.

I asked him the obvious question: are most marketers using data as a flashlight or a club? His answer should be printed on mugs: “Flashlights are weapons, right? It depends what you’re choosing to illuminate — and who you’re choosing to whack over the head with it.”

That’s the state of analytics today. Most teams aren’t searching for truth — they’re bludgeoning dissent. They call “noise” anything that contradicts the narrative and “insight” anything that flatters it.

When Simon and his co-host Neil Hoyne talk about “precision theater” on their Measure Up podcast, this is what they mean: our obsession with fake certainty has made marketers allergic to ambiguity. We’ve replaced curiosity with dashboards and called it strategy.

The irony? Measurement was supposed to make us smarter. Instead, it’s become our favorite way to cosplay intelligence. Simon’s diagnosis is merciless but spot on: most marketers don’t want better data; they want permission to believe their story.

So here’s the takeaway: stop worshipping models. Stop romanticizing CDPs. Stop iterating beige creative. And for the love of metrics, stop pretending you’re running a science lab when you’re just doing interpretive dance in Excel.

The model isn’t the sin. The inference is.

You just read the free version — the tasting menu. The full-course meal is inside ADOTAT+, where we publish the kind of deep-dive research no one else in adtech even attempts. Not recycled press releases. Not AI-summarized fluff. Actual reporting, sourcing, and frameworks built for people who build this industry.

ADOTAT+ is where we publish the Agentic Search Playbook 2025 in full — complete with templates, calibration maps, and implementation guides for marketers trying to brief and manage AI agents before those same agents start managing them.

We don’t chase clicks. We chase clarity. We build the field guides that help the people running adtech today survive what’s coming tomorrow.