The Channel That Was Right All Along
ChatGPT's ad play isn't a threat to programmatic. It might be its biggest moment in a decade.
There’s a particular kind of vindication that comes slowly, then all at once.
Display advertising has spent fifteen years being the unloved middle child of the media mix. Measurable enough to be scrutinised, unfashionable enough to be defunded, and technically sophisticated enough that nobody outside adtech really understood why it kept surviving. B2B marketers in particular treated it like a necessary evil, something you did when search inventory ran dry and you needed to justify the impressions line on a media plan.
That era is ending. And the trigger is the last thing anyone expected.
Display Was Never The Problem
Let’s be honest about something the industry spent years dancing around: display was never a weak channel. It was a channel with a format problem.
The data exhaust from display has always been extraordinary. Browse behaviour. Dwell time. Scroll depth. Return visits. Content affinity signals that tell you more about genuine purchase intent than any self-reported form fill. When programmatic scaled in the early 2010s, the infrastructure built to activate that signal was genuinely world-class — second only, arguably, to search.
But the creative format never kept pace. The banner ad arrived with Web 1.0 and, structurally, never really left. The 728x90. The 300x250. Formats conceived when broadband was a novelty and screen real estate was rationed. As the web scaled into immersive, social, video-first experiences, the banner sat in its corner — tolerated, banner-blocked, ignored.
The signal was premium. The execution was not.
The Native Promise That Wasn’t
I was at the sharp end of native advertising when it emerged as the supposed solution to this. The logic was compelling: match the form and function of the environment, earn attention rather than interrupt it, and suddenly display becomes a format that people might actually choose to engage with.
And it almost worked. For about eighteen months in the mid-2010s, native felt like the recalibration the industry needed.
Then reality arrived. Scale required aggregation. Aggregation required standardisation. Standardisation killed the native premise. You can’t be contextually matched to an environment when the same unit is running across three thousand publishers on a single insertion order. The UX became inconsistent. The quality signals became unreliable. Brands lost the editorial control they’d been promised, and found their assets sitting next to content they’d never knowingly have associated with.
Native chased scale by hijacking the 300x250, and morphed itself back into good ol’ display all over again.
More fundamentally: native tried to solve a format problem without solving the underlying tension between reach and relevance. It papered over the structural issue rather than resolving it.
The format problem remained unsolved. And then something larger happened.
The Zero-Click Inversion
The rise of LLMs did something to the open web that nobody in the ad industry quite reckoned with in time. It didn’t just change how people searched. It changed where intent forms.
For decades, the implicit contract of the internet was linear: user has a question, user goes to a search engine, user visits a website, publisher gets traffic, advertiser buys access to that attention. The entire programmatic ecosystem — the SSPs, the DSPs, the data platforms, the attribution models — was built around that flow.
LLMs broke the middle of that funnel. Not instantly, and not completely, but structurally and irreversibly. A meaningful cohort of users — skewed younger, skewed professional, skewed exactly toward the decision-makers B2B brands most need to reach — began resolving their queries inside the interface. Zero-click behaviour throttled traffic into the premium open web. Publishers felt it in their analytics before they knew how to talk about it publicly. They have been fighting it ever since, and the direction of travel is accelerating it. LLM’s were adopted about 4x faster than the internet.
The downstream effect for display was supply compression. Less premium inventory reaching the audiences that matter most. Higher CPMs for what remained. And a growing suspicion in CMO offices that open web display was a channel in structural decline.
It wasn’t. The intent was just moving upstream. The often forgotten truth is that even with this going on, the underlying open internet IS still growing.
The Irony At The Heart Of This Moment
Here is where it gets interesting. And where, if you’ve been watching programmatic for long enough, it starts to feel like a kind of poetic justice.
The same technology that drove zero-click behaviour and starved the open web of high-quality traffic is now, potentially, about to become the most significant demand driver display advertising has seen in a generation.
ChatGPT’s move into advertising and the signal that The Trade Desk could power the buy side of a display CPM model within that environment, reframes the entire equation. Suddenly, the question isn’t “where did the intent go?” It’s: “what happens when you can buy against it?”
If GPT is where modern purchase intent is being formed — and the data increasingly suggests it is, particularly among the millennial and Gen Z professionals who now hold real B2B buying authority — then ads served within that environment carry a different kind of weight. This isn’t a user who’s been retargeted twelve times and glazed over. This is a user at the precise moment of active research, forming a view, building a shortlist.
That’s not a display impression. That’s a search impression in a new coat. It is the best of search married with the best of programmatic. Real intent married with an open, data rich operating system.
Why The Trade Desk Angle Matters
I’ll admit my prior assumption was that OpenAI would build this in-house. They have the capital, the data, and the incentive to own the monetisation layer. A homebrew ad server, a proprietary auction, a walled garden that controls signal and inventory simultaneously — it would have been the logical move. I thought we’d be looking at Google 2.0 here
If they’re instead opening the buy side to TTD’s infrastructure, that changes everything.
It means the intent signals originating in GPT become actionable through the same programmatic pipes that B2B brands are already using to run CTV, audio, DOOH, and open web display. It means a DMP or CDP integration that already talks to TTD can, in principle, extend its reach into the highest-intent environment on the internet. It means the category doesn’t fragment further — it consolidates around infrastructure that serious B2B marketers already understand and can buy efficiently.
The best data ecosystem suddenly meets one of the best ad canvases - a huge moment of evolution. I have always said a lot of the programmatic infrastructure has been wasted on low rent sites with often unviewable placements. In a heartbeat the tide could turn here
And for programmatic as a category? This is the legitimacy moment. CMOs who have been steadily moving budget from display into performance channels, from awareness into last-touch, will be compelled to reconsider the model. Not because display suddenly got better at claiming credit — but because the audience and the moment are now undeniable.
A Rising Tide For The Whole Ecosystem
The fuller implication, though, goes beyond ChatGPT inventory itself.
Think about what happens to the intent trail. A user researches a solution inside GPT. They form a view. Maybe they click through to a vendor — maybe they don’t. But that interaction is a data event. That intent signal, if it flows back into the programmatic ecosystem — through TTD’s ID infrastructure, through first-party data matching, through the kind of clean room arrangement that makes sophisticated B2B marketers salivate — suddenly enriches every subsequent touchpoint. The data story is either born in ChatGPT or seriously fortified by it.
Find that same user on CTV. On LinkedIn. On the open web. On audio. The signal you’re activating against isn’t a third-party cookie approximation or a lookalike model built on last-quarter’s data. It’s live, declared, contextual intent from the moment they were actively researching.
Solutions like FunnelFuel, already adept at recognising buying committee members within accounts, can suddenly take a committee members research and amplify the message, knowing the moment is perfect, to the wider buying committee or even the wider org.
This is the thesis: GPT doesn’t just become an ad channel. It becomes the top of an intent graph that makes every channel below it more effective. And because the data all lives in programmatic pipes like TradeDesk, why would a CMO not work to maximise this intent layer FIRST, rather then last as they currently do?
For B2B specifically, where the buying cycle is long, the committee is complex, and the difference between an in-market account and an out-of-market one is everything, that signal enrichment is worth more than the display impression itself. It’s the data that makes your CTV spend smarter. It’s the trigger that makes your LinkedIn retargeting relevant rather than repetitive. It’s the signal that lets your programmatic partner — and yes, this is where FunnelFuel’s position becomes acutely relevant — run account-level progression scoring against real, current intent rather than modelled proxies.
Display was never just a reach channel. It was always a signal channel with a format problem. If ChatGPT’s ad model resolves the format question by placing the unit at the moment of intent formation, and TTD resolves the activation question by connecting that signal to the rest of the addressable ecosystem, then what we’re watching isn’t a minor product update from OpenAI.
It’s the structural rehabilitation of an entire media category.
What This Means For How You Allocate Budget
The practical implication for B2B CMOs is uncomfortable but important: the channel hierarchy you built your media mix around was constructed during a period of structural distortion. Display was undervalued because its format underdelivered. Search captured disproportionate budget because it sat closest to the moment of declared intent. Social inherited the brand dollars that used to go to TV.
If the moment of declared intent is now inside ChatGPT — and it is, increasingly, for the professionals you most need to reach — then the channel hierarchy needs rebuilding from scratch.
Display, powered by GPT intent signals and activated programmatically across a converging addressable ecosystem, doesn’t look like the unloved middle child anymore.
It looks like the critical pillar.
The brands that recognise this first, build the infrastructure to activate against it, and resist the temptation to wait for the case study before moving — those are the brands that will own the consideration set when budgets tighten and every other CMO is fighting for the same search inventory at inflated CPMs.
The tide is coming in. The question is whether you’ve built anything worth floating.
Mike Harty is a co-founder of FunnelFuel.io, a B2B-native programmatic advertising business, and the author of The B2B Stack — operator-led intelligence for modern B2B marketers. If you want to talk to me directly, simply reply to this email or email mike at FunnelFuel.io