That sentence is not a keyword. It’s a situation, a budget, a frustration, and a set of requirements — all in one breath. A week ago, that conversation would have stayed entirely private. As of August 24, 2026, it might also be worth money to an advertiser.

That’s the quiet, genuinely significant thing that just happened. On August 18, OpenAI announced it was expanding ChatGPT Ads across Europe, and by August 24 — confirmed live by ad-tech partner Criteo — the rollout had reached 31 European markets, including Germany, France, Spain, Italy, Sweden, Norway, Denmark, the Netherlands, and Austria. Access, for now, runs through OpenAI directly, agencies, and technology partners, with a self-service Ads Manager promised to follow later. Ads are showing up on the Free and Go plans; paid Business-class subscriptions stay ad-free, and OpenAI is explicit that advertising doesn’t influence or bias ChatGPT’s actual answers.

On the surface, this reads like “another ad network launched.” It isn’t. And the reason why is worth sitting with.

01The four platforms that taught marketers how to think

Every major ad platform has, at some point, forced marketers to learn a new mental model:

• Google taught us to think in search queries — explicit, typed intent.
• Meta taught us to think in attention and interest — who’s scrolling, who’s engaged.
• Amazon taught us to think in shopping intent — someone already in a buying mindset, wallet half-open.
• TikTok taught us to think in discovery — interest before intent even forms.

ChatGPT is introducing a fifth lens, and it’s a genuinely different one: contextual, problem-solving intent. Not “CRM software” as a keyword, but an entire paragraph describing a business’s constraints, its budget, its team size, and its frustration with the current tool. OpenAI’s ad system is explicitly built to weigh the context of the current conversation — not just a matched term — alongside the advertiser’s landing page, headline, copy, and something OpenAI calls “context hints.”

That single design choice is the real headline here, more than the market-count or the launch date.

02Why this breaks the old playbook

Here’s the part that should make performance marketers sit up: OpenAI has said, plainly, that these context hints are not exact-match keywords. You can’t simply port over your Google Ads keyword list and expect it to work.

Think about what that CRM company would traditionally build a campaign around:

CRM software · CRM platform · best CRM

Now compare that to what actually works in a ChatGPT-native structure — intent clusters, built around situations rather than terms:

• Replacing spreadsheets — small teams managing leads manually, looking for their first CRM.
• Switching CRMs — businesses frustrated with cost or complexity in their current tool.
• Automation-driven — sales teams specifically hunting for workflow or lead-nurturing automation.
• SMB-specific — small businesses wanting affordable CRM software for teams under 25 people.

That’s not keyword advertising with extra steps. That’s closer to Jobs-to-be-Done marketing — campaign architecture built around why someone is looking, not what they typed.

Which means the underlying structure of a campaign is shifting too. Traditional search runs:

Keyword → Ad → Landing page

ChatGPT-native advertising increasingly looks like:

Situation → Intent → Context → Ad → Landing page

The competitive edge stops being “who has the biggest keyword list” and starts becoming “who understands the actual situations that make a customer start researching this category in the first place.” That favors marketers who are strong at customer research, positioning, and conversion optimization — not just bid management.

03This isn’t an experiment anymore — it’s infrastructure

It would be easy to dismiss all of this as another beta test that fizzles out. The research says otherwise. ChatGPT Ads already has the plumbing of a real performance channel, not a placement pilot:

• Pricing models — CPM for reach, CPC for traffic, and oCPC (conversion-optimized CPC) for advertisers chasing actual downstream conversions.
• Measurement infrastructure — an OpenAI Pixel, a Conversions API, UTM tracking, custom conversion events, custom audiences, conversion attribution, and third-party measurement integrations.
• Targeting — platform-level (iOS / Android / Web) and geographic targeting.
• Auction mechanics — a relevance-weighted, second-price auction, meaning the highest bid alone doesn’t win. OpenAI currently suggests starting CPC bids somewhere in the $3–$5 range, though that will vary by category and geography.

That combination — pricing flexibility, real attribution tooling, and a relevance-weighted auction — is what separates “ad placement experiment” from “channel you can actually build a media plan around.”

04The e-commerce detail hiding in plain sight: product feeds

If there’s one feature that deserves more attention than it’s getting, it’s product feeds. Advertisers can now upload feeds containing product images, titles, prices, sale prices, ratings, brand, and landing pages — and OpenAI has said, specifically, that advertisers using product feeds have been among the strongest performers on the platform so far.

That’s worth pausing on. Picture someone typing:

“I need waterproof walking shoes under €120 for a trip to Norway in October.”

A traditional Google Shopping listing bids on “waterproof walking shoes.” A ChatGPT-native product match potentially reasons across budget, category, climate, use case, destination, and timing all at once — matching actual products against an actual situation. That starts to look less like sponsored text under an AI answer, and more like Google Shopping fused with conversational discovery.

05The early numbers — encouraging, but not gospel

Criteo, an ad-tech partner already live in the ecosystem, reported the following as of launch day:

Those are genuinely striking numbers. They’re also partner-reported results from one participant in the ecosystem — not an independently audited, platform-wide benchmark, and they shouldn’t be treated as one.

The counterweight is worth including too: one small, independent three-week test running a B2B WordPress offer reported cheap traffic but essentially no meaningful engagement or conversions.

High conversational intent does not automatically equal high commercial intent.

Someone describing a detailed problem might be actively buying — or writing a university paper, benchmarking competitors, researching on someone else’s behalf, or just idly exploring. Conversation depth gives the platform richer signals, but it doesn’t guarantee purchase readiness. Any evaluation of this channel needs to hold both facts at once.

06If the 80% number holds up, it changes what ChatGPT Ads actually is

That new-customer statistic — if it survives at scale, and it’s still early — is arguably more strategically important than the CTR figures. It would place ChatGPT Ads somewhere between:

• Paid search — which mostly captures demand that already exists
• Paid social — which mostly generates demand from scratch

into a category of its own: demand discovery. Users describing a problem often haven’t picked a brand yet — they’re still describing the situation. That’s precisely the moment a smaller, specialist advertiser has a real shot at entering the consideration set, before brand loyalty or SEO dominance has locked the door.

Imagine someone asking: “Which packaging supplier in Europe offers recyclable magnetic boxes for luxury cosmetics with low MOQs?” On Google, that query gets buried under whoever has the biggest SEO authority and PPC budget. In a conversational system built around matching a specific problem to a specific answer, a smaller advertiser who genuinely fits the description has a real opening. That’s very much in line with OpenAI’s own stated ambition for advertising to help smaller and emerging brands compete.

07What this means Monday morning

Not “go start running ChatGPT Ads.” That’s the wrong first move. The right sequence looks like this:

08Measuring it properly — four layers, not one number

A single “ChatGPT ROAS” figure won’t tell the real story. A more honest framework runs in four layers:

OpenAI’s own Ads Manager currently reports impressions, clicks, spend, CTR, CPC, CPM, and conversions — a solid delivery-through-conversion base. But the sharper, more useful question sits above all of that:

How much revenue would we not have captured at all without this channel?

That’s a materially harder question than “what’s the ROAS,” and it’s the one that actually determines whether this channel deserves a permanent line in the budget.

09Two things not to confuse

Being recommended by ChatGPT organically and paying to appear in ChatGPT are two entirely separate disciplines — the same way SEO and Google Ads have always been separate. Expect the marketing stack to expand accordingly: SEO, PPC, marketplace ads, social ads, generative/answer-engine optimization (GEO/AEO), and now conversational ads, all sitting alongside each other.

For agencies, that’s not a footnote — it’s a plausible new service line: conversational-intent research, context-hint architecture, feed optimization, AI-native creative, AI-specific landing pages, Pixel/CAPI implementation, and conversational attribution. The agencies that build this capability early, before there’s a mature certification ecosystem around it, get to keep the learning-curve advantage that everyone else will be paying tuition for later.

10The honest limitations

None of this makes ChatGPT Ads a Google-killer today, and it shouldn’t be sold as one. It’s genuinely early, and several real unknowns remain:

• Inventory scale — search advertising runs on enormous, predictable query volume; conversational commercial inventory may behave very differently.
• Intent consistency — not every commercially-worded conversation belongs to an actual buyer.
• Benchmark stability — the CPC, CTR, and conversion data available right now is still immature.
• Attribution complexity — AI-assisted buying journeys can hop between ChatGPT, Google, a marketplace, and a direct purchase.
• Regulation — European privacy and AI rules could still reshape targeting and personalization, especially in the EEA and Switzerland.
• User trust — OpenAI has to monetize an interface people frequently treat more like a trusted adviser than a media feed. Notably, OpenAI’s own ad policies already exclude advertising alongside many sensitive or brand-unsafe conversations.

11The real headline

The easy version of this story is: “ChatGPT Ads launched in Europe, here’s a market list.” The truer version is this:

ChatGPT Ads isn’t introducing another high-intent ad placement. It’s introducing a new targeting primitive — the actual problem a customer is trying to solve, described in their own words, with their own constraints attached.

Google Ads taught marketers to think in keywords. Meta taught marketers to think in audiences. Amazon taught marketers to think in products and purchase intent. ChatGPT Ads is going to force marketers to think in problems, situations, comparisons, and decision contexts — a fundamentally different unit of targeting than anything that came before it.

That’s why this launch is worth more attention than a line item on a media plan. It isn’t advertising moving from experiment to “real media channel.” It’s advertising moving from experimental placement to a measurable performance platform — and the infrastructure now in place (conversion optimization, context hints, custom audiences, product feeds, Pixel/CAPI, platform targeting, and third-party measurement) is what makes that claim defensible rather than hopeful.

The sales manager in Berlin, typing her CRM problem into a chat window at 4pm, has no idea any of this is happening behind the scenes. That’s rather the point.