Criteo Becomes First Ad Tech Partner in OpenAI's ChatGPT Ad Pilot. The $60 CPM Era of AI Advertising Begins.

Criteo is now inside ChatGPT. On March 2, the commerce media platform announced it has become the first advertising technology partner to integrate with OpenAI's advertising pilot, connecting roughly 17,000 advertisers to ChatGPT's Free and Go subscription tiers in the United States.

The numbers make the stakes plain. ChatGPT now reaches 900 million weekly active users. Ads will appear at the bottom of conversational responses, clearly labeled as sponsored content. The cost? Approximately $60 per thousand impressions, with a $200,000 minimum buy-in. Early data from 500 U.S. retailers observed in February 2026 shows that users referred from LLM platforms convert at roughly 1.5 times the rate of other referral channels, according to the company.

If you've spent the last two years wondering when AI would stop being a marketing buzzword and start being an actual media channel, this is that moment. Not a concept deck. Not a beta waitlist. A programmatic integration with the world's most popular AI interface, carrying real CPMs and real conversion data.

"This integration with OpenAI represents an exciting step forward in advancing advertising in an emerging AI experience," Criteo CEO Michael Komasinski said in the announcement.

The mechanics are straightforward, even if the implications aren't. Ads appear after ChatGPT's response to a query, not within the conversational flow itself. Each placement carries a clear "sponsored" label. OpenAI has implemented content restrictions: no ads touching health, mental health, or politics, and nothing shown to users under 18.

A clean tier structure creates a clean division between ad-supported and ad-free experiences:

Source: OpenAI advertising pilot documentation, March 2026

That $60 CPM demands context. Standard Meta advertising CPMs typically run under $20. Google Search ads range from $20 to $50 depending on industry vertical. Netflix's ad-supported tier launched at a comparable $60 CPM in 2022, though that figure has since come down. ChatGPT's premium reflects the same bet Netflix initially made: low ad clutter plus high user attention equals pricing power.

The 1.5x conversion rate is the number that will show up in every pitch deck for the next six months. It deserves both attention and scrutiny.

From a bullish perspective, the case writes itself. When someone asks ChatGPT to "recommend a laptop for video editing under $1,500," the commercial intent is explicit in ways that a Facebook scroll or even a Google search result page can't always match. The user is actively researching, comparing, and forming preferences inside the conversation. An ad placed at that moment reaches someone further down the funnel than most digital touchpoints can claim.

But the February 2026 data comes from just 500 retailers over a single month. These are early adopters who likely skew toward categories where conversational AI queries are already common: consumer electronics, travel, consumer goods. Whether those conversion rates hold across verticals like financial services or B2B SaaS is genuinely unknown. There's also the question of "AI ad blindness." Banner blindness took years to develop. Conversational ad blindness could develop faster, since users are trained to focus on the model's response and scroll past everything else.

Criteo's own Q4 2025 results add a wrinkle. The company reported a 2% revenue dip to $541 million, with Q1 2026 guidance of $245 million to $250 million in revenue (excluding traffic acquisition costs) that analysts considered soft. This partnership isn't just a strategic play for Criteo. It's a lifeline for a company that needs to prove its post-cookie, post-retargeting reinvention is actually working.

For performance marketing managers staring at budget allocation spreadsheets, the question isn't whether ChatGPT ads are interesting. It's whether $60 CPM delivers incremental value that justifies pulling dollars from channels with deeper measurement infrastructure.

Sources: eMarketer, Digiday, Ad Age, Criteo press materials, March 2026

The measurement gap is the real barrier. Right now, ChatGPT ads deliver impressions and clicks. That's it. No view-through attribution, no multi-touch modeling integration, no incrementality testing framework. For a CMO trying to justify a $200,000 minimum commitment to their CFO, "trust us, the conversions are real" isn't going to survive a quarterly review.

OpenAI's early direct-sales approach to advertising was, by most accounts, clunky. A $200,000 minimum with basic reporting doesn't exactly invite mid-market experimentation. Criteo changes the equation by sitting between OpenAI's inventory and the thousands of brands that already use Criteo's demand-side platform.

The company activates more than $4 billion in annual media spend across advertisers including Expedia, Best Buy, Target, and others. More critically, Criteo has spent two decades building recommendation models that understand product relevance, purchase probability, and cross-sell potential. That's the kind of commerce intelligence that makes an ad inside a "what laptop should I buy?" conversation actually useful rather than random.

Todd Parsons, Criteo's chief product officer, has described the shift as collapsing the manual planning steps that traditionally sit between an advertiser's objective and campaign execution. Through Model Context Protocol integrations, Criteo's APIs can now be queried by AI agents when generating product recommendations, essentially piping commerce signals like product relevance and retailer performance directly into the LLM environment.

For more than a decade, digital advertising has operated as a functional duopoly. Google owned search intent. Meta owned social attention. Everyone else competed for scraps. ChatGPT's entry doesn't immediately threaten that structure, but it introduces a third high-intent surface that CMOs now have to evaluate, budget for, and eventually measure.

The timing aligns with a broader fragmentation trend. Koah just raised $20.5 million to build monetization infrastructure for AI apps more broadly. Verve Group became the first ad platform to target using AI chat data. The IAB Tech Lab is proposing payment rails for AI content access. Individually, these are incremental moves. Together, they describe a new layer of the internet that will need its own advertising infrastructure, its own measurement standards, and its own budget line items.

Here is the most interesting question for your 2026 media plan isn't whether to test ChatGPT ads. It's how to account for a media landscape where the primary interface is a conversation, not a page. Attribution models built for clicks and pageviews don't translate cleanly to environments where a user asks a question, receives an answer, sees a recommendation, and buys, all without ever visiting your website.

Three things will determine whether this pilot becomes a permanent channel or an expensive experiment. First, measurement. Until OpenAI and Criteo offer view-through attribution and multi-touch integration, the $200,000 minimum will lock out all but the most adventurous (or flush) advertisers. Second, user reaction. The ad-free experience on Plus and Pro creates an escape valve, but if Free and Go users start perceiving bias in ChatGPT's recommendations, the trust problem could undermine the entire model. Third, competition. Google's AI Mode already reaches 75 million daily users and added checkout capabilities in early March. If Google places ads inside its own conversational search, the CPM premium for ChatGPT evaporates overnight.

The house cat that lived inside a research lab for two years just walked outside. Advertising inside conversational AI is no longer theoretical. The only question is how fast the rest of the industry follows Criteo through the door.

If you're a VP of Marketing or Director of Growth deciding whether to test this channel, the decision framework is clearer than it looks. The $200,000 minimum is a real barrier, but it's not insurmountable for mid-market e-commerce or B2B brands with $5 million-plus annual digital budgets. What matters is whether your product category maps to the kinds of queries that drive high-intent ChatGPT conversations.

Travel, consumer electronics, financial products, and education are natural fits. Complex B2B purchases with long sales cycles are probably not, at least not yet. The conversational format rewards products that benefit from comparison and recommendation, not ones that require a 45-minute demo.

The reporting limitations are the real constraint. If your attribution stack requires multi-touch visibility to justify spend, ChatGPT ads will frustrate your analytics team. Impressions and clicks are table stakes for any serious performance operation. Until there's a way to connect ChatGPT ad exposure to downstream conversions in your CRM or analytics platform, you're running on faith and directional signals.

Every time a new channel enters the media mix, measurement breaks a little more. ChatGPT ads accelerate an already painful trend. Your media mix model wasn't built for a channel where the "impression" is a contextual recommendation inside a conversation. Your attribution platform can't stitch together a user who asked ChatGPT a question, saw a Criteo-served ad, and then purchased on your website three days later through a branded search.

The companies that figure out measurement first will have a structural advantage. If you're running incrementality tests on your existing channels, start thinking about how you'd isolate the ChatGPT signal. Geography-based holdout tests are probably the most practical approach in the short term.

One genuinely surprising thing about this entire development: it took this long. ChatGPT has been the most-discussed technology product on Earth for over three years, and advertising didn't show up until 2026. The delay reflects OpenAI's caution, not a lack of advertiser demand. The demand was always there. The infrastructure wasn't.