On August 31, 2026, OpenAI opened self-service access to ChatGPT Ads in 31 European markets, including France. Before that date, a French business wanting to advertise inside ChatGPT conversations had to go through an agency or certified partner, a route that had only existed since August 24. The self-serve switch means an approved advertiser can now set up a campaign directly at ads.openai.com, the same way they already do on Google or Meta, without an intermediary.
For a B2B marketer who has already had to rethink how buyers search since AI Mode changed what shows up in Google Ads data, this is a familiar kind of shift: a new interface opens, and there is no established playbook for what works inside it. What makes the French launch worth a closer look right now is not the novelty of a new ad channel. It is a smaller, more specific fact: the tools to measure ChatGPT Ads performance were built and announced before most of Europe could even open an account.
From a managed US pilot to a French self-serve channel in seven months
ChatGPT Ads started as a managed pilot: OpenAI began serving ads to free and ChatGPT Go users in the United States on February 9, 2026, through a small group of launch advertisers that included Target, Albertsons and Williams-Sonoma, soon joined through agency WPP Media by Adobe, Ford and Mazda (Best Buy, Lowe’s and VistaPrint would not join the channel until the public self-serve launch on July 22). That pilot required a minimum spend of $250,000 at launch, lowered to $50,000 in April 2026, and ran only through agency partners, which kept it out of reach for most small and mid-sized businesses. The gate came down on May 5, 2026, when OpenAI launched ChatGPT Ads Manager, a true self-serve platform with cost-per-click bidding, no minimum spend, and a relevance-weighted second-price auction to decide which ad appears in a given conversation.
Europe followed in two separate steps, not one. OpenAI started serving ads across 31 European markets, the EU plus Switzerland, Norway, Iceland and Liechtenstein, on August 24, 2026, but only through agency and partner relationships (Publicis, Omnicom, WPP, Havas, Dentsu and MediaPlus among them). Direct self-serve access for approved advertisers in those same markets, France included, opened a week later, on August 31. Early categories were deliberately narrow: household and consumer goods, local services, travel and entertainment, and digital products and education. A French B2B software company or consultancy outside those categories cannot assume it is eligible yet simply because the channel exists in its country.
OpenAI used that same day, August 31, to disclose a different kind of number in a separate post: ChatGPT Ads had crossed $1 billion in annualized revenue in under 200 days since the February pilot, with what the company described as tens of thousands of advertisers active on the platform. OpenAI frames this as one piece of a diversified business ahead of a planned IPO, alongside subscriptions, enterprise contracts and API usage. The pace is real. What it proves about advertiser demand, as opposed to OpenAI’s own growth targets, is a separate question, and the case studies published so far do not resolve it either way.
The measurement layer arrived before the French market did
This is where the timeline gets interesting for anyone thinking about attribution. On June 10, 2026, roughly two and a half months before France had any form of access to ChatGPT Ads, call-tracking vendor CallRail announced a native integration that automatically attributes calls, text messages and form submissions to specific ChatGPT ad campaigns and ad groups, then sends that qualified conversion data back to OpenAI to improve targeting and bidding. Activation runs through an API key in CallRail’s existing settings, at no extra cost for eligible customers. CallRail describes itself as the first SMB and agency-focused provider to offer this kind of attribution.
A second vendor followed on July 20, 2026: CallTrackingMetrics (CTM) launched a native OpenAI Ads integration with closed-loop call attribution, plus a new in-platform agent called AskCTM for self-serve setup. CTM’s own press release does not claim a competitive “first” ranking: it frames closed-loop attribution, already available for Google, Meta and Microsoft, as extended “for the first time” to AI-powered advertising — a first claimed on the feature, not on rank among vendors. Only CallRail explicitly claims a first-place ranking in this specific market (“first SMB and agency-focused provider”), a claim neither confirmed nor disputed by an independent source. What matters more than who was first is the sequence itself: the infrastructure to measure a ChatGPT ad campaign existed well before a French advertiser could legally open one.
What’s actually been proven so far is retail, not B2B
Measurability and relevance are two different claims, and the second one is where the current evidence runs thin. Every advertiser OpenAI has publicly featured as a reference case, Best Buy, Lowe’s, VistaPrint, and separately Newegg, which reported a 3x return on ad spend over 28 days and 7x during a promotional period, sells directly to consumers. In that kind of journey, the stakes for an advertiser are catching the shopper at the precise moment of a home-improvement or electronics search, inside a chat window instead of a search bar. That is a retail buying pattern: short consideration window, single decision-maker, immediate purchase intent.
None of that maps cleanly onto a B2B sales cycle that runs for weeks, involves several stakeholders, and rarely converts on the first interaction, whether that interaction happens in a search box or inside a chatbot. A French marketer evaluating ChatGPT Ads today is being asked to extrapolate a B2B outcome from a set of consumer retail examples, not to replicate a proven result. That is not a reason to dismiss the channel, since the same extrapolation problem existed for most new ad formats when they launched. It is a reason to treat the current case studies as context rather than precedent.
Before testing from France, check what the attribution tool actually covers
There is a second, more practical gap between the headline announcements and what a French account can actually use. According to CallRail’s own support documentation, as relayed by several trade sources (the vendor blocks direct access to its support pages), native call tracking coverage would be limited to the United States, Canada, the United Kingdom and Australia; extending tracking numbers to other countries would require contacting CallRail support to activate international number features first, with the company reportedly noting it cannot forward calls to some UK and Australian mobile numbers due to carrier cost structures. Nothing in CallRail’s ChatGPT Ads announcement states that the integration itself carries a broader footprint than that base coverage.
CTM’s own positioning is different on this point: the company advertises call tracking numbers in more than 80 countries, a list that has historically included France and the rest of continental Europe. The two vendors named in this story are not on equal footing for a French advertiser, even though both describe their ChatGPT Ads integration in nearly identical terms. Before testing the channel from France, the more useful question to ask a vendor is not “do you support ChatGPT Ads” but “do you support my country natively, or does this require an activation request first.”
None of this is a reason to wait indefinitely. It is a reason to separate three claims that the current coverage blurs together: the channel is open in France, the attribution tooling for it existed early, and both are proven to work for B2B. Only the first two currently hold up against sourced, dated evidence. The third is still an open question, and a familiar one if you have already had to decide who answers for an automated bidding system making decisions inside a platform you do not fully control.
What this means before you brief anyone on it
The honest summary for a B2B decision-maker is that the market moved fast on access and even faster on measurement claims, but neither pace tells you whether a chat-native ad format converts a prospect who needs three internal sign-offs before buying. The same discipline that has started paying off in cleaning up B2B conversion data for Google Ads applies here before the first euro is spent: confirm the category you want to advertise in is already eligible, confirm the attribution vendor covers a French number without a support ticket, and watch for the first genuine B2B case study rather than assuming one is coming.
If you want to pressure-test what a first ChatGPT Ads trial would actually look like against your own sales cycle, we are glad to compare notes.
Photos: Pixabay/StartupStockPhotos.
