Sponsored placements began appearing inside ChatGPT across 31 European markets today, and the only quantified picture of what the channel does in its opening weeks comes from a search intelligence vendor that measured the two launches before this one.

Adthena published its read-across on August 21, 2026, three days after OpenAI confirmed the European expansion and three days before the ads themselves surfaced. The analysis, written by Serra Benstead, draws on almost 850,000 AI search queries collected in the United States and the United Kingdom. According to Adthena, the same launch sequence has now run twice: a narrow band of broad-catalogue advertisers absorbs most of the early visibility, ad frequency starts low, single placements dominate, and regulated categories stay absent.

None of the figures describe Europe. Adthena states this directly, noting that the numbers were captured before the European launch and are being used to predict behaviour rather than to report it. That distinction carries more weight than it might in an ordinary market entry, because the European rollout operates under constraints that did not apply in either of the markets Adthena measured.

What OpenAI confirmed on August 18

According to Adthena, OpenAI confirmed on August 18 that ChatGPT Ads was coming to 31 European markets, describing the move as its largest expansion to date. The same product update email that carried the European schedule also documented a campaign-level platform targeting control and confirmed that Brazil and Mexico had already activated, as PPC Land reported on August 21.

Three structural details from the announcement shape how the first weeks behave. Access is gated at the start, routed through OpenAI's ads team, agencies and technology partners, with a self-service Ads Manager described as arriving later in the summer. Delivery is confined to the Free and Go subscription tiers, leaving Plus, Pro and Enterprise accounts without advertising, a boundary that has held since the pilot's first day. And OpenAI characterises the platform as still early in its development, notwithstanding the conversion optimisation, geographic targeting, custom audiences and pixel-based measurement it has accumulated.

Adthena also reports observing a negative phrases field inside ChatGPT Ads Manager, a search-style exclusion control, before any public notice of it appeared. That pattern of documentation trailing deployment has recurred across the platform's build-out. View-through conversion reporting surfaced through documentation revisions on August 18 and 19rather than through an announcement, and automated bidding arrived preselected in new ad groups with a help centre page dated before the email that told advertisers it existed.

Where the early visibility concentrated in the United States

The sharpest number in the analysis concerns category concentration. According to Adthena, Retail and Fashion accounted for 24% of US query volume but 39% of all ad items observed, running at an ad frequency of 6.55% against a US average near 4.5%. The gap between share of demand and share of supply is the mechanism the firm expects to repeat: when the advertiser field is thin, the brands that match the widest range of conversational prompts capture a disproportionate share of the available slots.

Adthena identifies the first live ChatGPT ad it confirmed as Expedia, triggered on a single travel-planning prompt, with Best Buy close behind and landing two placements within one response. Both fit the profile the firm describes: broad catalogues, comparison-heavy intent, and a query surface wide enough to intersect with a large share of conversations.

Less predictable are the categories that ran hottest by frequency. According to Adthena, Logistics recorded 12.41%, Home and Garden 11.99%, and Beauty 10.03% in the United States. Those three sit well above the average and above Retail and Fashion, which suggests that the concentration story and the frequency story are not the same story. One measures who occupies the slots; the other measures how often a slot appears at all within a given vertical.

The format constraint is the second load-bearing figure. Across a March-to-May measurement window, ChatGPT served ads on 4.47% of US queries, and the average ad-bearing answer carried 1.06 ad items, according to Adthena. There was no row of three and no shopping carousel. A single sponsored unit sat below the response.

The comparison the firm draws is with Google, whose AI surfaces averaged 3.53 items per ad-bearing answer, behaving closer to a carousel than to a single placement. That difference has a direct auction consequence. One slot per answer means the number of winners per query is one, and the pool of bidders competing for it determines the clearing price. Adthena's argument is that a shallow pool and a single slot keep costs suppressed at the start, and that both conditions erode as more advertisers are approved in each market.

That framing is a vendor's framing, and it arrives from a company that sells competitive monitoring for the channel. Adthena launched AdBridge, a free tool that ports Google Ads campaigns into ChatGPT, on April 27, 2026, and an intelligence platform monitoring more than 300,000 daily prompts on May 11. The commercial interest does not invalidate the measurements, but it sits alongside them.

Ashley Fletcher, Adthena's chief marketing officer, framed the pattern in the published analysis. According to Fletcher, "the brands that establish position while a market is young are the hardest to dislodge." He described Europe's 31 markets as "the biggest early-mover opportunity we've seen in this channel yet."

The UK benchmark and the problem inside it

Adthena presents the United Kingdom as the closest available proxy for the 31 new markets. The UK went live on June 6, 2026, the first European market to receive the pilot, under the same gated, registration-only structure with delivery confined to Free and Go users and no open self-service at the outset. Self-serve buying reached UK advertisers on June 22, roughly sixteen days later.

A contradiction sits inside that proxy, and it originates with Adthena itself. The firm's Monthly AI Data Pulse for June 2026 recorded a 0.00 percent ChatGPT ad frequency across 169,560 United Kingdom scrapes, and zero UK ad items among 29,237 catalogued across all markets that month. The current analysis describes the UK launch as the best benchmark available for how the European markets will unfold, without reconciling that description against a June reading in which its own monitoring found no paid placements in the market at all.

The unreconciled readings extend further. Adthena separately counted 7,378 distinct ChatGPT advertisers during the week of July 13 to 20, with the United States accounting for 60.1 percent, and reported US ChatGPT ad frequency of 24.7% in that window. The figure published this week for the March-to-May period is 4.47%. Both are described as United States ad frequency. Neither the methodology difference nor the trajectory between them is set out in the current analysis.

Independent measurement does not settle the question. Similarweb recorded sponsored placements in 26 percent of ChatGPT responses at a 0.50 percent click-through rate in a dataset published on August 17, 2026. Different panels, different denominators, and no third-party measurement standard for conversational advertising surfaces produce numbers that cannot be laid against each other cleanly.

What does not travel from the US benchmark

The read-across assumes the European product behaves like the American one. On at least one documented dimension it does not.

OpenAI emailed European users at 08:29 on Saturday, August 15, 2026, confirming that advertising would reach the Free and Go plans before the end of the month. The accompanying help centre material stated that personalised advertising is not initially available in the European Economic Area or Switzerland. Selection at launch draws on the current conversation topic plus general location and device type, a narrower signal set than the past-chat and prior-engagement inputs that inform US delivery.

Every US frequency and concentration figure in Adthena's dataset was produced by a system with access to those additional signals. A market where selection runs on live thread context alone is not the same auction, and the direction of the difference is not predictable from the American numbers.

The legal footing sits in Dublin. The footer of OpenAI's European notification carried the registered office of OpenAI Ireland Limited, the controller named in the company's European privacy policy for users in the European Economic Area and Switzerland, which places lead supervisory authority with the Irish Data Protection Commission. PPC Land's review of three versions of that European privacy policy, dated November 4, 2024, February 6, 2026 and June 4, 2026, found no mention of advertising in any of them.

Measurement defaults compound the point. Automatic advanced matching, which transmits hashed customer information collected from website forms, became the default on existing web pixels on August 17, 2026 following a ten-day opt-out window. The ChatGPT pixel initialises consent as true unless explicitly set to false or a stored denial is found. In jurisdictions where prior consent is a precondition for tracking, the lawful basis for that transmission rests with the advertiser rather than with the platform, regardless of the default.

The stack European advertisers inherit

Six months separate the February 9, 2026 United States launch from today. In that window the platform accumulated a self-serve Ads Manager with cost-per-click bidding, opened to all US businesses on May 5 with the minimum spend removed, a measurement pixel and Conversions API, custom audience targeting, daily budgets, geographic targeting, conversion optimisation, product feeds, mobile attribution through AppsFlyer and Adjust, and a bulk API.

Entry pricing moved in the same period. The pilot opened at a 60 dollar CPM with a minimum commitment between 200,000 and 250,000 dollars. CPMs fell to as low as 25 dollars by mid-April as the threshold dropped to 50,000 dollars, before the minimum disappeared entirely on May 5.

Advertisers entering the 31 European markets therefore arrive at a platform considerably more built out than the one Adthena measured during its March-to-May window, and considerably less built out in one specific respect. The market count also complicates. Nine markets were live before today, following the Mexico and Brazil activation on August 17, which closed the five-country list OpenAI set out on May 7. The United Kingdom sits inside both the nine and inside common definitions of the European market set, which means the arithmetic of a total depends on which grouping is being counted.

Why this matters for the marketing community

The competitive asymmetry Adthena names is the part of its analysis that does not depend on any contested frequency number. United States advertisers have been running campaigns in this channel since February. UK advertisers have been running them since June. Advertisers in the 31 markets activating today start with no first-party learning about which prompts trigger placements, which creative formulations perform, or what a reasonable bid looks like in a single-slot auction.

That gap is measurable in months rather than in features, and features are the thing that has been arriving fastest. A market entering with automated bidding preselected, view-through conversions excluded from the optimisation loop, and no personalised delivery signal is a market where the learning curve runs on a different set of variables than the one the American data describes.

The regulated-sector exclusion holds across all of it. Dating, health, financial services and politics have been outside the test since it began, which removes a substantial share of European advertising demand from the auction before it opens. What remains is the broad-catalogue retail, travel and comparison inventory that Adthena expects to move first, in a channel where the number of paid slots per answer, on the most recent US measurement available, is approximately one.

Timeline

Summary

Who: OpenAI, operator of ChatGPT Ads, and Adthena, the London-founded search intelligence company whose analysis of the rollout was written by Serra Benstead and carries commentary from chief marketing officer Ashley Fletcher. The change affects media buyers, agencies and advertisers across 31 European markets, technology partners including Criteo, Kargo, Adobe, Pacvue, StackAdapt and AdRoll, and ChatGPT users on Free and Go subscriptions.

What: ChatGPT advertising became visible across 31 European markets. Adthena published a benchmark analysis drawn from nearly 850,000 AI search queries in the United States and United Kingdom, reporting that Retail and Fashion took 39% of observed US ad items on 24% of query volume, that ads appeared on 4.47% of US queries during a March-to-May window, and that the average ad-bearing answer carried 1.06 ad items against 3.53 for Google's AI surfaces. Logistics, Home and Garden, and Beauty recorded the highest US frequencies at 12.41%, 11.99% and 10.03%. Access is gated at the start through OpenAI's ads team, agencies and technology partners, with a self-service Ads Manager described as arriving later in the summer.

When: OpenAI confirmed the 31-market expansion on August 18, 2026. Adthena published its analysis on August 21, 2026. The ads began appearing today, August 24, 2026. The measurement window behind the headline frequency figures ran from March to May 2026.

Where: The 31 European markets, following the United States on February 9, 2026, Canada, Australia and New Zealand in March, the United Kingdom on June 6, Japan and South Korea on June 22, and Mexico and Brazil on August 17. Delivery is confined to Free and Go subscription tiers.

Why: Adthena's argument is that the auction dynamics of a new ChatGPT market are set in its first weeks, when a single sponsored slot per answer is contested by a shallow advertiser pool. The read-across carries two unresolved qualifications. Personalised advertising is not initially available in the European Economic Area or Switzerland, which means European selection runs on narrower signals than every US figure in the dataset was produced under. And Adthena's own June monitoring recorded zero paid ChatGPT placements across 169,560 United Kingdom scrapes, a result the current analysis does not reconcile with its presentation of the UK as the closest available benchmark.