ChatGPT ads are coming to Europe. On 18 August 2026, OpenAI announced that advertising in ChatGPT will expand to 31 European markets: the 27 EU member states plus Iceland, Liechtenstein, Norway and Switzerland. According to Digiday, ads start serving on Monday 24 August, bought through OpenAI's ads team and agency groups including Publicis, Omnicom, WPP, Havas, Dentsu and MediaPlus.
There is one big European difference: the ads will not be personalised, at least not at first. For marketers, that makes ChatGPT ads in Europe a contextual channel, sold on what someone is asking right now rather than on who they are. It also raises a measurement question that matters more than the launch itself: once ChatGPT sends you both paid and organic visitors, how do you tell them apart?
What OpenAI announced for ChatGPT ads in Europe
OpenAI's announcement and its help centre article Ads in ChatGPT set out the basics. As reported by Notebookcheck and Trending Topics:
| ChatGPT ads in Europe | |
|---|---|
| Announced | 18 August 2026 |
| Ads start serving | 24 August 2026 |
| Markets | 31: EU-27, Iceland, Liechtenstein, Norway, Switzerland |
| Plans with ads | Free and Go (Go costs €7.99 a month) |
| Plans without ads | Plus, Pro, Business, Enterprise, Edu |
| Personalised ads | Not initially available in the EEA or Switzerland |
| How to buy | OpenAI's ads team and partner agencies |
| Placement | Below the answer, labelled as sponsored, visually separate |
| What advertisers get | Aggregated data such as views and clicks, not chats |
OpenAI says ads do not influence ChatGPT's answers. Ads are not shown next to sensitive topics such as physical or mental health and politics, political ads are not allowed, and accounts OpenAI judges to belong to someone under 18 see no ads. Free users who don't want ads can switch to an ad-free option with lower limits, or pay for a higher plan.
How Europe fits into the global rollout
Europe is not the first market. ChatGPT ads began as a test in the United States in February 2026 and then reached Canada, Australia and New Zealand. The UK followed in June 2026 as the first European market, and Mexico, Brazil, Japan and South Korea came next. Digiday describes the European launch as the moment OpenAI's ads business reaches its six-month mark.
Two numbers explain why advertisers care. Speaking at Cannes, OpenAI's vice president of ads said that ChatGPT had passed 900 million weekly active users and that about 20% of queries showed direct commercial intent, according to Digiday. Digiday also reports that OpenAI introduced cost-per-click buying after launch and that CPC now accounts for most ad spend. That makes ChatGPT ads look a lot like search ads: you pay for clicks to your site.
OpenAI has said a self-serve buying platform for small businesses is coming, but at the time of writing it has not given a date for Europe. For now, if you want in at launch, you go through an agency or OpenAI's sales team.
Why ChatGPT ads are not personalised in the EEA
OpenAI's help centre is short on this point: personalised ads "are not initially available" in the European Economic Area or Switzerland. Instead, ads there are selected from the current conversation and broad signals such as approximate location and language. Past chats and saved memories are not used.
The background is the GDPR. Showing someone ads based on a profile of their interests means processing their personal data for advertising, and that needs a legal basis. According to Digiday's report of 18 August, OpenAI's new EU privacy policy, posted on 14 August, says it will rely "on consent as the legal basis" for personalised ads, rather than on legitimate interest. Trending Topics reports that OpenAI cites legitimate interest for the non-personalised ads it shows at launch.
Digiday points to Meta as the cautionary example: European regulators rejected Meta's attempts to base behavioural advertising first on its user contract and later on legitimate interest. OpenAI has chosen to start without personalisation and to ask for consent before adding it. When or whether that second step comes in Europe, OpenAI has not said.
Not legal advice. This section summarises how OpenAI describes its own approach, as reported by the sources linked above. It is not an assessment of whether that approach is compliant.
For advertisers, the practical consequence is simple. In Europe you are buying context and intent, not audiences. Your ad shows because someone is asking about something related to your product, not because a profile says they are in your target group. Your landing page should match the question, not a persona.
The measurement problem: paid vs organic AI traffic
Here is where ChatGPT ads get tricky for analytics. Many sites already receive visitors from ChatGPT without paying for them: ChatGPT cites pages in its answers, and people click through. We covered how to spot that traffic in how to track AI and ChatGPT traffic to your website.
From 24 August, a European visitor can arrive from ChatGPT in two very different ways:
- Organic: your page was cited in an answer and the user clicked it.
- Paid: the user clicked your sponsored ad below an answer, and you paid for that click.
Both clicks start on chatgpt.com. If your analytics groups visits by referrer alone, both land in the same "ChatGPT" bucket, and your organic AI visibility suddenly looks better than it is, or your ad campaign looks worse, because part of its result is credited elsewhere. On top of that, ChatGPT already adds utm_source=chatgpt.com to many of the links it cites, as Search Engine Roundtable reported. So "has a UTM source of chatgpt" doesn't mean "is an ad" either.
How to separate paid ChatGPT clicks from organic AI referrals
The reliable fix is the oldest one in online marketing: tag your own ad links. If you control the destination URL of every ad, you control how that traffic is labelled.
| Traffic | How it arrives | How to identify it |
|---|---|---|
| Organic ChatGPT citation | Click on a link in an answer | Referrer chatgpt.com, no campaign tag, or ChatGPT's own utm_source=chatgpt.com |
| Paid ChatGPT ad | Click on a sponsored ad | Your own UTM tags, for example utm_medium=paid-ai and a campaign name |
| Other AI assistants | Perplexity, Claude, Gemini, Copilot | Their own referrer domains |
| Copied link, no referrer | Link pasted into a browser or app | Often lands in Direct |
A few practical rules:
- Use a medium that no organic source will ever produce.
utm_medium=cpcworks if you report on it consistently; a dedicated value such aspaid-aimakes ChatGPT ads easy to separate from Google Ads too. Keep the source stable (utm_source=chatgpt) and put the detail inutm_campaignandutm_content. Our guide to UTM parameters for campaign tracking covers naming conventions. - Tag every ad, every time. One untagged ad pollutes your organic numbers. Agree on the convention with your agency before 24 August, because at launch the agency, not you, may be building the ads.
- Consider dedicated landing pages. A landing page that is only reachable from ads gives you a second check: if visits to that page don't match tagged ad clicks, something is off.
- Compare against the ad platform. OpenAI gives advertisers aggregated views and clicks. Your analytics should show roughly the same number of tagged visits; a big gap usually means missing tags or redirects that strip parameters.
- Watch the consent side of conversion tracking. If you add an advertising pixel to your own site to report conversions back to an ad platform, that is advertising tracking on your visitors' devices, which in the EU generally needs consent. Your first-party analytics and an ad platform's conversion pixel are separate questions.
What to report
Once paid and organic are separate, report them separately. Organic AI referrals tell you whether AI assistants consider your content worth citing, which is a generative engine optimisation question. Paid ChatGPT clicks tell you whether a new ad channel earns back its cost. Mixing the two hides both answers.
Should you advertise in ChatGPT?
That depends on your category and budget, and nobody has European benchmarks yet. Digiday quotes agency people calling the platform "very much unproven" and building fast. What is clear from OpenAI's own description:
- It suits intent-driven products. With no personalisation in Europe, your ad appears because of what the user is asking. Products people research by asking questions are the natural fit.
- The audience is limited to Free and Go users. Paying ChatGPT subscribers on Plus or higher never see your ad.
- Reporting is aggregated. You get views and clicks, not user-level data. What happens after the click is up to your own analytics.
- Access is managed for now. Without a self-serve tool in Europe at launch, small advertisers may want to wait.
What this means for your analytics
ChatGPT is turning from a pure referral source into an ad channel, and your analytics needs to keep the two apart. ClearAnalytics already recognises AI referrals from ChatGPT, Perplexity, Claude, Gemini and others, and reports UTM campaigns alongside them, so tagged ad clicks and organic citations show up as separate sources. It does this without cookies and with data stored in the EU. See how AI referral tracking works on the LLM visibility page.