Ads in ChatGPT (OpenAI Ads): how do they work in Europe?

Information checked: 3 September 2026 · about 14 minutes to read

Self-service OpenAI Ads Manager is now available to advertisers in supported markets across Europe. Ads in ChatGPT are paid modules shown below a ChatGPT answer, clearly labelled as sponsored and visually separated from content generated by the model. They are not part of the answer and do not constitute an OpenAI recommendation.

Three similar-sounding topics need to be separated from the outset. ChatGPT Ads means buying paid placement in ChatGPT. Creating ads with AI means using generative tools to produce copy or graphics. GEO, or optimising visibility in AI-generated answers, concerns organic presence. This article covers only the first area: how the new advertising channel works, what an ad looks like, who may see it, what advertisers pay for and when a test makes commercial sense.

Important: the platform is developing quickly. Country availability, features, policies and optimisation models may change. All facts and recommendations in this article reflect information confirmed on 3 September 2026. Tweetnij

# The essentials in 60 seconds

• Self-service OpenAI Ads Manager is available in 31 specified European markets, including Germany. Check the official country list again before planning a campaign.
• An ad appears below a ChatGPT answer, carries a sponsored label and remains visually separate from the answer.
• The module may contain the advertiser name, a favicon, headline, description, landing page and creative image.
• Ads may appear on Free and Go, and in logged-out experiences. They are not shown on Plus, Pro, Business, Enterprise or Edu, to accounts identified as under 18, in Temporary Chats or—during the current test—in ChatGPT Atlas. Free users may choose a no-cost Ads-Free option with lower limits and reduced tool access.
• Matching can draw on the intent and context of the current conversation, the ad copy, landing page, targeting settings and Context Hints supplied by the advertiser.
• Context Hints are not exact-match keywords and do not guarantee delivery alongside a particular conversation.
• Campaign objectives use CPM, CPC and oCPC models. With oCPC, the system optimises delivery for a conversion event, but the advertiser still pays for valid clicks rather than for the conversion itself.
• OpenAI's official $3–$5 starting suggestion applies to a maximum CPC bid. It is not an average click price for Germany, Europe or any other market, and it is not a promised result.
• OpenAI says advertisers do not receive conversations, chat history, memory or users' personal data. Reporting is aggregated and non-identifying.

# What are OpenAI Ads—and what are they not?

The official advertiser-facing product is called ChatGPT Ads, while campaigns are managed in OpenAI Ads Manager. Searchers also use terms such as OpenAI Ads, advertising in ChatGPT and ChatGPT advertising. In practice, they describe the same new paid channel for reaching people who use ChatGPT.

The most important difference from traditional ad inventory is the moment of contact. A user is not simply scrolling through content or entering a short query. They may describe a goal, constraints and selection criteria: planning a trip, comparing software, furnishing a home or trying to solve a problem. OpenAI presents ChatGPT Ads as a way to reach people while they explore options, compare them and make decisions.

That does not mean a brand can buy its way into the model's answer. According to the official FAQ about ads in ChatGPT, the advertising system is separate from the model that generates answers. Advertisers cannot influence the content, ranking or tone of a ChatGPT answer, and an ad impression is not an OpenAI recommendation or endorsement.

This distinction also prevents a common SEO mistake. A paid campaign in ChatGPT does not automatically increase organic brand mentions in answers. Conversely, strong organic visibility in AI systems does not replace a campaign when the goal is controlled delivery, a defined budget and click measurement. OpenAI Ads and GEO may complement each other, but they solve different jobs.

# What does an ad in ChatGPT look like?

An ad may appear below an answer and is visually separated from it. OpenAI's example shows a card marked as sponsored. According to the ChatGPT Ads format guide, an ad unit may include:

• the advertiser's name,
• a favicon or small brand icon,
• an ad headline,
• descriptive copy,
• a landing-page address,
• a creative image.

This is therefore not a link inserted into the answer or a hidden product placement. A user can see where model-generated content ends and the paid message begins. During testing, one or more units may appear below an answer, and a unit may present one or several offer items.

The format will continue to evolve, so a strategy should not be designed around a single screenshot. A more durable principle is a clear hierarchy: a concrete benefit in the headline, copy that fits the user's situation, a credible visual and a landing page that fulfils the ad's promise without surprises.

Users retain control over their advertising experience. They may hide or report an ad, check why it was shown and, in supported regions, manage personalisation settings. Turning personalisation off does not necessarily remove ads; they may still be matched to the context of the current conversation.

# Are OpenAI Ads available in Europe?

Yes. According to the current Ads Manager availability list, self-service access was available in Germany and 30 other specified European markets on 3 September 2026. This means an eligible advertiser can begin using Ads Manager directly rather than being required to enter through a sales team or technology partner. OpenAI nevertheless stresses that account creation and a country's advertising availability are separate questions.

This is a recent change. OpenAI announced the European expansion in August and updated its communication on 31 August to confirm self-service in 31 European markets. Older articles may therefore still say that European advertisers need to wait, join a list or work only through a partner. As of the date of this update, those broad claims are out of date for the listed markets.

Access to the advertiser tool does not guarantee a particular volume of impressions in Germany or elsewhere in Europe. Delivery depends on the number of eligible users, conversation contexts, targeting, relevance, bids, budget and the auction. Available means that a campaign can be run in a supported country; it is not a promise of scale or performance.

# Who can see an ad?

Ads may be displayed to users on the Free and Go plans in supported delivery regions. They may also appear in logged-out experiences; OpenAI says those ads are appropriate for all audiences, while mature-content ads are limited to eligible logged-in users. Ads are not shown on Plus, Pro, Business, Enterprise or Edu plans, to accounts identified as belonging to people under 18, or in Temporary Chats. During the current test, ads also do not appear in the ChatGPT Atlas browser.

Free-plan users in supported regions may choose a no-cost Ads-Free option, but with lower message limits and reduced access to some tools, such as image generation or deep research. Availability varies by region and over time. Turning off ad personalisation is not the same as Ads-Free and does not remove contextual ads.

That distinction matters when forecasting. The entire ChatGPT user base is not an addressable advertising audience. A company selling primarily high-value tools to teams on business plans should not assume that ChatGPT's overall popularity translates directly into an equally large campaign reach.

At the same time, a free-plan user may be at a very specific point in a decision. A longer conversation can reveal not only a product category, but also the use case, priorities and constraints. The channel's potential lies in matching an ad to that moment—provided the offer and landing page genuinely help the person continue the decision-making process.

# How ad matching and Context Hints work

The selection system considers predicted relevance and expected outcome. Signals described by OpenAI include:

• the context and intent of the current conversation,
• the ad's landing page,
• its headline and description,
• Context Hints supplied by the advertiser,
• targeting settings,
• selected signals from broader ChatGPT use only where ad personalisation is available and enabled.

Context Hints are contextual cues configured at ad-group level. They describe conversations, topics or words for which a product may be useful. A strong hint is more than a list of broad phrases. It captures the situation, problem and intent—for example, the need to choose a project-management tool for a small team rather than the single word software.

Context Hints should not be confused with exact-match keywords. OpenAI explicitly says they help guide matching but do not guarantee delivery in a specific conversation. An advertiser does not own a particular question, and the appearance of a phrase in a conversation does not determine whether an ad will be shown.

This changes how a campaign should be prepared. The complete path needs to be coherent: the situation described by a hint, the headline's promise, the argument in the description, the image and the landing-page content. If each part tells a different story, the system has less basis for assessing relevance and the user encounters a disconnected journey after clicking.

# What the European personalisation model changes

According to the current FAQ, personalised ads are not initially available in the European Economic Area or Switzerland. Advertisers in Germany and other EEA markets should therefore not build a plan on the assumption that earlier conversations or a user's memory will be used to personalise an ad.

This does not make advertising context-free. OpenAI says that when personalisation is off, matching may still use the current thread and basic information such as general location or language. Clear Context Hints, an accurately described product, legible creative and a relevant landing page consequently become even more important for European campaigns.

# ChatGPT Ads compared with Google Ads, Meta Ads and GEO

The useful comparison is about the type of signal and moment of contact, not a blanket verdict about which channel is better.

# ChatGPT Ads and Google Ads

In a conventional search campaign, the starting point is a query and a set of controls around keywords. In ChatGPT, a user may explain a goal over several sentences, refine requirements and continue the conversation. OpenAI Ads attempts to interpret the meaning of that current context, while Context Hints only support matching; they do not behave like guaranteed exact match.

Google Ads remains a mature channel with extensive benchmarks and a large pool of search intent. FAEO's overview of Google Ads campaign management explains the operating scope of that established search channel. ChatGPT Ads is a newer environment in which reaching someone during exploration and comparison may be particularly valuable. A sensible test should not automatically replace search campaigns. It should show whether conversational context brings additional high-quality traffic.

# ChatGPT Ads and Meta Ads

On social platforms, an ad usually competes for attention in a feed. In ChatGPT, the module appears beside a conversation that the user started to complete a task or make a decision. The audience may therefore be more focused on a problem, but the format also places a higher demand on usefulness and message relevance.

Creative should not simply be copied from one environment to the other. A broad brand slogan that stops a social scroll may fail to answer the precise need expressed in a conversation. In ChatGPT, it is worth testing genuinely different variants that explain who an offer is for, which problem it solves and when it becomes useful.

# ChatGPT Ads and organic visibility or GEO

GEO is intended to help a brand become a credible, citable source in generative-system answers. Results depend on factors such as information quality, crawler access, entity consistency and how easy the content is to cite. GEO does not, however, provide a panel for setting a daily budget or guaranteed delivery.

OpenAI Ads is a paid channel: it enables an advertiser to compete in an auction, direct traffic to a landing page and measure ad outcomes. It does not buy an organic recommendation and should never be presented as doing so. A sound strategy combines reliable content, technical site accessibility and a paid experiment instead of treating advertising as a shortcut to authority.

# How much do ads in ChatGPT cost?

There is not yet a stable, public benchmark for average CPC, CPM or customer-acquisition cost in Germany or across European industries. The market is young, self-service access is recent and results will vary by category, targeting, relevance, season, landing-page quality and objective. Publishing a single European OpenAI Ads rate today would create false precision.

Official materials describe three objectives:

• Reach / CPM — optimisation for reach and impressions, charged per 1,000 impressions,
• Clicks / CPC — optimisation for traffic, charged for a valid click,
• Conversions / oCPC — optimisation of clicks for a selected conversion event, while billing still occurs for a valid click.

For an account billed in euros, the current campaign-creation documentation states a minimum daily-budget setting of €15. A daily budget is an average spending target over seven days, not a hard cap for each day: spend may reach up to twice the setting on one day, while spend over the applicable seven-day period will not exceed seven times it, subject to proration after budget changes. This is an account-setting threshold, not a recommended test budget, performance benchmark or indication that €15 will generate a meaningful sample.

For CPC campaigns, OpenAI recommends starting with a maximum bid of $3–$5 per click. This is guidance for max CPC. It is not a price list, an average actual CPC, a minimum budget or a forecast for Germany, Europe or any other country. Ads Manager may also show bid strength and indicate whether the bid is restricting delivery.

With oCPC, the system uses predicted conversion likelihood to determine a per-click auction bid. The Bid Cap is the maximum amount an advertiser is willing to bid for a conversion, but it is neither the price charged for a conversion nor a guaranteed CPA. The official oCPC documentation states that advertisers still pay for valid clicks and that actual CPC is auction-determined. OpenAI does not currently publish a recommended oCPC bid.

Ad selection uses a relevance-weighted second-price auction. In simple terms, the highest bid alone may not be enough if the ad poorly fits the context. Commercially, the combination of cost, traffic quality and post-click outcome matters more than a single bid figure.

# How to measure performance without false certainty

Ads Manager Beta reporting includes impressions, clicks, spend, CTR, average CPC, average CPM and conversions. OpenAI Pixel and Conversions API can measure actions on the advertiser's site, while UTM parameters added to the destination URL remain present after a click and can be analysed in an existing analytics tool.

CTR alone does not reveal whether the channel acquires valuable customers. Evaluate the whole path:

1. Delivery: does the campaign generate enough impressions in the selected market?
2. Response: do distinct messages earn clicks without manipulative promises?
3. Visit quality: do users engage with the site and reach the relevant offer?
4. Outcome: are leads, registrations or purchases aligned with the objective?
5. Economics: do conversion cost and customer value support scaling?

At launch, label individual campaigns and variants in UTM parameters, define a conversion unambiguously and compare lead quality as well as volume. With a small sample, a single conversion can transform the apparent CPA, so the first few days of a test should not be presented as a market benchmark.

oCPC requires reliable conversion tracking and a supported standard conversion event. A very rare event gives the system little signal. If tracking is incomplete or a company cannot define the meaningful action after a click, it is better to fix analytics first than to expect automated optimisation to solve the problem.

# When to test—and when to wait

# A test makes sense when

• the offer addresses a specific need that a user would naturally describe in a conversation,
• the landing page quickly fulfils the ad's promise and works well on mobile,
• the business can measure at least one valuable micro- or macro-conversion,
• margin and customer value allow a test with uncertain CPC,
• the team can prepare several genuinely distinct messages and Context Hints,
• the product category complies with the current advertising policy,
• the result will be judged after a meaningful amount of data rather than a handful of clicks.

The strongest candidate does not have to be the largest advertiser. A specialised offer with a clear use case may have an advantage: software for a specific process, a local service, a product that supports a hobby or a solution useful while someone compares options.

# Wait or complete the groundwork when

• the company lacks reliable conversion tracking or consistent UTM parameters,
• the landing page is generic, slow or fails to match the ad promise,
• the team expects guaranteed delivery for a supposedly “purchased” question,
• the product is in a restricted category and eligibility has not been confirmed,
• the budget would generate only a few clicks from which no credible conclusion can be drawn,
• the only objective is to influence the organic ChatGPT answer,
• the company requires a guaranteed local benchmark before making any decision.

Waiting does not have to mean doing nothing. A business can prepare the landing page, strengthen measurement, map conversation intent, review policies and define test criteria. That turns a campaign launch into a controlled experiment rather than a bet on the novelty of the channel.

# Privacy, answer independence and brand safety

Conversation context helps match an ad, but it is not passed to the advertiser. OpenAI says it does not share users' conversations, chat history, memory, name, email address, precise location, IP address or sensitive information with advertisers. Advertisers receive aggregated, non-identifying performance data such as total impressions or clicks.

The distinction is important: the system may internally use the current thread to assess relevance, but the brand does not see an individual person's question. If a user deliberately starts a direct conversation with an advertiser through an ad feature, the advertiser sees only the messages that the user sends in that mode.

Personalised ads are not initially available in the EEA or Switzerland. Users in supported regions are also intended to receive controls such as the ability to hide or report an ad and manage advertising data.

A separate layer is the OpenAI advertising policy. The system is intended to prevent ads from appearing around sensitive conversations, including personal health, mental health and politics, as well as brand-unsafe content. In European and other markets outside the US, ads for financial and health services are generally prohibited, while ads for legal services are currently prohibited; limited exceptions may apply to general tools or educational materials that do not provide those services or make regulated claims. Political advertising remains prohibited.

Brand safety is not solely a platform filter. The advertiser remains responsible for truthful claims, category compliance, landing-page quality and its use of data after the click. Check the latest policies before every test because they may change as the product expands.

# What is confirmed—and what still needs testing?

As of 3 September 2026, official sources confirm the below-answer ad format, self-service access in 31 specified European markets, the core matching signals, three campaign objectives, billing mechanics, main reporting metrics and privacy principles. That is enough information to prepare a controlled pilot.

What remains unconfirmed is equally important: stable country-by-country industry benchmarks, the typical volume attainable in each segment, average CPA and any long-term advantage over mature channels. Those answers cannot be inferred honestly from a global bid suggestion or a single case study. They must come from a company's own properly measured tests.

The practical conclusion is straightforward. ChatGPT Ads should not be ignored, but they should not be promoted with promises unsupported by data. For a business with a suitable offer, reliable conversion measurement and an experimental budget, this is a real new channel. For a company without a prepared funnel, it will simply be another place to buy traffic without confidence in the outcome.

# How to prepare for a first test

At this stage, define six points without turning this article into a technical account-setup guide:

1. which problem the offer solves and at which point in the decision process it is relevant,
2. which products or services comply with advertising policies,
3. where the ad leads and whether the landing page matches its message,
4. which event represents a valuable outcome,
5. how much data is required before a pause-or-scale decision,
6. which channels provide the comparison for acquisition cost and quality.

A separate follow-up guide will cover creating an OpenAI Ads Manager account, configuring billing and measurement, and launching a first campaign. Keeping the topics separate lets this article answer how the channel works and whether it is worth testing; the next article will explain how to operate it.

If you would like to assess the channel's potential for your business, improve your measurement setup and define a pilot before that guide is published, contact FAEO. We start with the offer, unit economics and audience intent before recommending a budget or campaign model.

# Sources

• OpenAI Help Center: Ads in ChatGPT—principles, privacy and user controls
• OpenAI Help Center: Ads in ChatGPT—format, matching, pricing and measurement
• OpenAI Help Center: current Ads Manager availability by country
• OpenAI: ChatGPT Ads expands across Europe
• OpenAI Help Center: CPM, CPC and oCPC campaign objectives
• OpenAI Help Center: conversion-optimised oCPC campaigns
• OpenAI: advertising policies