ChatGPT Ads: What They Are and How Advertising in ChatGPT Works

ChatGPT Ads are paid placements shown inside ChatGPT, OpenAI’s AI assistant, sitting underneath an answer and clearly labelled as sponsored. They’re matched to the subject of the conversation rather than to a keyword someone typed, which makes them a different proposition from search advertising and is the reason Priority Pixels runs ChatGPT advertising for B2B marketing teams as a measured test of a new channel rather than an extension of an existing search account.

OpenAI began testing ads in the US in February 2026 and has since extended them to the UK, across Europe and to more than 40 countries worldwide. The product is still labelled a beta, so formats, targeting and policies change month by month and OpenAI continues to add new ways to buy and new ad formats on top of the original sponsored card.

Where ChatGPT Ads Appear and What They Contain

ChatGPT Ads sit below the end of a ChatGPT response, visually separated from the answer and marked as sponsored. One or more ad units can show beneath a single response, each carrying items from one advertiser or from several.

A standard ad is built from six elements set out in OpenAI’s guide to the basics of ChatGPT Ads, namely the advertiser name, a favicon or logo, a headline title, a short line of description copy, an image and the landing page the click goes to. Which ad appears is settled in a relevance weighted second price auction, so a high bid on its own won’t win a placement for an ad that’s a weak match for the conversation.

The format is closer to a compact sponsored card than a search text ad. With so little space on the card, the landing page carries more of the persuasion than it would on most paid search campaigns.

A ChatGPT Ads Example Described in Words

The simplest way to picture a ChatGPT ad is to follow a hypothetical conversation. Imagine an operations director at a facilities management firm asking ChatGPT to compare ways of scheduling engineers across several regional offices and ChatGPT replying with an answer that weighs up spreadsheets, packaged software and bespoke options.

Below that answer, set apart from it and labelled as sponsored, a card appears from a field service software vendor. It shows the vendor’s logo and name, a title such as “Engineer scheduling for FM teams”, a line about cutting travel time between jobs and a square product image. Clicking it opens a landing page about engineer scheduling instead of the vendor’s homepage, which is the approach OpenAI’s creative guidance recommends.

The example is invented, but the structure isn’t. Every element of the card is set by the advertiser and the answer above it is produced by ChatGPT with no input from any advertiser.

Who Sees Ads in ChatGPT and Who Doesn’t

Ads are shown to people using the Free and Go plans of ChatGPT. According to OpenAI’s help centre article on ads in ChatGPT, accounts on the paid plans don’t get ads and nobody identified as under 18 is shown them, based on account age information and OpenAI’s age prediction.

The split by plan decides the audience an advertiser can reach. The table below sets out each plan as OpenAI describes it.

For B2B advertisers this matters more than it first appears. A buyer on a company Business or Enterprise seat won’t see ads in that session, so the reachable audience is Free and Go users, which can include the same buyer researching on a personal account at another time.

How ChatGPT Matches Ads to the Conversation

ChatGPT Ads are matched to the context and intent of the current conversation, not to a keyword list. The ad system weighs what’s being discussed, the advertiser’s landing page, title and copy and the context hints set on the ad group and where a user has ad personalisation switched on it can draw on signals from their wider ChatGPT activity.

Context hints are the main lever advertisers have. They’re written descriptions of the conversations, topics or phrases an ad group is relevant to and OpenAI’s guidance on creating ad groups is clear that they aren’t exact match keywords or targeting rules, don’t guarantee delivery and can’t enforce geographic limits, schedules or exclusions.

That’s a real shift for anyone used to Google Ads management, where the search terms report shows which query triggered each click. In ChatGPT the advertiser describes a situation and the platform decides which conversations fit.

Priority Pixels writes context hints from a client’s real search term data and the questions their buyers ask, with one theme per ad group. Hints are changed one at a time so the effect of each change can be read.

Targeting Options Beyond the Conversation

Outside the conversation itself, targeting is limited to location, device and customer lists. Country targeting is set at campaign level according to OpenAI’s campaign setup guidance, with countries explicitly included instead of others excluded and campaigns can be limited to the Android app, Android web, a computer browser, the iOS app or iOS web.

Custom audiences built from customer lists can be used for inclusion, exclusion or bid adjustments. The custom audience guidance sets a minimum of around 25,000 matched users before a list can be used for inclusion or bid adjustments, with 100,000 or more recommended, which smaller B2B customer lists may not reach.

OpenAI’s help centre doesn’t document demographic targeting by age, gender or income, or job title targeting of the kind B2B advertisers know from LinkedIn Ads campaigns. That puts the weight on context hints and landing pages instead of on audience selection.

Measuring Results From ChatGPT Advertising

Conversions from ChatGPT Ads are measured with the OpenAI Pixel on the advertiser’s website, the Conversions API sending events from the server, or both. The conversion measurement documentation explains that the Pixel stores a click reference in a first party cookie and that reported conversions can include modelled figures where available.

Reporting covers impressions, clicks, spend, CTR, average CPC, average CPM and conversions, with attribution on a last touch basis. Conversions take 24 to 48 hours to appear according to OpenAI’s guide to measuring results and with no cross advertiser benchmarks yet, early results have to be judged against the account’s own baseline.

For UK advertisers the Pixel is a tracking technology like any other advertising tag. The ICO’s guidance on how the storage and access rules apply to online advertising says consent is needed for any storage and access technology used for advertising purposes, including measuring campaign effectiveness, so the Pixel belongs behind the consent banner.

Priority Pixels installs the Pixel and Conversions API before the first ad is submitted, with a single primary lead event and reports observed and modelled conversions separately. UTM tags keep paid ChatGPT clicks distinct from the organic visits GA4 now groups under its AI assistant channel group. Every account is visible on our reporting dashboard between monthly reports.

How ChatGPT Ads Are Bought

ChatGPT Ads are bought in an auction on impressions or clicks and campaigns optimised for conversions are still billed on clicks or impressions rather than per conversion. Buying started as CPM only, moved to a CPC option and a beta self serve Ads Manager on 5 May 2026 and now also runs through agency partners including Dentsu, Omnicom, Publicis and WPP and technology partners such as Criteo and StackAdapt.

The early test looked very different from today’s self serve platform. In January 2026, Adweek reported that OpenAI had confirmed a US$200,000 minimum commitment from select beta advertisers, a figure that described that early test and not the pricing that applies in the self serve platform now that the minimum has been removed.

Accounts run from campaign to ad group to ad and setting up advertising on ChatGPT involves business verification, a billing profile and crawler access before anything can run. Spend is paid directly to OpenAI and OpenAI hasn’t published average UK click or impression costs, so our guide to ChatGPT Ads costs covers bids, budgets and what a meaningful test needs in more detail.

Retail Giants Are Building Their Own Routes Into ChatGPT Ads

Amazon became the first major retail platform to plug its own advertising business into ChatGPT Ads. Amazon Ads and OpenAI launched a pilot on 10 September 2026 that lets select US advertisers buy ChatGPT ad inventory as a managed service through Amazon DSP, with Delta Vacations among the first named advertisers in the trial, as Digiday reported at the time.

The pilot is limited to the US and to a small group of advertisers for now, so it doesn’t change how a UK business buys ChatGPT Ads today. It does show that established ad platforms are now building bridges into ChatGPT instead of treating it as a closed system, which is worth watching if your organisation already buys programmatic media through a DSP.

OpenAI Is Testing Conversational Sponsored Agent Ads

OpenAI has started testing an ad format that turns a click into a conversation instead of a visit to a website. Reported by Digiday on 14 September 2026, the format lets an advertiser attach a branded AI agent to an ad, so when a user taps a call to action such as “Chat with us” they move into a branded chat inside ChatGPT where they can ask the agent questions before being handed off to the advertiser’s own website.

Wayfair confirmed to Digiday that it was trialling the Sponsored Agent pilot to explore how conversational commerce can help customers make decisions, with guardrails around product accuracy and service handoffs. OpenAI has also started rolling out AI assisted campaign creation tools and its first CRM integration, with Performance Marketing World reporting new HubSpot and Shopify integrations alongside the Sponsored Agents test. None of this is available through self serve Ads Manager yet and OpenAI’s own leadership has described the standard ad format beneath a ChatGPT answer as the “basic starting point” for the platform, so it’s a direction to watch and not an option most B2B advertisers can plan around today.

From a US Test to UK and European Rollout

ChatGPT Ads moved from announcement to international rollout in under eight months. The company began testing ads with logged in adult users on the Free and Go plans in the US on 9 February 2026, a few weeks after announcing the plan and the beta started as a focused pilot that ran from February to April.

UK users started seeing ads in early June. Digiday reported that OpenAI’s global head of ads announced the UK launch, with UK inventory sold through OpenAI’s own team at first before self serve buying opened to UK advertisers a couple of weeks later. Our guide to ChatGPT Ads in the UK covers what UK advertisers need to know about access, contracts and compliance.

Europe followed at the end of the summer. OpenAI’s ads business reached 31 European markets at the end of August, alongside India, the Middle East and North Africa, as part of the same push that took the product past the six month mark since the US pilot began.

Those figures describe the platform’s scale instead of results any single advertiser can expect. ChatGPT Ads is still described as a beta in the early stages of scaling, so it’s worth checking OpenAI’s own pages before every campaign build.

What Advertisers Can’t Do in ChatGPT

Advertisers can’t influence what ChatGPT says. Ads run on separate systems from the chat model and OpenAI’s published approach to advertising sets out principles covering answer independence and conversation privacy, so advertisers can’t shape, rank or change responses or see what people have been discussing.

What advertisers receive is aggregated performance data such as views and clicks. Chats, chat history, memories, names, email addresses, precise locations and IP addresses stay with OpenAI and an ad appearing in ChatGPT doesn’t mean OpenAI endorses the advertiser.

Organic visibility in AI answers is its own discipline, covered in our guide on optimising for ChatGPT, Gemini and Copilot. On the advertising side, OpenAI’s ad policies keep ads away from sensitive conversations such as mental health, politics and self harm, political advertising isn’t currently allowed and some categories are restricted or disallowed outright.

Eligibility for a particular business and offer comes down to the policy check and our guide to which businesses can advertise on ChatGPT sets out the categories and restrictions in detail. Priority Pixels checks each client’s category against OpenAI’s live ad policies before any work begins and records the policy version used.

Why ChatGPT Ads Matter for B2B Advertisers

ChatGPT Ads put a brand in front of people at the point they’re comparing options or asking what a service costs, inside a conversation where they’re already working through a decision. For B2B organisations with long sales cycles, that’s a different moment from a search query or a social feed.

The matching model suits B2B better than it might first seem. Context hints let an advertiser describe the situations their buyers are in, such as reviewing a supplier or preparing a business case, instead of guessing at the exact words they’ll type.

Trade press is already debating whether ChatGPT Ads deserve a line in next year’s media plan, which is exactly the question a measured test is built to answer rather than guess at. It’s a young platform and the buying routes and ad formats are still changing, so the sensible approach is a test with a success measure agreed before spend begins.

Our guide to ChatGPT Ads for B2B organisations looks at fit by sector and buying cycle and our guide to choosing a ChatGPT Ads agency covers the questions to ask before handing over an account. Priority Pixels has delivered paid media from the UK since 2016 and runs ChatGPT Ads programmes through six AI workflows, from eligibility and tracking to daily management and monthly audits, each reviewed by our PPC team. At the end of the read period we tell clients plainly whether the platform is working for them and recommend whether to continue, scale up or bring the test to an end.