Platform information: ChatGPT Ads are developing and platform requirements can change. This article was last reviewed on 11 September 2026. Shays Media rechecks material platform information before relying on it for client decisions.
What ChatGPT Ads actually are
A ChatGPT Ad is a paid placement that can appear underneath a ChatGPT response. It has an advertiser name, a small logo or favicon, a title, some ad copy, a landing page link and an image. That's the format, at the user-facing level: a clearly labelled sponsored unit shown beneath the response. The selection and delivery system behind it considers multiple relevance and outcome signals, which we come back to below.
What's different from most digital advertising isn't the ad unit itself. It's where that ad unit sits: at the end of a conversation, rather than next to a list of search results or in a social feed. The person has just asked ChatGPT something, read an answer, and is now looking at a suggestion related to what they were discussing.
That's worth pausing on, because it's easy to assume ChatGPT Ads work like search ads with a different interface. They don't, and the rest of this guide is really about that distinction.
Where ads appear and what's in one
Ads appear below ChatGPT's response — not inside it, and not instead of it. The response itself is generated normally; the ad is a separate, clearly labelled sponsored element beneath the response.
An ad unit currently includes:
- the advertiser's name and logo or favicon
- a title
- ad copy
- a landing page
- an image asset
There's no interruption of the answer and no blending of sponsored content into the model's own text. If you've used search advertising, think of the ad as visually adjacent to the answer rather than embedded in it.
Who sees them — and who doesn't
Ads can appear to users on ChatGPT's Free and Go plans. OpenAI's own consumer FAQ states that people on Plus, Pro, Business, Enterprise and Edu plans are not shown ads. Accounts identified as belonging to users under 18 don't receive ads either.
Ads Manager itself is a beta product, and OpenAI has been explicit that it continues to evolve. None of the plan or eligibility detail above should be treated as fixed — it's worth rechecking against current OpenAI documentation before you rely on it for planning, and definitely before publication of anything built on this article.
How ad selection actually works (and how it doesn't)
This is the part most likely to get misrepresented, so it's worth being precise.
OpenAI's own description is that ad selection and delivery consider multiple signals: the current conversation's context and intent, the advertiser's landing page, title and copy, the context hints the advertiser has written, and — where the user has ads personalisation enabled — a selected set of broader ChatGPT signals.
What that description does not say is that ChatGPT Ads work by matching exact keywords, the way search advertising has trained most marketers to think. There's no keyword list you bid on, and no guarantee that your ad appears whenever a particular word or topic comes up in a conversation. If someone tells you "your ad will show whenever a user mentions X," that's a simplification that goes further than OpenAI's own documentation supports.
It's also worth being direct about something else: ads do not influence ChatGPT's own response. The system generates its answer, and a separate ad may or may not appear underneath it. Advertising doesn't shape what the model says.
Campaigns, ad groups and ads — the structure
Ads Manager currently organises everything in a three-level hierarchy:
Campaign → Ad Group → Ad
The campaign is where the objective, overall budget and targeting sit. Ad groups sit inside a campaign and organise things by theme or intent — this is also where context hints live. Ads are the individual creative units — the actual title, copy, image and landing page a user might see — and they sit inside an ad group.
In practice, this means a single campaign can cover more than one distinct customer need, with meaningfully different needs split into separate ad groups. Each ad group has its own context hints; Shays Media may also recommend separate landing pages where the offers or customer situations differ enough to justify them. Cramming several unrelated offers into one ad group tends to produce vague, unhelpful hints — more on that below.
Context hints, briefly
Context hints sit at the ad group level. They're short, natural-language descriptions of the situations, needs or topics where that ad group might reasonably be relevant — written as phrases a person might actually think or say, not as a list of search terms.
A useful hint describes what the product or service does, who it helps, or when it's useful. It should describe a genuine use case, not a broad audience label bolted on for coverage.
Two things worth stating plainly, because they get confused constantly: a context hint is not an exact-match control, and it is not an audience-targeting rule. Writing one doesn't guarantee your ad will appear for that scenario, and it isn't a promise that only people in that scenario will see it. It's an input the system uses alongside everything else.
This deserves its own article, because context hints are one of the signals OpenAI uses to help determine when an ad group may be relevant — getting them right is worth the attention it takes, even though it's one input among several rather than something Shays Media has an independent, proven causal performance effect to report on. We've covered context hints in more depth in ChatGPT Ads Context Hints: What They Are and How They Work.
How buying and objectives currently work
At a high level, current OpenAI documentation supports more than one way to buy:
- Impression-based (CPM) campaigns, where you pay based on how many times the ad is shown.
- Click-based (CPC) campaigns, where you pay based on clicks.
- Conversion-optimised campaigns, which optimise delivery toward a defined conversion event rather than clicks or impressions alone. These require conversion tracking to be set up and a supported standard conversion event selected. Two billing options currently exist: oCPC (conversion optimisation with click billing) and oCPM (conversion optimisation with impression billing), which OpenAI currently documents as a beta feature with access expanding — worth checking directly before assuming it's available on a given account.
It's a mistake to describe CPC as the only route to a performance-based outcome on this platform, and an equally common one to assume every account has every option switched on from day one. Current OpenAI guidance also notes that the objective and billing model, and the conversion event selected, have restrictions on being changed once a campaign is created — which is a good reason to get the setup decision right the first time rather than treating it as something to adjust freely later. That detail should be rechecked immediately before publishing anything more specific, because it's exactly the kind of implementation rule that shifts as the beta develops.
Why conversational intent is commercially interesting
The reason this format is worth a serious look — not the reason to assume it will work — is the context a conversation can contain. Someone asking ChatGPT to help them compare options, plan a purchase, or work through a decision is doing something different from typing a short query into a search box: a conversation can contain considerably more stated context than a short search query — what they're trying to achieve, what they've already ruled out, what matters to them.
That's genuinely interesting. It's also not, by itself, a reason to advertise. Interesting context doesn't override weak offer economics, a landing page that doesn't match what was promised, or a business that can't yet measure whether a click became a sale. The conversational environment is one input into a suitability decision, not a replacement for the rest of it.
What ChatGPT Ads does not guarantee
Worth stating clearly, because a beta platform with a genuinely new placement invites overclaiming:
- It doesn't guarantee your ad appears for any specific word, topic, conversation or user.
- Context hints don't function as delivery guarantees.
- It doesn't guarantee any particular volume, cost, or conversion outcome.
- It doesn't guarantee approval — OpenAI retains control over advertiser eligibility, campaign acceptance, and delivery.
- It doesn't guarantee that a business currently eligible today will remain eligible under future policy.
None of that is unusual for advertising. It's just worth saying directly rather than leaving it implied.
What to check before you spend anything
Before committing budget, it's worth having honest answers to:
- Eligibility. Is your business and product type currently permitted under OpenAI's Advertising Policies? This changes, so check the live policy rather than relying on what was true when someone last looked.
- Offer economics. Does the maths of your offer support paid acquisition at all, before you even test a new channel?
- Landing page. Does the destination match what the ad promises, and is it reachable and crawlable?
- Measurement. Can you actually tell whether a click became a lead or a sale? If not, any cost figure you generate will be a number without a meaning attached to it.
- Willingness to stop. Are you testing, or are you expecting a guaranteed result? Those are different starting positions, and they lead to different decisions three or four weeks in.
If those answers are shaky, that's useful information on its own — it's better to know before spending than after.
OpenAI requirement vs Shays Media judgement
Everything in this guide splits into two kinds of statement, and it's worth keeping them separate. OpenAI defines the mechanics and eligibility — what an ad unit contains, how selection works, what campaigns and ad groups are, what's permitted under its policies. Shays Media's job is separate: judging whether the format makes commercial sense for a specific business. Neither one substitutes for the other, and a guide like this can only really help with the first.
Where this fits into a decision
ChatGPT Ads is a new placement with a genuinely different environment from search or social. That's a reason to understand it properly, not a reason to assume it belongs in every media plan. The mechanics in this guide are the starting point; whether it makes commercial sense for a specific business is a separate question, and it's the one that actually matters.
Check Your Fit — the Fit Check looks at eligibility, offer economics, landing page and measurement readiness before anything is switched on. Prefer to talk it through first? Contact us via the Contact page.
A few questions people ask at this stage
Do ChatGPT Ads use keywords? Not in the way search advertising does. There's no keyword list to bid on. Ad selection considers conversation context and intent, your landing page, copy and context hints together — context hints describe situations and needs, not exact-match terms.
Can you guarantee my ad appears for a particular conversation? No — and any answer other than that isn't accurate. Context hints and the other signals influence relevance; none of them are a delivery guarantee, and OpenAI's own documentation is explicit about that.
Is Shays Media part of OpenAI? No. Shays Media is an independent advertising service and is not affiliated with, endorsed by or an official partner of OpenAI.