Quick Answer: ChatGPT ads are labelled units shown below a ChatGPT answer, bought through OpenAI Ads Manager on CPM or CPC bids. Ads Manager Beta suggests a starting bid of $3 to $5 per click and runs a relevance-weighted second-price auction. Only Free and Go users see them, and the ad never changes the answer itself.
ChatGPT Ads went from a February 2026 pilot in the United States to a $1 billion annualised run rate in under 200 days, per OpenAI's 31 August update. For a marketer the practical questions are narrower: where the unit sits, what it costs, and who sees it.
The question that matters most is whether paying changes what the model says about you. It does not, and that is why the paid unit and the earned citation now compete on the same screen.
Where do ChatGPT ads appear and who sees them
ChatGPT ads appear below the response, as a clearly labelled block with the advertiser name, favicon, headline, description, landing page and one image. OpenAI's ad basics describe exactly those six elements and nothing inside the answer text.
Only users on the Free and Go plans are eligible. Plus, Pro and every Business plan are ad-free, and accounts identified as under 18 never see ads. In practice that means a B2B buyer on a company Enterprise seat will never meet your ad, while a founder on a free account will.
At a glance
| Question | Answer as of September 2026 |
|---|---|
| Placement | Below the answer, labelled, separate from the response |
| Who sees ads | Free and Go plans only, 18 and over |
| Buying models | CPM (Views), CPC (Clicks), conversion-optimised with click billing |
| Suggested starting CPC | $3 to $5, per OpenAI Ads Manager guidance |
| Auction | Relevance-weighted second-price |
| Targeting | Context hints at ad-group level, geo, platform, custom audiences, product feeds |
| Measurement | Impressions, clicks, CTR, CPC, CPM, conversions, OpenAI Pixel, Conversions API, UTM parameters |
| Self-serve markets | US, 31 European markets, India, the Middle East and North Africa from 31 August 2026 |
Does the ad influence the answer
No. OpenAI's ads principles state that advertising does not influence the answers ChatGPT gives, and advertisers get no access to the conversation. The ad is selected on the context of the current chat and, where the user has personalisation on, signals from their wider ChatGPT use.
That separation is the fact that decides your strategy. A paid unit can sit under an answer that recommends a rival, and a brand with no ad budget can still be the one the answer names. How ChatGPT chooses which brands to cite is a different system from the one that sells the slot beneath it.
How much do ChatGPT ads cost
ChatGPT ads cost whatever the auction clears at, and OpenAI recommends a maximum bid of $3 to $5 per click to start. Billing is by impression for Views campaigns, by valid click for Clicks campaigns, and by valid click for conversion-optimised (oCPC) campaigns as well.
Bidding and budgets
Advertisers set a maximum bid at the ad-group level: a CPM ceiling for Reach campaigns or a CPC ceiling for Clicks campaigns. Ads Manager shows bid-strength guidance so you can see whether a bid is likely to win delivery or throttle it.
Daily budgets are averages, not caps. OpenAI's own FAQ states that spend can reach twice the daily figure in a single account-local day and seven times it in a Sunday to Saturday week, per the Ads Manager FAQ. A campaign-total budget limits the whole flight instead.
What OpenAI will not tell you yet
There are no published benchmarks for CTR or conversion rate by industry. The FAQ says so directly: as an early platform, results depend on objective, creative, landing page and relevance. Treat any agency deck quoting an "average ChatGPT ads CTR" as a single account's number, not a market figure.
How targeting works without keywords
ChatGPT ads are matched on conversation context, not exact-match keywords. At the ad-group level you supply context hints, which describe the topics, conversations or terms where your product is relevant, and the system uses them to guide matching rather than to trigger it.
Since the August expansion the platform also supports geo-targeting, platform targeting, custom audiences and product feeds, and CPC plus outcome-optimised bidding now account for most campaigns, according to OpenAI. The mental model is closer to a social feed than to a search campaign: you describe the buyer's situation and the system decides the moment.
Why context hints change your copy
A search ad is written against a query. A ChatGPT ad is shown after the user has explained a goal, constraints and what matters to them, so the headline has to answer a decision the person is already halfway through. Copy that restates a keyword performs like a banner; copy that resolves the trade-off the user just described performs like an answer.
How to measure ChatGPT ads next to your earned citations
Measure paid and earned on the same prompt set, because they now compete for the same reader. Ads Manager reports impressions, clicks, CTR, average CPC, average CPM and conversions, and it supports the OpenAI Pixel, the Conversions API and static UTM parameters on landing pages.
The three numbers to keep side by side
- Paid share of the prompt. For each buyer prompt you track, note whether your ad served under the answer during the flight.
- Earned share of the prompt. Whether the answer itself named or linked you, measured on the live ChatGPT surface rather than an API sample, since the two can differ.
- Referred sessions by source. Ad clicks carry your UTM tags; earned clicks arrive as
chatgpt.comreferrals, so tracking AI-referred sessions needs both columns.
When the ad serves but the answer names a rival, you are renting attention on a page that is recommending someone else. When the answer already names you, the ad is defending a slot you have won, and that is a much cheaper decision to justify.
How to read the paid-versus-earned table
Build a simple weekly table with one row per tracked prompt and four columns: ad served, brand named, brand linked, sessions. Unveilr's tracking reads the live ChatGPT surface for the "named" and "linked" columns on a fixed schedule, so a vanished mention shows up as a real regression rather than sampling noise.
The pattern to watch for is a prompt where paid impressions rise while earned mentions fall. It usually means a rival's page has overtaken yours as the source the model quotes, and the fix is on your page, not in your bid.
That is the job a managed engagement exists to do. Unveilr's proprietary agents track the prompt set, its IIT-graduate team writes and refreshes the pages that win the earned column, and engagements start at ₹65,000 a month, scoped on prompt and content volume.
It is the sole VC-backed AEO agency registered in India, funded by AJVC. The other AI-native managed agency is US-registered and starts at $900 a month on its public pricing page (verified 16 September 2026).
What ChatGPT ads mean for organic AI visibility
ChatGPT ads change the economics of the answer page, not the answer. The answer still comes from retrieval and the model's own knowledge, so the levers that earn a citation are unchanged: a page ChatGPT can fetch, a claim it can attribute, and the answer-first structure it already trusts.
The free-tier audience is the earned-citation audience
More than a billion weekly users keep ChatGPT free, and those free users are the only ones who see ads. They are also the users most likely to act on the answer's recommendation without opening a second tab. The paid unit sits beside the moment of decision; the citation sits inside it.
Where the budget argument lands
Paid buys a slot for the flight. Earned compounds, because the page that got cited this week is still there next week, and the AEO budget question is really about which of the two you want to own. Most teams that run both find the ad is the fastest way to learn which prompts convert, and the citation is the cheapest way to keep converting them.
What to do this quarter
Run a small Clicks campaign against the ten prompts you most want to win, with UTM tags and a conversion event. Then compare the cost of each paid conversion with the cost of the content that earns the citation on the same prompt, and move budget toward whichever one is cheaper per lead in month three, not month one.