Short answer: Yes, you can advertise on ChatGPT. OpenAI opened a self-serve advertising platform in May 2026. The minimum is $25 per day per campaign, you pay per click rather than per impression, and you target by geography and by conversation topic. There is no keyword bidding, which is the single biggest thing to understand before you start. Ads only reach users on the Free and Go plans, and buying ads does not make ChatGPT recommend you.
Platform details current as of September 2026. This product has changed several times since launch, so verify specifics against OpenAI’s documentation before you commit budget.
What ChatGPT Ads are
OpenAI started testing ads in ChatGPT in early 2026, initially in the United States on the Free and Go tiers. Through the spring it expanded to the United Kingdom, Mexico, Brazil, Japan, and South Korea. In May it opened a self-serve Ads Manager, which was the moment the platform stopped being reserved for large brands going through agency holding companies.
In an announcement dated 31 August 2026, OpenAI said ChatGPT Ads had reached a one billion dollar annualised revenue run rate, was being used by tens of thousands of advertisers, and was available in more than forty countries, with self-service access rolling out across India, Europe, the Middle East, and North Africa.
That trajectory matters for one practical reason. Advertising platforms that reach that scale that quickly tend to keep changing, and this one has. Anything you read about ChatGPT Ads that is not dated should be treated as unreliable, including this post in six months.
Who actually sees the ads
This is the first thing to check and the thing most likely to disqualify you.
Ads appear only for users on the Free and Go plans. According to OpenAI’s help documentation, Plus, Pro, Business, Enterprise, and Edu accounts do not have ads. OpenAI also states it does not show ads to accounts identified as belonging to users under 18.
If you sell to software engineers, agency people, consultants, or anyone else whose job involves heavy AI use, a large share of your market is on a paid tier and is therefore unreachable here. If you sell to a broad consumer base, the addressable audience is much larger. Neither situation is good or bad, but establishing which one you are in should happen before anyone builds a campaign.
What it costs
The minimum daily campaign budget is $25. Billing runs on a maximum cost-per-click model, so you pay when someone clicks, you set the ceiling you are willing to pay, and impressions cost nothing.
We are not going to publish a cost-per-click benchmark, because any number we quoted would be an average across categories and geographies that behave nothing alike, and it would be out of date quickly. What we can tell you is that the arithmetic is the same as any other paid channel:
Cost per lead = cost per click ÷ conversion rate
Cost per customer = cost per lead ÷ close rate
Run that with your own numbers before you spend. If a customer is worth $3,000 to you, a wide range of click prices works. If a customer is worth $200, the tolerance is narrow and you will know quickly. We went through this calculation in more detail in our post on what Google Ads actually cost, and the method transfers directly.
The removal of the large fixed minimum is the genuinely notable part. During the invite-only pilot, reported entry requirements were in the tens of thousands of dollars. At $25 a day, a single-location business can run a real test for the cost of a modest print ad.
The big difference: there is no keyword bidding
If you take one thing from this post, take this.
Google Ads is built on a keyword auction. You decide which searches you want, you bid on them, and your skill shows up in keyword selection, match types, and negative keyword discipline. None of that exists here.
ChatGPT Ads matches ads to the topic of a conversation, combined with geographic targeting. You choose where, and you influence the contextual alignment, but you do not choose the phrase. That means an experienced Google Ads manager cannot simply port their playbook across. The levers that remain are campaign structure, creative quality, geography, and landing page relevance.
Practically, this cuts both ways. You lose precision, which is a real loss. You also lose the need to guess every phrasing a person might use, because the system is matching intent at the topic level rather than the string level. Whether that trade favours you depends on how well-defined your category language is.
What the platform cannot do yet
Worth stating plainly, because most coverage of this platform omits it.
- No keyword bidding. Covered above.
- No demographic targeting. No age or gender selection.
- No detailed interest targeting. Nothing resembling Meta’s interest graph.
- No behavioural retargeting. You cannot build an audience of previous site visitors and pursue them.
- No cross-platform profiles. Targeting works without cookies or personal data, which is a privacy positive and a precision negative.
- Limited reporting. Advertisers receive aggregated figures such as views and clicks. OpenAI states advertisers do not have access to chats, chat history, memories, or personal details.
If your paid programme currently runs on retargeting warm audiences, that capability is simply absent. Plan around its absence rather than assuming it arrives next quarter.
How to set up a first campaign
Seven steps, in the order that avoids wasted work.
- Confirm your category is eligible. Check OpenAI’s current advertising policies against your industry first. Some verticals, notably healthcare and legal services, have appeared in both limited and restricted lists as policy evolved. Establishing this before you build creative saves a rejection cycle.
- Create an advertiser account. You will need a business email, company details, a tax ID or EIN, a billing method, and a landing page. The business must be registered in an available country. OpenAI directs interested businesses to openai.com/advertisers.
- Set up conversion tracking before launch. Create a conversion data source for web, iOS, or Android and confirm it fires. Launching without it means judging a brand new channel on clicks, which tells you almost nothing.
- Define geography. Country, region, DMA, or postal code, with exclusions where useful. Narrow it to the area you genuinely serve. This is the single easiest place to waste money.
- Configure contextual targeting. Align the campaign to the conversation topics your buyers actually raise while researching. This is the nearest thing to keyword strategy available.
- Build the ad. Business name, image, headline, description. Write for someone mid-research, not someone ready to buy. Reusing a search ad written for high commercial intent is the most common creative mistake here.
- Launch small with a defined threshold. Agree your acceptable cost per lead before you launch, then judge against it. New channels get either over-funded on excitement or killed on impatience, and a pre-agreed number prevents both.
Point three deserves emphasis. A purpose-built landing page matters more here than on most platforms, because the traffic arrives colder. Someone researching a category is not someone who typed your service plus your city into a search bar.
Does buying ads make ChatGPT recommend you?
No, and this is the most commercially important thing on this page.
OpenAI states that ads do not influence ChatGPT’s answers, and that sponsored units are clearly labelled and visually separated from the organic response. Appearing beside an answer and being named inside one are two completely different outcomes.
Being cited in the answer itself comes from entity clarity, structured data, content a model can parse, and corroborating mentions across the web. That is AI SEO, along with its narrower relatives answer engine optimisation and generative engine optimisation. It compounds, it feeds traditional search at the same time, and it cannot be bought. We break down the distinctions in GEO vs AEO vs SEO.
If anyone selling you ChatGPT Ads implies that spend improves how the model describes your business, that is worth pushing back on hard.
ChatGPT Ads compared to Google Ads and Meta
Google Ads catches people at the moment of commercial intent. Someone types what they want. Keyword targeting is mature, conversion data is rich, retargeting exists. It remains the best channel for capturing demand that already exists. See Google Ads.
Meta ads interrupt rather than answer. Strong at demand creation and audience targeting, weaker at catching someone who needs a service today. We compared the two in Google Ads vs Facebook Ads.
ChatGPT Ads sits earlier than Google and differently from Meta. It reaches people in the research and comparison phase, in a context where they are asking open-ended questions rather than typing two words. Less targeting control, earlier in the journey, and a genuinely new inventory pool.
The reasonable position is additive rather than substitutive. Keep the channel producing measurable leads. Fund a contained test on the new one.
Who should test it, and who should wait
Worth testing if your buyers research before deciding, if your category involves comparison, if you have a longer decision window, or if you are already visible in traditional search and want to reach people earlier.
Worth waiting if you sell urgent-need services where people open a search engine rather than a chat window, if your buyers are mostly on paid ChatGPT tiers, if your programme depends on retargeting, or if you do not currently have working conversion tracking. That last one is not a platform problem, but it makes any test unreadable.
The summary
ChatGPT Ads is a real, live advertising platform with a low enough entry point that a small business can test it honestly. The two things to internalise before you start are that there is no keyword bidding, and that only Free and Go tier users will ever see your ad.
Do the cost-per-lead arithmetic first. Get conversion tracking working before you spend. Set a threshold and judge against it. And keep the organic AI search work running separately, because that is the part that decides whether the model recommends you, and no amount of ad spend changes it.
MPH Marketing Solutions is a full-service agency in Berkley, Michigan, working with businesses across Metro Detroit and Michigan. If you want an honest read on whether this platform can reach your buyers before you budget for it, see our ChatGPT Ads page or get in touch. Sometimes the answer is that your audience is not on the free tier, and that is worth knowing in advance.
