For years, paid acquisition has meant two big choices: catch people while they search on Google, or interrupt them while they scroll on Meta. ChatGPT Ads add a third place where buyers now spend time, and it works differently from both. People ask ChatGPT long, specific questions ("which carpet cleaner handles pet stains in a rental flat?"), and OpenAI now shows paid units next to some of those conversations.
Clients have started asking me whether they should be there. This guide explains what ChatGPT Ads actually are, how ads get matched to conversations, which campaign objectives exist, and who I think should test them now. The platform is new and changes often, so I flag where things are still moving and where you should check OpenAI's current documentation yourself.
What are ChatGPT Ads?
ChatGPT Ads are paid placements that OpenAI shows inside the ChatGPT app, below an assistant response, when an advertiser's offer is relevant to the conversation. They are labelled as sponsored and kept visually separate from the answer itself.
OpenAI began testing ads in the US in early 2026 and then opened a self-serve Ads Manager (still labelled beta at the time of writing) where businesses can sign up, set budgets and bids, upload ads and read reports. Availability has expanded country by country since then. OpenAI's own announcements have reported a steadily growing list of countries, so check OpenAI's country availability page before you plan a launch outside the US.
A few rules from OpenAI's help centre are worth knowing up front:
- Ads appear for users on the Free and Go plans. Plus, Pro and the business tiers stay ad-free.
- Accounts identified as belonging to people under 18 do not see ads.
- Users can turn off ad personalisation, hide or report ads, and ask ChatGPT about an ad they see.
Where do ChatGPT Ads appear?
ChatGPT Ads appear below a ChatGPT response, as one or more ad units, and never inside the answer text. OpenAI states that the ads system runs alongside answer generation and that advertisers cannot shape, rank or change what ChatGPT says.
That matters for expectations. Buying ads does not make ChatGPT recommend your brand in its answer. Organic mentions inside answers are a separate topic (often called generative engine optimisation), and paying for ads does not buy them.
How does ChatGPT Ads targeting work?
ChatGPT Ads are matched mainly by conversation context rather than by keywords you bid on. OpenAI describes matching advertiser ads to the topic of the current chat, the user's past chats and their past interactions with ads, unless the user has turned personalisation off.
OpenAI has also published a technical write-up of its ad stack: conversations are summarised into short text, turned into semantic embeddings, matched to candidate ads, filtered for eligibility and quality, and then ranked. In plain terms, the system tries to work out what the person needs and finds ads that fit.
What you control as an advertiser is narrower than in Google Ads:
| Control | What it means in practice |
|---|---|
| Country targeting | Choose which supported countries a campaign runs in |
| US location targeting | State, designated market area (DMA) and ZIP code options in the US |
| Ad creative | Your headline, description, image and landing page tell the system what you offer |
| Objective and bid | How you pay and what delivery optimises for |
| Budget and dates | Daily or lifetime limits and flight dates |
There is no keyword list to manage and no search terms report in the Google sense. Your creative and landing page do much of the work that keywords do in search.
What do advertisers get to see?
Advertisers receive aggregated performance data such as impressions and clicks, not anyone's conversations. OpenAI says advertisers never receive users' chats, memories, names, emails, precise locations or IP addresses. Your own measurement (pixel, Conversions API, analytics) is where conversion detail comes from.
What campaign objectives can you choose?
ChatGPT Ads campaigns currently use one of three objectives: CPM, CPC and oCPC. Many guides describe these as Reach, Clicks and Conversions, which is a fair way to think about them.
- CPM (cost per thousand impressions). You pay for visibility. Useful for awareness or for a new category where people do not yet click much.
- CPC (cost per click). You pay for clicks to your site. The simplest option for a first test.
- oCPC (optimised cost per click). You still pay per click, but delivery optimises towards a conversion event you choose, such as a purchase or a lead. This needs conversion tracking in place.
Two details from OpenAI's help centre catch people out. First, the objective cannot be changed after the campaign is created, so a switch means building a new campaign. Second, conversion-optimised campaigns use one standard conversion event per campaign, and custom events cannot be the optimisation goal.
Third-party blogs report various CPM and CPC figures, minimum budgets and recommended starting bids. These numbers move quickly and differ by country and currency, so I do not repeat them here. Check the current figures in Ads Manager and OpenAI's help centre before you build a budget.
How are ChatGPT Ads different from search ads?
ChatGPT Ads differ from search ads mainly in how intent is read and how much control you get. Google reads intent from a short query and lets you bid on it directly; ChatGPT reads intent from a whole conversation and decides for you which ads fit.
- Intent signal. A search query is short; a conversation can include budget, location, constraints and follow-up questions. That can mean richer context, but you cannot pick the exact moments.
- Control. Search gives you keywords, match types and negatives. ChatGPT Ads give you creative, location and objective.
- Maturity. Google Ads has two decades of tooling, Smart Bidding and integrations. ChatGPT Ads Manager is in beta, and features change month to month.
- Measurement. Both rely on your own tracking. ChatGPT Ads use an OpenAI measurement pixel and a server-side Conversions API, much like Meta's Pixel and CAPI.
I go deeper on this in ChatGPT Ads vs Google Ads, with a side-by-side table.
Who should test ChatGPT Ads now?
ChatGPT Ads are worth a test now for businesses whose buyers research before they purchase, and who already have reliable conversion tracking. They are a poor first channel for a business that has not yet proven its offer on Google or Meta.
Good candidates:
- Considered purchases. Software, courses, B2B services, higher-ticket ecommerce, where people ask "which one should I choose?"
- Service businesses with clear offers. US home services where people ask ChatGPT for help comparing quotes or understanding a job, if your location options cover your area.
- Brands with tracking already working. If your purchases or leads are measured correctly today, adding a new channel is far less risky.
Wait a little if:
- You have no working conversion tracking on Google or Meta yet.
- Your budget only just covers your best-performing channel.
- Your country is not yet supported for self-serve.
What should a first test look like?
A first test should be small, time-boxed and measured properly from day one. This is the order I would follow:
- Install the OpenAI measurement pixel and confirm your key event fires (I cover this in my ChatGPT Ads conversion tracking setup guide).
- Add UTM parameters to the landing page URL so the traffic shows clearly in GA4.
- Start with a CPC campaign on one clear offer and one landing page.
- Move to oCPC once the conversion event is recording reliably.
- Compare results in Ads Manager and in GA4, and expect the numbers to differ for the usual reasons explained in why GA4 and ad platforms disagree.
Final thoughts
ChatGPT Ads are a real new channel, but they are still early and they reward advertisers who measure carefully. Treat them as a structured test, not a replacement for search. If you want your tracking ready before you spend, my ChatGPT Ads tracking service covers pixel, Conversions API and GA4 setup, or you can book a free tracking audit.
