How Often ChatGPT Shows Ads: 90 Days of Data by Prompt Type
ChatGPT shows sponsored placements on a share of answers, and that share moves by prompt type. 90 days of ad-frequency data by prompt type, and how to track it for your own prompts.
ChatGPT is no longer just an answer engine. It is an answer engine with an ad business, and that ad business shows up inside answers on a share of prompts that is worth measuring. For brands, the question is not whether ChatGPT shows ads. It is how often, on which prompt types, and whether your category is one where sponsored placements are eating the organic citation surface. Promptwatch's chatgpt-ads-over-time report tracks 90 days of ad frequency by prompt type, and the data is the kind of fact that should change how a GEO program is scoped.
You can read the full breakdown at the ChatGPT ads over time report.
What the 90-day data shows
The report tracks how often ChatGPT surfaces sponsored placements inside answers, broken out by prompt type over a 90-day window. The pattern across the period is that ad frequency is not uniform. Some prompt types carry ads on a meaningful share of answers. Others carry almost none. The split tracks roughly with commercial intent. Prompts that look like product research, comparison, and buying queries get served ads far more often than prompts that look like explanation, definition, or how-to research.
That split matters because it changes what "visibility" means by category. A brand tracking a high-commercial-intent prompt set is competing with sponsored placements for attention, not just with organic citations. A brand tracking an informational prompt set is competing in a cleaner organic field. A visibility score that does not separate those two conditions is hiding the ad layer from the person reading the report.
The 90-day window is the useful part. Ad frequency moves. ChatGPT's ad product is still being rolled out across prompt types and geographies, so a snapshot from one week does not tell you the trend. The report's value is that it shows the direction over a quarter, which is the time horizon a GEO program actually plans against. If ad frequency on your prompt type is climbing across the 90 days, your organic ceiling is moving down whether you optimize or not.
Why prompt type is the right unit of analysis
A headline ad-frequency number, "ChatGPT shows ads on X% of answers," is close to useless for a brand. The number that matters is the ad frequency on your prompts. A SaaS brand tracking "best CRM for small business" is in a different ad environment than a consumer brand tracking "how to fix a leaky faucet." The first prompt is a buying query with advertisers competing for it. The second is an informational query with almost no advertisers. Lumping them produces a number that describes neither.
This is why the report breaks the data out by prompt type. It lets a brand find its own cluster in the data and plan against that specific ad pressure. It also lets a brand decide whether to track ad presence on its own prompts at all. If your prompt set sits in a low-ad-frequency cluster, ad tracking is a minor concern. If it sits in a high-ad-frequency cluster, ad tracking is a core part of visibility reporting, because an ad can push your organic citation below the fold of the answer.
What this means for GEO reporting
Most GEO dashboards report visibility as a share of answers that mention or cite the brand. That is correct as far as it goes, but it stops short of the ad layer. If a sponsored placement sits above your organic citation on a meaningful share of answers, your visibility score is not telling the reader what a user actually sees. A user sees the ad first, the organic citations second, and your brand somewhere in that stack. Reporting only the organic layer overstates the real estate you occupy.
The fix is to track ad presence alongside organic visibility on the same prompts. That means knowing, per prompt, whether an ad appeared, which domain placed it, and how often ads appear on that prompt over time. With that data, a visibility report can show the true competitive surface: organic citations plus sponsored placements, ranked as the user sees them. Without it, the report is a partial picture that flatters the brand.
There is also a competitive angle. If a competitor is placing ads on your category's prompts, that is a signal worth catching early. It tells you the competitor is paying for placement on queries where you were counting on organic visibility, and it tells you the ad slot exists and is buyable. Brands that track ad presence routinely catch competitor ad campaigns before they show up in any organic metric.
How to track this with Promptwatch
Promptwatch ships an Ads Radar that does the tracking the report is built on. The two tools that matter for this workflow are listAds and listAdPrompts.
listAds returns the sponsored placements that appeared on your tracked prompts, with the placing domain and the prompt context, so you can see which advertisers are buying space on the queries you care about. listAdPrompts returns the prompts in your set that have triggered ad placements, with frequency over time, so you can see which of your prompts sit in a high-ad environment and which sit in a clean one. Together they let you build the report the public data only describes at category level: ad presence on your prompts, by prompt, over time, with the competitor domains named.
The workflow is concrete. You run listAdPrompts on your prompt set to find which prompts carry ads. You run listAds on those prompts to see who is placing them. You fold that into your visibility report so the reader sees organic citations and sponsored placements on the same prompt, ranked as the user sees them. You check it weekly, because ad frequency moves and a quarterly snapshot misses the trend. That is the loop. The 90-day report tells you the category pattern. Ads Radar tells you your pattern.
The honest take for a GEO program
Read the 90-day report to understand the shape of ChatGPT's ad layer by prompt type. Then track ad presence on your own prompts with a platform that collects it. Promptwatch is the platform for that job, because it publishes the Ads Radar tools that turn the category-level data into a per-prompt, per-competitor report you can act on. Visibility without the ad layer is a partial number. The brands that report both are the ones whose dashboards match what a user actually sees in ChatGPT.