This article was sponsored by Trends. The opinions expressed in this article are those of the sponsor.
Are my competitors serving ChatGPT ads?
Is there an ad library for ChatGPT sponsored results?
How do I know who is advertising in AI responses?
Your most intentional buyers are currently asking ChatGPT about your category.
A sponsored placement appears below the response, and if a competitor has purchased it, it intercepts clicks at the precise moment buyers are ready to make their decision.
Unless you run every relevant prompt yourself, competitors are harming the visibility of your AI in the most important moments, and you don’t see it.
What this walkthrough covers
This is an overview of the manual process of finding out who is bidding on your category and where you can see exactly who is buying ads in your customers’ ChatGPT responses without doing it yourself.
OpenAI launched ChatGPT ads for US Free and Go users on February 9, 2026.
In the spring, more than 600 advertisers had placements served in response to high-intent prompts:
- Software comparisons.
- Planning a weekend.
- “What is the best CRM tool? »
These queries lived on Google; now they are presented inside ChatGPT as ads.
ChatGPT ads appear in the reply experience as a sponsored card below the reply.
After ChatGPT responds to a prompt, a sponsored card appears below the response, visually separated and clearly labeled “Sponsored.” The sheet includes the name of the advertiser, a favicon, a short title, a precise description (around 19 words on average) and a link to a landing page.
OpenAI does not currently publish an ad library equivalent to that of Meta or Google, and there is no central searchable database of every active ChatGPT ad. To see who is serving ads, you need to run prompts in eligible US sessions and capture what appears.
For monitoring purposes, four data points define what a competitor is doing in a given ad:
- Ad title: the title that a competitor broadcasts
- Description of the announcement: the sentence(s) in the body under the title
- Final URL: the destination to which they send traffic
- Share of impressions: how often a competitor’s ad is shown at a given prompt across multiple broadcasts
You need all four to read the competition table.
The title and description tell you how they are positioned.
The final URL tells you if they’re sending to a generic homepage, category page, or comparison.
Impression share, the percentage of total ad impressions on a given prompt that went to a specific advertiser, transforms “I saw them once” into “they own this prompt.”
For competitive intelligence This matters more than raw impression count, because it normalizes across prompts with different ad fill rate.
Step 1: Map the queries your buyers are already asking
Create a prompt list that represents how your buyers actually talk to ChatGPT. You’re not optimizing for impressions in general terms. You surface competitor activity in conversations that lead to your category.
Start with questions that you already know convert in paid search and high-intent organic.
Then translate them the way someone would express the same need in ChatGPT. People don’t search for ChatGPT the same way they search on Google. They write complete sentences with context, constraints and intent.
A list of job prompts for a paid search manager in any business category should contain 30-50 prompts and cover:
- Direct comparisons (“best (category tool) for (use case)”, “(Brand A) vs (Brand B)”).
- Recommendation prompts (“I need a (tool) for (job to be done), what should I look at?”).
- Prompts for change (“alternatives to (brand)”).
- Use case fit prompts (“which (tool) is best suited for (specific small team/business/industry)”).
- Pricing prompts (“affordable (tool) for (public)”).
- Long edge case (“(tool) that fits into (niche stack)”).
Leverage your brand and category SQL datathe top organic keywords and any customer-facing inputs you have (support tickets, sales calls, on-site search logs, review mentions), so the list represents actual buyer language, not what you assume they’re saying.
If your competitors bid on prompts you haven’t mapped, you’ll never see them; Your ad library begins and ends with your own list of prompts.
For a tip: To use Advertising radar to extract your prompt list and run it continuously.
Step 2: Run each prompt in a ChatGPT session
Once you have the list of prompts, run it and pay attention to session setup, where the data becomes useful or becomes noise.
Run each prompt and capture the response, including any sponsored cards that appear below the response.
Don’t run each prompt just once.
ChatGPT ad auctions do not serve the same ad to every user at the same prompt; different sessions surface different advertisers based on bidding, relevance signals, and rotation.
A single run captures an auction result, not the competitive set.
To get usable playback on a given prompt, plan on at least 20 to 30 runs over several days.
Vary the session: clear cookies between batches and the pace extends to morning, afternoon and evening. Run all 30 in 10 minutes from the same session and you’ll sample a slice of the auction.
Step 3: Capture the four data points that define a competitor’s ad
For each sponsored placement that appears, record the same four fields, in the same place, every time. Otherwise you can’t compare executions.
The four data points to capture per print:
- Ad title: the exact copy of the title in the sponsored card. Copy character for character. The headlines change.
- Description of the announcement: the sentence(s) in the body under the title. About 19 words on average right now, but the range varies. Capture the full text.
- Final URL: the destination URL to which the map links. Remove UTMs to identify the canonical landing page, but keep the full URL in a secondary column so you can analyze tracking patterns later.
- Share of impressions: calculated, not directly observed.
For each prompt, count the number of times each advertiser appeared out of total broadcasts. If you ran a prompt 25 times and Competitor A showed up in 12 of them, that’s a 48% impression share on that prompt for the run window.
Mark each line with the prompt that triggered the announcement, the date and time of execution, and the session details (Free or Go, Location). Set up your spreadsheet so you can pivot impression share per prompt, per competitor, and per week.
Advertising copy iterates quickly. The same advertiser might show three or four different titles in response to the same prompt in a single week while their team tests the creative. The final URLs also change; a competitor can alternate between a homepage, comparison page, and category landing page to test conversion. Capture just the title and you’re missing the iteration patterns and URL strategy, which are the bulk of what tells you what your competitor is doing.
Step 4: Repeat often enough to see share of voice over time
A casual reading of competitors’ advertising activity will mislead you. You will catch the one who won the auction on the day you performed the prompts and miss the rotation that happens every other day. Decide the budget from a single day’s snapshot and you decide the noise.
To see share of voice, i.e. who actually belongs to this category in ChatGPT, you need a recurring cadence. The minimum that gives you the signal:
- Daily executions on your 5-10 most profitable prompts (the queries closest to purchase intent)
- Weekly Runs on Full List of Prompts 30-50
- Monthly trending allows you to see how competitors are gaining or losing share over 30-day rolling windows.
For a tip: Use Ad Radar to run this cadence automatically and get a continuous reading of competitors’ advertising activity in ChatGPTwithout the spreadsheet overhead.
Stop Flying Blind in Paid AI Search
Paid search managers have bidding insights, ad libraries, and dozens of third-party monitoring tools for Google. For ChatGPT ads, they don’t have any of that yet. ChatGPT Ads are a new auction that addresses the same buyer intent, and right now, most teams don’t have visibility into who is bidding against them. If competitors are already in your customers’ ChatGPT responses, you’ll find out through your own monitoring or through a gap in the pipeline that you notice too late to take action.
Advertising radar runs prompt monitoring continuously and surfaces every announcer, every prompt, every creative iteration. Gain continuous visibility into competitors’ ChatGPT advertising activity in your category.
Image credits
Featured image: Image from Shutterstock. Used with permission.
