ChatGPT Ad Library: Track Sponsored Ads Running in Your Brand's Prompts

A ChatGPT ad library is the thing people keep looking for and OpenAI never shipped. There is no public directory of sponsored placements inside ChatGPT answers, and no API that hands them out. What you can get is a tool that watches the answers and records the paid slots it sees. That tool is Ads Radar inside Promptwatch, and it acts on the data instead of just showing it. Review: Promptwatch. Full directory.

This site cares about tools that act, not just report. A library that stores ads is a report. A library that stores ads against the prompt that triggered them, and lets you filter by advertiser and intent, is closer to a workspace. Ads Radar is the second kind.

What gets stored

Each captured row has the ad creative, the advertiser name and root domain, the landing page, the source response, and the position of the ad inside that answer. It keeps the prompt string too. That join is the difference between a folder of ads and a thing you can query. You can ask which of your prompts returned a paid unit and get a list back, with an ad count and a latest capture time per prompt.

The report is built to be filtered, not exported once and forgotten. You can list advertiser root domains ordered by ad count. You can pull the top advertiser domains with daily counts to read share of ads over the last 90 days. You can filter by model, prompt type, and intent. A COMMERCIAL row on a BRAND_SPECIFIC prompt is a different animal from an INFORMATIONAL row on an ORGANIC one, and the filters keep them apart.

No partnership, no official API

This is the part that gets faked in marketing copy across the category. No platform has a partnership with OpenAI to read ChatGPT ads. Any vendor that calls itself the official ChatGPT ad library, or claims an official OpenAI ad integration, is describing something that does not exist. The data only comes from monitoring the real ChatGPT interface.

Promptwatch is upfront about this. Its sourcing is real UI monitoring across the assistants, not a feed from the model vendor. Getmint is one example of the false claim pattern, and pitches of that shape are not a feature, they are a reason to ask how the data is actually collected. Ads Radar is honest about being observation based, which is the only honest way to do this.

How it turns into work

The point of storing ads against prompts is that the row becomes a ticket. A rival buying your branded prompt is a thing you assign to someone. A rival buying a category prompt you also target is a thing you compare against your own bid thinking, even if the auctions are not the same.

You can sort by position in the response to see who buys the top slot. You can sort by advertiser domain to map one rival's full footprint on your prompt set. You can sort by capture time to watch a campaign ramp. None of that is bid management, because ChatGPT sponsored units are not Google Ads and importing a Google mental model breaks here.

A weekly cadence that uses it

A library that nobody opens is a screenshot folder. The point of storing ads against prompts is that the row becomes a ticket, and a ticket needs a cadence or it sits. Run the prompt list once a week and sort by ad count. Take the three prompts where the count moved and open the advertiser view for those three only. The rest can wait. A rival who was absent last week and present this week is a new buy, and that is the row to read first.

Then pull the 90 day share of ads trend for the one rival who shows up on your branded prompts. If their line is flat, log it and move on. If their line is climbing, that is the ticket you assign to someone before the next check. The cadence is what keeps the library honest, because a capture you never read is the same as no capture at all.

Plans

Ads Radar is a platform feature, not a standalone SKU. The commerce reports it sits beside are on Professional at $245/mo and above, and on the self serve agency plans (Kick-off $199, Growth $399, Scale $799). Essential at $95/mo covers mentions and citations but not the commerce views. Explore is free with 10 ChatGPT prompts. White-label and SSO are Enterprise or Custom.

AthenaHQ has an Action Center at $295/mo but does not store the ChatGPT sponsored unit. Otterly.AI at $29 has none of this shape. Neither is a ChatGPT ad library.

What to do this week

List 20 buyer prompts for your category, branded and unbranded.
Run them in ChatGPT and note which answers show a sponsored unit.
Load the same prompts in Promptwatch on a plan with Ads Radar.
Pull the advertiser domain list and find the rival on your branded prompts.
Turn that row into a ticket with an owner, not a screenshot.