A Promptwatch Ads Radar Workflow to Monitor ChatGPT Sponsorships

A dashboard you check once is a report. A dashboard you check on a cadence is a program. The difference between ad tracking that produces work and ad tracking that produces nothing is the workflow around it, not the data. Ads Radar inside Promptwatch gives you the rows. This is a workflow for turning those rows into a weekly habit that catches a rival buying into your prompts, and the point of the cadence is to end on a ticket, not a chart. Review: Promptwatch. Full directory.

This site ranks tools that act. A workflow that ends on a dashboard is a report. A workflow that ends on a ticket with an owner is a workspace. The cadence below leans toward the second. One fact first: there is no official ChatGPT ads API and no vendor has an OpenAI partnership for this. Ads Radar monitors the real ChatGPT interface and stores what it sees, which means captures depend on which prompts you track and how often they run. A vendor claiming an official integration is misrepresenting how it works, and getmint is one example of that false claim pattern. A workflow built around observation is honest. A workflow built around a complete feed breaks the first time a capture is missing.

Monday: the prompt view

Open Ads Radar and list the prompts whose answers contained sponsored ads this week. Each row has an ad count and a latest capture time. Sort by ad count descending. The prompts at the top are where the auction is active. The prompts with zero captures are where nothing happened, and you skip them this week.

The move on Monday is narrow. Take the three prompts where the ad count moved the most and ignore the rest. A prompt that went from zero captures to five is a new auction. A prompt that went from two to twelve is a rival spending harder. Both are worth ten minutes. A prompt that held steady is not.

Tuesday: the advertiser view

For the three prompts you flagged Monday, open the advertiser view. List advertiser root domains ordered by ad count for those prompts only. This is the set of brands buying into the queries that moved. The rival you are looking for is the one who was not there last week and is there this week, or the one whose count jumped.

Filter the full ad list to that one domain and read every ad they have run on your prompts. Each row has the ad creative, the advertiser name, the landing page, the source response, and the position of the ad inside that answer. Save the exact copy, not a paraphrase. A paraphrase is a guess, and a guessed ad will not survive a client check.

Wednesday: the trend view

Pull the top advertiser domains with daily ad counts for the last 90 days, filtered to the rival you found Tuesday. This is the share of ads trend, and it is the view that tells you whether to escalate. A rival whose daily count is climbing on your branded prompts is a different problem from one whose count is flat. A competitor who held steady for a quarter and then doubled in the last month is running something new, and you cannot tell that from a single capture.

The decision on Wednesday is binary. If the line is flat, log it and close the loop. If the line is climbing, that is the row you escalate, and you escalate it before the next Monday, not on it.

Thursday: sort by position and intent

Two sorts keep the rows comparable. Sort by position in the response to see who buys the top slot, because the top sponsored unit and the third one are not the same buy. Then filter by intent. A COMMERCIAL row on a BRAND_SPECIFIC prompt is a rival buying your own demand, and that is the row to escalate first. An INFORMATIONAL row on an ORGANIC prompt is a rival buying awareness, which is a different response and usually a lower one.

The intent filter stops you treating every captured ad as the same kind of event. A rival buying your branded prompt is a ticket. A rival buying a broad category prompt you do not target is context. Blending them wastes the week.

Friday: ticket or close

Not every captured ad is a ticket. A rival buying a category prompt you do not target is context, not work. A rival buying your branded prompt is a ticket. A rival whose ad copy contradicts your docs is a ticket. Assign an owner to each ticket and save the exact ad string with it. Close the rest. A backlog of ads we saw that nobody owns is the thing that kills the workflow, because by the third week nobody opens it.

This is not bid management. ChatGPT sponsored units are not Google Ads, the auction is different, and importing a Google campaign mental model breaks here. Ads Radar is a monitoring tool for the ChatGPT surface. Treat the output as observation with capture gaps, because no feed exists.

Ads Radar is a platform feature 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 does not include it. Explore is free with 10 ChatGPT prompts. AthenaHQ at $295/mo and Otterly.AI at $29 do not store the sponsored unit against the prompt, so they do not support this workflow.

What to do this week

Run the Monday prompt view and flag the three prompts where ad count moved.
Tuesday, open the advertiser view for those three and find the new rival.
Wednesday, pull the 90 day trend for that rival and decide if the line is climbing.
Thursday, sort by position and filter by intent to find the branded demand buy.
Friday, ticket the branded prompt row with an owner and close the rest.