How to track sponsored ads in local AI answers

Local prompts have obvious commercial intent. "Emergency plumber near me," "best family dentist in Bristol," "same-day boiler repair in Leeds." When AI answers to prompts like these carry sponsored ads, the advertiser sits right next to the organic recommendation. If that advertiser is a local rival or a lead-gen marketplace, it's competing for the same customer you are.

You can't see this from a normal ad account. You only see your own campaigns there. To see who else is buying around your local prompts, you have to watch the answers. This guide covers how to do that with Ads Radar in Promptwatch and how to point it at specific places.

What Ads Radar captures

Ads Radar records the sponsored ads that appear inside ChatGPT Search, Google AI Overviews and Google AI Mode answers to the prompts you track. For each ad it captures the title, the description, an image or price, the advertiser name and domain, the target URL, and the prompt that triggered it. Promptwatch describes the use as showing which advertisers bid on your prompts, their ad copy, claims, landing pages and positioning.

The views are:

Ads per day, for the overall trend.
Top advertisers, to see who buys most across your prompts.
A By Ad table, one row per captured ad.
A By Prompt table, showing each prompt's ad count, your average visibility on it and the latest ad.

You can filter by model, advertiser domain, prompt type and date range.

Two limits matter. Ads Radar doesn't cover Perplexity. And it doesn't report on your own OpenAI campaigns; that's what OpenAI Ads Manager is for. Ads Radar is a view of the answers, not of your ad account.

Ads Radar is included on Business ($579/mo) and on every agency plan: Kick-off ($199/mo), Growth ($399/mo) and Scale ($799/mo). It isn't on Essential or Professional.

Make the prompts local

Ads Radar reads whatever prompts your monitors track, so local ad tracking is mostly a monitor setup job.

City and state monitors start on Business and are on every agency plan, which conveniently matches where Ads Radar starts. Create a monitor per city you care about with the same prompt set in each. Write the prompts the way a local buyer would type them, both with the city name ("dentist in Bristol open Saturday") and without it ("dentist open Saturday near me"), because "near me" only means something relative to where the question is asked from.

That second kind of prompt is where coordinate targeting helps. Local GEO/SEO coordinate targeting pins a monitor to exact coordinates with a radius from 20 m to 5 km, so ChatGPT, Google AI Overviews and Google AI Mode answer as if asked from that spot. Those are the same three engines Ads Radar covers, so a coordinate monitor gives you ad captures as they'd appear near a specific branch. Coordinate targeting is on Business, Agency Scale and Enterprise. Agency Kick-off and Growth have Ads Radar and city tracking but not coordinates.

Whether ads differ by city or neighbourhood in your category is something to observe, not assume. Run the same prompts in two city monitors, or two coordinate monitors a few kilometres apart, for a couple of weeks. Then compare the Top advertisers view for each. If the lists match, one location is enough to watch. If they differ, you've found local buyers worth tracking separately.

Read the By Prompt table first

For local work, the By Prompt table is the most useful view. It puts each prompt's ad count next to your average visibility on that prompt. Look for two patterns:

Prompts with ads where your visibility is low. A rival or marketplace is paying to appear where you don't appear organically. That's where an organic gain would matter most, and where you might consider your own ad budget.
Prompts with ads where your visibility is high. Someone is buying attention next to your recommendation. Check the ad copy for claims aimed at you or your category.

Then open the By Ad rows for the advertisers that matter. Read the copy and follow the target URL. A local competitor sending ad clicks to a city landing page tells you how they're positioning in that city. A marketplace bidding on your branded local prompts is a different conversation, and possibly one for your agreements with that marketplace.

Use the filters to separate engines. An advertiser active in Google AI Mode but absent in ChatGPT Search may simply be buying through one platform.

Put an agent on it

Reading ad tables every week gets old. Agent Chat can do the first pass. For a hypothetical project:

"For each city monitor, list advertisers that appeared in Ads Radar this week but not last week, the prompts they appeared on, and our average visibility on those prompts. Quote one ad title per new advertiser."

Save that as a Scheduled Chat, and our guide to automated local reporting covers delivery in the app or Slack. Turn anything that needs a response into an Action Board item with the captured ad attached. Agencies can pull the same data over the hosted MCP server into their own reporting.

What to do with it

Ad tracking is context, not a fix by itself. The responses are ordinary marketing work: improve the location page so you earn the organic mention, check the claims a competitor makes in its ad copy, and decide whether your own campaign belongs on the prompts where others are buying. What Ads Radar adds is the evidence of who's buying which local prompts and what they say. Watching answers by city or by coordinates is also how you'd notice a new local advertiser at all.

If you need city-level ad visibility for your own locations, that's Promptwatch on Business, which also adds coordinate targeting. Agencies get Ads Radar with city tracking on any agency plan, and coordinates on Scale.