Best GEO Software
All posts
By Best GEO Software TeamChatGPT adsbuyer's guidecompetitor tracking

Setting up alerts for new ChatGPT ads from your competitors: a buyer's guide to the tools

Want to know when a competitor starts running ChatGPT ads? What to evaluate in a monitoring tool, what each option costs, and how to pick one without overbuying.

The request usually lands the same way. Someone in paid search notices a rival's name in a sponsored unit under a ChatGPT answer, forwards a screenshot, and asks whether you can be told the next time it happens. You can, roughly. But the alert has to be assembled from a monitoring tool, because the ad platform itself won't tell you.

This guide covers what OpenAI provides, the criteria that separate useful ad monitoring from a screenshot habit, the real prices of the options in our catalog, and a short way to choose.

Start with what OpenAI does and doesn't publish

ChatGPT ad testing started in the U.S. on February 9, 2026. Ads show for Free and Go users, below a response. Plus, Pro, Business, Enterprise and Edu accounts get none, and neither do accounts identified as under 18. By October 21, 2026 OpenAI said the program covered 89 countries and over 100,000 advertisers, bought through technology partners or the self-service Ads Manager.

For buyers of monitoring tools the important part is the gap. OpenAI's Ads Manager reports your own impressions, clicks, spend and conversions, segmented by device, country and platform. OpenAI says advertisers only get aggregated, non-identifying performance data and have no access to chats. We found no official ad library, no transparency center for ChatGPT ads, and no OpenAI feature for watching competitor ads. Third-party sites calling themselves a "ChatGPT ad library" exist, but they aren't official, and we didn't evaluate them.

What to evaluate in a monitoring tool

Which answers it reads

An ad only exists inside a real answer. Ask which engines the tool captures ads from, and whether it reads them from the live product. Promptwatch Ads Radar covers ChatGPT Search, Google AI Overviews and Google AI Mode, pulling ads out of the responses your monitors collect anyway. It does not cover Perplexity.

What it records per ad

A name in a list isn't enough to act on. Look for the creative (title, description, image or price), the advertiser name and domain, the landing page, and above all the prompt that triggered the ad. Without the prompt you can't tell whether a competitor is bidding on your category or on your brand.

Who controls the prompt set

If the vendor picks the prompts, you're watching their sample. You want to load your own commercial and comparison prompts, because the coverage of any ad tracker equals the prompts it runs.

How often it runs

Ad delivery swings from day to day. Promptwatch's report "How Often Does ChatGPT Show Ads?" covers May 20 to August 17, 2026. It found a 90-day average of 20.1% of web-search responses carrying an ad and a seven-day average of 32.4% at the end of the window. Single days reached 43.9%. A weekly sample can miss a short test entirely.

Promptwatch chart of ChatGPT ad-bearing responses split by organic, brand-specific and competitor comparison prompts

Promptwatch data report "How Often Does ChatGPT Show Ads?", prompt-type chart, captured November 14, 2026.

That chart is a volume mix: on an average day 73.1% of ad-bearing responses came from organic prompts, 17.9% from brand-specific ones and 9.1% from competitor comparisons. The report says plainly that this shows where ads appeared, not how likely a prompt type is to trigger one.

How you hear about changes

Ask the vendor to show you the alert, not describe it. Promptwatch doesn't list a push notification for a new advertiser. What it has is a By Prompt view sorted by newest ad, an advertiser-domain filter, and Scheduled Chats that Agent Chat can deliver in-app or in Slack. For most teams a fixed weekly review of those views is the honest version of an alert.

What the options cost

Prices below come from our catalog entries and Promptwatch's own pricing, monthly billing.

OptionStarting priceWhat it does for competitor ads
Promptwatch Business$579/mo (7-day trial)Ads Radar: creatives, advertisers, landing pages and triggering prompts across ChatGPT Search, AI Overviews and AI Mode
Promptwatch agency plansKick-off $199/mo, Growth $399/mo, Scale $799/moAds Radar on every tier, unlimited projects and prompts
Cognizo PlatformFrom $499/moLists "ChatGPT Ads"; scope not documented in our entry
BrandlightCustom, not publishedHomepage lists an AI ads module
Rank Prompt$49 monthly (Starter)ChatGPT Ads management for your own campaigns
OpenAI Ads ManagerYour ad spendYour campaigns only; no competitor data

Essential ($95/mo) and Professional ($245/mo) Promptwatch plans don't include Ads Radar. Annual billing on Promptwatch is 12 months for the price of 10.

How to choose

If you only need a yes or no on whether a rival is in the auction, run a few manual checks from a Free account in a country where ads run, outside Temporary Chat. Expect inconsistent results: OpenAI uses conversation context and, with personalization on, signals from past chats, so your test account isn't a neutral sample.

If you need to answer the question every week, for a known set of prompts, buy a tracker that captures the triggering prompt. Brands land on Promptwatch Business; agencies managing several clients get Ads Radar from Kick-off, which also has unlimited projects.

If what you really want is to run your own ChatGPT campaigns, that is Ads Manager or a campaign tool like Rank Prompt, and a different purchase.

Before you sign anything, check one claim with each vendor: can it show you the exact prompt where a competitor's ad appeared last Tuesday? A vendor that can't do that is selling a dashboard of logos.

Where to start

Take the Promptwatch 7-day free trial on Business or Kick-off, put twenty or so of your own buying questions into a ChatGPT monitor, and open Ads Radar after a week of daily runs. If Top advertisers is empty, that's an answer too: nobody is paying for those prompts yet.