See the new Writesonic live. Fact Check, Sentiment, Outreach Agent, and more →Reserve my seat

How we collect the data

What the numbers in the AI Ads Library mean, and where they stop.

Capture

Writesonic runs a large daily sample of real prompts through ChatGPT and Google Search in several countries and records every sponsored unit in the ChatGPT answers and Google AI Overviews: the advertiser, image, headline, description, price and landing page. Organic results the answer cites on its own are not ads and aren't counted.

It's a sample, not a census, so treat counts as indicative rather than a full count of a brand's spend.

Definitions

Seen
How many times we captured the ad across our prompts.
Active
We captured it in the last 7 days.
Days running
Days between our first and last capture of the ad.
Countries
Countries whose prompts returned the ad.
Position
The ad's usual slot among the sponsored results in the answer, 1st being the first.
Prompts
Prompts whose answers showed the brand's ads, with how many ads and answers.
Ads in rotation
Ads whose first and last capture span that day. We record a first and last sighting per ad, not every day it ran, so this is an estimate of what was live.

Which advertisers get a page

A brand gets a page once its data is deep enough to read on its own: at least 10 distinct ads on ChatGPT or Google AI Overviews, 7 days between its first and latest ad, and an ad in the last 90days. Ad networks and some sensitive categories (gambling, pharmacy, clinical care) don't get pages.

Corrections

If this is your company, or we've filed an ad under the wrong advertiser, email [email protected]. Writesonic isn't affiliated with OpenAI, Google or any platform we track.