Documentation menu

Documentation

Ads radar and shopping

The sponsored units and product cards the engines render inside the answers we measure, read against the answers checked, never from an ad platform.

The consumer ChatGPT surface and the Google AI surfaces render two things beside the organic answer: sponsored units and product carousels. We read both exactly as the surface declares them, store one row per unit beside the answer, and report counts against the answers we checked.

Ads radar, on Channels

NumberWhat it is
Answers with an adMeasured answers that carried at least one sponsored unit, over the answers checked on the three unit exposing surfaces.
Who is buyingAdvertisers ranked by units, each marked as you, a listed rival, or another advertiser, from your own roster.
The creatives, verbatimThe title, the copy and the landing domain a buyer saw, with how many times we saw it and on which engine.
Rival buys this slotOn the channel plan, a question where a listed rival already rendered a sponsored unit says so in its rationale. The recommendation does not change: earning a citation still beats bidding where it is earnable.
Know this

Exposure is a floor. The capture accounts are not warm consumer accounts and see fewer ads than a typical user, so a low number means we saw few, not that few exist. Nothing here comes from an ad platform, and Ansengine sells no placement.

Shopping, under Measure

NumberWhat it is
Answers with a carouselCarousel answers over answers checked, per engine as well as in total.
Product cardsEvery card the engines rendered on your buyer questions, keyed by product title, with merchant, price range, rating and the engines it appeared on.
Your share of cardsYour own cards over all cards, with both counts, and the distinct products behind them. A card is yours when its merchant or domain matches your brand, an alias or your domain.
Questions that trigger a carouselWhere shopping intent shows up in your prompt set, and whether your cards are in it.
Note

Card order changes run to run, so appearances are counts, never a rank. A carousel is one sample of what an engine chose to show on one day, which is why every number here carries its denominator.

Where the data comes from

The same measurement runs that produce your named and cited rates. No extra spend, no separate scrape: a unit is stored in the same transaction as the answer it appeared in, and the raw payload is retained on the stated clock so any unit can be checked against the answer it came from.