AI Citation Meter Check whether AI answers cite your site, engine by engine, with the cost of each check
The AI Citation Meter measures whether AI answers cite a site, engine by engine, with each call's cost written down and a grid separating crawled, cited and paid. Our own 2026 readings: Copilot, 1,039 citations in five days; a fixed grid on two engines, 4 of 20 cells cited. $199. Not on sale yet.
What it does: Measure, per engine and per campaign, whether an AI answer cites a site, with the cost of every API call written down and the engine always named.
Who it is for
Anyone who has to answer, with a number, whether AI answers cite a site, and to say which engine said so.
Not for: Anyone looking for a way to rank in an AI answer. Nobody can promise that, and this product does not.
What it solves
- "The AIs cite us" is said without an engine, a date or a count. One check per engine, each naming its engine, with the cost of every call written down. Copilot: 1,039 citations in a five-day window in 2026, 97% on vendor-brand queries; last Copilot reading we hold.
- A visibility number that flatters is worse than none. Readings at zero ship with the product. Perplexity and Google: 0 of 12, on five dates in a row.
- Being crawled gets mistaken for being cited, and cited for being paid. The grid separates the three, with our own readings filled in. Two readings, nine days apart, both in 2026.
Price and access
Files and their updates are delivered in English.
Waiting list: AI Citation Meter
No ship date is set: a system goes on sale once the points under “What we have not verified yet” that hold it back are closed and a live purchase of it has been tested. You get one dated note that day, with the price and the proof attached. No other emails, no countdown.
On sale now: The Domain-Only Website Kit, $129.
What is inside
The procedure for each check, one per engine, with what each call cost us on the date we paid it. The grid “crawled ≠ cited ≠ paid” with our own readings filled in. The cost record template. The monthly routine that produced our numbers. The list of what the checks cannot see (the Bing report has no API for share of authority; Google AI Mode is sampled, not counted).
What we measured on the sites we manage
Not on sale yet: this product goes on sale once the points under “What we have not verified yet” that hold it back are closed and a live purchase of it has been tested.
Measured with: Bing Webmaster AI Performance and Clarity for the Copilot side (same Microsoft data, counted once; read by export, not by script); a scripted browser check for Perplexity and Google; a paid fixed grid, one call per cell, for ChatGPT Search and Google AI Mode. “The AIs cite us” is not a sentence we write; the engine is always named.
The proof is kept in 2 dated audit folders and shared on request; every figure is also on our figures page.
What we have not verified yet (unverified)
- Third dated test: Two dated tests exist, nine days apart, both in 2026. The third is planned with the monthly review; the three-test check stays partial until then.
AI Citation Meter: questions
Does it tell me how to get cited?
It tells you whether you are, by which engine, on which query, and what it cost to find out. The grid separates crawled, cited and paid; what you do with the reading is your strategy.
Why one check per engine?
Each engine shows its citations through a different door (a database of AI mentions, collected ChatGPT answers, Bing's AI performance report, Google's AI mode), and none of them agree. One cost record, the engine named on every line.