AI-visibility tracking ChatGPT · Gemini · Claude
You cannot fix an answer you have never seen.
A scenario. Somebody asked ChatGPT last night to recommend a business like yours. You will never see that conversation. What we can do is ask the same AIs the question a customer asks, grade what they say about you, and keep the result on a dated record. Then we ask again, on a schedule you pick. If your grade changes, we email you when that scan runs. If you fix something on your site, the next scan tells you whether it worked.
| Scanned | Cold | Grounded | Change |
|---|---|---|---|
| Sep 11 | A- | A | +12 |
| Sep 04 | B | A- | DRIFT |
| Aug 28 | B | A- | +16 |
| Aug 21 | C+ | B | +11 |
| Aug 14 | D | B | -4 |
COLD = WHAT AI SAYS WITH NO HELP · GROUNDED = WHAT IT SAYS READING YOUR SITE · THE GAP IS THE WORK, THE TREND IS THE PROOF
Nothing to install, nothing to grant us access to. We just ask the AIs the way your customers already do.
We ask ChatGPT, Gemini and Claude the question your customers ask them: what is this business, and should I hire them? Each answer is graded against your own website.
Two grades for every answer. Cold is what the AIs say about you from memory, no help. Grounded is what they say after reading your website, which is about as good as it can get today. Most businesses see a wide gap between those two numbers the first time they look. Closing it is the work.
The AIs change their minds without telling anyone. So when your grade moves, it goes on your dated record, and we email you when that scan runs.
Every plan runs the same instrument: scheduled scans, a grade for every answer, and a dated record of every change. The bigger plans watch more businesses.
The Fix
When the scan shows the AIs getting you wrong, this is what we do about it. We go through your record and list exactly what is wrong or missing. Then we write the fix: a plain facts page that says who you are, what you do and who you serve, plus the labels the AIs read behind the scenes. You put it on your site, or we hand your web person the files. Then we scan again, and the record shows what the AIs made of it, before and after. Nobody can promise what an AI will say tomorrow, us included. What we promise is the page, written and handed over, and the moment captured on your record.
Log in with your email. We send a one-time link, no password, any device.
Log in →Mirror Monitor - on watch
A snapshot tells you once. The watch keeps telling you.
Next scan: -
Cold grade at each scan. A rising line means the AIs are learning you; a drop is drift worth a look.
| Scanned | Cold | Grounded | Change |
|---|
Straight off the scan record: how the answer scored, who got recommended instead of you, and what AI got wrong.