We do not stop at measuring —
we stay until you get found.
We ask the same question across ChatGPT, Gemini, Claude and Perplexity and measure whether your brand comes up. Then we tell you why it did not — and what to do about it.
Five things it does
Measure
We put numbers on whether AI mentions your name and whether it cites your site. Not just present or absent — where you appeared, and who you were listed alongside.
- Mention rate (how often your name appears) and citation rate (how often you are the source)
- Position distribution — 1st / 2nd / 3rd or lower. What we cannot determine stays "unknown"
- Co-mentions — which competitors showed up in the same answers as you
Fact check
Is AI saying something false about you? We cross-check the answers against your own site — stale pricing, services you dropped, things you never offered. Left alone, that becomes your description.
- Anything we cannot ground is reported as "unverifiable" — never asserted as wrong
- If there is nothing to report we show nothing — "no issues" and "nothing to check" are not the same
Discover
We surface the questions your customers actually ask AI, tuned to how each language phrases them rather than translated literally. And we keep the questions you never appeared in — that is where the room to grow is.
- Questions generated for each stage: searching, comparing, deciding
- Missed prompts — the list of questions you never came up in
Prescribe
We do not stop at "0% mention rate". We tell you why, and what to do next. If a monitoring tool is a thermometer, we consider the job unfinished until the prescription is written.
- GEO site audit — structured data, FAQ, llms.txt and seven more
- Fixes tied to the questions you actually missed and to who is being cited instead
- A report you can hand over by link, or print as-is
Cross-lingual
We run the same questions in Japanese, English, Korean, Chinese and Cantonese. AI answers differently in each, so being first in one language and absent in your next market is entirely normal. We put that gap side by side.
- A FOUND / not-found breakdown for every language
- Territory single-language tools never look at
What running it actually needs
How many AIs, how many languages and how many questions a month depends on your plan. The free plan measures with one language and one AI, both of your choosing. See plan limits
Accuracy, published
Detection accuracy, position extraction error and report fabrication rate are published — including where we are still weak. If we sell a tool that produces numbers, we should not hide how sound they are.
See how we measure →