Every morning we ask ChatGPT, Gemini, Google’s AI and Perplexity the questions your customers really ask — then show you who gets recommended, why, and what to fix. Every answer is kept, so every number can be checked.
You set it up once in a few minutes. After that it runs by itself.
We read your website — what you sell and where — and write the questions buyers ask AI, like “who does panel upgrades in Burnaby?”. You pick which to track.
Each question goes to every AI system on your plan, anonymously, from your city or country — just as a stranger looking for a supplier would ask.
Did it name you? Did it read your website? Who did it recommend instead, and which websites did it trust? Every answer is kept word for word.
A monthly action plan says which page, topic or listing to work on. Mark a fix done, and we show whether it worked.
Plain answers to the questions owners actually ask.
One number out of 100, with an honest range, by AI system, by topic and by buying stage — and how it moved since last month.
The businesses AI recommends for your questions, how often, and the exact answers where they won.
The directories, review sites, forums and Google listings AI read before answering — so you know where to be listed.
For each question: your Google position, whether Google’s map shows you, whether AI names you, and how many people search Google for it each month — rising or falling.
Which pages AI reads, which it credits you for, and which pages could win answers — add any of them to your plan.
Visitors who came from ChatGPT and others, a before-and-after for every change you mark done, a morning alert when something moves more than usual — and team seats, so colleagues work on it with their own logins.
Every answer gets one of three verdicts — the same colours you’ll see on the dashboard.
Your business is named in the answer itself. This is the result that brings customers.
Your page was a source for the answer. If AI read you but named someone else, that page is one of your quickest fixes.
We record who it named instead and which websites it read — that’s where your plan comes from.
Being named counts most; being read as a source counts too.
Score = 70% × how often AI names you + 30% × how often it reads your site.
AI answers change from one hour to the next, so every score comes with a range — where your real position very likely sits. More answers, narrower range. Failed checks are never counted as zeros.
Some tools fill gaps with guesses. We show only what the AI actually said.
For the curious — and for agencies who need to explain every number to a client.
A customer does not ask an AI “tell me about your business”. They ask “who does panel upgrades in Burnaby” or “best lift chair I can order online”. Those are the questions worth measuring, and they are different for every business.
So we start by reading your website — the pages, what you sell, the city you are in and the areas your pages say you serve. From that, the question finder writes the questions a customer in your market would actually ask, grouped by topic. Questions that only contain your own name are dropped: of course an AI mentions you when someone asks about you by name.
Each question is tagged by buying stage — ready to buy, comparing, learning, or looking you up — so you can see whether you are named where money changes hands or only in the background reading. Each is also checked against the places your site says you serve; a question naming somewhere we could not find on your site is flagged, never deleted. Our reading of your site is a guess. You decide.
You tick the ones to track. Those are asked, unchanged, every time.
Every plan asks the first four. Pro and Agency add Perplexity, asked the same way every morning and counted in the score like the others. Perplexity is asked through its own API, using its web-searching Sonar model — not the perplexity.ai page itself, so an answer can differ from what the app shows — and the pages it read come back with each answer.
Each tracked question is put to each system once per check — every morning on paid plans, once a week on Free. The question is asked anonymously: a fresh session, no account, no history, no location beyond the country or city the customer would be in. That is the position a stranger looking for a supplier is actually in, and it is the only fair one to measure from.
The full text of every answer is kept. Click any question on the dashboard and you read exactly what each system said that morning and which pages it read to say it.
Each answer is read for three things:
An answer with no AI Overview (Google often shows none) is recorded as exactly that. A check that failed — a timeout, a blocked request — is recorded as a failure. Neither is ever scored as a zero.
The score is a percentage from the answers that actually arrived. Being named counts for most of it; being read counts for some.
A single morning is one sample. The same question can score 13 one hour and 25 the next, because the systems change their answers. So the score always carries a range — the band your true position very likely sits in given how many answers it rests on:
The score is then placed in one of four bands:
The dashboard shows the score, the range, the band, and the same three numbers broken down by AI system, by topic and by buying stage. Every one of them links to the answers it was computed from.
One check tells you almost nothing. The line through many checks is what you can act on, and it only exists if the measuring keeps happening. That is why the product is a daily measurement rather than a one-off audit.
Each check is plotted as a dot, with your average drawn through them — not a smoothed line that hides how much the answers move. Comparing this month with last month is done on the same set of questions; if you changed the set in between, the dashboard says so instead of comparing two different things. A period with too few answers to compare reads as “not enough measured yet”, not as a change.
Being named is the cause; visits are the effect. If you connect your Google Analytics (read-only, one click) or paste a one-line snippet, the dashboard shows the sessions that arrived from ChatGPT, Perplexity, Gemini, Copilot, Claude, Grok and the rest — by AI, by landing page, by week — and how those visitors behaved against everyone else.
One thing nobody can count: clicks from Google AI Overviews and AI Mode. Google reports them as ordinary Google traffic, with no way to tell them apart. The card says so rather than guessing.
Find the questions your customers ask AI, free, in about twenty seconds — or look around a real customer’s dashboard first.
Think a number on your dashboard is wrong? Write to [email protected] — every answer is on record, so it can be checked.