Ten real buying questions
Written from your own website, in your buyers’ language, for the market you sell into. No brand names in them, so nothing steers the answer.
Free visibility check
AI Visibility Check
The scaile AI Visibility Check reads your website, writes the ten questions your buyers ask an assistant before they buy, and puts every one to ChatGPT, Gemini and Perplexity. The questions are free; the report adds who was named.
Reading your site and writing your questions …
A few seconds. We’re reading your pages to work out what you sell, then writing the ten questions your buyers would ask.
The ten questions we’ll ask
Every one of these goes to ChatGPT, Gemini and Perplexity, asked from your market in the language your buyers search in.
Starting your check for …
One moment. We’re reading your site and writing the questions.
Your report is being produced
The check is running as you read this. In a few minutes it lands in your inbox: the ten questions, who was named in every answer, the sources behind them and what to fix first. Want it read against your pipeline?
Book a CallTrusted by 40+ Enterprises
What you get
The answer your buyers actually see, and every name in it.
The full competitor leaderboard
Every brand the answers put forward, ranked by how often it is named and how early, across all three assistants. Your row is in there, whatever it says.
The sources behind the answers
The pages the assistants read before they answered, and whether yours was one of them. That is where an answer about your category is decided.
How the AI Visibility Check works
Your homepage and the pages behind it, to learn what you sell, who you sell it to and which market you sell it in.
The questions someone asks with their wallet out, in the language your buyers use. Telling us your core product sharpens them.
Every question goes to ChatGPT, Gemini and Perplexity, asked from your market the way a buyer would ask it. We record who gets named and what was cited.
A score, the leaderboard, what every assistant answered to all ten questions, and the fixes, as a PDF in your inbox a few minutes later.
How the questions are written, and what the check does not do
The result depends entirely on what is asked, so what is asked is stated here in full rather than left as a black box.
What runs on this page
- Your homepage and the pages behind it are read, to learn what you sell and who you sell it to.
- Ten buying questions are written from that, in the language your buyers search in.
- No brand name appears in any question, including yours. A question that names a brand answers itself.
- The questions are shown to you before anything is asked for.
What it does not do
- It does not measure traffic. Being named in an answer and being clicked are different things.
- It is a snapshot, not a trend. Assistants disagree with themselves between runs, which is why the report prints what was actually returned.
- It does not ask about you by name, so a low result is not evidence that nobody knows you; it is evidence about your category.
- It cannot see what a specific person is shown. Answers vary by account, history and region.
What the full report adds
- All ten questions put to ChatGPT, Gemini and Perplexity, with what each one answered.
- The competitor leaderboard: every brand named, ranked by how often and how early.
- The pages the assistants cited before answering, and whether any of them were yours.
- The five articles we would write first, each against a question from your own run.
Being named by an assistant, in plain terms
What this check measures, and what the number does and does not tell you.
How do I know whether ChatGPT recommends my company?
Ask it the way a buyer would, without naming yourself. "Best supplier of X for Y in Germany" is a buying question; "what is company Z" is not: the second answers itself and tells you nothing about whether you would ever come up. Then ask the same question several times, and in more than one assistant, because the answers are not stable. That is the whole method here, run ten times across three assistants so a single lucky answer cannot flatter the result.
Why do ChatGPT, Gemini and Perplexity give different answers?
They read different things. Perplexity retrieves live pages for almost every answer and cites them. ChatGPT in search mode retrieves for some questions and answers from training for others. Gemini leans on Google’s index and on AI Overviews. So a brand strong in one can be absent from another, and an average across the three hides exactly the gap worth acting on. The report breaks the result down by assistant for that reason.
What is a good AI visibility result?
There is no absolute number, because it depends entirely on how many credible suppliers your category has. What is readable is your position relative to the same set of competitors in the same ten questions: named in most answers and named early is strong; named late or only when the question is narrow is a real gap; never named at all in your own category is the finding that matters, and it usually has a structural cause rather than a content one.
Why is my competitor named and not me?
Usually one of three reasons, and they are separable. Either the assistant cannot read your site; a blocked crawler is the single most common cause and takes one line in robots.txt to create. Or it can read you but you never answer the question being asked, so there is nothing to quote. Or a third-party page (a directory, a review site, a forum thread) carries your competitor and not you, and that page is what gets cited. The report shows which of the three is happening by printing the sources behind each answer.
Does being named in AI answers actually bring customers?
It brings fewer visits than a search ranking and, in our own client data, better-qualified ones: the assistant has already done the shortlisting, so the person who arrives has fewer alternatives open. The honest caveat is that this is hard to attribute: many assistants send no referrer, and some send none at all. Treat it as a shortlist position rather than a traffic channel, and measure it as share of the answer rather than as sessions.
Method last reviewed Reviewed by Simon Wilhelm, Co-Founder and CEO of scaile.
SimonCo-Founder and CEO
Not in the answer? Let us fix that.
Related free tools
The others run on their own too. No registration, no credit card, no call required.
Health Check
An instant visibility score across 103 technical, structured-data, AI-crawler and E-E-A-T checks.
Sitemap Visualiser
See your site the way AI crawlers see it: your structure drawn as a map, with the pages they cannot reach.
Frequently asked questions
Need more detail? Book a Call
How long does it take?
The check itself runs in a couple of minutes, and the report is sent the moment it finishes, as a PDF, to the address you gave us.
What do the ten questions look like?
They are buying questions for your category, not brand searches: "which supplier does X for Y", "best X for Y". None of them contains a brand name, including yours, because a question that names a brand answers itself.
Why do you ask what I sell?
Most companies sell several things, and the questions can only be about one of them. Tell us the one that matters and the check measures the market you actually care about. Leave it blank and we read the main offering off your site.
Which assistants does it ask?
Three: ChatGPT and Gemini in their search modes, plus Perplexity, all asked from your own market and in your own language. They disagree more than you would expect, which is why the report breaks the result down by assistant.
What language is the report in?
The language of the page you asked from: English here, German on the German site. The questions themselves are asked in the language your buyers search in, and printed in the report exactly as they were asked.
Is this the same as the Health Check?
No. This measures whether you are named in AI answers today. The Health Check scores whether your site is built to be cited at all.