Free tool · AEO and GEO readiness
Can AI Engines Actually Read, Verify and Cite Your Page?
Paste a URL. In a few seconds this AEO checker runs 32 checks on whether ChatGPT, Gemini, Claude and Perplexity can reach your page, work out what your business is, and lift an answer out of it. It is an AI search readiness test, not a ranking report. The score and the failures are free, with no email required.
Run it on a page you actually want an assistant to cite, such as a service or guide page. A contact page will score poorly and that is not a fault. Nothing is stored unless you ask for the detailed report.
AI visibility result
Now test it yourself, in two minutes
This tool measures whether AI engines can read and verify you. Whether they currently do mention you is something only the assistants can answer, so here are five questions written from what is on your page. Edit them to match how your buyers actually ask, then paste them into ChatGPT, Gemini, Claude or Perplexity and see who gets named.
Want the full breakdown?
The score and the failures above are yours either way. If you want every check with the specific fix for each one, send it to your inbox and we will also take a look ourselves.
What This Tool Can and Cannot Tell You
Most AI visibility tools answer the question you want answered by guessing. This one tells you where the line is.
What it measures
- ✓Whether AI crawlers are allowedChecks GPTBot, OAI-SearchBot, ClaudeBot, PerplexityBot and Google-Extended against your robots.txt for this exact URL, then splits the answer by assistant so you can see whether ChatGPT, Perplexity, Google and Claude are each allowed in.
- ✓Whether your server actually serves themRequests the page as GPTBot and compares what comes back with what a browser gets. Firewalls block crawlers that robots.txt permits.
- ✓Whether you are machine-readableStructured data, business identity, sameAs links, entity cross-references and author markup.
- ✓Whether content survives without JavaScriptMost AI crawlers do not run scripts. If your copy is injected client side, they see an empty shell.
- ✓Whether answers can be liftedQuestion-shaped headings, a direct answer under each one, definitions, lists and tables.
- ✓Whether the page is worth citingNamed author, dates, outbound citations, sourced statistics and contact details.
What it will not pretend to know
- ✗Whether ChatGPT mentions you todayThat needs live API calls to every assistant on every scan. It costs real money per check and would be trivially abused on a public page. We would rather hand you the questions than invent a number.
- ✗Your position in AI answersThere is no stable ranking to measure. Answers vary by phrasing, by account, by session and by model version, often within the same hour.
- ✗How often you are mentioned across the webThat needs a paid third-party index. Any tool showing you this for free is either estimating or selling your data.
So the tool scores readiness, then writes you five questions from your own page. Paste those into ChatGPT, Gemini, Claude or Perplexity and you will know within two minutes whether you get named. That is the honest version of the answer everyone wants.
What gets scored
The Five Things It Scores
Thirty two checks, weighted so the ones that actually stop a citation count for more.
AI Crawler Access
Eight checks on whether the major AI crawlers can reach this URL at all, including a live request made as GPTBot and a look for llms.txt. This is the group most sites fail without knowing.
Machine-Readable Identity
Seven checks on your structured data: is it present, does it parse, does it declare a real business, and do the entities reference each other so an engine can resolve who you are.
Readable Without JavaScript
Four checks on whether the content exists in the raw HTML or only appears once a browser runs your scripts, and how much of what you serve is content rather than markup.
Answer-First Structure
Seven checks on whether the page is shaped like something an assistant can quote: real questions, short answers directly beneath them, definitions, lists and tables.
Citability and Trust
Six checks on the signals that make a page safe to cite: a named author, visible dates, outbound sources, sourced numbers, contact details and a consistent business name.
Three Things Can Block an AI Crawler, and Most People Check None of Them
This is the part of the check that surprises people. A site can be perfectly built, rank well, and still be completely invisible to an assistant, because something between the crawler and the content said no.
- ›robots.txtThe file most people set once and never revisit. A single Disallow line aimed at GPTBot removes you from ChatGPT, and nothing in your analytics will ever tell you.
- ›Your firewallCloudflare, Sucuri and most bot-fighting plugins block AI crawlers by default, silently, regardless of what robots.txt says. This is the one that catches people out, because the site owner has done everything right and is still invisible.
- ›Your CMSSome hosts and security plugins serve a challenge page to anything that is not a known browser. A crawler sees the challenge, not your content, and moves on.
The checker tests all three. It reads your robots.txt for this exact URL, and then makes a real request as GPTBot and compares what comes back against what a browser receives. If those two differ, something is filtering, and the tool will say so.
What Your AEO and GEO Readiness Score Means
85 and above
Strong
The page is readable, verifiable and quotable. Keep it current, and put your effort into being mentioned elsewhere on the web rather than into more on-page work.
70 to 84
Workable
The foundations are sound. Work through the failures in order of weight. Usually it is the answer structure, not the technical setup.
50 to 69
Patchy
An assistant can see the page but struggles to confirm what you are. Fix crawler access and structured data before touching the copy.
Below 50
At risk
There is very little an assistant could safely say about you from this page. Start at the top of the list and work down.
AI Visibility Checker: Questions
Does this tell me if ChatGPT mentions my business?
No, and no free tool can do that honestly. Checking it properly means calling the OpenAI, Google, Anthropic and Perplexity APIs on every scan, which costs money per check and would be abused within a day on a public page. What this tool does instead is measure whether those systems can read, verify and quote you, then hand you five questions written from your own page. Paste them into the assistants yourself and you have the real answer in two minutes.
Can ChatGPT see my website?
Usually yes, but not always, and the failure is silent. Three separate things can stop it: a Disallow line in robots.txt aimed at GPTBot or OAI-SearchBot, a firewall or CDN rule blocking AI user agents regardless of robots.txt, or a site that renders its content in JavaScript that the crawler never runs. This checker tests all three, including a live request made as GPTBot so you can see what a crawler actually receives rather than what your settings claim.
What is a good AI visibility score?
Above 85 is strong. 70 to 84 means the foundations work and the gaps are in how the content is structured. Below 50 means an assistant has very little it could safely say about you. The score is weighted, so a failure on crawler access or structured data costs far more than a long average sentence.
Which page should I test?
A page you genuinely want an assistant to cite: a service page, a guide, a comparison. Testing your contact page or home page will give you a low score that is not really a fault, because those pages are not trying to answer a question.
Why does my page score badly when it ranks well on Google?
They measure different things. Google rewards authority, links and relevance. An assistant has to understand and restate you, so it needs structured data, a clear entity, and answers it can lift without context. Plenty of pages that rank on page one are almost unusable to an assistant.
Is blocking AI crawlers always a bad idea?
No. If your content is your product, or it sits behind a paywall, blocking is a legitimate commercial decision. The problem is doing it accidentally. Most of the sites we scan that block AI crawlers never chose to, usually because a security plugin or CDN did it for them.
What is llms.txt and do I need one?
It is a plain text file at the root of your site naming what you do and linking your most useful pages. It is an emerging convention rather than a standard, and no engine is confirmed to depend on it. It takes twenty minutes and costs nothing, so we treat it as cheap insurance, not a ranking factor. We score it as a warning rather than a failure for that reason.
Do you store the URL I scan or the results?
The scan result is cached on our server for thirty minutes so that repeat checks are fast, and then it is gone. Nothing is attached to you unless you ask for the detailed breakdown and give us your email. If you do, we see your name, email, the URL and the score.
Can I save or share the result?
Yes, and neither needs an email address. The result panel has a copy-link button. The link carries the URL you tested, so anyone you send it to gets the same check run fresh against that page, which also means the link never goes stale. Next to it is a print button that produces a clean PDF through your browser, with the input box, the form and the buttons stripped out.
How is this different from a normal SEO audit?
An SEO audit asks whether Google can rank you. This asks whether an assistant can understand and repeat you. There is overlap on the technical basics, but the parts that matter most here, entity resolution and answer structure, barely feature in a standard SEO report. We have a separate SEO checker if you want the traditional view.
Can I check a competitor?
Yes, it works on any public URL. It is a reasonable way to see whether the firms getting named in AI answers are doing something you are not. Often they are simply easier to parse.
The tool says my firewall is blocking crawlers. What now?
In Cloudflare, look under Security for bot-fighting mode and any WAF rule matching user agents. Most security plugins have a bad-bot or crawler list where the AI agents have been added by default. Allow the ones you want citing you, then re-run this check to confirm the block is gone.
Scored Badly and Want It Fixed?
Send us the result. We will tell you which failures are worth paying to fix and which you can do yourself in an afternoon, before anyone talks about a retainer.