AI Visibility Score
Paste your URL. This checker fetches your site the way an answer engine does, with no JavaScript, and scores what comes back out of 100. You get the number, the five things behind it, and a specific fix for every point you lost. Free, and no email.
Takes about ten seconds.
What to fix
What gets checked
Five bands, weighted by how much each one can cost you. The scan reads only what your server returns. No JavaScript runs, because for the crawlers that matter, none does.
Retrievability
30 pointsIs there any text in the HTML your server sends, before a single script runs? Are the answer-engine crawlers allowed in? Is there a sitemap that parses? This band carries the most weight because a site can pass every other check on this page and still hand a crawler an empty document. We have watched that happen on two live sites that scored 100% on their own schema and robots audit.
llms.txt
15 pointsPresent is not the test. The test is whether the file describes your site. A file that returns a clean 200 and talks about your website builder instead of your business scores zero here. Any check that only asks whether the file exists will call it fine. No engine has confirmed it reads llms.txt, and we say so on the report.
Answerability
25 pointsOne H1. Headings shaped like the question a buyer actually asks. An FAQ block whose visible text matches its markup. A first paragraph that stands on its own, because that is the chunk most likely to get lifted into an answer.
Structured data
20 pointsJSON-LD that parses, and the four blocks worth having: who publishes the site, the site itself, an FAQ, and a type for the page. A block with a trailing comma is discarded silently, so it does nothing while looking present. The report tells you which ones you actually have.
Citability
10 pointsCanonical URL, Open Graph tags, a named author, a date. Small things, but an unattributed and undated page is a weaker source than a signed one, and engines weigh that.
What this score doesn't tell you
It doesn't tell you that traffic is coming. Most AI answers resolve inside the chat window and send nobody anywhere. A site can be cited hundreds of times in a month and see almost no visits from it. That is normal, and any tool that implies otherwise is selling you something.
What a good score does mean is that you can be read and quoted accurately. That is the precondition. If a crawler gets an empty document, nothing else you do about AI search matters, because there is nothing to cite.
Once you clear that bar, the next question is a content one: does your page give an assistant something it can't just absorb and paraphrase away. A calculator, a checker, a dataset, a real number nobody else publishes. That's the part worth a conversation.
Frequently asked questions
What is an AI visibility score?
It's a 0 to 100 read of whether an AI assistant can fetch your site, read it, and quote it. Five things get scored: whether the page is retrievable at all without JavaScript, whether your llms.txt actually describes your site, whether the content is shaped like an answer, whether your structured data is present and valid, and whether the page can be attributed to someone. The scan reads only what your server returns, which is what the crawlers read.
Do AI crawlers run JavaScript?
Mostly no. Google renders JavaScript. The answer-engine crawlers generally do not, and none of them promise to. If your text only appears after a script runs, you're betting your visibility on behavior nobody has committed to. This is the most common way a site scores well on every other checker and still can't be quoted.
Does llms.txt actually do anything?
No major engine has confirmed it reads the file. It's scored here for two reasons. It's the only place you get to state which pages you want quoted, and a wrong one is worse than none. A file that returns a clean 200 and describes your website builder instead of your business scores zero on this checker. Any check that only asks whether the file exists will call it fine.
Should I block GPTBot?
That's a real choice and this tool doesn't penalize it. Training crawlers like GPTBot, ClaudeBot and CCBot collect content for model training. Answer-engine agents like OAI-SearchBot, Claude-SearchBot and PerplexityBot fetch a page because a person asked a question right now. Blocking the training crawlers doesn't remove you from live answers. Blocking the answer engines does. People conflate the two constantly.
Will a better score bring me more traffic?
Not on its own. Most AI answers resolve inside the chat window and never send a click. A good score means you can be read and quoted, which is the precondition. Whether anyone leaves the chat to visit you is a separate problem, and it usually comes down to offering something the assistant can't just summarize away.
How is this different from an SEO audit?
An SEO audit asks whether Google will rank you against everyone else. This asks a narrower question first: can a machine read your page and repeat it accurately. A site can be well optimized for Google, render entirely in the browser, and be invisible to every assistant that fetches it.
Want the fixes done, not just listed?
The scan tells you what's broken. A call tells you which two things are worth your time and which eight are noise. 30 minutes, no pitch deck.
Book a CallThe front door at Veteran Vectors is done-for-you lead generation: I book you 30 qualified sales calls in 60 days on LinkedIn. Guaranteed. You record short daily videos at a pace we set on your audit call, I run everything else, and if I miss I keep working free for up to 30 more days. Search visibility work like this is what we build once your pipeline is full.
See how lead generation works →