How it works
Six checks. One score.
We fetch your site the way AI crawlers do — then grade what they find.
Crawler access
What your robots.txt says, and what crawlers actually get. We read your robots.txt to see which AI crawlers you've told to stay out: GPTBot, ClaudeBot, PerplexityBot and the rest. Then we fetch your page and check whether each one can really get in. Your report shows both side by side, so you can spot a site that welcomes AI on paper but blocks it in practice.
Structure
One main headline, sub-headlines in order, and the page's sections labeled — header, menu, main content, footer — so AI can tell what the page is about and pull answers from it.
Metadata
Your page title and description, the one web address you want the page known by, how the page looks when it's shared, and your business facts in a format machines read (schema) — the signals AI uses to understand and quote a page.
Content
Content depth, and whether your about, contact, and pricing pages exist and are linked — what gives AI something worth citing.
FAQ signals
Question-and-answer content with FAQ schema — the format AI assistants cite most.
Performance
Load speed and page weight — slow pages get crawled less and trusted less.
New to these terms? Our help guides explain each one in plain English. See the help guides →
A scan, step by step
We fetch your page
We fetch your page with a plain HTTP request, the way AI crawlers do, and score what that fetch returns.
We read your robots.txt
We read your robots.txt for the ten AI crawlers we test, then check with a live fetch whether each one can really get in.
We also render the page
We also render the page in a real browser and compare it with the plain fetch, so your report shows how much of your content depends on JavaScript.
We score the six checks above, each 0–100, and combine them into one 0–100 score.
The weights are on the methodology page.
You get the score and every finding ranked by impact.
A token is used only when the scan completes.
The report
Ranked by impact.
Your report opens with a 0–100 score, then lists every finding in priority order. The biggest wins come first — each with the exact change to make and, where useful, ready-to-paste code like robots.txt rules or JSON-LD schema. One token pays for one completed scan. If a scan errors out, the token stays in your balance.