Free tool
Reads the URL the way a crawler that does not run JavaScript would. No page is stored, no model reads your content, and the scoring rules are published.
GEO Score
Answer engines quote sentences, not pages. This counts the statements on a page that could be lifted and attributed, and tells you which ones could not — scored the same way for every page, with no model in the loop.
Arithmetic, not opinion
Every number is a count of something on your page. The same page always scores the same.
The rules are published
Every weight, threshold and cap is readable at the API. A tool that asks you to show your work should show its own.
A shell scores nothing
If the page returns markup with no text, you get no score and a reason — not a low one dressed up as a verdict.
How the score is built
Each category is a count of something on the page, expressed as a percentage. The weights below are fixed and do not change with the kind of page you scored.
25%
Citability — the share of statements short enough to lift whole and long enough to say something
20%
Structure — one H1, sections that earn their headings, and at least one list
20%
Definitions — sentences that name their own subject, with a concrete example nearby
20%
Data and citations — figures with units, named entities and outbound sources
15%
Takeaways — a summary section an engine can find, and a closing passage worth quoting
Method
The page is fetched once, server-side, with no JavaScript executed — the same view an AI crawler gets. Site navigation and footers are removed before anything is counted, because a nav label is not a statement.
What it will not tell you
Whether a claim on the page is true, and whether any engine has actually cited you. This measures how liftable your sentences are, which is a precondition for citation and not a substitute for it.
Check our working
Every weight, threshold and cap the score uses is published, and the report is available as JSON at the same address. Nothing about the scoring is hidden.