Skip to main content
Where Site Audit asks is this site technically sound?, GEO Audit asks is this site something an AI engine would confidently cite? Different question, different findings.

​
Running it

GEO Audit runs automatically the first time you open the page for a site that’s never been audited. After that, Re-run starts a fresh scan. A running scan shows live progress; the finished report can be downloaded. If a workspace has no site yet, the page prompts you to add one first.

​
What it scores

The report opens with an overall citability score and a band — Excellent, Strong, Fair or Needs work — then breaks the score into sections.
The mechanics an answer engine depends on: clean markup, structured data, and content it can parse without executing your application.
Whether the crawlers behind the major AI platforms are permitted to fetch your pages, and whether anything in your configuration blocks them.
How well your content is shaped for the way each platform assembles answers.
The corroborating pages that make a brand safe for a model to cite: an About page, contact information, a LinkedIn presence and similar signals. Models prefer sources they can verify.
Whether your brand exists as an entity where models look it up — Wikipedia, Wikidata, Crunchbase. Missing entries are called out with a Claim listing action.
Which intent types your content covers, shown on a radar chart. Gaps here map directly to prompts you’re likely to lose.
Content decay — pages stale enough that a model may prefer a more recently updated source.
Things that actively work against you:
  • Hallucination bait — vague or contradictory claims a model may reword into something wrong.
  • Keyword stuffing — patterns that read as manipulation.
  • Prompt-injection surface — page content that could be interpreted as instructions by a model reading it.
The ranked list of changes that would most improve your citability. Start here.

​
Using the results

1

Clear the blockers

AI crawler access and prompt-injection surface first. Nothing else matters if engines can’t or won’t read you.
2

Build the trust stack

An About page, real contact details and a knowledge-graph entry are cheap wins with outsized effect.
3

Close intent coverage

Use the radar chart to find the intents you don’t answer, then send those to the Blog Agent.
4

Re-run and compare

Re-run after publishing and watch the band move.

​
If a scan fails

A network or scan error stops the audit and the page offers Retry. A previously completed report stays available in the meantime.

AI Visibility (GEO)

Readiness is the input; visibility is the outcome.

Blog Agent

Fill the intent coverage gaps this audit found.