Skip to main content
The GEO Score is a 0–100 measure of how well your website is technically optimised for AI-powered search engines — ChatGPT, Perplexity, Gemini, Claude, Grok, and Google’s AI surfaces (AI Overviews and AI Mode). It evaluates 15 signals across 3 pillars — Accessibility, Readability, and Understandability — to determine whether AI crawlers can find, parse, and comprehend your content.
The GEO Score measures technical readiness, not actual AI visibility. A brand can score low on GEO Score but still be highly visible in AI answers due to brand authority. For actual visibility measurement, request a GEO Audit.
Each signal links to a detailed reference page covering exactly what is measured, how it is scored, how to verify it independently, and how to fix it.

​
The 3-Pillar Framework

​
Accessibility (33 points)

Can AI crawlers find and access your content?

​
Readability (37 points)

Can AI parse and understand your content’s structure?

​
Understandability (30 points)

Can AI extract meaning, context, and authority?

​
Scoring

The GEO Score is calculated by summing all 15 signal scores. Each signal is scored independently — the total is additive, not weighted or gated.

​
Score Bands

​
What the GEO Score does NOT measure

  • Actual AI visibility — whether your brand is mentioned in AI-generated answers
  • Brand authority — how well-known your brand is to AI models
  • Content quality — whether your content is accurate, helpful, or well-written
  • Competitive position — how you compare to competitors in AI citations
For these dimensions, request a GEO Audit, which queries AI platforms directly and measures real mention rates.

​
How scanning works

The scanner runs in three stages:
  1. Static fetch. robots.txt, sitemap.xml, llms.txt, and the homepage are fetched directly via HTTP.
  2. Browser crawl. A headless browser renders up to 5 pages and extracts DOM-level signals — headings, internal links, author info, structured-data blocks, content patterns.
  3. Signal evaluation. Each of the 15 signal analysers runs against the collected page data, and the scores are summed.
Page selection uses hierarchical sampling so the crawl reflects what a real visitor (or AI crawler) would encounter: the homepage, a blog/content page, a product/service page, an about/team page, and one additional page discovered via sitemap or internal links. If the browser stage can’t render a page (rare, but it happens — heavy bot protection, network failures), the signals that depend on rendered DOM are flagged as unscored rather than counted as failed. The denominator of your score is reduced accordingly, and a warning is surfaced on the result so you know which signals were skipped. A low score never silently includes unmeasured signals as zeros.

​
How to improve your score

Run a free scan at getcited.in/geo-score to see your score breakdown by pillar. The result identifies your top failing signals and links straight to the per-signal reference page for fix guidance.