The four pillars
Each pillar answers a different question an AI engine asks before it cites you. Each is scored 0–100.
Technical
Can AI bots reach your pages? Do they see real content rather than empty React shells? Are schema.org markers present? Is the site fast enough to crawl deeply?
Content
Do your headings answer real questions? Are direct answers above the fold? Do you cover the topic comprehensively across related sub-questions?
Entity
Brand mentions, About page, author bios, schema.org Organization markers. AI engines need to know who you are before they'll cite you.
Trust
E-E-A-T signals. Author bylines with credentials, references to authoritative sources, third-party mentions of your brand. AI engines won't quote sources they don't trust.
How the GEO score is built
The four pillar scores are weighted equally, 25% each, into a single 0–100 GEO score. Below 60 on a pillar needs work, 60–80 is OK, and 80+ is strong. Because every scan uses the identical model, an Index score and your own free-scan score are directly comparable.
Alongside the GEO score we compute two sub-scores: SEO (how well the page would rank in classic blue-link search) and AEO (Answer Engine Optimization - how cleanly an engine could quote the page verbatim). They share signals with GEO but are reported separately so you can see where the gap is.
Citation rate is measured, not estimated: it's the share of tested buyer queries where an AI engine cites your own domain - a link or explicit reference - across ChatGPT, Perplexity, Claude and Google AI Overview. We measure citation, not mention.
What the score does - and doesn't - claim
A GEO score is a point-in-time measurement of the signals on your site. It is not a guarantee of a citation: AI engines change their models and retrieval continuously, so a snapshot is exactly that. Scores and placement cannot be bought - the only way the number moves is to ship the structural fixes AI engines reward.
We report the framework openly because a methodology only earns trust if it survives being challenged. If a score looks wrong, the fastest way to test it is to run the scan and read the per-pillar findings.
Frequently asked questions
What is the CITE framework?
CITE is Osoro GEO's scoring model for AI citation readiness. It measures four signal groups - Technical, Content, Entity and Trust - and combines them into a single 0–100 GEO score. Each pillar answers a different question an AI engine asks before it cites a source.
How is the GEO score calculated?
Each of the four CITE pillars is scored 0–100, and the pillars are weighted equally at 25% each to produce the overall GEO score. Below 60 on a pillar needs work, 60–80 is OK, and 80+ is strong.
How is this different from an SEO score?
SEO measures whether you rank in the blue-link results. GEO measures whether AI engines cite you in their answer. They overlap, but a site can rank #1 on Google and still be invisible in ChatGPT - because AI engines weight question-led structure, entity clarity and trust signals that classic SEO never scored.
Which AI engines does Osoro GEO test?
ChatGPT, Perplexity, Claude and Google AI Overview - the four answer engines your buyers actually use. Citation rate is the share of tested buyer queries where an engine cites your own domain, not merely mentions your name.
Can you buy a higher GEO score?
No. The score is a measurement of your site's own signals; placement and scores cannot be bought. The only way the number moves is to ship the structural fixes - schema, answer-shaped content, entity clarity, trust references - that AI engines actually reward.
Score your own site on all four pillars
Free, about 30 seconds, no account. You'll get your GEO score, the four pillar scores, and the prioritised fixes that move each one.