How Gemini evaluates authorityE-E-A-T is the public framework; specific Gemini ranking is not published.
Google publishes the Search Quality Rater Guidelines (E-E-A-T: Experience, Expertise, Authoritativeness, Trustworthiness) as the canonical public framework for source quality. It does not publish a specific Gemini ranking algorithm separate from this. This page describes the documented surface — Google-Extended, E-E-A-T, and the structural signals operators can verifiably influence — without claiming knowledge of internal scoring.
By Martin Yarnold · UpdatedDocumented framework vs inferred ranking
Documented
Inferred
Frequently asked questions
Does Google publish Gemini's authority-evaluation algorithm?
Not in full. Google publishes the Search Quality Rater Guidelines (the canonical E-E-A-T framework — Experience, Expertise, Authoritativeness, Trustworthiness), which informs Google Search quality and is the most authoritative public guidance on how Google evaluates sources. Google does not publish a specific Gemini ranking algorithm separate from this. Operationally, Gemini and Google Search share retrieval infrastructure and quality signals; AI Overviews surface in Search results when retrieval-augmented answers are deemed appropriate.
Is E-E-A-T the right framework for Gemini visibility?
It is the best public reference. Google's Search Quality Rater Guidelines apply to Search ranking and inform AI Overview and Gemini behaviour because Gemini grounds against Google's retrieval surface. The four E-E-A-T pillars — Experience, Expertise, Authoritativeness, Trustworthiness — are signals Google explicitly says quality raters use to assess sources. Specific ranking weights inside Gemini are not published, but the E-E-A-T framework is the closest public documentation operators have.
How does Google-Extended relate to Gemini?
Google-Extended is a robots.txt usage-control token (with no HTTP fetcher of its own) that controls whether crawled content may be used for Gemini Apps and Vertex AI Gemini training and grounding. The crawling itself is performed by Googlebot. Disallowing Google-Extended removes content from Gemini Apps / Vertex AI Gemini grounding and training but does not remove pages from Google Search and does not by itself remove pages from AI Overviews or AI Mode in Search — those are Search features controlled separately via Googlebot plus preview controls (nosnippet, data-nosnippet, max-snippet, noindex). Disallowing Googlebot removes content from Google Search.
Which structural signals can operators verifiably influence?
For Gemini Apps / Vertex AI Gemini grounding eligibility: allow Google-Extended in robots.txt. For Google Search ranking (which includes the underlying retrieval for AI features in Search): allow Googlebot, and use Google's preview controls (nosnippet, data-nosnippet, max-snippet, noindex) to manage AI Overview and AI Mode inclusion at the snippet level. Independent of either: valid JSON-LD structured data, complete metadata, substantive E-E-A-T-aligned content, and clean internal linking. These are documented by Google or are industry-standard. Specific Gemini or AI Overview ranking weights beyond these are not published.
Does AI Overview citation behaviour differ from the Gemini app?
Observably yes. AI Overviews (in Google Search) tend to cite sources more visibly — there is a documented source-card surface. The standalone Gemini app produces answers that sometimes cite, sometimes do not, depending on whether the query triggered retrieval. Google does not document a complete behavioural contract for either surface. Treat AI Overviews as the higher-citation-visibility surface and the Gemini app as conditional.
How do I measure Gemini citations?
Two surfaces to measure separately. (1) AI Overview presence in Google Search — easier; tools like Semrush surface this. (2) Gemini app citation rate — manual against a fixed query set, checking for source-card appearance per query. Promagen Sentinel automates structural-signal measurement (Google-Extended access, schema validity); the per-query AI Overview / Gemini app measurement is currently best done manually or with specialised tools.