Use this when
geo, ai citation, ai optimization, citation audit, aeo, chatgpt citation.
Process
This is one SEO discipline with classic Google search, not a separate playbook with its own rules — Google's own 2026 guidance frames generative-AI optimization as ordinary SEO: no special markup or llms.txt requirement for Google visibility.
Score across five categories, each checked against concrete criteria, then scaled to a 0-100 display score:
- Evidence-backed citability (raw /4, scaled to /27): what share of sections between headings are context-independent (make sense extracted alone), contain a specific claim plus supporting evidence plus source attribution, and answer a question completely without needing adjacent sections. 80%+ of sections = 4pts, 60-79% = 3, 40-59% = 2, 20-39% = 1, below 20% = 0.
- Purpose fit and reader utility (raw /3, scaled to /20): clear purpose in the intro, sections that state their point without throat-clearing, and heading/format choices that actually fit the material.
- Entity clarity (raw /3, scaled to /20): one unambiguous primary topic, consistent entity naming throughout, a clear topic statement in the intro, and a title that matches the content focus.
- Content structure for extraction (raw /3, scaled to /20): standalone summary, comparison tables with real headers, ordered lists for processes, clear definition formatting, and evidence-backed explanations for reusable claims — 4-5 present = 3pts, 3 present = 2, 1-2 present = 1, none = 0.
- AI crawler accessibility (raw /2, scaled to /13): content and schema present in the rendered DOM and reachable by the target crawler; normal Google crawlability/indexability; for non-Google visibility, robots.txt treatment of GPTBot/ClaudeBot/PerplexityBot matching the site's stated goal.
Rating bands: 90-100 Excellent, 70-89 Good, 50-69 Needs Work, below 50 Poor.
Give platform-specific notes rather than one blanket recommendation: for ChatGPT, check that claims are source-backed and useful independent of format; for Perplexity, check crawlability and source fidelity, especially for time-sensitive queries; for Google AI Overviews and AI Mode, apply ordinary SEO hygiene (helpful, crawlable, indexable) rather than any special markup, and measure Search and AI-feature visibility separately since organic overlap in a sample doesn't guarantee inclusion; for Claude/Gemini/Copilot/You.com, evaluate clarity, source accessibility, and documented crawler-access policy.
Never cite a numeric AI-citation benchmark without a full source block (URL, publisher, methodology, sample size, retrieval date); if any field is missing, label the number directional or drop it.
Report the score breakdown, a per-section citability table, platform-specific recommendations, and up to three self-contained evidence improvements for the weakest sections.