geo-content-auditlisted

# GEO Content Audit Principles distilled from GEO research (Aggarwal et al., KDD 2024), Ahrefs' citation-length studies, and generative-search verifiability research (Liu et al., 2023), applied to check whether website content **will get cited by AI** and, if cited, **will get restated correctly** — with concrete, actionable fixes. Every cited figure's original source is listed under Sources at the end. **Output language**: Write the report in the language the user is conversing in; default to English. ## Scope: content-semantics layer only This skill focuses on **how the words themselves are written** — paragraphs, phrasing, information completeness, evidence quality. The following technical-layer items are **never checked, scored, or written into the report**: robots.txt and crawler accessibility, JSON-LD/schema correctness, meta tags, sitemap, llms.txt, performance. Those are better handled by rule-based tools (e.g. the geo-optimizer CLI) — an LLM guessing at them item by item is more likely to get it wrong. The one exception: if the audit **incidentally** surfaces a technical-layer dealbreaker that would undermine everything else (e.g. the whole site blocks AI crawlers), add a single-line note at the very end of the report flagging it — "Also noticed: ..., recommend checking with a technical tool" — without expanding on it or folding it into the scorecard. The audit target is "the visible text a non-JS-executing crawler would see." Text inside JSON-LD counts only as