AI SEO & GEO Audit

What is a GEO audit?

Generative Engine Optimization (GEO) is the practice of making a page easier for AI systems — chat assistants, answer engines, and AI-powered search features — to crawl, understand, and potentially reference. A GEO audit checks the technical and structural signals that make that possible: whether AI crawlers can actually reach the page, whether its content is readable without executing JavaScript, whether it declares clear structured data about who published it, and whether it's organized the way answer engines tend to extract information (clear questions, direct answers, concise summaries). See our GEO vs SEO and Generative Engine Optimization guides for more background.

What this tool checks

This tool runs 31 automated, technically verifiable checks against a single URL, grouped into six categories: AI Crawlability (robots.txt rules for AI crawlers, noindex, HTTPS, canonical), Machine Readability (server-rendered content, title/description, charset, language, internal links), Entity & Structured Data (JSON-LD validity and the entity/content types it declares), Citation Readiness (author, dates, references, organization signals), Answer-Engine Content Structure (headings, questions, FAQ blocks, intro paragraph), and LLM Discovery Files (llms.txt / llms-full.txt). Every check is fully automated against real, fetched data — nothing here is a manual placeholder, and nothing here queries any AI model directly.

AI crawler access

The centerpiece check parses robots.txt for rules affecting Googlebot, GPTBot (OpenAI), ChatGPT-User (OpenAI's on-demand browsing fetcher), ClaudeBot and Claude-SearchBot (Anthropic), PerplexityBot (Perplexity), and Google-Extended — a separate control for whether certain Google AI training and grounding uses (Gemini model training, AI Overviews grounding) may use this content; it does not affect Google Search crawling or indexing, which Googlebot's own rule controls. The audit reports three distinct outcomes per crawler — blocked (an explicit Disallow rule won), allowed (an explicit Allow rule won, or nothing in the crawler's rule group disallows this path), or no specific rule found (the crawler isn't named anywhere in the file, including no wildcard fallback) — and deliberately does not treat "no rule found" as confirmed permission. This reflects how this tool's own parser interprets the file, using the precedence rules Google's robots.txt spec documents; other crawlers are not guaranteed to apply identical precedence or tie-break rules, and robots.txt directives are advisory — this cannot confirm how any specific crawler actually behaves once it fetches the page.

Structured data and entity signals

AI systems rely heavily on explicit, machine-readable signals rather than inferring everything from prose. This audit checks for valid JSON-LD, the schema.org types it declares (Organization, WebSite, Person, Article, BlogPosting, FAQPage, BreadcrumbList, Product, LocalBusiness), and whether an Organization/Person entity includes a name and sameAs links to official profiles. It never claims Google Rich Results validation — only that the structured data parses and declares a recognized type.

Citation readiness

These checks look for deterministic, pattern-based "citation readiness signals" — a visible author/byline, a publication or modified date, outbound reference links with descriptive anchor text, an About/Contact path, and schema author/publisher attribution. This does not determine E-E-A-T or actual trustworthiness, and presence of these signals does not mean the content is accurate — only that common attribution and sourcing patterns are present.

Machine readability

Most AI crawlers do not execute JavaScript, so this audit checks how much meaningful text is present directly in the server-returned HTML (not after client-side rendering), alongside charset, language, title/description, and internal linking. A page that appears mostly empty in its raw HTML is flagged as a warning that "content may depend heavily on client-side rendering" — not a confirmed statement that any specific crawler failed to render it.

llms.txt, honestly

llms.txt and llms-full.txt are an emerging, community-proposed convention for giving AI/LLM tools a curated, linked summary of a site. They are not an official ranking or citation factor for any AI system — no major AI provider has confirmed using them that way. This audit checks whether the files exist, are non-empty, and look like a real maintained manifest (markdown links and headings) rather than a placeholder. A missing file is scored as the lowest-severity "opportunity," worth 5% of the overall score, specifically so it can't meaningfully damage a page's score.

GEO vs traditional SEO

Traditional SEO optimizes primarily for search engine ranking algorithms and human readers browsing a results page. GEO adds a parallel goal: making content easy for an AI system to fetch, parse, and potentially summarize or cite directly in a conversational answer — often without the reader ever visiting the page. The two overlap heavily (crawlability, structured data, and clear content structure help both), which is why several checks here are shared with our SEO Audit tool. For a deeper comparison, see GEO vs SEO, AEO vs SEO, and GEO vs AEO.

How to improve AI search readiness

Start with anything marked Critical — these are crawl or index blocks that make every other signal irrelevant for the affected crawler. Then work through Warnings (typically a blocked AI answer-engine crawler), followed by Opportunities (entity, citation, and content-structure gaps). Use the linked tool on each finding — Robots.txt Generator, Schema Markup Generator, FAQ Schema Generator, Google Index Checker, and Internal Linking Assistant — to act on each finding directly.