How Do You Audit a Website to See If AI Search Engines Can Cite It?

What does an AEO audit cover that SEO doesn't?

An AEO audit evaluates whether AI search engines can access, understand, and cite your content — a fundamentally different question from whether Google can index and rank it. SEO audits check crawlability for Googlebot, keyword targeting, and link equity. AEO audits check whether GPTBot, ClaudeBot, and PerplexityBot receive usable HTML, whether schema enables entity resolution, and whether external signals corroborate your authority.

A site can score 95/100 on a traditional SEO audit and fail an AEO audit completely — for example, a React SPA with strong backlinks but no server-rendered content. A 2025 Ahrefs study found that 44% of sites ranking in Google's top 10 for informational queries were not being cited by any AI search engine.

The 6-dimension scoring framework

This framework assigns weighted scores across six dimensions, totaling 100 points. Citability carries the highest weight because it measures the end goal directly — whether your content is structured in a way AI engines can extract and attribute. The weights reflect observed correlation with actual AI citation rates across 1,500 sites in a SCALEBASE benchmarking study.

Each dimension is scored 0-100 internally, then multiplied by its weight to produce a final composite score. In the benchmarking study, sites scoring above 70 composite were cited by at least one AI platform. Sites scoring above 85 were cited by three or more platforms. The median score across all audited sites was 41.

Step-by-step AEO audit process

A thorough AEO audit follows a fixed sequence: access first, then structure, then signals. Starting with access prevents wasting time analyzing content that crawlers cannot reach. Expect 4-6 hours for a 50-page site. The process below covers each dimension in order.

Phase 1: AI crawler access (30-45 minutes)

• Check robots.txt for Disallow rules targeting GPTBot, ClaudeBot, PerplexityBot, or wildcard blocks
• Fetch 5 key pages with curl and verify the HTML contains full content, not empty divs
• Test with AI crawler user agents: curl -H 'User-Agent: GPTBot' — confirm 200 status and same content
• Check for llms.txt at /llms.txt and /llms-full.txt
• Verify XML sitemap is accessible and contains all canonical URLs

Phase 2: Content structure (60-90 minutes)

• Review 10 key content pages for question-based H2 headings
• Check whether each H2 section opens with a direct 40-60 word answer
• Look for TL;DR or summary boxes at the top of long-form content
• Count data points per section — target at least one concrete statistic, date, or measurement per H2
• Evaluate FAQ sections: do they exist, do they match real user questions
• Assess content length: 1,200-2,000 words for pillar content, 600-1,000 for supporting pages

Phase 3: Schema and structured data (30-45 minutes)

• Validate JSON-LD on 5 key pages using Schema.org Validator
• Confirm Organization schema with sameAs links on the homepage
• Check Article schema with author, datePublished, dateModified on content pages
• Look for FAQPage schema on pages with FAQ sections
• Test speakable property presence and CSS selector accuracy
• Verify all schema is server-rendered (curl test, not browser test)

Phase 4: Entity signals (45-60 minutes)

• Search Wikidata, Wikipedia, LinkedIn, Crunchbase for the brand
• Check 3-5 industry directories relevant to the company's vertical
• Google the brand name excluding the company's own domain
• Count total distinct entity signals and score completeness
• Identify disambiguation issues if the brand name is shared with other entities

Phase 5: Platform-specific checks (30-45 minutes)

• Query the brand name and key topics in ChatGPT, Perplexity, Gemini, and Google AI Overviews
• Document which platforms cite the brand, which mention it without citation, and which ignore it
• Check citation accuracy — is the information attributed correctly
• Note competitive citations: which competitors appear for the same queries

Phase 6: Scoring and report (30 minutes)

• Score each dimension 0-100 based on findings
• Calculate weighted composite score
• Document critical findings (blockers), high-priority improvements, and medium/low optimizations
• Prioritize actions by impact-to-effort ratio

How to prioritize findings

Audit findings fall into four tiers based on their impact on AI citability. Critical findings are blockers — they prevent AI engines from accessing content entirely. Address these before anything else. High-priority findings significantly reduce citation likelihood but do not block access completely.

In a typical audit, 15% of findings are critical, 30% are high, 35% are medium, and 20% are low. The most common critical finding across SCALEBASE audits is AI crawler blocking via robots.txt — present in 28% of all audited sites. The second most common is JavaScript-only rendering with no server-side content.

Free tools for a basic AEO audit

A basic AEO audit does not require paid tools. The following free tools cover approximately 70% of what a full audit examines. The remaining 30% — entity signal depth, competitive citation analysis, and platform-specific optimization — requires manual research or specialized tooling.

• curl (command line) — fetches raw HTML to verify SSR, schema presence, and AI crawler access. The single most important AEO audit tool.
• Google Rich Results Test (search.google.com/test/rich-results) — validates schema syntax and rich result eligibility. Note: executes JavaScript, so not a substitute for curl.
• Schema.org Validator (validator.schema.org) — checks JSON-LD against the full Schema.org vocabulary without Google-specific restrictions.
• Google Search Console — shows which pages are indexed, flags crawl errors, and lets you submit sitemaps.
• Wikidata (wikidata.org) — search to check if your entity exists and verify property completeness.
• ChatGPT, Perplexity, Gemini — manually query your brand name and key topics to see current citation status.
• Screaming Frog (free up to 500 URLs) — crawls your site and flags missing schema, broken pages, and rendering issues.
• PageSpeed Insights (pagespeed.web.dev) — measures server response time, which affects whether AI crawlers timeout.

For a full AEO audit with scoring, competitive analysis, and prioritized action plan, see SCALEBASE's AEO services. For the specific schema and entity checks referenced above, see schema markup for AEO, entity signals for AI search, and what is AEO.