AI Visibility Audit: How to Check if ChatGPT Can Find and Recommend Your Website

A traditional SEO audit checks whether Google can find, crawl, and rank your pages. An AI visibility audit goes further: it checks whether AI platforms like ChatGPT, Claude, Gemini, and Perplexity can find your content, extract reliable facts from it, and confidently recommend your site in generated answers.

The audit covers four pillars: crawl access (can AI bots reach your pages?), content structure (can AI models extract facts from your pages?), machine-readable signals (do you have structured data, llms.txt, and proper meta tags?), and action readiness (can AI agents interact with your site beyond just reading it?).

A site can score perfectly on a traditional SEO audit and still be invisible to AI platforms. The most common scenario: a site that ranks well on Google but blocks GPTBot and ClaudeBot in robots.txt, has no structured data, and serves JavaScript-rendered content that AI crawlers can't parse.

The audit starts with the most fundamental question: can AI crawlers physically access your pages?

This checks robots.txt rules for every known AI crawler user agent (GPTBot, ClaudeBot, PerplexityBot, Google-Extended, and others). It also tests whether your CDN or WAF silently blocks bot traffic — a surprisingly common issue where robots.txt allows access but the firewall returns 403 or a CAPTCHA challenge.

Sites using Cloudflare Bot Fight Mode, AWS WAF with aggressive bot rules, or custom rate limiting often block AI crawlers without realizing it. The audit flags these invisible blocks that standard SEO tools never check.

Once a crawler can access your pages, the audit evaluates whether the content is in a format AI models can reliably parse.

This includes checking whether key content is in the initial HTML response or hidden behind JavaScript rendering. AI crawlers have varying levels of JavaScript execution capability — some render pages fully, others only see the initial HTML. Critical information (pricing, product details, contact information) that requires JavaScript to appear may be invisible to certain AI platforms.

The audit also checks content quality signals: does the page have a clear heading hierarchy, are facts presented in structured formats (tables, lists, definition pairs), and is the most important information near the top of the page rather than buried after walls of marketing text?

This pillar audits the structured metadata that helps AI models interpret your content with confidence.

JSON-LD structured data is the highest-impact signal. The audit checks for relevant schema types (Organization, Product, SoftwareApplication, FAQPage, HowTo, BreadcrumbList) and validates that the data is accurate and complete — not just present but actually matching the visible page content. A Product schema with a price that doesn't match the displayed price is worse than no schema at all.

llms.txt is checked for existence, format validity, and link accessibility. The audit verifies that every URL in llms.txt resolves correctly and isn't blocked by robots.txt.

Open Graph and meta description tags are checked for presence and quality. While primarily used for social sharing, these tags are also consumed by AI models as content summaries.

The most forward-looking pillar: can AI agents actually do things on your site, not just read about them?

As AI moves from answering questions to completing tasks (booking appointments, comparing products, adding items to cart), sites that offer structured API endpoints or WebMCP-compatible actions become more valuable in AI recommendations. A hotel site where an AI agent can check availability and book a room is more likely to be recommended than one that only provides a phone number.

The audit checks for the presence of documented API endpoints, WebMCP configuration, and action schemas that enable AI agents to interact with your site programmatically. For ecommerce sites, this includes product search, cart, and checkout endpoint readiness.

Most AI visibility audits produce a score or grade. Understanding what different score ranges mean helps you prioritize fixes.

Below 40: Critical issues are blocking AI visibility entirely. Usually a robots.txt or CDN block that prevents crawlers from accessing the site at all. This is the highest-priority fix because everything else is irrelevant if bots can't reach your pages.

40 to 60: Crawlers can access your site but content isn't structured for AI extraction. Common issues: missing structured data, no llms.txt, key content rendered only via JavaScript. These fixes have the best ROI because they make already-accessible content usable by AI models.

60 to 80: The fundamentals are in place. Improvements at this level involve richer structured data, better content formatting for extraction, and expanding coverage to more page templates. Returns diminish above 75.

Above 80: Your site is well-optimized for AI visibility. Focus shifts from technical fixes to content quality and authority building.

After auditing thousands of sites, the most impactful issues follow a clear pattern.