AI Discoverability Check: Is Your Site Visible to ChatGPT, Claude & Perplexity?

For most of the web's history, "can people find my site?" meant one thing: does it rank on Google. That's no longer the whole question. A growing share of discovery now happens inside AI assistants — someone asks ChatGPT, Claude, or Perplexity for "the best tool to do X," and the model answers with a shortlist. If your site isn't visible to those models, you're invisible in an entire, fast-growing channel.

An AI discoverability check answers a simple question: can AI assistants actually reach, read, and recommend your pages? This guide covers what to check, why each part matters, and how to fix the gaps.

Why AI discoverability matters now

AI assistants don't invent recommendations — they pull from content they've crawled or can fetch live. When a model suggests a product, tool, or source, it's surfacing pages it was able to read and trust. Two things follow:

• If AI crawlers can't reach you, you can't be recommended. No access, no citation.
• If your content isn't clear and machine-readable, you get skipped in favor of a competitor whose page states the same facts more plainly.

This is the same shift SEO went through decades ago — except the "search engine" is now a language model, and the "result" is a sentence in an answer with your name in it.

The 5 things an AI discoverability check looks at

1. Are you blocking AI crawlers?

The single most common — and most self-inflicted — problem. Your robots.txt might be quietly disallowing the very bots that feed AI answers. The ones that matter:

• GPTBot, OAI-SearchBot, ChatGPT-User (OpenAI)
• ClaudeBot, Claude-Web, anthropic-ai (Anthropic)
• PerplexityBot (Perplexity)
• Google-Extended (Google's Gemini / AI training)

Some sites block these deliberately; many block them by accident (a copied robots.txt, a security plugin, a CDN rule). If you want to be discoverable, make sure you're not disallowing them.

How to fix it: review your /robots.txt, remove any Disallow rules targeting those user-agents, and — if you want to be explicit — add Allow rules welcoming them.

2. Do you publish an llms.txt file?

llms.txt is an emerging convention: a plain-text file at your domain root that tells AI models what your site is, what matters, and where to look — the way robots.txt guides crawlers and sitemap.xml guides indexers. It's a low-effort way to hand AI a clean, curated summary instead of making it guess from your HTML.

How to fix it: publish an /llms.txt. We wrote a step-by-step guide with a copy-paste template: How to Create an llms.txt File.

3. Is your content actually in the HTML?

Many AI crawlers fetch your page but don't execute JavaScript. If your content is rendered client-side — a single-page app that ships an empty shell and fills it in with JS — a crawler may see a blank page. Your beautiful content simply isn't there as far as the model is concerned.

How to fix it: server-render (SSR/SSG) your important content so it's present in the initial HTML. View-source on a key page: if you can't find your headline and body text in the raw HTML, neither can most AI crawlers.

4. Is your content clear and quotable?

Models favor content that states facts plainly. A page that buries "WebScore scans your site for SEO, security, and performance in 60 seconds" under three paragraphs of throat-clearing is harder to cite than one that says it in the first line. Structured data (JSON-LD), clear headings, FAQs, and direct answers all make your content easier for a model to extract and reuse.

How to fix it: lead with the answer, use clear headings, add FAQ and structured data, and write like you're explaining to someone who needs the fact, not selling to them.

5. Is the site technically healthy?

AI discovery still rides on classic fundamentals. A site that's slow, broken, non-HTTPS, or riddled with errors is harder to crawl and less trustworthy to cite. Performance, security, and crawlability all feed into whether — and how confidently — a model surfaces you.

How to run an AI discoverability check

You can check each item by hand (open your robots.txt, view-source a page, look for /llms.txt), and you should understand it well enough to. To do it all at once, run a free WebScore scan: the SEO module reports AI crawlability — whether you block AI bots, whether an llms.txt exists, and whether your content is server-rendered — alongside performance, security, and the rest, each with a specific fix.

WebScore also is an MCP server, so if you work with an AI coding assistant, it can audit a site's AI-readiness for you directly from the chat.

The takeaway

AI discoverability isn't a separate discipline you bolt on — it's the same crawlable-HTML, clear-content, technically-healthy foundation as good SEO, plus three AI-specific moves: don't block the AI crawlers, publish an llms.txt, and write quotable facts. Get those right and you're eligible to show up where a growing number of people now start their search: inside the answer.

Want to know where you stand? Run a free scan and see your AI crawlability score in about 60 seconds.