How do I make my website readable by ChatGPT and Perplexity?

TL;DR

Serve content as static or server-rendered HTML, ship FAQPage and Article JSON-LD on every meaningful page, write each page with a definitive 1–2 sentence answer in the first paragraph, and make sure your robots.txt does not block GPTBot, ClaudeBot, or PerplexityBot. The goal is a page a crawler can read in one fetch and a model can quote in one extract.

Three failure modes account for most "ChatGPT doesn’t see my site" complaints. First, the page renders body content via client-side JavaScript that the crawler doesn’t execute. Second, the page has no structured data and no answer-shaped section, so even when the crawler fetches the HTML, a model has nothing clean to extract. Third, the site’s robots.txt blocks the AI crawlers explicitly.

The fix for the first issue is hosting choice, static or server-rendered HTML, not pure-client-side React or Wix-style JavaScript rendering. The fix for the second is content shape, write pages that lead with the answer, list the facts, and ship the matching schema. The fix for the third is one line: don’t add `Disallow` rules for GPTBot, anthropic-ai, ClaudeBot, or PerplexityBot.

For ongoing visibility, ship an llms.txt file at the root of your domain pointing crawlers at your highest-signal pages, keep your dateModified within 90 days on important pages, and re-test your domain in ChatGPT and Perplexity every 30 days to catch regressions.

Key facts

GPTBot is OpenAI’s crawler for ChatGPT; ClaudeBot is Anthropic’s; PerplexityBot is Perplexity’s.
All three crawlers respect robots.txt and standard HTML, they do not execute JavaScript reliably.
A page modified within the last 90 days carries more freshness signal than one modified two years ago.
A site with llms.txt at the root is indexed faster and more accurately by AI crawlers than one without.
AI engines re-fetch high-signal pages every few weeks; small content updates compound over months.

Common follow-ups

Should I block GPTBot from training on my content?

Mostly no. Blocking GPTBot prevents OpenAI from using your content for training but also reduces your visibility in ChatGPT search results, which use a different crawler path. If you don’t mind being part of training data, leave GPTBot allowed.

How do I check if ChatGPT can see my site?

Open ChatGPT with web search on and ask "What does [your domain] do? Cite sources." If your site is visible, you should see your domain in the citation list. If you see only third-party references (review sites, directories), your own site isn’t making the cut and one of the failure modes above is in play.

Does pagespeed matter for AI search?

Less than for traditional ranking, but still. AI crawlers time out on slow pages just like human users. A page that loads in under 2.5 seconds (LCP) is reliably crawled; a page over 5 seconds is unreliable.

Sources

By Isaiah Grant, Founder, Rebuilt StudioUpdated Apr 27, 2026