GEO / AI visibility
Why ChatGPT Doesn't Cite Your Website (7 Technical Reasons)
ChatGPT, Perplexity, and Gemini skip your site for fixable technical reasons: blocked AI crawlers, missing structured data, no llms.txt, thin E-E-A-T signals, and more. Here's how to check and fix each one. The good news: every one of these is fixable, most in under an hour — and none require paying for a tool.
Last updated: October 2026 · Based on crawler behavior across ChatGPT, Perplexity, Gemini, and Claude
Check all 7 in 30 seconds
Our free AI crawler check tests crawler access, robots.txt AI rules, llms.txt, and structured data — no signup required.
Run the free AI crawler checkREASON 1
Your robots.txt (or CDN) blocks AI crawlers
Before ChatGPT can cite you, its crawler (GPTBot) has to fetch your pages. Many sites block AI bots without knowing — a legacy Disallow rule, a WordPress security plugin, or a Cloudflare 'bot fight mode' that treats GPTBot, ClaudeBot, and PerplexityBot as threats. If the crawler gets a 403 or is disallowed, you are invisible to that AI system. Full stop.
How to check
Open yoursite.com/robots.txt and search for GPTBot, ClaudeBot, PerplexityBot, Google-Extended, or CCBot. Then check your CDN/firewall bot rules.
How to fix
Allow the major AI crawlers in robots.txt unless you have a deliberate reason not to. Whitelist them in Cloudflare bot management.
Free AI crawler check →REASON 2
You have no structured data
AI systems don't 'read' pages like humans — they parse them. Schema.org markup (Article, FAQPage, HowTo, Product, Organization) is the machine-readable layer that tells an AI what your content is, who wrote it, and when. Pages without structured data force the AI to guess, and it would rather cite a competitor whose markup makes the facts unambiguous.
How to check
View source on your key pages and search for 'application/ld+json'. Or run Google's Rich Results Test.
How to fix
Add JSON-LD for your content type. At minimum: Organization (site-wide), Article or WebPage (per page), FAQPage where you answer questions.
Free SEO audit tool →REASON 3
Your content is thin — nothing worth citing
AI answers cite sources that add information: original data, statistics, step-by-step procedures, definitions, comparisons. If your pages are 300 words of generic marketing copy, there is nothing for the model to quote. Princeton's 2026 GEO research found original data and statistics are the single strongest predictor of AI citation — and most sites have none.
How to check
Ask yourself: does this page contain any fact, number, or procedure that exists nowhere else on the internet? If not, it's uncitable.
How to fix
Publish original data: survey results, benchmark numbers, teardown analyses, methodology notes. Add statistics with cited sources and dates.
Free AI visibility audit →REASON 4
Weak E-E-A-T signals — AI doesn't trust you
AI systems are tuned to prefer authoritative sources. The signals they read: named authors with credentials, publication and update dates, an Organization entity with logo and sameAs links, outbound citations to primary sources, and a contactable real-world entity behind the site. Anonymous, dateless, authorless content reads as low-trust — and gets skipped.
How to check
Do your articles show a real author byline, datePublished/dateModified, and link to an Organization schema? Is there an about page with real people?
How to fix
Add author bylines with bios, visible dates, Organization schema with sameAs links to your social profiles, and cite primary sources.
Free SEO audit tool →REASON 5
Your site is too slow for crawlers to bother
AI crawlers have budgets. A page that takes 6 seconds to deliver its HTML — or that requires heavy JavaScript rendering to show content — may get partially crawled or skipped entirely. Server-rendered, fast HTML wins. If your content only appears after client-side JavaScript executes, some AI crawlers will never see it.
How to check
Test your key pages with the free performance check. Look at time-to-first-byte and whether content is in the initial HTML.
How to fix
Server-render critical content, compress images, defer non-critical JavaScript, and use a CDN. Aim for LCP under 2.5s.
Free performance check →REASON 6
No llms.txt — AI doesn't know your rules
llms.txt is the emerging standard (proposed 2024, widely adopted 2025–2026) that tells AI systems what your site is about, which pages matter, and how to use your content — like robots.txt, but for language models. Sites with a good llms.txt give AI crawlers a clean, curated entry point. Sites without one make the crawler work harder, and harder means less cited.
How to check
Open yoursite.com/llms.txt. If it 404s, you don't have one.
How to fix
Create an llms.txt with your site summary, key page links, and usage guidance. Keep it under a few thousand words, markdown-formatted.
Free AI crawler check →REASON 7
Competitors are cited instead of you
AI answers have limited citation slots — usually 3 to 8 sources. Even if your site is technically fine, you lose the slot to competitors with stronger signals: more backlinks from cited domains, Reddit threads mentioning them, review-site profiles, YouTube transcripts. Citation is zero-sum, and right now someone else is winning your slots.
How to check
Ask ChatGPT and Perplexity your target questions and note which domains get cited. Compare their technical signals against yours.
How to fix
Earn mentions where AI looks: Reddit threads, review sites, directories, YouTube. Then make sure your own site's technical foundation converts those mentions into citations.
Free brand protection check →The 30-day citation plan
Days 1–3: Remove the blockers
Run the AI crawler check. Fix robots.txt, whitelist AI bots in your CDN, publish llms.txt. These are the highest-leverage fixes — a blocked crawler makes everything else irrelevant.
Days 4–14: Make yourself parseable
Add Organization + Article/FAQPage JSON-LD site-wide. Add author bylines and dates. Compress the slowest pages. This is the machine-readability layer AI depends on.
Days 15–30: Become worth citing
Publish one piece of original data — a survey, a benchmark, a teardown with numbers. Earn 2–3 mentions on Reddit, a review site, or a directory. Then re-run the checks and watch crawlers pick it up.
FAQ
How do I know if ChatGPT can crawl my website?
Check your robots.txt for blocks on GPTBot, OAI-SearchBot, and ChatGPT-User. Then verify your server actually serves those user agents a 200 response — some CDNs block them at the firewall even when robots.txt allows them. AIClicks.net's free AI crawler check tests both in about 30 seconds.
Does ChatGPT cite websites in every answer?
No — only when the model decides citations add value, typically for factual, current, or contested claims. Your job is to be citable when it does: structured data, original facts, clear authorship. You can't force a citation, but you can remove every technical reason to skip you.
Is llms.txt actually used by AI systems?
Adoption grew rapidly through 2025–2026. Anthropic documents llms.txt support, and multiple AI crawlers fetch it when present. It's a low-cost, high-signal file: even where support is partial, it costs nothing and helps the systems that do read it.
How long does it take to start getting cited after fixing these issues?
Crawl-dependent fixes (robots.txt, llms.txt, structured data) can take effect within days to weeks once AI crawlers re-fetch your pages. Authority signals (backlinks, mentions) compound over months. Most sites see first citations within 30–60 days of fixing the technical blockers.
Do I need a GEO tool to get cited?
No — the 7 fixes above are free and technical. GEO/prompt-tracking tools (like aiclicks.io) are useful for monitoring which prompts mention you over time, but they can't fix a blocked crawler or missing schema. Fix the technical foundation first; add tracking when you want measurement.
Which AI crawlers should I allow?
At minimum: GPTBot and OAI-SearchBot (OpenAI/ChatGPT), ClaudeBot (Anthropic), PerplexityBot (Perplexity), Google-Extended (Gemini/AI Overviews training signal), and CCBot (Common Crawl, feeds many models). Block only if you have a deliberate commercial reason — blocking by default is the most common self-inflicted visibility wound.
Related: GEO audit checklist · llms.txt guide · Best GEO tools
