Not Cited in AI Answers? How to Diagnose and Fix It
The short answer
If AI assistants never cite you, work through three layers in order: access (can AI crawlers fetch your pages?), extractability (is your content clean, structured, and quotable?), and authority (do trusted third parties confirm your brand?). Most invisible sites fail at access — a robots.txt or WAF rule silently blocks GPTBot or ClaudeBot.
Why isn't my brand cited in AI answers?
There are only three reasons an AI engine doesn't cite you: it can't reach your pages, it can't cleanly extract an answer from them, or it doesn't trust your brand as a source. Diagnose them in that order — fixing extractability is pointless if crawlers are blocked.
The most common cause is the quietest: a robots.txt rule, a Cloudflare "block AI bots" toggle, or a WAF rule returns a block to GPTBot, ClaudeBot, or Google-Extended. Your site looks fine in a browser while being invisible to the engines.
Layer 1 — Can AI crawlers actually fetch your site?
Check your live robots.txt for Disallow rules targeting AI user-agents. Then confirm your CDN or WAF isn't blocking them at the edge — Cloudflare and some hosts inject AI-bot blocks that override your own robots file.
- ▪Fetch yoursite.com/robots.txt and search for GPTBot, OAI-SearchBot, ClaudeBot, PerplexityBot, Google-Extended.
- ▪Test a real fetch with each user-agent — a 200 in a browser can still be a 403 for a bot.
- ▪Check CDN/WAF dashboards (Cloudflare AI Crawl Control, bot-fight rules) for managed AI-bot blocking.
Layer 2 — Can the engine extract an answer?
Even reachable pages fail if the answer is buried. AI engines lift short, self-contained passages under clear headings. JavaScript-only content, walls of text, and 90% navigation chrome all reduce what gets quoted.
- ▪Render content in raw HTML — not client-side only.
- ▪Use one H1 and question-shaped H2s; put the answer in the first paragraph after each heading.
- ▪Keep paragraphs short and self-contained; add FAQ and structured data (Organization, FAQPage).
- ▪Publish llms.txt and keep the content-to-chrome ratio high.
Layer 3 — Does the engine trust your brand?
Engines ground answers in entities they can verify. If competitors get cited for your category, they usually have third-party confirmation you lack: a Wikipedia/Wikidata entity, reviews on G2 or Trustpilot, consistent descriptions across the web, and visible authorship and dates.
How do I measure all of this at once?
Manual checking is slow and easy to get wrong. An AEO scanner automates it: it probes live AI answers with your buyers' real questions, then audits crawler access, extractability, schema, and trust — and tells you exactly which layer is failing.
cited? was built for this exact diagnosis. It generates the questions your customers ask, asks live answer engines, and reports who got cited instead of you, with a copy-paste fix prompt per failed check. It's free and needs no signup.
Frequently asked questions
+How do I know if AI crawlers are blocked?
Fetch your live robots.txt and look for Disallow rules on AI user-agents like GPTBot, ClaudeBot, and Google-Extended. Then test an actual request using each user-agent — your CDN or WAF can block bots even when robots.txt allows them. A page that loads in a browser can still return 403 to a crawler.
+How long until I'm cited after fixing access?
Search-style engines (Perplexity, ChatGPT web search, Google AI Overviews) can pick up changes within days to a few weeks as they recrawl. Model-training visibility is much slower. Fixing crawler access and extractability first gives the fastest measurable improvement.
+Who can help diagnose AI visibility problems?
AEO scanners like cited? automate the diagnosis end to end — crawler access, extractability, schema, trust, and live citation testing — and return prioritized fixes. For deep brand-authority work (earning a Wikipedia entity, third-party coverage) you may also need content and digital-PR help.