AI assistants — ChatGPT, Perplexity, Google AI Overviews, Claude — answer questions by fetching pages and citing sources. AI visibility is whether your page gets cited. This guide explains how answer engines pick sources and how to make your pages quotable.
How AI search engines choose sources
Four signals decide citation: bot access (GPTBot, ClaudeBot, CCBot, Google-Extended must not be blocked), machine-readable structure (schema, clean headings, lists), factual clarity (a self-contained 40 to 60 word answer the model can lift), and freshness carried by sitemap lastmod dates.
Writing for answer engines
Put a direct answer to the page's main question in the first paragraph. Turn section headings into the actual questions users ask. Keep paragraphs short and factual. Add FAQ schema where the content genuinely answers questions — structured data is the strongest measurable citation lift per hour spent.
Technical requirements
The page must return a clean 200 status with one canonical URL and render without JavaScript, because most AI crawlers do not execute scripts. An llms.txt file at your domain root helps compliant crawlers discover your key pages faster. Test everything with the AI visibility checker.
Measuring progress
Ask the assistants directly: query your topic in ChatGPT and Perplexity and see which sources get cited. Track whether your pages appear. Re-run the AI visibility checker after every template change, since silent regressions are common.
Measuring citation share
Direct measurement beats speculation. Query your ten most important topics in ChatGPT, Perplexity, and Google AI Overviews, and record which domains get cited. Repeat monthly. The list of cited competitors is your outreach and content benchmark — their pages share whatever structural and clarity properties the models currently favor.
Content patterns that get cited
Pages that win citations share four traits: a self-contained answer in the opening paragraph, descriptive headings phrased as real questions, short factual paragraphs under each heading, and at least one list or table. Marketing language that hedges and delays the answer is consistently passed over in favor of pages that state things plainly.
Technical checklist for AI crawlers
Confirm robots.txt allows GPTBot, ClaudeBot, CCBot, Google-Extended, and PerplexityBot. Verify the page returns a clean 200 status with one canonical URL. Ensure the content exists in the raw HTML, not only after JavaScript execution. Keep the page in your XML sitemap with an honest lastmod date. Each item takes minutes and compounds across every page on the site.
Free tools in this hub
Every tool below is free, requires no signup, and runs instantly in your browser:
- AI visibility checker — Audits bot access, structure, and quotability of any page
- AI robots.txt explainer — Explains what your robots.txt means for AI crawlers
- AI canonical decision helper — Resolves canonical questions that confuse answer engines
- AI content brief reviewer — Reviews briefs for answer-engine friendliness
Related guides
Keep AI crawlers welcome in the robots.txt guide, and keep content fresh via the XML sitemap guide.
Frequently asked questions
Why does AI search ignore some pages?
The usual causes are blocked crawlers, JavaScript-only rendering, missing structured data, or writing that never states anything quotable in a complete sentence.
How long are AI-cited answers?
Cited passages usually run 40 to 60 words. Writing that tight summary near the top of each page is the cheapest AI visibility improvement available.
Does llms.txt replace robots.txt?
No. llms.txt is a hint for AI crawlers; robots.txt remains the standard access-control file. Keep both aligned and permissive for answer engines you want cited.
