How AI search engines decide which sites to cite

Classic SEO was a ranking contest: be #4 instead of #9 and you still got clicks. AI search is a selection contest: the answer quotes three to six sources, and everyone else gets nothing. That changes what "optimization" means.

What selection actually rewards

AI engines assemble answers from passages, not pages. Across vendor documentation and published GEO research, the same traits keep coming up:

Trait	Why it wins citations
A direct answer near the heading	Engines lift self-contained passages. If your answer needs three paragraphs of wind-up, someone else's cleaner paragraph gets quoted
Numbers, definitions, comparisons	Concrete claims are quotable; vibes are not. Tables are extraction-friendly
Visible freshness	Dated content loses to updated content on time-sensitive queries
Entity clarity	Consistent naming of who you are and what you do (helped by schema markup) — engines cite sources they can attribute
Crawlability by the right bots	None of the above matters if the engine's crawler never saw the page

The step everyone skips: can AI engines see you at all?

Each engine has its own index and its own crawler. Perplexity needs PerplexityBot. ChatGPT search needs OAI-SearchBot. Google's AI features draw on Googlebot's index. Block one — or have a WAF silently challenge it — and you're invisible on that engine no matter how citable your content is.

Most sites have never checked. Robots.txt files accumulate "block AI" rules from 2023-era advice; some CDN bot protections challenge crawlers by default. The symptom is quiet: you just never appear in AI answers, and there's no Search Console to warn you.

Where VisitorType fits

VisitorType shows you which AI crawlers and search bots actually visit — GPTBot, OAI-SearchBot, PerplexityBot, and 30+ others, each verified against the vendor's published IP ranges. If PerplexityBot hasn't hit your site in a month, that's a crawlability problem to fix before rewriting a single headline. And the AI agent directory tells you what each bot does before you allow or block it.

Check your baseline first: create a free account and see which AI engines are already reading you — and which never show up.