LESSON 17 · 1 MIN READ

How AI decides who to recommend

Understand AI search retrieval, source selection and citation limits. Learn how accessible pages, clear answers and supporting evidence can help readers.

Ask ChatGPT or Perplexity for a recommendation and it doesn't flip through ten links like Google. When web search is enabled, it can retrieve sources and combine their information into an answer. So the game isn't "rank number one". It's "be one of the few it decides to quote".

And it's fussier than you'd think. It reaches for clean, self-contained paragraphs that nail the question, from sites it already trusts. Make the main answer easy to find and support it with evidence. This helps readers evaluate the page, but it does not guarantee selection by an AI system.

AI answer layouts and source selection vary by product, query, and whether web search is enabled. Keep a dated sample of prompts and citations rather than treating one answer as a stable ranking.

One more thing worth knowing: a single question quietly splits into several. Ask "best CRM for a small agency" and behind the scenes it's also checking pricing, integrations, reviews and alternatives. Every one of those you answer well is another shot at getting your name in the reply.

Further reading: Google explains query fan-out and AI search.

One step closer.

Keep track of what you have learned in this browser.