AI Overviews: how Google picks who to quote

What actually gets cited

Studies and our own probes agree on the pattern: most AI Overview citations come from pages already on page one for the query or a sibling of it. The overview is not discovering hidden gems; it is compressing the results Google already trusts. Which means the unglamorous prerequisite for AI Overview visibility is ordinary ranking strength. There is no shortcut around it, whatever a tool subscription promises.

Structure decides who gets lifted

• Answer-first blocks. The complete answer in the opening paragraph of the section, elaboration after. Overviews quote openings.
• One question per section. Headings phrased the way people ask, each section self-contained enough to survive being quoted alone.
• Facts in parseable form. Numbers, prices and comparisons in tables and short sentences, backed by schema. Overviews love sourced specifics because they reduce Google's own risk.
• Freshness signals. Dated updates on pages targeting queries where the answer changes.

What does not work

Blocking Google-Extended does not remove you from AI Overviews (that is a Gemini training control, a common confusion). Keyword-stuffing "AI optimized" pages does nothing. And publishing thin FAQ farms hoping to be quoted mostly earns nothing, because the ranking prerequisite still applies.

Measure it or you are guessing

Whether an overview appears for your buyer queries, and who it cites, changes month to month. We measure it with live probes and source retrieval in the AEO Visibility Audit, then fix what it finds with AEO Optimization, and watch the drift with monthly monitoring. Related reading: what AEO is and how to get cited by ChatGPT.