Query Fan-out

Query Fan-out is an AI Search mechanism that automatically splits a single user question into 5–10 sub-queries, each searching the index independently. The query 'how to cook rice' generates sub-queries: water-to-rice ratios, cooking time, rice in pot vs rice cooker, how to make fluffy rice, steamed rice.

Each sub-query is a separate citation opportunity — fan-out is a reach multiplier: if you cover 7 out of 10 sub-queries, you have 7x greater chance of citation than a competitor covering only one. Three decomposition types: semantic (entity + attributes), intentional (follow-up questions along customer journey), and verification (cross-checking facts across multiple sources). Different platforms vary in fan-out aggressiveness: Perplexity generates 10–20 sub-queries, ChatGPT uses Bing with broad fanout, Gemini directs fanout to Knowledge Graph and YouTube.

In practice, use Frame Semantics as reverse engineering for fan-out — a semantic frame gives you a sub-query map BEFORE writing the article. AI Search is an investigative journalist, not a student with a textbook — it breaks down the topic into 10 questions and calls 10 sources.