Query Fan-Out

Query fan-out is the technique of taking one seed query and expanding it into the full cluster of adjacent questions, comparisons, and follow-ups a large language model is likely to generate when a user asks about a topic. LLMs internally reformulate a single prompt into multiple sub-queries before synthesizing an answer, so the sources cited are the ones that satisfy the whole fan-out — not just the literal phrase. Why it matters: Optimizing for the fan-out rather than a single keyword multiplies citation surface area, because one well-structured page can win multiple sub-queries at once. It is the LLM-era replacement for keyword clustering.