Definition
Query fan-out is the pattern where an AI system turns one user question into several parallel searches, each targeting a different facet of the request. The sub-queries run at once, and their results are merged before the model writes an answer. Where decomposition is the splitting, fan-out is the parallel search that follows.Why It Matters for AI Search
Fan-out is how an assistant covers a compound request in a single turn. Each fanned-out search is its own competition, with its own set of retrieved sources, and a business appears in the final answer only if it wins one of them. The lesson for content is coverage. A page that speaks to one facet clearly can be pulled into an answer built mostly from other sources. Businesses that document their specifics get fanned into answers they were never asked about directly.Related Terms
Query decomposition
See also
Query understanding
See also
Multi-step retrieval
See also
RAG
See also
Agentic search
See also