Fanout queries - how AI assistants expand a single question

Ask an assistant "what should I pack for a week in Portugal in October?" and it doesn't run one search. Under the hood, it decomposes the question: Portugal October weather, lightweight layering pieces, packable rain jackets, travel capsule wardrobes, linen versus cotton in humid climates. Each sub-query retrieves pages; the answer is composed from all of them. That decomposition is fanout - and it quietly rewrites the targeting logic of content strategy.

Why fanout changes the game

Classic keyword research targets queries humans type, sized by search volume. Fanout queries are machine-generated - specific, constraint-laden, combinatorial - and they have no volume data because no human types them. "Breathable overshirt for 15-18 degree weather packable" will never appear in Keyword Planner. But pages that answer queries like it get retrieved into fanouts constantly, and each retrieval is a chance to be the cited source in the composed answer.

The strategic consequence: long-tail specificity, which classic SEO deprioritized as low-volume, becomes premium inventory. The page too specific to rank for anything big is exactly the page a fanout retrieves.

Anticipating your fanouts

You can't see the actual sub-queries, but you can reason about them, because decomposition follows the structure of the buying decision. Take a question your customer would ask an assistant and break it the way the machine would: by constraint (budget, shipping destination, size range, sustainability), by context (occasion, weather, activity - the dimensions of use), by comparison (material vs. material, this type vs. that type), and by attribute (every structured fact a product could be filtered by). Each branch is a potential retrieval target. A practical exercise that takes an afternoon: write your ten most plausible customer questions, decompose each into 5-8 sub-queries, and check honestly - does any page on your store answer each one? The empty cells are your fanout content map.

Building for fanout

Two content moves follow. Cover the intersections: fanout sub-queries live where attributes meet context - "linen for office wear," "wool weights for Scandinavian winter." Your collection content, buying guides, and comparison pieces (Phase 4's formats, unchanged) should deliberately address these intersections rather than staying at category level. Make each answer self-contained and extractable: a fanout retrieval reads your page for one specific answer; the page that states it cleanly - under a clear heading, facts explicit - wins the citation (last two lessons' principles, applied).

And underneath it all, the recurring foundation: intersections are made of attributes, and attributes come from your catalog. A store whose product data includes climate-relevant, use-relevant, constraint-relevant facts can generate fanout-ready content systematically. A store with thin data is guessing. Phase 2 keeps paying rent.