Query fan-out
The hidden sub-queries AI models research before answering.
When an AI answers a complex prompt, it rarely searches once. It silently decomposes your prompt into several sub-queries, researches each, and synthesizes an answer from the results. This behavior is called query fan-out.
Why it matters: you can be visible for the main prompt yet absent from the sub-queries that shaped the answer. Covering fan-out sub-queries in your content means being present at every step of the AI's research, not just the final question.
What platforms expose
- Gemini reliably emits fan-out, typically three to six sub-queries per call
- ChatGPT typically echoes your prompt as a single sub-query
- Claude and Perplexity rarely expose any sub-queries
This is why fan-out data concentrates on Gemini answers.
Where you see it in SEOAI
- The execution result page lists the fan-out queries for that answer
- The Query Fan-out page aggregates sub-queries per prompt with the platforms that emitted them
- The Gaps area counts fan-out coverage: a sub-query is covered when you appeared in at least one answer that ran it, and uncovered sub-queries are topics the AI researches where you do not exist yet
Keywords AI surfaced
The Query Fan-out page also mines two-to-three-word phrases from answer text, excluding your own prompts, your brand and your tracked competitors. These are the terms AI assistants reach for when talking about your market, and they make good candidates for new prompts or content.
Query Fan-out page
Prompts: choosing what to track