When you ask an assistant a broad question, it does not search for that question. It breaks it into several narrower ones, gathers what it finds, and composes an answer from the pieces.

Those narrower questions are fanout queries, and they are where the citation decision actually happens.

Why this matters more than the original prompt

You can write a page aimed perfectly at "best rank tracker for agencies" and still not be cited for it, because the engine never searched that phrase. It searched something like "rank tracker white label reporting", "rank tracker multiple clients pricing" and "rank tracking API access", then assembled an answer from whatever those returned.

If your page covers the headline question but none of the sub-questions, it does not enter the answer. A competitor whose page happens to address three of the four does.

What the view shows

For each prompt you track, the sub-questions that fed the answer, and what was found for each. It is the difference between knowing you lost and knowing which part you lost.

How to use it

  1. Take one prompt you care about and lose

    Open its fanout list. You will typically see between three and eight sub-questions.

  2. Mark which ones your site answers

    Be strict. "Mentions it in passing" is not answering it. The test is whether a reader looking only for that sub-answer would find it on your page without hunting.

  3. Look at the ones you do not cover

    That list is your content brief. It is more specific and more reliable than any keyword tool output, because it came from the engine's own decomposition of the question.

  4. Cover them on one page, not five

    The engines reward a page that answers the cluster. Five thin pages, one per sub-question, perform worse than one page that handles all of them properly.

This is where the content calendar should come from

Fanout gaps make better briefs than keyword volume does. Automations schedules work from them, and content score checks the result before you publish.

A caution

Fanout is generated per answer, so the sub-questions shift between runs. A sub-question that appears once might be noise; one that appears in most runs of the same prompt is real.

Look at the pattern over several scans before rewriting a page around it. Prompts refresh on your plan's schedule, weekly or daily, so give it a few refreshes before you decide.

Need help?

Write to us and a person answers. See the frequently asked questions first if your question is a common one, or get in touch for anything else.