AI search engines rarely answer a question with one search. Google’s AI Overviews and AI Mode may run several related searches across subtopics, a technique Google calls query fan-out, and ChatGPT does the same. Query Fan-Out shows those searches for a query and which of your pages answer each one.
Running it
- Open Content → Query Fan-Out and enter a query.
- Choose the data: Real: what ChatGPT and Google ask, or AI estimate (Gemini).
- Optionally pick an audit under Coverage check to match every query against that site’s pages.
- Click Analyze Query.
Real data or an estimate
Real data runs on your DataForSEO account, about $0.01 per query. Each query in the result is labelled with where it came from:
- ChatGPT: the searches ChatGPT itself ran while answering, with the pages it cited.
- People Also Ask and related searches from Google’s results.
- Google AI Mode’s answer and the pages it cites.
With only a Serper key you get the People Also Ask questions. The AI estimate uses your Gemini key and is labelled as a guess: it is not what the search engines do.
Coverage of your pages
With an audit picked, every query gets a verdict and the page that answers it best:
| Verdict | Meaning |
|---|---|
| Strong | A page answers it: its title or headings match, or a passage of its text does. |
| Weak | A page touches it, but only in part. |
| Gap | No page answers it: a candidate for a new section or page. |
The agent reads the text of up to 20 of your pages for this, and each verdict comes with a short why, so you can see what matched.
In Page Strategy
Page Strategy plans one page at a time: target query, title and description, headings, FAQ and content gaps. Tick Real AI fan-out (about $0.008 per page, off by default) and it adds ChatGPT’s real searches to the plan, each labelled with its source.
Using the results
- Turn each Gap into a section that answers it plainly, or into a new page when the topic is big enough.
- Strengthen Weak pages with a clear heading and a direct answer below it.
- Google says pages need no special markup or files to appear in AI features: they need to be indexed and eligible for a snippet.
The research-driven Content Writer uses the same real fan-out when it researches an article.