Query fan-out is how AI search answers a question. Instead of running one search, Google’s AI Mode and AI Overviews (and assistants like ChatGPT) split the question into several related searches, gather results from all of them, and build one answer. That’s why the pages AI cites often aren’t the pages ranking on page one for the original query. The citation pool is wider than the results page, and that changes how you should plan content.
This guide explains how fan-out works, what the research shows, and how agencies and businesses can plan for it without chasing every possible question.
Key takeaways
– Google confirms AI Overviews and AI Mode use query fan-out, running related searches at once to find more sources (Google’s AI optimization guide).
– Moz’s study of nearly 40,000 queries found only 12% of AI Mode citations matched URLs in the organic top 10 for the same query.
– Ahrefs’ March 2026 update found 37.9% of AI Overview citations ranked in the top 10, down from about 76% in its July 2025 study.
– AirOps found ChatGPT ran two or more fan-out searches on 89.6% of prompts.
– Plan around topics and the questions buyers ask next, not single keywords.
What is query fan-out?
Query fan-out is when an AI search system breaks one question into several sub-questions, searches for each, and combines the results into a single answer.
Picture someone asking AI Mode: “What’s the best way to finance a kitchen remodel?” Behind the scenes, the system might also look up typical remodel costs, home equity loans vs. personal loans, credit score requirements, contractor payment schedules and local permit costs. The final answer cites sources from all of those searches, not just the one the person typed.
Google describes this directly. Its guide to generative AI features says AI Overviews and AI Mode rely on retrieval from the Search index and on query fan-out, which runs related queries at once to pull in additional results.
Why doesn’t ranking first guarantee an AI citation?
Because the AI isn’t only looking at the results for your keyword. It’s looking at the results for a whole set of related searches, and picking the sources that best answer each part.
The research lines up:
| Study | Surface | Finding |
|---|---|---|
| Moz, Feb 2026 (~40,000 queries) | Google AI Mode | 12% of citations matched organic top-10 URLs; 96% of responses included at least one citation, most pulling 10+ URLs |
| Ahrefs, Mar 2026 | Google AI Overviews | 37.9% of cited URLs ranked in the top 10; a large share ranked outside the top 100 |
| AirOps (15,000 prompts) | ChatGPT | 32.9% of cited pages found in any top-20 results were found only through fan-out searches |
Ahrefs suggested the drop from about 76% to 38% may reflect AI Overviews leaning more on fan-out results since newer Gemini models arrived. Moz’s Tom Capper put it simply: AI Mode is “branching out to a broader set of queries and topics rather than just the exact one you typed in.”
Rankings still help. AirOps found that 43.2% of pages ranking first in Google were cited by ChatGPT, about 3.5 times the rate for pages outside the top 20. Moz found a higher rank “clearly increases the odds” of an AI Mode citation. Ranking gets you into the pool. Covering the related questions is what gets you picked.
What does fan-out mean for content strategy?
Three shifts.
1. Think in topics, not keywords
A single keyword is now the starting point of a search, not the whole search. Plan content around a topic and the questions that sit around it: costs, comparisons, steps, risks, alternatives, local factors, “is it worth it?”
2. Answer the next question
Fan-out searches tend to follow the questions a buyer would ask next. If your page answers the main question but skips the obvious follow-ups, a competitor’s page that covers them may get the citation for that part of the answer.
3. Make each answer easy to lift
Each section should open with a clear, self-contained answer that still makes sense if it’s quoted alone. AirOps also found pages whose titles closely matched the query had about twice the citation rate of pages with little overlap, so clear, descriptive headings help.
How do you find the fan-out questions for a topic?
You can’t see Google’s exact fan-out queries, but you can get close:
- Ask the assistants. Run your main question in AI Mode, ChatGPT, Perplexity and Gemini several times. Note the sub-topics each answer covers and the sources cited for each.
- Check “People also ask” and related searches in Google for the main query and its close variants.
- Mine your own data. Sales calls, support tickets and Search Console queries show the follow-up questions real customers ask.
- Look at who’s cited. For each sub-topic, which pages get cited? Those are your real competitors, even if they’re not on page one for your keyword.
- Keep the repeats. AI answers vary between runs. Questions that come up again and again are the ones worth covering.
Should you write one big page or many small ones?
Usually a strong core page plus a few supporting pages, not dozens of thin ones.
Google’s guide recommends against chasing every variation, and its spam policies warn against mass-producing pages with little value. A practical structure:
- One core page that answers the main question and briefly covers the most common follow-ups.
- Supporting pages only where a follow-up deserves real depth (a detailed cost guide, a comparison, a local page with genuinely local information).
- Internal links both ways, so search systems and readers can move between them.
If a sub-question can be answered well in two paragraphs, it belongs on the core page.
Illustrative example: mapping a fan-out cluster
Illustrative example (placeholders, not client data): An accounting firm targets “[city] small business tax help.” Testing the question in AI Mode and ChatGPT across several runs surfaces recurring sub-topics: quarterly estimated taxes, deductible expenses, S-corp vs. LLC, deadlines and penalties, and what an accountant costs. The firm’s existing page covers only “we offer tax services.” The plan: rewrite the core page to answer each sub-topic briefly with first-hand guidance, add one in-depth page on entity choice where the firm has real expertise, and add a clear cost range based on its own pricing ([X]–[Y]). Success is measured by AI impressions on those pages in Search Console and by citation checks for the recurring sub-topics.
How should you measure fan-out coverage?
Move from keyword rank to topic visibility:
- AI impressions by page in Search Console’s Generative AI performance report. Watch whether pages in the cluster gain visibility.
- Citation share by topic. For each cluster, run a fixed set of prompts monthly and record how often you’re cited for each sub-topic.
- Rankings for the follow-up questions, not just the head term.
- Leads from the cluster’s pages in GA4 or your CRM.
What should agencies and businesses do differently?
Agencies: stop reporting single-keyword positions as the main proof of value. Build topic maps that include follow-up questions, and report citation share by topic alongside rankings. Price work by coverage of a topic, not by number of posts.
Businesses: ask your agency which follow-up questions your pages don’t answer yet, and who gets cited for them today. If the answer is “we only track the main keyword,” the plan is missing most of the citation pool.
How does fan-out change technical and off-site work?
Fan-out isn’t only a content issue. A few non-content checks make a difference:
- Indexing of supporting pages. If your cost guide or comparison page isn’t indexed, it can’t be pulled into a fan-out search. Check Search Console’s Pages report for important URLs marked “Crawled, currently not indexed.”
- Internal links with clear anchor text. Link from the core page to supporting pages using descriptive text (“kitchen remodel cost breakdown,” not “click here”). It helps search systems connect the cluster.
- Fast, readable pages. Important answers should be in plain HTML text, not locked in images, tabs that fail to load, or PDFs only.
- Third-party presence. Fan-out searches often land on review sites, forums, trade publications and YouTube. If those sources cover your topic and you’re not mentioned, that’s an off-site gap worth a digital PR or partnership plan.
- Bing and AI crawler access. For ChatGPT and other assistants, check that you’re not blocking their search crawlers and that your key pages are indexed in Bing.
Content, technical and outreach teams should share the same topic map, so each gap gets fixed by the right team.
What are the common mistakes?
- Spinning up a page for every variation. Thin pages compete with each other and risk spam issues.
- Trusting one AI answer. Answers change between runs. Look for repeated patterns.
- Ignoring non-website sources. Ahrefs found YouTube made up a notable share of AI Overview citations, and Moz found it was AI Mode’s second most-cited external source. Video and third-party coverage can be part of a cluster.
- Burying the answer. If the key point sits in paragraph six, it’s harder to cite.
- Dropping rankings entirely. Rankings still raise the odds of being cited.
Frequently asked questions
What is query fan-out in Google AI Mode?
It’s the process where AI Mode breaks a question into several related searches, runs them at the same time and combines results from all of them into one answer. Google confirms AI Overviews and AI Mode both use it.
Why is my page ranking first but not cited in AI Overviews?
AI Overviews pull from the results of several related searches, not just yours. Another page may answer a specific part of the question more clearly. Ranking raises your odds, but covering the follow-up questions matters too.
How many sources does AI Mode cite?
Moz found 96% of AI Mode responses included at least one citation, with most pulling from 10 or more unique URLs.
Does ChatGPT use query fan-out?
Yes. AirOps found ChatGPT ran two or more fan-out searches on 89.6% of the 15,000 prompts it studied.
Should I create a page for every related question?
No. Cover common follow-ups on a strong core page and create separate pages only where a topic needs real depth. Mass-producing thin pages can run into Google’s spam policies.
How do I track whether fan-out coverage is working?
Watch AI impressions by page in Search Console’s Generative AI report, run a monthly set of prompts to track citation share by topic, and tie the cluster’s pages to leads in GA4 or your CRM.
Want a content plan built for the whole citation pool?
We map topics, follow-up questions and citation competitors for businesses and agency partners, then build content that covers what buyers actually ask.
Prefer to talk? Call (813) 397-3665 or email strategy@seooutsourcing.com.
