Milan Bogojevic Blog

Query Fan-Out for Nonprofits: A 30-Day Sub-Query Plan

6 min read updated 10 October 2026

Getting Cited in AI Search
Getting Cited in AI Search

Query fan-out expands a search question into related questions across several topics or sources. A nonprofit can use a 30-day plan to map those questions, close gaps on three priority pages, and record whether AI answers name the organisation accurately. Start with eligibility, location, schedules and required checks: these are often the details a visitor needs before acting.

Before you start

Read the AI visibility guide and run its 90-minute audit if access and identity have not yet been checked.

This guide assumes three things are already true: AI crawlers can reach your site, your organisation's name and facts are consistent across your own pages, and you have at least one page per topic you care about. If any of those is not true yet, fix it first. Sub-query coverage and good writing cannot compensate for a page a crawler cannot reach.

What happens between the question and the AI answer, in one paragraph

An AI system rarely answers a question from one page. It expands your question into related sub-questions (query fan-out), retrieves sources for each, and assembles an answer from whichever pages cover those sub-questions most clearly. A food bank's volunteer page can rank on page one and still lose the AI answer to three other organisations, because each of them answered one sub-question (eligibility, schedule, background checks) that the top-ranked page left out.

Mapping the sub-queries for your own topic

Most nonprofit content answers one question ("how do I volunteer") and leaves related details unstated. Google describes query fan-out as multiple related searches across subtopics and sources. iPullRank is a search marketing agency whose AI Mode webinar explores fan-out planning. The eight categories below are a practical mapping exercise; they are not an exhaustive or verified taxonomy of Google's internal queries:

Sub-query type Example for "how do I volunteer at a food bank"
Equivalent "how can I help at a food bank"
Follow-up "what do volunteers actually do"
Generalization "how do I get involved with hunger relief"
Specification "how do I volunteer at a food bank on weekends"
Canonicalization "[organisation name] volunteer sign up"
Translation the same question in the languages your community speaks
Entailment "do I need a background check to volunteer"
Clarification "is this for one-time or recurring volunteering"

Exercise (30 minutes): Take your single most important page. List every sub-question a real visitor would ask around that topic. For each one, check: does this page answer it in a sentence someone could lift out and understand on its own? If not, that is your first content gap, not a brand-new page topic.

One strong page per topic beats ten thin ones. If you already have five short posts that each touch volunteering, merge their sub-question coverage into one page an AI system can treat as the authority on the topic, rather than splitting your own evidence across pages that compete with each other.

How should the answers be written?

Give each answer a named subject and enough context to stand alone. Keep facts, sources and dates together, and update them when the facts change. Use the paragraph-level checklist to test the passages you rewrite.

A 30-day plan for a small comms team

Week Focus What you do
1 Map List sub-queries for your top 3 pages using the table above; flag the gaps
2 Write Close the gaps on page 1; rewrite its weakest paragraphs so each stands alone
3 Write Repeat for pages 2 and 3; merge any thin duplicate posts into their parent page
4 Check Run your first prompt panel (below); note what's missing; schedule the next round in 30 days

This is deliberately scoped to three pages. Trying to fix the whole site in a month is how a two-person team abandons the project in week two.

What I'd skip

I would not spend a developer's time on llms.txt unless it is a one-click setting in your CMS. No AI system currently requires it, and the time is better spent closing a sub-query gap on your volunteer page. I would not add noai meta tags expecting them to function as a technical block. Treat them as a polite request, not a wall. And I would not chase every AI SEO tool that promises a proprietary "citation score". Your own prompt panel, run consistently, tells you more than a vendor dashboard that cannot show its methodology.

How to tell whether it's working

The pillar's four-layer measurement model is the board-level view. Week to week, you need something lighter: a monthly prompt panel. Write five to ten real questions your audience asks, the sub-queries you mapped in week 1 are a ready-made list, and run them in two AI systems on the same day each month. Log three things per question: were you named, was the answer accurate, and which competitor or source filled the gap when you weren't. That third column is what tells you which sub-query to close next.

Check your server logs once a quarter for AI crawler activity (OAI-SearchBot, PerplexityBot, and similar user agents). An absence of visits from these user agents is a reason to investigate access, rather than proof that a page can never be cited through another index or source, which usually points back to the crawler-access check in the pillar rather than anything in this guide.

Frequently asked questions

Is this worth doing for a small organisation with no dedicated SEO staff?

Yes. The queries that trigger AI answers tend to be specific and low-volume, exactly the kind a focused local nonprofit already answers better than a large generic competitor. The barrier is time, not budget.

Does old content still get cited?

It can, especially evergreen how-to content. Update it when the facts change, not on a fixed schedule. Accurate evergreen information can remain useful for years. An updated date alone does not prove the content is better or more likely to be cited.

How many sub-queries should one page try to cover?

As many as a real visitor would plausibly ask about that exact topic, usually five to ten. Past that, split into a second page on a genuinely different sub-topic rather than padding one page with marginal questions.

Sources

About the author

Milan Bogojevic works on nonprofit visibility across search and AI, with a focus on clear content, organisational identity and verifiable evidence.

Last updated: 10 October 2026.

Next in AI SEO & GEO