Query fan-out & internal linking

Vadim Kravcenko
Vadim Kravcenko
Nov 16, 2025 · 3 min read

When someone asks an AI "What's the best X?", the AI doesn't look for one page. It breaks the question into many small checks first. That split is query fan-out. Think of the main question as a tree: price, features, setup, mistakes to avoid, and alternatives. If your site already answers those small parts — and those pages link to each other — the AI can pull a safer, clearer answer from you.

The tree metaphor holds up surprisingly well — until it doesn't. For straightforward commercial queries like "best espresso machine under 300 euros," the branches are obvious: price, noise level, grinder quality, cleaning, warranty, alternatives. But for ambiguous queries like "is espresso healthier than drip coffee?" the fan-out is messier. There's no clean tree. The AI checks nutrition, caffeine content, preparation method, personal health conditions, cultural context — and the branches overlap in ways a tree diagram can't represent. It's more like a root system. The metaphor breaks down precisely when the query gets interesting.

Which is useful to know, because the queries where fan-out gets messy are the ones where having comprehensive, interlinked content gives you the biggest advantage. If your content handles the clean tree queries well, any competitor can replicate that structure. The root-system queries — where sub-questions feed into each other and there's no single authoritative order — are where depth and internal linking actually differentiate you.

Example of the clean version: someone asks, "best espresso machine under €300." The AI checks things like: what counts as "best," which models are actually under €300 in the EU, which grinders are included, how loud they are, how easy they are to clean, warranty rules, and common failures. One question becomes many mini-questions.

Internal links make this work. They help crawlers and LLMs discover related pages fast, understand how each piece fits, and trust your site as a complete source on the topic. A good hub page that links to focused child pages (and children that link back to the hub and to siblings) gives machines a clean path to follow.

"GEO" (getting visible in AI search) isn't a new trick, even though a lot of people are screaming that SEO is dead and GEO is the new King. Good GEO = good SEO. It's solid SEO basics with sharper edges: cover a topic deeply, write a short summary at the top of each page, and connect related ideas with clear, human anchors. Do that, and both Google and AI systems can assemble your pages into the answer people see.

Internal linking for fan-outs

I want to show you why internal links make a big difference when Google or an AI tool tries to answer a question. This isn't abstract theory — I've watched the impact directly in Search Console data across dozens of sites that implemented hub-and-spoke linking.

Discovery. Links are the roads on your site. Google and AI follow those roads to find related pages fast. More clear roads = more of your good pages get seen. A page with zero internal links pointing to it might as well not exist. We've found pages on client sites that had been live for over a year with no internal links — Google had never crawled them.

Context. Pages that sit next to each other add meaning. One page gives a definition, another gives an example, a third adds limits or costs. Linked together, they give stronger, clearer answers. Without the links, each page is an orphan making claims without context.

Authority. A hub page that points to many focused child pages says, "we cover this topic properly." Consistent links in and out of that hub tell machines you're a reliable source.

Freshness. When you link new posts from your hub, crawlers find them sooner. Fresh pages get tested faster, which helps you show up earlier. I aim to link every new post from its parent hub within 24 hours of publishing. The indexing speed difference is measurable.

Answer assembly. AI often pulls small parts from several pages. If your subpages are connected, it can stitch those parts into one useful answer. If they're not connected, the AI has no way to know that your pricing page and your comparison page are part of the same story.

Here's a simple picture to keep in mind:

Root intent
├─ Budget
├─ Specs
├─ Maintenance
├─ Where to buy
└─ Alternatives

Your job is to cover these branches and connect them with clear links. But here's where most guides stop, and where the real work starts: some branches will need sub-branches. "Maintenance" might split into "daily cleaning," "descaling schedule," and "when to replace parts." If those sub-branches are substantial enough, they deserve their own pages. If they're not, they stay as sections within the parent. Making that judgment call well is what separates a useful content structure from a bloated one.

Anchors & page structure that AIs understand

Two things make your pages easy for Google and AI tools to use: clear link text (anchors) and a clean page structure. Here's the combined, plain-English playbook.

1) Anchors that tell the truth (no spam)

  • Be descriptive. Use anchors that match the promise of the page:

    • Good: eligibility rules 2025, maintenance costs, download form (PDF)

    • Not great: click here, read more, best guide

  • Vary the wording, keep the intent.

    • eligibility rules 2025who qualifies this yearrequirements & documents
      All three point to the same idea without repeating the same phrase. This matters more than people realize — if every internal link to your pricing page says "pricing," Google learns less about what that page covers than if your anchors say "monthly plans," "what SEOJuice costs," and "pricing for agencies."

  • Write for humans first. Ask: "If I click this, do I know exactly what I'll get?" If yes, it's good for LLMs too.

  • Avoid stuffing. One clear anchor beats five keywordy ones. I've seen sites where every other sentence has a bolded anchor text link — it reads like a Wikipedia parody and search engines treat it about as seriously.

2) Page structure for answerability (copy this)

Give machines a neat, predictable layout so they can lift answers safely.

  • Top box (2–3 sentences): a plain-language answer to the main question. This is the "extractable" bit AIs love.

  • Body with clear headings: each H2 maps to a sub-question (the "fan-out" branches).

  • One-sentence summary under each H2: opens the section with the takeaway before details.

  • Helpful links between sections: use the descriptive anchors above to connect related answers.

  • Schema where it fits: add FAQ, HowTo, or Article schema to reinforce structure.

  • Cite sources when needed: short references/footnotes if you quote rules, data, or regulations.

3) A simple template you can paste

Title: Heat Pump Incentives 2025 — Who Qualifies & How to Apply

Summary box (2–3 sentences):
Most households qualify if income is below X and the unit meets standard Y. Apply online with forms A and B; decisions take ~3 weeks.

H2: Who qualifies in 2025?
One-sentence summary: Households under income X with approved installers qualify.
Details…
Link: eligibility rules 2025

H2: Documents & forms
One-sentence summary: You need ID, proof of residence, and form A; download below.
Link: download form (PDF)

H2: Costs & maintenance
One-sentence summary: Expect €Z upfront and €Y/year maintenance; see breakdown.
Link: maintenance costs

H2: Deadlines & timeline
One-sentence summary: Submit by March 31; typical review is ~3 weeks.

FAQ schema / references (if applicable)

Are AIs using your content?

How to tell you're showing up in AI search (GEO proxies)

There's no official dashboard for this. No "AI Overview impressions" metric in Search Console. So you're working with proxies, and I want to be upfront about what's reliable and what's a guess.

  • Answer cards / featured snippets: You win more snippets on the hub and its children. Track weekly; rising wins = good fan-out coverage. This is the most reliable proxy because the data comes straight from GSC.

  • Brand mentions in AI overviews/aggregators: Your name/domain appears inside "AI summary" boxes or roundup sites. Save screenshots; note the query and page used. This is manual and tedious. I spend about 30 minutes a week checking our top 20 queries in incognito. It's not scalable, but it's the only way to know for sure right now.

  • Engagement lift after query spikes: When a topic trends, your hub + children see impressions/clicks rise together in Search Console. That sync is a strong hint AIs are pulling/contextualizing your pages.

  • Your "answerable paragraphs" get reused: Short top-box answers are quoted or paraphrased (with or without a link). Set alerts and keep examples.

Common mistakes (and how to fix them fast)

  • One mega "complete guide," no children → Split into a hub + focused child pages. Link hub → child and child → hub. Add a short summary box to each. I know this feels like creating more work. It is. But a 10,000-word mega-guide with no internal structure is harder for AIs to extract from than five 2,000-word focused pages that link to each other.

  • Anchors like "click here" → Replace with intent-rich anchors (e.g., "eligibility rules 2025", "maintenance costs", "download form (PDF)").

  • No links to new posts → From the hub, link every new post within 24–48h. Also add new → evergreen links inside the fresh post.

  • Thin comparison pages → Add a pros/cons list, specs table, and a "when to choose X vs Y" section. Link to the detailed child pages.

  • Dead links left unfixed → Redirect 404/410s or relink to a live alternative. Run a monthly sweep so you don't leak authority.

Bottom line: When your topic is covered by a hub with connected child pages — and each page has clear, descriptive anchors — Google and LLMs can assemble your content into answers. Watch the signals above and keep closing the gaps.

Final thoughts

GEO isn't magic dust, and it's nothing new. It's the same solid SEO you already know — done with a bit more care. If you cover a topic's fan-out (the real sub-questions people have), write a clear 2–3 sentence summary at the top of each page, and connect those pages with straightforward internal links, both Google and AI systems can stitch your work into confident answers.

If you want one simple plan: pick a topic, make a hub page, create 5–10 focused child pages, add descriptive anchors between them, and keep things fresh (fix dead links, link new posts from the hub within 48 hours). Then pick the next topic and do it again. The compounding effect is real — each new hub reinforces the authority of your existing hubs through cross-linking.

Same fundamentals — new distribution. Keep it tidy, keep it human, and you'll show up where it matters.

Focus on good SEO — and GEO will follow.

Related reading: