Topical authority and internal-linking silos

In short: a topical silo is a tightly interlinked cluster of pages — one pillar hub plus its supporting spokes — built around a single subject so that internal link equity and topical depth stay concentrated where they earn rankings. The structure signals expertise twice over: to a search ranker following your links, and to an AI model reading your cluster as one coherent body of knowledge. The failure mode is leakage — links that bleed authority out to unrelated pages — and the cure is discipline, not more links.

What a topical silo actually is

A silo has two roles. The pillar is a broad hub that covers a subject at overview depth and links down to everything under it. The spokes are narrow pages that each answer one specific query, link back up to the pillar, and link sideways to their closest siblings. Together they form a hub-and-spoke shape where authority flows inward toward the pillar and coverage fans outward across every subtopic.

A content cocoon is the stricter version: pages link almost only within their own cluster, so a reader — and a crawler — keeps circling the topic instead of wandering off to unrelated content. Contrast both with a flat blog, where every post links to every other post regardless of subject. That pattern spreads link equity thin and tells a ranker nothing about which pages belong together.

Hub and spoke: the shape of authority

Axis	Flat blog	Hub-and-spoke silo	Content cocoon
Link pattern	everything to everything	spokes to pillar, pillar to spokes	within-cluster only
Equity flow	diluted evenly	concentrated on the pillar	tightly recirculated
Topical signal	weak / mixed	clear cluster	very strong cluster
Crawl behavior	shallow, scattered	deep within topic	deep, self-contained
Cannibalization risk	high	low	lowest

The three are a spectrum, not rivals. Most sites want silos for their money topics and can relax toward a flatter graph for one-off posts that don’t belong to any cluster.

Why depth signals expertise — to the ranker and the model

For the ranker, internal links carry equity. The original PageRank model treats a link as a vote, and votes flow along the graph; a dense cluster pooling its links on one pillar concentrates that flow instead of scattering it. Covering a subject’s subtopics thoroughly also reads as topical completeness — the site clearly treats this subject seriously, not as a single thin post.

For the model, the signal is different but aligned. An AI answer engine reading your cluster sees the same entities repeated consistently, a full subtopic map, and internal links that confirm which pages belong to the topic. That coherence raises its confidence that you are an authority on the subject — which is exactly what makes it more likely to cite you when it fans a question out into subtopics. (More on that in query fan-out.) Topical depth is the one signal both audiences read the same way.

Anti-cannibalization: one intent, one URL

The most common silo defect isn’t structure — it’s overlap. When two spokes target the same search intent, they compete with each other, split their equity, and force the ranker to pick between them (often picking the weaker one). The rule is blunt: one intent, one URL. If two pages answer the same question, merge them or sharpen each to a distinct angle, then point specific anchor text at each. Two pages both anchored “best running shoes” are a cannibalization bug waiting to surface. See semantic cannibalization for how to spot overlap before it costs you a ranking.

Link equity and orphan pages

You don’t fix leakage by banning all outbound links — a genuinely relevant link to another cluster is fine and often helpful. You fix it by making the default link target something inside the silo, and treating cross-cluster links as deliberate exceptions, not accidents.

Building a silo without leaking

• Every topic has one clear pillar page that links to all its spokes.
• Every spoke links back up to its pillar and sideways to its closest siblings.
• Each search intent maps to exactly one URL — no two pages compete.
• Internal anchor text is specific and page-distinct, never the same phrase twice.
• No orphan pages: every page worth ranking has a relevant internal link in.
• Cross-cluster links are deliberate exceptions, not the default target.
• The whole cluster is reachable within two clicks of the pillar.

Our silo and link-equity toolkit does the tedious part — it maps the actual internal-link graph, flags leaks and orphans, and measures how much of each cluster’s equity stays inside the silo.

Does silo structure actually move rankings?

Honestly: structure is a supporting signal, not a multiplier you can bolt on for an easy win. What it reliably does is stop you wasting the equity you already have and stop two of your own pages from fighting each other. The chart below is illustrative — the point is the direction, not the decimal: a leaky cluster lets a large share of its internal links point away from the topic; a tight silo keeps almost all of them inside.

Leaky cluster 60%

Tight silo 95%

источник: internal audits, 2026

The gap is the whole argument: equity that stays inside the cluster compounds on the pillar; equity that leaks out is authority you paid for and gave away.

In short

• A topical silo is a pillar hub plus supporting spokes, built around one subject.
• It concentrates internal link equity and topical depth where they earn rankings.
• Topical depth is the one signal both search rankers and AI models read the same way.
• One intent, one URL — overlapping spokes cannibalize each other and split equity.
• Leakage and orphans are the silent killers; audit the link graph, not just the pages.

Sources

• Google Search Central — “Make your links crawlable” / links best practices — how internal links are discovered and how they pass signals between pages.
• Page, Brin, Motwani & Winograd, “The PageRank Citation Ranking” (Stanford InfoLab, 1999) — the model behind internal link equity: a link is a vote, and votes flow along the graph.
• Aggarwal et al., “GEO: Generative Engine Optimization”, KDD ‘24 — arXiv:2311.09735: why coherent, complete topical coverage raises a passage’s odds of being cited by an answer engine.
• Google Search Central — “Creating helpful, reliable, people-first content” — topical depth and demonstrated expertise as quality signals.