Main area
Agency
on-page-seo

Content Strategy with Topic Clusters — Hub, Spoke, and Internal Linking

Redaktion ·

When every article fights alone — and none of them wins

You’ve published 40 blog posts. Each one is well researched, each targets a distinct keyword. Rankings hover around positions 15–30, with a few lucky shots on page one. The classic pattern: standalone articles without architecture. Every post is an island, Google sees no context, no theme, no expert.

Topic clusters fix exactly this problem. Instead of 40 islands you build five to ten topical areas, each with one central hub page (the pillar) and several detail articles (the spokes), all interlinked. After this article you’ll know how the architecture works, where it typically fails, and how to retrofit it into an existing site step by step — without starting from scratch.

Core mechanics — hub, spoke, pillar

A topic cluster is a group of pages that together cover one bounded topical area. Three roles divide the work.

Pillar page (hub)

The central page for the broader topic. Wide in scope, often 2,000–4,000 words, covering the area once in full — at a level that works for someone just getting oriented. Example pillar: “Search Engine Optimization — the Complete Guide.” It answers mid-funnel questions itself and links to spokes for depth.

The pillar carries the contested short-head term (seo, content-marketing, running-an-online-shop). On terms like these a smaller site has no real chance — except through the internal backing of its spokes.

Spoke pages (cluster content)

Detail articles on sub-aspects, each answering exactly one question and aiming at the long tail. Per pillar typically 8–25 spokes. Example spokes off the SEO pillar: “Writing the title tag,” “Internal linking for online shops,” “Measuring Core Web Vitals.” Every spoke links back to the pillar — and the pillar links to every spoke. That’s the namesake hub-and-spoke shape.

Spokes also link to each other, but only where reading flow makes it natural — not as duty. Someone reading “Writing the title tag” who hits a mention of meta descriptions gets a link to the meta-description spoke. That turns the cluster from a star into a web.

Why this actually works

Three mechanisms combine.

Topical authority. Search engines don’t rate single pages alone — they rate how comprehensively a domain covers a topical area. A site with 25 connected SEO articles reads as a more credible source than a site with three scattered ones. That advantage feeds back into rankings even for pages that, in isolation, would never compete.

PageRank distribution through internal links. Every inbound link distributes link equity. In a cluster, internally distributed equity concentrates on the pillar (every spoke links back) — which then strengthens the spokes in turn. External backlinks pointing at spokes flow through cross-links and the pillar back into the cluster. One strong backlink turns into broad effect.

Clear search intent per page. When every spoke answers exactly one question, keyword cannibalization disappears — that case where two pages of the same domain compete for the same keyword. Cannibalization is one of the most common silent issues in older blogs without cluster logic.

Pitfalls — where clusters fail

Wrong pillar scope

A pillar pitched too narrow (“SEO for local hairdressers in Hamburg-Eimsbüttel”) gives the central page too little volume; pitched too wide (“marketing”) it goes so generic that it loses to brand sites and produces no coherent spoke list. Rule of thumb: the pillar should cover a topic for which an interested reader can comfortably formulate ten to twenty concrete follow-up questions.

Spokes without search volume

Planning spokes purely along a logical topic map without keyword data ships articles nobody searches for. Cluster logic is often used as an excuse to skip that check — it isn’t one. Verify search volume per spoke first, then consolidate the plan.

Linking as duty work

Spoke links to pillar with anchor text “learn more” at the bottom of the page. Doesn’t work. Internal links carry weight when they sit inside body text, use a meaningful anchor (the actual target keyword or a natural variant), and live where a reader would actually click. A footer link out of obligation counts structurally but barely registers in user signals.

Pillar never gets updated

Pillar pages age faster because they’re broad. An SEO pillar from 2022 with no GEO section is materially off in 2026. A cluster without an update rhythm for the pillar decays over time. Plan at least one major revision per pillar per year.

Clusters in name only

Existing blog archives sometimes get rebranded as “clusters” without actually reworking internal linking. That’s theater, not a cluster. Without consistent hub-spoke links it stays a stack of standalone articles with a new label.

Levers — how you build clusters

Cluster research: from keyword to topic map

The starting point isn’t the topic map but the keyword database. Practical sequence:

  1. Pick the short-head keyword that will carry the pillar. Check search volume, competition, and business relevance.
  2. Collect long-tail keywords around it — via SERP features (People Also Ask, Related Searches), via tools (Ahrefs, Semrush, Sistrix), and via your own Search Console.
  3. Cluster the collected keywords by search intent — not by lexical similarity. “Learn SEO” and “SEO basics” belong in the same spoke; “SEO agency Hamburg” doesn’t belong in the cluster, it belongs on a service page.
  4. Per cluster, build a spoke list with keyword, intent, working title, and a note on whether the pillar already has a section for it.

Twelve to twenty spokes per pillar is a healthy band. Fewer than eight: the cluster looks thin. More than 25: usually two clusters in a trench coat — better split.

Writing the pillar — wide, navigable, linked

The pillar has to serve two reading modes at once: linear top-to-bottom reading and scanning (table of contents, jump to the right section). In practice that means a clear TOC up top, descriptive H2 headings, compact H3 blocks. Every major pillar section contains a pointer to the matching spoke — typically a sentence inline and/or a “More on this in Writing the title tag” cue at the section’s end.

The pillar isn’t a 4,000-word textbook on everything. It’s the front door with twenty clearly signposted rooms.

Internal linking — the actual work

The architecture only works if the links are placed well. Three rules:

  • Every spoke links back to the pillar at least once, in body copy, with a descriptive anchor (not “click here”). Often in the intro, often again in the conclusion.
  • Cross-links between spokes emerge organically as you write — when spoke A naturally references the topic of spoke B, link it. Don’t force, just take it along.
  • Vary anchor texts but keep the core keyword recognizable. “Writing title tags well,” “Title tags and their effect” — both fine. “Click here” — wasted.

Tools like Ahrefs Site Audit, Screaming Frog, or a small custom script will show pages with too few internal links (orphans or near-orphans). That’s the list to attack.

Programmatic SEO as a cluster amplifier

Programmatic SEO builds many structurally similar spoke pages from a data source — typically long lists like “X for Y” (“SEO agency in city,” “CRM for industry”). It only works when paired with a strong pillar and clean data hygiene. The frequent anti-pattern: 5,000 generated pages with no topical hub, no per-page value, no internal linking. Google now reliably devalues those. Anchored to a pillar, with a clear data foundation and at least one non-generated block per page (experience, context, original data), it becomes a legitimate lever.

Comparison — cluster types at a glance

| Type | Pillar | Spokes | When useful | Risk | |---|---|---|---|---| | Classic hub-and-spoke | 1 broad pillar, 8–25 spokes | Long-tail articles | Established topic stem, B2B content marketing | Pillar maintenance, ages quickly | | Nested subcluster | Pillar → sub-pillar → spokes | Three levels | Very wide topics (whole disciplines) | Architecture grows complex, link upkeep balloons | | Programmatic cluster | 1 pillar, many generated pages | Data-driven sub-pages | Structured data, city/industry/product lists | Without per-page value → thin content | | Topic hub without a pillar | Just an overview hub with link list | Existing articles | Quick consolidation of legacy content | Wastes pillar effect, no real authority concentration |

Practice — retrofitting an existing site

Three realistic scenarios.

Scenario 1: blog with 60 articles, no cluster

Inventory: every article with main keyword, monthly traffic, and top-ranking keyword in one table. Form topic buckets — 60 articles typically fall into 4–7 sensible clusters. Per cluster, check: is there already a broad “SEO basics” article that could serve as the pillar? If yes: expand it, add a TOC, link to all spokes. If no: write the pillar from scratch. Then walk every spoke and place one clean link back to the pillar. Realistic effort for 60 articles: two to three weeks of focused work, plus the new pillar pages.

Scenario 2: shop with categories as pillars

For shops, category pages are often the natural pillars. Spokes are blog guides that link to the category. Example: pillar “Running shoes” (category) — spokes “Running shoes for heavier runners,” “How to measure your shoe size correctly,” “Pronation explained.” Important: the category page must carry editorial content (not a filter strip with two sentences), otherwise it can’t carry the pillar role.

Scenario 3: pillar exists, spokes are missing

Common with younger sites: there’s a big guide but only three detail articles. Run the long-tail research per pillar section first, define a 12–20-spoke roadmap, then ship two to four spokes per month. Important: maintain the pillar in parallel — every new spoke gets linked from the right pillar section.

FAQ

How is a topic cluster different from a sitemap structure?
The sitemap shows URL hierarchy, the cluster shows topical hierarchy. A cluster needs no URL depth of its own — pillar and spokes can sit flat under /blog/. The connection is built through internal links, not through paths.
How many clusters should a site have?
Three to ten is a healthy range for most content-driven sites. Fewer than three: the business is narrow, often one cluster plus service pages is enough. More than ten: usually fragmented, better to merge two.
Does every article have to be in a cluster?
No. News posts, company updates, and personality pieces stay outside. Clusters need topics with search volume and long-term relevance — daily-business writing doesn't have that.
How long until a cluster shows results?
Three to nine months, depending on domain authority and competition. The effect rarely arrives linearly — often nothing visible happens for a long stretch, then the pillar suddenly pulls rankings, and the spokes follow over weeks.

Conclusion

Topic clusters aren’t SEO magic, they’re a structural principle: topics instead of single keywords, hubs instead of islands, linking instead of stacking. The lever lives in disciplined execution — maintain the pillar, build spokes around real long-tail keywords, actually wire up the internal links. Sites that hold this line build authority in a topical area that no lucky-shot single article ever reaches.

For most sites the right sequence is: inventory existing content, define three to five clusters cleanly, work out the pillar pages for the two most important clusters first, then pull the link structure together. Only then does it pay to write new spokes — otherwise you’re building on architecture that doesn’t yet hold. And for everything that touches AI search (see the GEO whitepaper), the cluster logic is the precondition for being recognized as a coherent source at all.

Themenuebersicht

On-Page SEOContent-Struktur