Guide

Pillar pages and content clusters, done right

Content clusters break when the pillar is just a link list or spokes compete for one query. Here's how to scope a pillar and link spokes that actually work.

Key takeaways
  • Scope a pillar by listing 8 to 15 distinct questions it must answer, each owned by one spoke; if you can only name three, write one page instead of a cluster.
  • Link the pillar to every spoke and each spoke back to the pillar, and allow spoke-to-spoke links only where a reader's next question is that specific page.
  • Before publishing any spoke, check that no existing URL already owns its search intent, and merge any two pages that compete for the same query.

Most content clusters on the web are a pillar page with a list of links under the headline, surrounded by a dozen posts that say roughly the same thing with different keywords in the title. The diagram looks right. In search it fails for two reasons. The pillar never answers anything on its own, so there is nothing substantial to rank. And the spokes compete with each other for the same queries, so Google keeps choosing between them and none of them wins.

A cluster is worth building when it does two jobs. It gives one topic a clear, complete home, and it makes your internal links say something true about how your topics relate. Get the scope and the linking right and the structure does real work. Get either wrong and you’ve built a lot of pages that cancel each other out.

What content clusters are (and what they aren’t)

A content cluster is a group of pages organized around one broad subject. One page, the pillar, covers the whole subject at overview level and links out to a set of spokes. Each spoke takes one narrower subtopic and goes deeper than the pillar can. The spokes link back to the pillar, and the pillar links to every spoke. That’s the hub-and-spoke model, and it’s the whole shape.

The term comes from marketing, and it describes an information architecture, not a Google feature. Google has never documented a “cluster” ranking signal, and publishing the right shape won’t flip a switch. The benefit is indirect, and it’s real. A cluster concentrates internal links on one topic. It assigns each search intent to exactly one URL. And it forces you to cover a subject in depth instead of scattering posts across it. If you expect the diagram itself to rank, you’ll be disappointed. If you use it to make your coverage of a topic legible to crawlers and to readers, it pays.

This is where the idea overlaps with topical authority. Topical authority is the outcome: your site becomes the obvious source on a subject. Clusters are one of the most practical ways to get there, which is why people treat the two as the same thing. The broader strategy is in how to build topical authority that ranks. This piece is about the structure itself: how to scope it, how to link it, and how it breaks.

Why does the shape help at all? Think about what a search engine and a reader each have to work out. The engine has to decide which of your pages best answers a broad query, and which pages are the detailed answers underneath it. A reader has to move from the general to the specific without hunting. A cluster answers both questions by making the relationships explicit in the link graph. The pillar’s links say “these pages belong to this topic.” Each spoke’s link back says “this is one piece of that topic.” The anchor text carries the rest.

Write the invoiceInvoice numberingPayment termsOverdue invoicesRetainer billingInternational clientsPillar pageFreelance invoicing
One pillar, six spokes. Solid lines: the pillar links to every spoke, and each spoke links back. The dashed line is the one spoke-to-spoke link, drawn because a reader who finishes one page needs the other next.

Scope the pillar before you write a word

Most cluster failures start here. The pillar gets scoped too broadly, so no set of spokes can satisfy it. “Marketing” is a subject, not a page. Or it gets scoped too narrowly, so the pillar has no reason to exist and the “cluster” is two posts that happen to be related. The right size sits between those extremes, and three tests will find it.

  1. The question list test. Write every question a buyer asks that falls under the pillar’s heading. You want roughly 8 to 15 distinct ones. If you can name only three, you have a long article, not a pillar. If you can name sixty, you have two or three pillars sharing a name.
  2. The distinct-intent test. Each question needs its own page-level answer. “How do I write an invoice” and “what should a freelance invoice template include” look different, but they only justify two pages if the answers diverge. If both answers would be the same paragraph, they are one spoke.
  3. The overview test. Read the pillar’s overview alone. Does it answer the broad question well enough that a reader could stop there and feel served? A pillar that only teases its spokes fails this test, and failing it is the most common mistake in cluster design.

Scope rule: if you can't write a one-line question for a spoke, that spoke doesn't have a job yet. Cut it or rethink the pillar.

Before you commit, check demand. Run the pillar’s head term and a sample of its questions through your Google research process and confirm that people actually search them. A cluster built around queries nobody types is a conversation nobody is having.

Build the pillar as a map and an answer

A pillar has two jobs, and most pillars fail at one of them. It must answer the broad question completely at overview depth. It must also act as a map, routing readers to the specific page for each subtopic. A page that only does the second job is a table of contents. A page that only does the first is an article with no reason to link anywhere.

A structure that works looks like this:

  • Open with the framing. Name the topic, say who it’s for, and give the single most useful way to think about it. For a pillar on invoicing, that might be “every invoice is a request for money you haven’t received yet, and everything else is about making that request easy to say yes to.”
  • Give each subtopic a real overview section. A few solid paragraphs that answer the subtopic at its own level, so the reader is served without clicking. Then one contextual link to the spoke for the full treatment.
  • Summarize the map. A table listing each spoke, the question it answers, and its link. Readers use it to navigate, and it gives you a clean check that every spoke is linked from the pillar.

The pillar will usually run longer than a standard post, and that’s expected. But if it needs ten thousand words to cover its overview, the scope is wrong and the spokes are doing the pillar’s job. Review the pillar every time you add a spoke, because its map has to stay accurate.

Here is how the two page types differ in practice:

Pillar Spoke
Job Answer the broad question and map the topic Answer one specific question in depth
Target The head term, such as “freelance invoicing” One query and one search intent
Depth Complete at overview level Complete at full depth, with mechanism and edge cases
Links out To every spoke, once each Back to the pillar, plus at most one or two siblings
Links in From site navigation and high-traffic posts From the pillar, and from sibling spokes where relevant
Fails when It reads as a list of links It only restates the pillar

Build spokes that stand on their own

A spoke is an article. It targets one query with one intent, and it has to pass the same test as any standalone post. Someone who lands on it from search, with no context from the pillar, should get a complete answer to the question in the title.

That changes how you write it. The spoke goes deeper on its subtopic than the pillar does. It explains the mechanism, not just the steps. It includes a worked example with real numbers, and it names the edge cases where the standard advice breaks down. A spoke that ends with “see our guide for more” has failed. It is the pillar’s content under a different headline.

Write every spoke with the same brief and voice as any other piece on the site. There is no lower bar for pages that happen to sit in a cluster. If anything the bar is higher, because the spoke’s title has to match a query closely and its opening has to deliver on that promise immediately.

Linking rules between pillar and spokes

The link graph is where a cluster either works or turns into noise. Four rules cover most of it.

  1. The pillar links to every spoke. Each spoke gets a contextual link from the matching section of the pillar, plus an entry in the summary table. An unlinked spoke is an orphan. Search engines find orphans slowly, if at all, and readers almost never reach them.
  2. Every spoke links back to the pillar. Place one clear link where the reader needs the wider context, usually within the first few paragraphs. Vary the anchor from spoke to spoke. Use something like “our full guide to freelance invoicing” on one page and “the complete invoicing overview” on another. Stamping the identical exact-match phrase on every page reads like a template, and it looks manufactured.
  3. Link spokes to siblings only where the next question is that sibling. “How to write an invoice” should link to “how to number invoices,” because a reader who has finished one usually wants the other next. “Overdue invoices” should link to “payment terms,” because late payment is often a terms problem. Don’t build a mesh where every spoke links to every other spoke, usually from a footer widget. A mesh makes every link look equally important, which means none of them carries much signal, and it produces link blocks readers skip.
  4. Describe the destination in the anchor. “Read more” and “click here” say nothing. “the payment terms that make late payment less likely” tells a reader and a crawler exactly what they’ll find on the other side.

Put links in the body text wherever you can. A “related posts” block at the bottom is a fine addition, but it can’t carry the structure alone, because it’s generated and carries no context about why two pages connect. Keep every in-body link to something a reader would actually click. For how link equity moves through a whole site, read an internal linking strategy that actually moves rankings. The rules above are the cluster-specific part of that picture.

A worked example: freelance invoicing

Say you sell invoicing software to freelance designers. Your pillar targets “freelance invoicing,” and its scope is everything between creating an invoice and getting paid, for a solo service business. Here is how you would build the cluster from there.

Start with the question list. A realistic version looks like this:

  • How to write a freelance invoice, and which fields must appear on it
  • How to number invoices so the sequence still works in January
  • Payment terms: net 15, net 30, due on receipt, and when to take a deposit
  • How to chase an overdue invoice without damaging the client relationship
  • Late fees and interest: how to state them in your terms
  • Invoicing international clients: currency, tax IDs and VAT
  • Retainers and milestone billing for ongoing project work
  • Free freelance invoice templates

That’s eight questions, inside the range. Two things stand out.

First, “invoice software for freelancers” is missing on purpose. It’s a commercial comparison page, and mixing it into an informational cluster muddies what the cluster is for. Keep it on your product site and link to it from the pillar only where a reader would find it useful.

Second, two entries overlap. “How to write a freelance invoice” and “Free freelance invoice templates” serve the same person: a freelancer who needs an invoice they can use today. If you publish both as separate posts, you have two URLs competing for one need. Google will pick one, the other drifts, and the links you spent building get split between them. The fix is to decide which page owns the need. Here the template is a utility, so it belongs inside the how-to spoke as a downloadable block. The standalone template post goes away, and its URL redirects to the how-to. That’s a real judgment call, and it’s the one teams most often skip.

Two URLs, one intent: when two pages compete for the same query, search engines have to choose between them for every search, and the choice can flip. Merge the weaker page into the stronger one and 301-redirect the old URL.

Third, a thin spoke. Suppose someone proposes a page called “What is an invoice?” At 250 words it’s a definition. It fails the standalone test, because a reader who wants a real answer gets a dictionary entry and leaves. It belongs in the pillar’s opening section, where the framing already lives. Merge it there.

Once the list is cleaned up, you have six or seven spokes, each with a distinct intent and each linked from the pillar. Publish the two or three with the strongest demand first, and let the data decide which of the rest deserve a full treatment. That sequencing matters more than how many pages you launch with.

Where content clusters fail, and how to spot it

The failure modes below share a pattern. They look fine in a diagram and break when you check the data. Each has a symptom you can find and a fix you can apply.

Thin spokes

The symptom is a spoke that earns impressions in Search Console but almost no clicks, shows no ranking movement over months, and reads like a paragraph that has been stretched to fill a page. Usually it’s a near-duplicate of a pillar section. The fix is honest: rewrite it to answer its question fully, or merge it into the pillar. If an honest rewrite still can’t pass the standalone test, it was never a spoke.

Keyword overlap

The symptom is two of your URLs appearing for the same query, alternating or sharing the ranking. Find these in Search Console by filtering queries and checking which URLs they map to. The fix depends on the cause. If the intents really are the same, pick the stronger URL, merge the content, and redirect. If the intents really differ, make the difference visible. Rewrite the titles and first headings so each page’s job is clear from the outside, and update the internal links so each one points at its own query.

The pillar that is a link list

The symptom is a pillar with a strong title, a short intro, and forty links. Traffic lands on it and leaves almost immediately. Search engines have little real content on the page to evaluate. The fix is to write the overview sections described earlier. A map with no terrain is not useful to anyone.

Premature clusters

The symptom is a plan for fifteen spokes, all published in one week, several of them aimed at queries that barely get searched. The fix is to check demand for each spoke before writing it, publish the strongest first, and treat the rest as a backlog you activate on evidence.

The honest wrap-up

A content cluster is an honest way to organize a topic. It gives the subject one clear home, gives depth to the questions people actually ask, and uses links that say what they mean. It won’t make a weak site strong, and it won’t rank a thin page just because a pillar sits above it. What it does is make your work easier to see, both for a reader moving through the site and for a search engine trying to work out what you know.

Most teams that struggle with content clusters are really struggling with research. They picked a pillar and its spokes from instinct instead of from demand. AutoSEO Stack starts from what buyers actually search, writes each page as a standalone article with its own intent in a human voice, and publishes on a schedule, so the same quality bar applies to every URL, not just the pillar. That’s the discipline this article describes, applied page by page. If you want to see it running on your own site, start with AutoSEO Stack.