Topical Authority: Building Content Clusters That Rank

Ten years ago you could rank a single page on a strong domain for almost anything. Write the post, point some links at it, wait. That playbook is mostly dead. Today, Google — and increasingly the AI assistants that sit on top of the web — decide who to trust on a topic by looking at how completely a site covers that topic. That trust has a name: topical authority. And the practical unit for building it is the content cluster.

TL;DR: Topical authority is Google's measure of how completely you cover a subject. You build it through content clusters — a pillar page plus tightly-linked supporting posts — not through writing one great article. Coverage beats volume, and internal linking is what makes the cluster work.

We're building clusters on this very blog right now, so this isn't theory recycled from someone else's case study. This post covers what topical authority actually is, why it matters more now, how to map and structure a cluster, how to measure it, and the two failure modes that quietly kill most cluster projects.

What Topical Authority Actually Is

Topical authority is a search engine's confidence that your site is a real source on a subject — earned by covering that subject completely, not by domain age or raw backlink counts.

The distinction matters. Classic SEO treated authority as a site-wide score: get enough links, and every page you publish inherits the power. Topical authority is narrower and, honestly, fairer. A five-year-old site with 40 interlinked posts about email deliverability can outrank a 20-year-old news domain's single article on SPF records — because the smaller site has demonstrated, page by page, that it actually understands the topic.

Search engines infer this from signals you control:

• Breadth: do you answer the head term and the long-tail questions around it?
• Depth: do individual pages go further than restating what already ranks?
• Structure: do your pages link to each other in ways that show topical relationships?
• Consistency: have you kept publishing on this topic, or did you write three posts in 2023 and stop?

None of those require a big domain. All of them require a plan.

One thing topical authority is not: a score you can read in a tool. Third-party metrics are reverse-engineered guesses. The only honest measurement is behavioral — does Google show you for more of the topic over time? More on that below.

Why It Matters More Now

Two shifts pushed topical authority from "nice framework" to "the main game."

The Helpful Content Update changed the math

Google's Helpful Content system (rolled into the core algorithm in 2024) introduced site-wide classifiers. If large portions of your site look like content produced for search engines rather than for people with real questions, the whole site gets dragged down — including the good pages.

The inverse is also true. Sites that demonstrate genuine subject focus get evaluated more generously. A blog where every post connects to a coherent topic reads as "people who know this field" to a classifier. A blog that chases whatever keyword tool spat out that week — recipe roundups next to crypto guides next to CMS comparisons — reads as content farming, even if individual posts are decent.

Post-HCU, your worst off-topic content actively taxes your best content. That alone justifies clusters over scattershot publishing.

AI search cites sources that cover topics deeply

The second shift is bigger. When ChatGPT, Perplexity, Claude, or Google's AI Overviews answer a question, they cite a handful of sources. Watch which sources get picked: it's rarely the single viral post. It's the site that has the whole topic — the definition page, the comparison, the how-to, the edge cases.

This makes sense mechanically. An LLM assembling an answer about, say, headless CMS migration needs material on pricing, hosting, API differences, and migration steps. A site that covers all four in interlinked pages is one retrieval source that satisfies the whole answer. A site with one good post is a fragment.

So clusters now pay twice: classic rankings on each post, plus citation share in AI answers. If you want the second payoff, make your content easy for crawlers and LLMs to ingest — clean HTML, structured data, and an llms.txt file that maps your content for AI crawlers all help the machines find the cluster you built.

Anatomy of a Content Cluster

A cluster has three parts: one pillar page, a set of cluster posts, and the links between them. Get the proportions wrong and the whole thing underperforms.

The pillar page

The pillar targets the head term — the broad, high-volume query like "headless CMS" or "email deliverability." It's usually 2,500–4,000 words and covers the entire topic at medium depth: every subtopic gets a section, every section links out to the cluster post that goes deeper.

The pillar's job is twofold. For readers, it's the orientation page — the one you'd send a colleague who asked "explain this topic to me." For search engines, it's the hub that declares "this site owns this subject" and distributes relevance to every spoke.

The cluster posts

Each cluster post targets one long-tail query at full depth. Not three queries. One. "Headless CMS vs traditional CMS" is one post. "How much does a headless CMS cost" is another. "Best headless CMS for Next.js" is a third — even though all three could plausibly live as H2s in a mega-post.

Plan for 8–15 cluster posts per pillar. Fewer than 8 and you haven't demonstrated coverage; the topic still has obvious unanswered questions. More than 15 and you're usually either splitting hairs (cannibalization risk — see below) or you've actually found two topics wearing one trench coat, and should split into two clusters.

The interlinking

This is the part most teams do badly. The rules are simple:

• Every cluster post links up to the pillar, in the body, with descriptive anchor text — not just "click here."
• The pillar links down to every cluster post from the relevant section.
• Cluster posts link sideways to 2–4 sibling posts where genuinely relevant.

Dense, deliberate interlinking is what turns 12 posts into a cluster instead of 12 orphans. It concentrates link equity, helps crawlers understand the relationships, and — practically — keeps readers on your site for a second and third page. On UnfoldCMS we lean on categories as cluster hubs: each category page acts as a structural index of its cluster, sits in the XML sitemap, and gives every post a machine-readable home that mirrors the topical structure we want search engines to see.

How to Map a Cluster Before Writing Anything

Mapping is the most valuable hour of the entire project. Skipping it is how you end up with 60% of a topic covered and three posts competing for the same query.

Step 1 — Fix the head term. Pick one. It should be broad enough to support 10+ subtopics, narrow enough that you can plausibly own it. "Marketing" is too broad. "Headless CMS" is about right for a CMS company.

Step 2 — Enumerate everything around it. Pull every question, comparison, how-to, and modifier you can find: autocomplete, People Also Ask, Reddit threads, your own support inbox, sales call notes. You're not filtering yet — just collecting. A real enumeration for a mid-size topic produces 40–80 raw queries.

Step 3 — Group by intent. Collapse the raw list into distinct intents. "Headless CMS pricing," "headless CMS cost," and "how much does a headless CMS cost" are one post. This grouping step is where cannibalization is prevented or created.

Step 4 — Check what already ranks. Search each intent. If page one is all domains you can't realistically compete with and the content is genuinely good, deprioritize it for now. If page one is thin, outdated, or generic — that's your opening. Most topics have more openings than you'd expect, especially on comparison and "alternatives" queries where the ranking pages are often stale listicles.

Step 5 — Sequence. Write the pillar first or in parallel with the first few cluster posts, then fill the map. Don't publish the pillar alone and let it sit linkless for months.

A typical mapped cluster looks like this:

Coverage Beats Volume

If you remember one thing: 30 posts on one topic beat 100 posts scattered across ten topics. Every time.

This feels wrong to teams trained on output metrics. A hundred posts looks like more surface area, more lottery tickets. But search engines don't reward lottery tickets anymore — they reward demonstrated expertise, and expertise is legible only when content concentrates.

The scattered site sends a confusing signal: who is this site for, what does it know? The concentrated site sends a clean one. It also compounds — post #14 in a cluster ranks faster than post #1 did, because it launches into an established topical context with a dozen internal links already pointing at its subject area. Post #87 on a scattered blog launches into nothing.

Cadence and Patience

Topical authority builds over months, not sprints. Expect 3–6 months before a new cluster shows clear movement, longer in competitive niches. Consistency is part of the signal: a burst of 12 posts in one week followed by silence looks different from 12 posts over 12 weeks, and the steady version wins.

So pick a cadence you can hold — one or two cluster posts per week is plenty — and hold it. This is a workflow problem more than a writing problem. Batch the writing when energy is high, then schedule the releases. We do exactly this with UnfoldCMS's scheduled publishing: posts are written in batches and queued weeks out, so the blog ships on rhythm even when nobody touched it that week. The CMS also handles the plumbing that keeps a growing cluster healthy — JSON-LD on every post, and slug history redirects so that renaming a post mid-cluster doesn't orphan the internal links you spent months building. If you're evaluating platforms for this kind of work, our CMS SEO checklist covers the ten features that matter.

The patience part is real. The first six weeks of a cluster usually look like failure: impressions trickle, rankings bounce around page four. This is normal. Google is still discovering the structure. The teams that win are the ones that keep publishing through the flat part of the curve.

Measuring Topical Authority

Don't measure single-post rankings. Measure the cluster.

The metric that matters is total impressions across all cluster URLs in Google Search Console, tracked monthly. Impressions are the leading indicator — Google showing your pages for more queries, more often, is the literal definition of growing topical trust. Clicks and rankings follow impressions by weeks or months.

The practical setup:

• In Search Console, filter Performance by page path or regex matching your cluster URLs.
• Record total impressions and total distinct queries monthly.
• Healthy cluster: impressions grow month over month, and the query count grows too — your pages start appearing for long-tails you never targeted. That second effect is the clearest sign of authority: Google trusting you with queries you didn't even write for.
• Watch the pillar specifically. Pillars are usually last to rank. When the pillar starts moving for the head term, the cluster has matured.

A single post jumping to position 3 tells you that post is good. Forty percent impression growth across 12 URLs tells you Google now treats your site as a source on the topic. Only the second one compounds.

The Two Failure Modes That Kill Clusters

Cannibalization: two posts, one intent

The most common failure. Two cluster posts target queries that look different but carry the same intent — "headless CMS benefits" and "why use a headless CMS," say. Google can't decide which to rank, so both float between positions 8 and 20 forever, swapping places weekly.

The fix is discipline at the mapping stage: one intent per post, decided before writing. If you already have cannibalizing posts, merge them — combine the content into the stronger URL and 301 the weaker one. The merged page almost always outranks both originals within weeks.

The diagnostic is easy: in Search Console, check whether one query returns impressions for two of your URLs. If both get meaningful impressions for the same query and neither ranks well, you've found cannibalization.

The pillar that's just a link list

The second failure: a "pillar" that's 800 words of intro plus a wall of links to the cluster posts. Google treats it as a doorway page, readers bounce off it, and the head term never moves.

A real pillar must be worth reading on its own — someone who reads only the pillar and clicks nothing should still come away with a working understanding of the topic. The cluster links are exits for readers who want more depth, not the content itself. If your pillar would be useless with the links removed, it isn't a pillar yet.

A Realistic Example: A CMS Company Clusters "Headless CMS"

Here's how this plays out concretely — and yes, this is roughly the shape of what we're doing.

A self-hosted CMS company wants to own "headless CMS" territory. The pillar is a full guide to what headless CMS means, when it's the right architecture, and when it isn't. Around it, the cluster posts each take one intent:

• Decision posts: "benefits of a headless CMS," "is headless overkill for a blog?"
• How-tos: connecting a frontend framework, migrating from a monolithic CMS.
• Comparisons and alternatives — historically the fastest movers, because buyers search them with high intent and the incumbent results are often outdated. Our Strapi alternatives and Sanity alternatives posts are exactly this layer of the cluster: each targets one tool's switchers, each links to siblings and up to the broader headless content.
• Edge cases: self-hosting a headless CMS, API authentication, content modeling.

Every post links up to the pillar, the pillar links down to all of them, and siblings cross-link where the topics genuinely touch. Twelve to fifteen posts, shipped over a quarter at a steady cadence, measured as one unit in Search Console.

Notice what's not in the plan: no off-topic traffic grabs, no "10 productivity tips for developers" because a keyword tool said it had volume. Everything serves the topic.

Where to Start

Pick the one topic your business most needs to own — the topic your best customers research before they find you. Map it fully before writing anything: head term, 40+ raw queries, grouped into 8–15 single-intent posts plus a pillar. Then ship at a cadence you can sustain for six months, link every post into the structure the day it publishes, and judge the work by cluster-wide impressions, not individual rankings.

Coverage beats volume. Structure beats luck. And six months of focused publishing beats five years of scattered output — we've bet this blog on it.

Frequently Asked Questions

How long does it take to build topical authority? Typically 3–6 months after publishing a complete cluster. Google needs time to crawl all posts, process the internal links, and observe user engagement signals. Sites with existing domain authority may see results faster; new domains take longer.

How many posts do I need for a content cluster? A minimum viable cluster is one pillar page plus 5–8 supporting posts covering distinct subtopics. Quality and completeness matter more than volume — 6 well-researched posts beat 20 thin ones.

Does topical authority apply to new websites? Yes, but it takes longer. A new site can build topical authority by going deep on a narrow topic fast, rather than spreading thin across many topics. Pick one cluster, publish it completely, then expand.

What is the difference between a pillar page and a supporting post? A pillar page covers the main topic broadly (2,000–4,000 words, links out to all supporting posts). Supporting posts go deep on one subtopic (1,500–3,000 words, link back to the pillar). The pillar is the hub; supporting posts are the spokes.

Can internal linking alone build topical authority? Internal linking signals the relationship between pages, but topical authority also depends on content quality, backlinks, and user signals. Internal linking is necessary but not sufficient — you need genuinely useful content that people engage with.