SEOProfessional services

Building a Content Cluster Around a Service

How a cluster is planned before a word is written, and why the link map comes first.

This is an implementation example, not a client case study. It describes how this work is actually carried out — the method, the sequence and the reasoning. No client is named and no result is claimed, because inventing either would make it worthless as evidence.

Topic cluster plan from pillar through supporting articles to service page

Most content programmes are a publishing schedule with no architecture — an article a week, chosen from a keyword tool, linked to nothing. They produce a pile of pages that individually rank for nothing and collectively tell Google the site has no subject.

Project type
Topic cluster build supporting a commercial service
Sector
Professional services

The challenge

A small site cannot out-authority an established competitor across many topics. It can be the best resource on a narrow one — but only if the coverage is deliberate, connected, and finished rather than abandoned halfway.

The objective

A cluster where a reader with any question about the topic finds an answer, every article supports the others, and the whole thing routes to the service page it exists to support.

Approach

Plan the whole cluster — topics, intents, internal links, commercial destination — before writing anything. Writing first and linking afterwards produces the incomplete, stale hand-added linking that makes most content programmes underperform.

Implementation

1. Collect the real questions

From sales calls and support enquiries, not a keyword tool. The questions people are slightly embarrassed to type are the ones with the least competition and the clearest intent.

2. Assign one intent per question

Group them. Anything that resolves to the same intent is one page, not two — two pages on the same intent compete with each other and both lose.

3. Pick the pillar

The broadest question that still has a single coherent answer. It will not rank first; the supporting articles will. Its job is to be the page that eventually becomes competitive once the cluster has established relevance.

4. Map the links before writing

Every article declares:

  • the pillar it supports
  • two or three siblings it links to
  • the service page it routes to
  • any worked example worth referencing

Recording this as data rather than as prose means the links can be generated and verified, instead of being remembered.

5. Write in order of commercial value

Not alphabetically, not easiest-first. The pages closest to a purchase decision get written while enthusiasm is highest.

6. Measure query coverage, not rankings

The signal that a cluster is working is growth in the number of distinct queries generating impressions — including queries you never targeted. That appears months before headline rankings move, and it is the difference between "this is working slowly" and "this is not working".

Technology

  • Google Search Console
  • Content model
  • Internal link map
  • JSON-LD

Decisions worth explaining

Internal links declared as data, not written by hand

Hand-added links are incomplete the day they are written and stale within a year. Declared relationships can be generated, validated and audited.

One page per intent, strictly

Two pages on the same intent cannibalise each other. Google picks one — usually not the one you would have picked — and both perform worse than a single good page.

Query coverage as the leading indicator

Rankings are lagging and noisy. Distinct-query growth moves first and tells you whether the topic is being understood.

What it produces

A cluster with a pillar, supporting articles covering the real questions, declared internal links that can be verified automatically, and a clear commercial destination. Traffic outcomes depend on competition and time.

What it teaches

  • Ten articles on one topic outperform fifty across ten.
  • The link map is the part that makes a cluster a cluster rather than a folder.
  • Publishing to hit a number produces filler, and filler drags the cluster down.
  • Query coverage tells you it is working long before rankings do.

Related services

Related guides

Worked examples

All case studies