TL;DR. A topic cluster is one pillar page (3,000+ words, the definitive local answer) surrounded by 8 to 15 supporting pages that each answer one specific question, all linked to each other. Engines trust depth on a topic: ten-plus pieces earn real citations, three scattered posts earn silence. By the end of this project you can scope your first cluster on the topic your audit showed you are losing.
- Watch us check our own cluster's link map (the receipts)
- What is a topic cluster (and a pillar page)?
- What does a real cluster look like?
- Why is Step 6 the honest chapter?
- How do you scope your first cluster, step by step?
- How do you make this your own?
- How do you train someone else on this?
- Where do you share your result?
- Frequently asked questions
Watch us check our own cluster's link map (the receipts)
This is Project 6 of 7 in the AI SEO playbook, and it is the honest chapter: this step is the part-time job. That is exactly why it is a moat.
What is a topic cluster (and a pillar page)?
A pillar page is the definitive answer to your market's biggest question: long, structured, answer-shaped, the page you would want an AI to read first. The cluster is the 8 to 15 supporting pages around it, each answering ONE specific buyer question, every one linking up to the pillar and across to its siblings.
The linking is not decoration. It is how machines learn that you do not just have an opinion on a topic, you have the topic covered. One good blog post is a stranger with a good answer. A cluster is the local expert.
What does a real cluster look like?
Here is the honest scale. The client work this playbook is distilled from involved roughly 226,000 words across 69+ pages for one local business: city pages, service pages, condition pages, and 113 FAQs, built and interlinked over months. That is what "win the whole market's questions" costs at full volume, and it produced the results in the case study.
You do not need all of that to move. You need ONE cluster, done properly, on the topic your Project 1 audit showed you are losing. And you are inside a live example right now: this curriculum is itself a cluster, a pillar with projects interlinked around it. We run the method we teach.
Why is Step 6 the honest chapter?
Because this is where the do-it-yourself line usually sits. Steps 1 through 5 are evenings and habits. Step 6 is sustained production: a pillar plus eight-plus pages, written to the answer-shaped standard, interlinked, with schema. The method is not the hard part; the volume is.
That is not a reason to skip it. It is a reason to scope it honestly: one cluster, on your one losing topic, at a pace you can sustain. Depth on one topic beats thin coverage of five, every time, with machines and with buyers.
How do you scope your first cluster, step by step?
The drill. This session is planning, not writing; give it 90 minutes:
- Pick your losing topic. From your Project 1 audit: the money question where the AI named a competitor and not you.
- Define the pillar question. The single question whose definitive local answer would win that topic ("cryotherapy in Modesto: what it is, what it costs, who it is for").
- List 8 to 15 supporting questions from your Project 2 prompt list: one page each, one question each, in the buyer's words.
- Draw the link map. Every supporting page links UP to the pillar and ACROSS to at least two siblings; the pillar links DOWN to every page.
- Plan the schema while you are in there. FAQ schema on the FAQs, article schema on guides, organization schema sitewide.
- Schedule it honestly. One supporting page a week is a strong pace for an owner. Put the dates in the calendar; a cluster half-built is a cluster.
Write the pillar first, using everything Project 4 taught, then produce the supporting pages in order of buyer intent (prices and "best for" questions first).
The pages you are reading are the worked example: one pillar, seven projects, six field guides, 14 pages on one topic, and we checked our own link map mechanically the day this page went live. The pillar links down to all 7 projects, all 7 link up, and every one links across to at least two siblings. Our page of original data is the whole confession series: 48 Search Console queries, 119 mapped citations, 1 of 5 GBP checks passing. Numbers only we have.
The honest part, honestly: this cluster took us 48 hours because building them is our whole craft. Your pace is one supporting page a week, exactly as the drill says, and a scheduled cluster beats a rushed one. And depth alone is not the win: 14 linked pages and our share of voice is still 2 of 10, because the engines also want time, reviews, and citations. That is Project 7.
The receipts live in the Project 6 post in our build-in-public thread on X.
How do you make this your own?
Your upgrade: one page of original data. Somewhere in your cluster, publish a number only you have: what your average client actually experiences, what something really costs across a season, the results of a simple customer poll. Original data is the hardest thing on the internet to copy and the easiest thing for an engine to cite, because you are the primary source.
Building clusters for a software company? The page types swap: the category definition page, vs pages, alternatives, and use cases. The B2B SaaS upgrade maps them. Selling products? The commerce cluster is Brand-vs-Brand pages and cost guides with real numbers: the eCommerce upgrade has the shapes.
How do you train someone else on this?
Divide the cluster. Give your manager or partner three supporting questions, teach them the answer-shaped drill from Project 4 on the first one, and review the two-sentence test together on everything they draft. Teaching the structure while sharing the load is how a part-time job becomes a team habit instead of a founder's burnout.
Where do you share your result?
Publish the pillar and say what it is: "We just published the most complete answer to [topic] in [city]. Every question we get asked, answered in one place." The pillar is the rare piece of content that is worth announcing on its own, and the announcement post becomes another mention pointing at it. Link back here and tag us; strong pillars get featured on the playbook.
Score yourself: this project plus Project 5 earns the citation point on the playbook's 0 to 5 AI-awareness score, because a real cluster is what gets you ONTO the pages AI cites. Next and last: Project 7 makes the whole system measurable.
Frequently asked questions
What is a pillar page?
A pillar page is the long, definitive, answer-shaped page on your market's biggest topic: 3,000+ words covering what it is, what it costs, who it is for, and every major question underneath, with links down to the supporting pages that go deep on each one.
How many posts does a topic cluster need?
One pillar plus 8 to 15 supporting pages, each answering one specific question, all interlinked. Ten-plus pieces on one topic is the depth threshold where engines start treating you as the authority; three scattered posts read as noise.
How long does a cluster take?
At an honest owner's pace of one supporting page a week, a full cluster takes two to four months. The pillar comes first and starts working immediately; the cluster compounds as each page joins the link structure.
Should I write about many topics or one?
One, first. Depth on one topic beats thin coverage of five, for machines and buyers. Win the cluster your audit says you are losing, measure it, then start the second.
Two ways to run this playbook. Get your free AI Visibility Score in about 30 seconds: 50 prompts, 3 AI engines, scored live, plus a free 20-minute walkthrough with Matthew. Or get The Playbook and keep doing it yourself.
