A topical authority content cluster is one pillar page covering a subject broadly, surrounded by eight to twelve pieces that each answer one specific question inside it completely, all linked together. If you already have scattered posts, building one is mostly an audit and a merge, not a publishing project. You will usually end up with fewer articles than you started with, and rank better for it.
That last part is the bit nobody sells you, because there’s no invoice in deleting things.
Why do scattered blog posts underperform?
Because they compete with each other, and because none of them is the obvious answer to anything.
Here’s the mechanism. You’ve written three posts over two years that all partially cover “how to choose a CRM”. None was written as the definitive answer, so each covers about 40% of it. Google can’t tell which to rank, splits signals between them, and all three sit around position twelve.
Merge them into one complete piece and you have a single page answering the whole question, with every internal and external link pointing at one URL instead of three. That page can realistically reach the first page. Nothing was added.
The second problem is context. When an engine decides whether you’re a credible source on CRM selection, one article looks like an opinion. Twelve connected articles look like a body of work.
How do you turn existing posts into a topical authority content cluster?
Work through these in order. Steps 1 and 4 do most of the work.
- Pick one pillar topic you can plausibly own. Scope it to where your specificity beats a bigger competitor’s authority. A six-person consulting firm choosing “digital transformation” as a pillar has picked a fight with McKinsey. “ERP migration for mid-size Indian manufacturers” is winnable.
- Write your pillar’s scope in one sentence. “Everything a 50 to 200 person manufacturer needs to decide, plan and survive an ERP migration.” Anything outside that sentence isn’t in this cluster, however good the idea. That sentence is your editorial filter.
- Map eight to twelve subtopics from real buyer questions. Take them from sales calls and support tickets, not a keyword tool. Each must be a question someone would type or ask an AI, and answerable in one page.
- Inventory every existing post against that map, and sort into four buckets. Keep, merge, rewrite, retire. Be ruthless. Most firms find half their archive belongs in merge or retire.
- Merge the duplicates before you write anything new. Take the posts covering the same ground, write one complete piece that genuinely answers the question, and publish it at the strongest existing URL.
- 301 redirect everything you retire or merge to its replacement. Never leave a URL returning a 404 when a relevant destination exists. This is where the accumulated value transfers.
- Rewrite each keeper so it answers exactly one question, completely. Open with a standalone answer in the first forty words. A cluster piece that half-answers three questions is the problem you’re trying to fix.
- Write the pillar page last. Standard advice says write it first. You can’t synthesise a subject well until you’ve written the pieces underneath it, which is why so many pillar pages read like padded introductions.
- Wire the internal links deliberately. Covered in the next section, and it’s the step most people do wrong.
- Fill the gaps in commercial order, not alphabetical order. Whichever missing subtopic sits closest to a buying decision gets written next.
- Measure the cluster, not the posts. Track combined impressions, combined enquiries, and how many subtopics you now rank on page one for.
How should the internal links actually run?
In a hierarchy, not a web. This is the difference between a cluster and a pile.
- Every cluster piece links up to the pillar, using consistent anchor text. Same phrase every time. That repetition signals what the pillar is about.
- The pillar links down to every cluster piece, each with anchor text matching that piece’s specific question.
- Cluster pieces link sideways only when genuinely useful. If the ERP data-migration post naturally references the downtime-planning post, link it. Don’t manufacture links.
The common failure is linking everything to everything with anchor text like “read more”. That flattens the hierarchy and tells an engine nothing about which page is the authority.
One point specific to how AI engines read you: use the same terminology throughout. Call it AEO in one post, answer engine optimisation in the next and GEO in a third, and you fragment your own entity across three names. Pick one primary term, define it once in the pillar, and stay consistent. Synonyms belong in body text, not headings and anchors.
How many pieces, and how fast?
Eight to twelve cluster pieces plus a pillar is the working range for most B2B firms. Fewer than six rarely reads as authority on anything.
Speed matters more than people expect. A half-built cluster performs worse than none, because you’ve published four thin pieces referencing a pillar that doesn’t exist and cross-linking to articles you haven’t written. Google sees an incomplete structure; a reader sees a dead end.
The realistic plan: do the merge and redirect work in one focused fortnight, then publish the gaps at one solid piece a fortnight until the cluster closes. Six months, not two years. If you can’t sustain that, build eight rather than fourteen.
The honest trade-off in a pillar page and cluster model: while you’re building cluster one, you publish nothing about anything else. Founders find this genuinely hard. It is also the entire point, because scattered publishing is what got you here.
What goes wrong with content clusters?
Deleting without redirecting. Someone clears out forty old posts, the URLs 404, and every link and ranking signal those pages accumulated evaporates. Redirect to the closest relevant page, or keep the URL.
Picking a pillar nobody buys from. A cluster on an interesting adjacent topic earns traffic and no enquiries. Test it against one question: does someone researching this eventually need to pay somebody? If not, it’s a hobby.
The pillar becomes a 6,000-word monster. Its job is to define the subject, state the shape of the answer, and route people to depth. A hub, not an encyclopaedia. Twelve hundred to two thousand words is usually enough.
Rebuilding the cluster every quarter. New advice arrives, the scope sentence gets rewritten, half the cluster no longer fits. Lock the scope for a year.
Judging it in month two. A cluster compounds as it completes. The pieces published first typically do nothing until the pillar and the internal links exist, and then several move at once. This is the most common reason people abandon a content cluster strategy two-thirds of the way through, which is the worst possible moment to stop.
FAQ
How many articles do I need for a content cluster?
Eight to twelve cluster pieces plus one pillar page works for most B2B firms. The real test isn’t the count, it’s coverage: could a buyer research the whole subject without leaving your site? Below about six pieces, a cluster rarely reads as authority.
Should I delete my old blog posts when building a cluster?
Merge rather than delete. Combine overlapping posts into one complete piece at the strongest existing URL, then 301 redirect the others to it. Deletion with a 404 throws away accumulated links and signals. Retire a post only when nothing relevant exists to redirect it to.
Do I write the pillar page first or last?
Last, or write a rough version first and rewrite it once the cluster exists. A pillar is a synthesis, and you can’t synthesise before working through the detail. Pillars written first tend to read as padded introductions, which is why so many underperform.
How long does topical authority take to build?
Expect six to nine months from starting the restructure to seeing the cluster rank as a group, assuming you finish it. Individual pieces may move sooner. The compounding effect arrives when the cluster is complete and interlinked, not while it’s half-built.
Can I have more than one cluster?
Yes, and eventually you should, but finish one before starting the second. Two half-built clusters is the same scattered-publishing problem with better labelling. The second goes faster anyway, because you’ve learned how your buyers actually phrase things.
Where to start
Export a list of every URL on your blog into a spreadsheet, then write your pillar scope sentence at the top. Go down the list and mark each post keep, merge, rewrite or retire against that one sentence. Most people find their cluster is already three-quarters written and just badly organised.
If you want a second pair of eyes on the map before you start merging, get in touch.
