You have a page about "refrigerator repair" that gets three visits a month, an article about "compressor" that nobody finds, and a services page stuck on the second page of Google results. Your problem isn't a lack of content. The problem is that Google doesn't understand these three pages belong to the same topic. Until it does, none of them gain topical authority, and your services page won't rise either.
Topic clusters are the solution: a pillar page that covers the topic from above, and several cluster articles that each explore a subtopic in depth and all link to the pillar. But doing it right involves three hard decisions, and most sites get stuck on one of them.
Choosing the pillar topic: the criterion that actually works
The first mistake is choosing the pillar topic based on interest or keyword search volume. The pillar topic must be something under which you can define at least eight to twelve independent subtopics, each of which alone is worth an article. If you can't, it's not a pillar topic; it's a long article.
A practical test: write down the topic name and list ten article titles beneath it, each answering a different user question, not a rewrite of each other. If the titles start repeating, make the topic bigger or smaller. "SEO" is too big; "technical SEO for small stores" might be exactly the right size.
Second criterion: the pillar topic must connect to a commercial page. If your cluster is about "houseplant care" but you have nothing to sell, you're building topical authority that doesn't reach your pocket. In SEO, a cluster that doesn't connect to a services page only produces traffic.
What a pillar page looks like and what it doesn't
A pillar page is not a list of links. It must itself provide a complete, independent answer; someone who reads only this page should get their initial answer and then go to the cluster articles for more depth. Its length usually comes out between 2,000 and 4,000 words, but the number isn't important. What matters is that each section summarizes a subtopic and links to its full article.
Put the pillar page on a short, clean path, not three levels deep. The structure /seo-technical/ for the pillar and /seo-technical/crawl-budget/ for the cluster is more readable both for users and for search engines. If your current site structure is messy, before anything else use a technical domain and DNS check to make sure redirects and canonical tags work correctly; a cluster built on duplicate URLs is wasted from day one.
Dividing the topic into cluster articles: the "one question, one page" rule
Each cluster article must answer exactly one question. If your article title has two "and"s, it probably should be two articles. This simple rule prevents the next biggest headache: two pages competing for the same keyword and keeping each other down.
Before writing, assign one primary keyword and two to four secondary keywords to each article and save them in a file. Then put all the titles side by side and see whether two titles could appear for the same search. If yes, remove or merge one. This half hour of work eliminates months of confusion.
Depth versus breadth: which to build first
Here you have a real choice. You can write the pillar first and then add the clusters one by one, or write three or four cluster articles first and then build the pillar. I choose the former, but on one condition: don't publish the pillar with links to articles that haven't been written yet.
My suggested order is this: write the pillar, but don't publish it. Complete the first three cluster articles. Then publish the pillar with links to those three and add the rest week by week. This way the pillar page has live links from day one, and when Google comes back to it, it sees real growth rather than a half-finished page.
If you're forced to get results fast, there's another order: first publish the cluster articles closest to purchase, then the pillar. This approach makes more sense for sites that need immediate commercial traffic, but its internal link structure stays incomplete for several weeks.
Publishing order and internal linking from day one
Internal linking isn't something you add at the end. If you write articles without a link map, you'll later have to go back and edit everything manually, and half of them will be missed. Before writing the first article, draw the link map: which article links to which article and with what anchor text.
The basic rule: every cluster article has at least one link to the pillar page, and the pillar page links to all clusters. Also link between clusters, but only where it genuinely helps the reader. A forced link between two unrelated articles is both annoying for users and transfers no value to Google.
Write descriptive anchor text, not keyword-stuffed. "A guide to reducing image size" is better than "image optimization," because it reads more naturally and preserves anchor text diversity. If you want to see how competitors do this, SEO and webmaster tools are a good starting point for examining link structure.
This is where they go wrong
The most common mistake I've seen is this: all cluster articles link to the pillar page, but the pillar page doesn't link back to any of them. The result is that the pillar page shows up in reports as a page with many inbound links but its ranking doesn't budge, because its authority doesn't flow back to the clusters, and the clusters aren't connected to each other either. The site looks like a star with one center and planets that know nothing of each other.
You see the sign in Search Console: the pillar page has high impressions but low CTR, and the cluster articles get stuck at positions 15 to 30 after a few weeks. The fix is adding links from the pillar to each cluster and two or three horizontal links between related clusters. An hour of work, no more.
How to assess the link structure
After publishing five or six articles, you need to see data, not feelings. In the Links section of Search Console, check the number of internal links for each page. A page with zero inbound internal links is effectively orphaned from Google's view. Also see which page has the most inbound links; if it's anything other than the pillar page, your structure has deviated from your goal.
A technical point that gets less attention: internal links in global navigation (menu and footer) carry less weight than in-text links, because they're repeated on every page. If you only link to clusters from the menu, your cluster structure effectively doesn't exist. The link must be in the body of the text, where the reader is actually looking for it.
To compare two approaches, see this table:
| Approach | Advantage | Cost |
|---|---|---|
| Complete cluster before publishing the pillar | Complete link structure from day one | Several weeks without a reference page |
| Pillar first, gradual cluster | Faster results in search | Incomplete links in the first month |
| Menu links only | Simple and fast | Effectively no topical authority |
If your site is large and you have many clusters, you'll also need an intermediate layer: a category page that gathers several related clusters. But for a site under a hundred pages, this layer only adds complexity. Stay simple.
When the cluster doesn't work
Sometimes you've set everything up correctly and still don't rank. Before rewriting the content, check speed and user experience. A pillar page that takes 4 seconds to deliver its first byte won't stay in top positions even with the best link structure. The article What is TTFB and why does your site seem slow? explains exactly this.
If your site is in Persian and you load fonts from an external CDN, you probably have text shift and render delay affecting Core Web Vitals. Read optimizing Persian website fonts before any content rewrite; often the ranking problem is a speed problem, not a structure problem.
And if you're sure everything technical is correct, it's time to hand the work to someone who does only this. ServerNet's SEO services are for exactly this stage, when you've drawn the map but its regular execution takes time.
One last warning: don't confuse topic clusters with "mass content production." Ten thin articles that each answer with one paragraph ruin the cluster structure, because Google doesn't count a page that has no independent value as part of the cluster. Three deep articles are better than ten shallow ones. Always.
Frequently asked questions
What is a topic cluster and how does it differ from a regular site category?
A topic cluster is a set of pages surrounding a pillar page, connected to each other and to the pillar through internal links. Its difference from a regular category is in purpose: a category organizes content, a topic cluster concentrates authority on one topic.
In a regular category, pages are only connected through the menu. In a cluster, links are in the body of the text with descriptive anchor text, and each article covers exactly one subtopic.
How many articles does a topic cluster need?
At least five to eight cluster articles for the structure to have a noticeable effect. With fewer than that, the pillar page looks like a long article with a few links, not a topical reference.
More than fifteen articles is usually a sign that you've made the pillar topic too big and it should be split into two separate clusters.
Are links from the pillar page to all clusters necessary?
Yes, and this is the part most sites miss. If clusters link to the pillar but the pillar doesn't link back, authority flows in one direction and the clusters get stuck in middle positions.
Put the return link in the body of the text, not at the bottom of the page as a list of links. A sentence that guides the reader to a deeper article is both more useful for users and more meaningful for search engines.
What effect does the publishing order of cluster articles have?
Order affects how quickly you reach results, not the final result. If you publish purchase-adjacent articles first, commercial traffic comes sooner, but the link structure stays incomplete for several weeks.
If you build the pillar first and then the clusters, the structure is complete from day one but results come later. For a new site, I recommend the latter; for a site that needs immediate traffic, the former.
Comments 0
No comments yet — be the first!