Topic Clusters and Pillar Pages
Topic clusters are an industry pattern, not a Google mechanism: the words "pillar" and "cluster" do not appear in the documentation. What is documented is what a cluster actually buys: link analysis systems work out how pages link to each other to determine what those pages are about; without direct links from a hub its children may not be found by crawling at all; and a site's link structure is what Google builds sitelinks from. This lesson covers what a cluster really gives you, and where it turns into a pile of duplicates.
It rests on the ranking systems guide and the site structure guidance.
What internal links do, per the documentation
The ranking systems guide states: "we have various systems that understand how pages link to each other as a way to determine what pages are about and which might be most helpful in response to a query". The same page names PageRank as one of those systems and notes it continues to be part of the core ranking systems.
That is an important correction to the usual metaphor: links are not a pipe through which "authority" flows into a chosen page. They help determine what a page is about and how helpful it is. So inside a cluster what matters is not the number of links but their meaning: from which page, in what context, with what anchor. That is what internal link analysis shows.
The first thing a cluster buys: discovery
The most mundane and most expensive one. The site structure guidance puts it literally: add links from menus to category pages, from category pages to sub-category pages, and from sub-category pages to all product pages. And the consequence: if category pages don't include direct links to all products in a category, Googlebot might not find all of your products by crawling alone.
Transfer that to a cluster: a hub that lists forty subtopics but links to six is a hub whose other thirty-four pages are on their own. What was actually reachable by links is what the site crawler reports, and a link only counts when it is an a element with an href attribute.
The second: the observable payoff is sitelinks
Sitelinks are links from the same domain clustered together under a text result; Google's systems analyse the link structure of your site to find shortcuts that save users time. And the other side of it: if the structure of your site doesn't allow the algorithms to find good sitelinks, they are not shown.
The same guidance says to create a logical site structure that is easy to navigate and to link to your important pages from other pages. The hierarchy is further clarified by breadcrumbs with structured data.
The third: where a cluster breaks
If the subtopics are sliced too thin the pages grow alike — and then it is no longer a cluster at work but canonicalization: Google clusters pages whose primary content is very similar and picks a canonical itself. A cluster of forty pieces retelling one paragraph collapses into a single page of Google's choosing. That is worked through in the lesson on the keyword map, and conflicts in live data are what the cannibalization check finds.
What is required of the overview page
An overview page does not exist for length. The documentation asks whether a piece provides a substantial, complete or comprehensive description of the topic — that is its job: a map of the topic from which it is clear where to go next. Coverage is what entity coverage helps estimate.
Assembling a cluster
- Tasks. Write out the reader tasks in the topic — one per page, as in the map.
- The overview page. A full description of the topic linking to every subtopic without exception.
- Links back. From each subtopic, a meaningful link to the overview and to its nearest neighbours — not a site-wide "related" block of twenty items.
- Addresses. The URL structure reflects the hierarchy — see the lesson on architecture.
- Verify by crawling. A crawler should reach every page of the cluster by following links.
- Check against data. Which sections earn impressions shows in folder performance, and individual pages in the pages report.
Mistakes
- Believing links "pour authority": they determine what a page is about.
- A hub that lists its subtopics without linking to them.
- Subtopics that differ by phrasing rather than by task.
- An overview page that is a five-line table of contents.
- A site-wide link block instead of contextual links.
- A cluster never checked against data: some subtopics will be dead — as covered in content strategy.
What comes next
One lesson of the module is left: how to write a page so that it answers the task rather than accumulating length.