Content clusters

What does a content cluster do in practice?

It gives each question one page to quote and gives crawlers a path between them. Answer engines now recommend clusters as standard practice. On 3 October 2026, ChatGPT answered "how to rank in Perplexity for B2B SaaS" with a 90-day plan, read over the Attensira MCP. Comparison pages, FAQs and original research filled the first 60 days. "Create content clusters" came in the last 30, under the heading "Scale".

That ordering is right. A cluster is how you scale coverage once the pages that answer the high-intent questions exist. It is not a substitute for them. And the advice is generic: in the 30 days to 4 October 2026 the engines gave 12 answers to that question, and none of them named or cited Attensira.

Our own site shows what a cluster gets you on the crawling side. The feature pages are spokes, one question each: how prompts are tracked, how content is optimised, what the crawler logs record. The features hub links to every one of them. In the same 30 days, by the user agent each request claimed, AI crawlers requested /features/prompt-tracking 5,805 times, /features/content-optimization 5,468 times and /features/crawler-logs 5,202 times. Only four pages on the site drew more: /contact-sales, the homepage, /pricing and /about.

Clusters also have a cost we are paying now. The glossary had grown pages that overlapped: two terms for the same idea, several web-performance terms with no AI search angle. In the current rebuild we deleted those and pointed their links at the one page that owns each question, because two pages answering the same question compete with each other.

What is a content cluster?

A set of pages that together cover one topic:

• A hub that covers the subject broadly, answers the main question and links to every spoke. Our glossary index is the hub for this cluster.
• Spokes, each answering one narrower question in depth and linking back to the hub.
• Cross-links between spokes where a reader would naturally want the next one, such as this page to internal linking.

The rule that makes it a cluster rather than a pile: every question has exactly one page that owns it.

How do you build a content cluster?

1. List the questions. Every question a buyer asks about the topic, in their words. Group near-identical phrasings; they are one question.
2. Assign one owner page per question. If two existing pages answer the same question, merge them and point links at the survivor.
3. Write the hub last. It answers the main question in brief and routes to each spoke with a link that says what the spoke answers.
4. Link deliberately. Every spoke to the hub; spokes to each other where the next step is obvious. Use descriptive link text, not "click here".
5. Keep shared facts in one place. A price or a definition stated in several spokes drifts. State it once and link to it.

Our AI crawler guide is built this way: a pillar page, chapters for each sub-question, and a crawler directory whose entries each point back to the chapter that explains what allowing that kind of agent costs.

Does site structure affect crawling?

At scale, yes. Google's starter guide notes that on sites with more than a few thousand URLs, how you organise content may affect how Google crawls and indexes it. Its crawl budget guidance addresses very large sites, with a million or more pages, or ten thousand or more that change daily. Most sites are far below that, so for them a cluster's value is clarity: crawlers find spokes through the hub, and every page is a few links from the homepage. See crawling and indexing.

Do clusters help AI answers?

Indirectly, in three ways:

• One passage per question. Retrieval picks passages. One owner page means one strong candidate instead of two weaker ones splitting the match.
• Discoverability. A linked cluster gets crawled; an orphan page may not.
• Evidence about the publisher. A body of related pages on one subject says more about who wrote them than any single page can. That is topical authority in action.

What are the objections to building content clusters?

"More spokes means more traffic." More spokes that answer distinct questions, yes. More spokes that restate each other split your own signals.

"Clusters need a special URL structure." Links matter far more than paths. A shared folder is tidy, not required.

"Pillar pages should be huge." The hub needs to answer the main question and route well. Depth belongs in the spokes.

"Deleting overlapping pages loses traffic." On a low-traffic site, one clear owner per question is worth more than two half-owners.

How does Attensira build clusters?

Attensira maps the questions your buyers ask AI to the pages that should own them, finds questions with no owner and pages that compete for the same question, and has the crew write, merge and link pages to fix it. Each change ships after your approval, and the crew measures which pages the engines cite before and after. Content optimization shows how the crew decides which page owns a question and what it changes on it.