Topic cluster
A topic cluster is a set of pages on one subject, built around a main pillar page and linked to each other, so a site covers the subject fully.
A topic cluster is a group of pages on one subject, arranged around a main page. The main page, often called a pillar page, gives the overview. Each of the other pages goes deep on one question, and they all link to each other.
The idea comes from SEO. It carries over to AI search because AI platforms rarely search for just one thing. They run several searches around a question, and a site with a good page for each of them gets read more than once.
How a cluster is built
Start with a subject your customers care about. Write one page that covers it broadly. Then list the questions people ask around it and give each one a page of its own, linked from the main page and back to it.
For example, Fernhill Coffee builds a cluster on coffee subscriptions. The main page is "The guide to coffee subscriptions." Around it sit pages on choosing a grind, how often to get a delivery, pausing and skipping, subscriptions for offices, and subscriptions as gifts. Each answers one question properly, and each links back to the guide.
Why it helps in AI search
When someone asks "Which coffee subscription should I get for my office?," a platform might search for office coffee, delivery frequency and prices. With a cluster, Fernhill has a page that fits each of those searches. Without one, it has a single general page competing against more specific pages on other sites.
Clusters also make it easier for search engines to see what your site is about, because the links tie the pages together.
Don't build it for its own sake
A cluster only helps if each page is worth reading on its own. Twenty thin pages that say the same thing in different words won't be cited, and they make the whole site look weaker.
In AnswerOrbit
Your topics in Topics and prompts are a good map for clusters, since each topic is a subject your customers ask about. The Pages tab in Sources shows which of your pages the answers use, so you can see which parts of a cluster are doing their job.