How to Build a Topic Cluster for AI Search

Most topic cluster advice stops at the diagram: one main hub in the middle, a few cluster pages around it, and lines connecting them. That diagram is useful, but it is only the surface-level version of the work. The hard part is deciding which subtopics deserve their own URL, which ideas should be merged, where supporting articles fit, and how to make the structure clear enough for readers, search engines and large language models to understand.

That matters even more now that discovery is happening across classic Google Search, Google’s AI summaries, conversational search, ChatGPT, Perplexity and other answer-style platforms. In those environments, visibility is not always about one page ranking for one keyword. It is often about whether your site has enough structured content around a subject to be selected, cited or used as supporting evidence when a system breaks a broad question into smaller parts.

This is the method I use when mapping content clusters for clients. I am using our own AI SEO cluster at eCBD as the working example, because it makes the structure easier to follow and keeps the advice practical rather than theoretical.

BOOK A FREE STRATEGY MEETING

Key Takeaways

• A strong cluster needs more than a pillar and a handful of posts. It needs a clear hierarchy, specific subtopic pages, supporting articles and deliberate internal links.
• Topical authority comes from covering a subject properly across connected pages, not from publishing several similar articles around the same keyword.
• Keyword research is still useful, but it should be combined with search intent, user intent, SERP data, Google Search Console queries, competitor gaps and AI result testing.
• Content clusters work best when each page has a distinct job: define the topic, compare options, explain a process, answer a narrow question or support a buying decision.
• AI tools can help with semantic analysis, keyword clustering and content briefs, but AI-generated content still needs original experience, clear judgement and quality control.
• You should measure the cluster at both page and topic level, using organic traffic, rankings, overview-result visibility, conversational result testing, LLM citations and conversion signals.

What Is a Topic Cluster?

A topic cluster is a group of connected pages built around one broad subject. The central page covers the main topic at a high level, while the surrounding pages go deeper into specific subtopics. Internal linking then ties those pages together so users and crawlers can understand how the content fits.

For example, a broad SEO services page might act as the pillar. Around it, you might have pages about technical SEO, local SEO, keyword research, content strategy, link building and reporting. Each page has its own search intent, but together they show that the site has depth across the wider subject.

In traditional SEO, that helps a search engine crawl the site, understand relationships between pages and assess topical authority. In AI-led discovery, the same structure also helps answer engines and large language models find specific, source-worthy pages when they need evidence for a more complex response.

Why Content Clusters Matter More in AI-Led Search

AI systems rarely answer broad questions by looking for one perfect page. They often break a prompt into smaller questions, retrieve information from different sources, and then combine the findings into one response. Google uses query fan-out in AI Mode, and similar retrieval logic appears across other generative search experiences.

That is why content clusters are useful. If your site only has one broad page, you give the system one chance to understand your expertise. If your site has a connected set of quality content covering definitions, comparisons, steps, risks, use cases and examples, you create several possible entry points for citation.

This does not mean you should create hundreds of thin pages just to chase long-tail keyword variations. That would create content bloat and weaken the cluster. The goal is to build enough depth to answer real questions clearly, while keeping each URL distinct from the others.

Pillar Pages, Cluster Pages and Supporting Articles

I prefer to map clusters in three tiers rather than two. The three-tier model is cleaner because not every page deserves to sit directly under the pillar with the same weight.

Tier	Job	Example from an AI SEO cluster
Pillar page	Introduces the whole topic, explains the main offer or subject, and links to the major subtopics.	AI SEO services
Cluster page	Goes deep on one major subtopic with its own intent, examples and practical advice.	How to optimise for AI visibility
Supporting article	Answers a narrower question that supports one parent subtopic without competing with it.	AI visibility audit framework

The mistake I see most often is treating every URL as equally important. That creates a flat structure where the pillar, subtopic pages and supporting articles all compete for the same role. A good content strategy makes the hierarchy obvious.

BOOK A FREE STRATEGY MEETING

Step 1: Define the Pillar Topic and Its Boundaries

Start by writing a one-sentence definition of what the pillar covers and what it does not cover. If that sentence is vague, the whole map will become vague as well.

For our own AI SEO pillar, the boundary is: how businesses can improve visibility, citations and recommendations across AI-powered discovery platforms, as distinct from ranking-only traditional SEO. That boundary allows us to include pages on Google’s AI summaries, conversational search, citation tracking and AI visibility audits, while excluding broader AI business topics that do not belong in the cluster.

This step also protects against chasing every adjacent keyword opportunity. A topic can be commercially relevant without belonging inside this specific cluster. The boundary is what lets you say no.

Step 2: Use Keyword Research Without Letting Keywords Run the Strategy

Keyword research is still useful, but it should not be the only input. A keyword export from Ahrefs, Semrush, Google Keyword Planner or the Keyword Magic Tool can show search volume, long-tail keyword variations, target keywords and keyword opportunities. What it cannot always show is whether two searches deserve separate pages.

That is where search intent and user intent matter. If two queries have different wording but the same underlying need, they probably belong on the same page. If they represent different stages of the journey, different levels of knowledge or different decision points, they may need separate URLs.

I usually start with a spreadsheet that includes:

• Primary and secondary keywords
• Search volume data, where available
• Intent type: informational, commercial, navigational or transactional
• Likely user intent in plain English
• Current ranking URL, if one exists
• Best-fit page type: pillar, subtopic page, supporting article or FAQ section

This stops keyword targeting from becoming messy. You are not simply matching one keyword to one URL. You are building a useful map of questions, intents and page roles.

Step 3: Group Keywords Into Meaningful Clusters

Keyword clustering is the process of grouping related queries into page-level themes. A keyword cluster might include several phrases that all point to the same topic, such as “how to get cited in AI”, “AI citation SEO”, “LLM citations” and “how brands appear in ChatGPT”.

AI tools can speed up this work by running semantic analysis across a keyword list and suggesting groups of clustered keywords. That can be helpful, especially when the export is large. But I would not let the tool make the final decision. A clustering tool may group terms by language similarity, while a human SEO still needs to judge intent, business value and overlap.

The test I use is simple: could one strong page satisfy the main reason behind all of these searches? If yes, keep them together. If no, separate them. That keeps content clusters clean and avoids splitting the same topic across too many weak pages.

Step 4: Find Content Gaps Before Creating More Pages

Once the early keyword cluster map is in place, look for content gaps. This is where many cluster plans become much stronger, because the missing pages are often not obvious from keyword tools alone.

I would check:

• Google Search Console queries where impressions exist but the current page is not a great match
• SERP data for the target keywords, including what kind of pages are ranking and which formats Google seems to prefer
• Competitor content clusters to see which subtopics they have covered better than you
• Questions appearing in People Also Ask, featured snippets and AI-generated summaries
• Manual tests in conversational search and other AI chatbots to see which sources are being cited
• Sales calls, client emails and internal team questions that reveal real user concerns

This step prevents the cluster from becoming too tool-led. Some of the best supporting articles come from actual customer questions, not from high-volume keywords. A low-volume question can still be valuable if it influences trust, comparison, risk or conversion.

Step 5: Build the Cluster Map

Now turn the research into a working structure. I usually map this as a table before anything gets written.

Page role	Page topic	Main intent	Internal linking notes
Pillar	AI SEO services	Explain the offer and the wider strategy	Links to all major subtopics
Cluster	How to optimise for AI search	Teach the practical optimisation process	Links back to the pillar and across to citation and measurement pages
Cluster	How to get cited in AI	Explain what makes a page citation-worthy	Links to audit, measurement and technical SEO pages
Cluster	How to measure AI SEO	Show how to track visibility beyond rankings	Links to Google’s AI summaries, conversational search and reporting content
Supporting	AI visibility audit checklist	Give a narrow diagnostic framework	Links up to the optimisation and services pages

This kind of map makes production easier because every page has a job before a brief is written. It also makes cannibalisation easier to spot. If two rows have the same intent, same audience and same internal linking destination, they probably should not be two separate pages.

Step 6: Create Content Briefs Before Writing

Content briefs are where the cluster becomes practical. A good brief should explain the purpose of the page, the target keywords, the search intent, the page angle, the internal links, the required examples and the questions the page must answer.

For AI-focused content production, I would add a few more fields:

• Which sub-question this page should answer better than competitors
• Which claims need evidence, screenshots, examples or first-hand explanation
• Which related pages should be linked from the body content
• Which entities, tools, platforms or concepts should be mentioned naturally
• Which sections need to be updated when Google or the wider search platform changes

This is also where you control AI-generated content risk. Generative tools can help with content generation, outlines and drafting support, but they should not be used to mass-produce shallow pages. The final page still needs original judgement, useful examples, Australian English, accurate terminology and a clear reason to exist.

Step 7: Build Internal Links Like a System, Not an Afterthought

Internal links are what turn a group of related posts into a working cluster. Without them, you have a pile of pages. With them, you have a structure that readers, crawlers and answer engines can follow.

The internal linking rules I use are straightforward:

• The pillar links down to every major subtopic page.
• Each subtopic page links back up to the pillar using descriptive anchor text.
• Related subtopic pages link sideways where the relationship genuinely helps the reader.
• Supporting articles link up to their parent subtopic page and, where useful, across to closely related supporting content.
• Anchor text should describe the destination naturally. Avoid generic labels like “read more” when a clearer phrase would help.

This kind of internal linking helps search engines understand which page is the broad authority, which pages cover subtopics, and how the cluster should be crawled. It also supports users by giving them a logical next step instead of leaving them at the end of a single article with no pathway forward.

Step 8: Strengthen the Cluster With External Authority Signals

Topical authority is mostly about depth and relevance, but domain authority still matters in competitive spaces. Domain authority is not the same thing as topical authority, but it can strengthen a content marketing programme when the site also has a clean topic structure. If two sites have equally useful content clusters, the one with stronger brand signals, mentions, backlinks, expert authorship and third-party trust will often have the advantage.

That is especially relevant for LLM citations. Large language models may draw on search results, trusted publications, review sites, forums, knowledge graph relationships and widely referenced pages. Your own cluster gives you the content foundation, but it should ideally be reinforced by off-site signals that show your business is a credible source in the category.

For a service business, that might include expert commentary, original research, case studies, industry listings, podcast mentions, conference profiles, partner pages, client reviews and relevant media coverage. Those signals do not replace quality content, but they can make your content easier to trust and cite. In competitive categories, stronger domain authority can be the extra layer that helps an already-useful cluster earn attention.

Step 9: Measure the Cluster at Page and Topic Level

You need two layers of reporting. Page-level reporting tells you how each URL is performing. Topic-level reporting tells you whether the cluster is building authority as a whole.

At page level, track rankings, impressions, clicks, engagement and conversions. Google Search Console is still one of the most useful tools here because it shows real queries and helps you find pages that are getting impressions without earning enough clicks.

At topic level, look at broader signals:

• Whether organic traffic is growing across the cluster, not just on one page
• Whether the pillar improves after supporting pages go live
• Whether multiple pages are ranking for related search engine results
• Whether your brand appears more often in AI Overviews and AI Mode responses
• Whether LLM citations are increasing for priority prompts
• Whether LLM citation rates improve after content updates, internal links or new supporting articles are added

This is the part where the cluster either proves itself or shows you where the structure is weak. If you publish more pages but the pillar does not improve, the problem may be internal linking, page overlap, weak briefs, poor intent matching or a lack of authority signals.

Common Mistakes When Building Content Clusters

The first mistake is building from keywords only. Keyword lists are helpful, but they can push content creators into producing several pages that all say the same thing. Start with user intent, then use keyword data to support the structure.

The second mistake is making the pillar too broad. A pillar page should be broad enough to organise a topic, but not so broad that every page on the site can technically fit underneath it. If the pillar has no boundary, the cluster will not have one either.

The third mistake is publishing without consolidation. If a new page overlaps heavily with an existing one, improve the existing URL instead. More pages do not automatically create more topical authority. Sometimes they just create more confusion.

The fourth mistake is treating content clusters as a one-off project. Search behaviour changes, AI tools change, competitor pages improve, and new content gaps appear. The cluster should be reviewed regularly, especially in fast-moving topics.

A Simple Workflow You Can Use

If you are building your first cluster, keep the workflow simple:

1. Choose the pillar topic and write the boundary in one sentence.
2. Collect keywords, search volume data, GSC queries, SERP data and customer questions.
3. Group the terms into a keyword cluster map using semantic analysis and human review.
4. Assign each idea to a page role: pillar, subtopic page, supporting article or FAQ.
5. Consolidate anything with overlapping search intent.
6. Create content briefs for each priority page.
7. Publish in stages, starting with the pillar and the highest-value subtopics.
8. Add internal links as each page goes live.
9. Measure rankings, organic traffic, AI visibility and LLM citations.
10. Refresh the cluster as new gaps, queries and platform behaviours appear.

That process is not complicated, but it does require discipline. The win comes from making clean decisions before content production begins, rather than trying to fix a messy cluster after twenty pages have already been published.

Frequently Asked Questions

What is a topic cluster in SEO?

A topic cluster is a group of related pages built around one main subject. It usually includes a pillar page that covers the topic broadly, subtopic pages that go deeper, and internal links that connect the pages together.

What is the difference between a pillar page and a cluster page?

The hub page gives the broad overview and acts as the main hub. A cluster page focuses on one specific subtopic and links back to the pillar. The pillar organises the topic, while the surrounding pages provide the depth.

How many pages should a content cluster have?

There is no fixed number. A small cluster might have one pillar and three or four subtopic pages. A larger cluster might include ten or more pages, plus supporting articles. The right number depends on the topic, intent range, content gaps and commercial value.

How do content clusters help topical authority?

Content clusters help topical authority by showing that your site covers a subject from several useful angles. When the pages are well written and internally linked, they make it easier for search engines and AI systems to understand your expertise.

Should I use AI tools to build a topic cluster?

Yes, but as support rather than as the strategist. AI tools can help with keyword clustering, semantic analysis, content briefs and content gap discovery. The final structure still needs human judgement, especially around user intent, overlap, accuracy and business value.

Can AI-generated content be used in a cluster?

It can be used carefully, but it should not replace expertise. AI-generated content still needs fact-checking, editing, original insight and a clear purpose. Thin pages created at scale can weaken the cluster instead of helping it.

How do I know if my cluster is working?

Look at the full topic, not just one page. Track rankings, organic traffic, impressions, conversions, internal link performance, AI Overviews visibility, conversational result testing and LLM citations. If the pillar and related pages improve together, the structure is usually working.

What should I do if two pages in the cluster overlap?

If two pages answer the same search intent, consolidate them. Keep the stronger URL, move useful sections across, update internal links and make the remaining page clearer. Overlap can confuse users, search engines and AI systems about which page is the best answer.