A list of keywords only becomes a plan when each keyword has a home. That means grouping similar keywords into clusters, deciding which page should rank for each cluster, and making sure your own pages don't compete with each other. This lesson covers clustering methods, building a keyword map, and diagnosing and fixing keyword cannibalisation.

Why cluster at all?

Search engines understand that many phrasings express the same need. "seo agency sydney", "sydney seo company", "seo services sydney" and "seo firm in sydney" are, for practical purposes, the same request. Creating a separate page for each would produce near-duplicate pages that compete with each other and risk looking like doorway pages. Instead, one strong page should target the whole group.

On the other hand, "seo agency sydney" and "how much does seo cost" are different needs (hiring versus understanding pricing). They deserve different pages, linked together.

Clustering answers the question: how many pages do we need, and what is each one for?

Clustering methods

1. By meaning and intent (manual)

Group keywords that ask for the same thing, using your judgement, the modifiers covered in the intent lesson, and knowledge of the business. This works well for smaller lists (up to a few hundred keywords) and catches nuances tools miss, such as a product name that means something different in Australia and the US.

2. By SERP overlap (data-driven)

If two keywords share several of the same top-ranking URLs, search engines are treating them as closely related, so one page can probably rank for both. A common rule of thumb is to group keywords that share at least three or four URLs in the top ten. Many SEO tools automate this; you can also check manually for important terms.

SERP overlap is powerful because it reflects how the engine actually behaves in your market. For example, you might expect "accountant" and "tax agent" to cluster together, but in Australia "tax agent" results may focus on registered tax agents and individual tax returns, while "accountant" results include business advisory firms. If the results differ substantially, consider separate pages.

3. By topic hierarchy

Above clusters sit topics: broader themes made of several clusters. A typical structure is a hub page (for example "Commercial cleaning") linking to more specific pages ("office cleaning", "medical centre cleaning", "strata cleaning") and supporting guides ("commercial cleaning checklist", "how often should offices be cleaned"). The hub links down to each page and each page links back up and across where relevant.

Building a keyword map

A keyword map is a table with one row per page. Typical columns:

Column Example
URL /services/office-cleaning-brisbane/
Page type Service page
Primary cluster office cleaning brisbane
Secondary terms office cleaners brisbane, commercial office cleaning brisbane cbd
Questions to answer how much does office cleaning cost, how often, after-hours cleaning
Intent Transactional, local
Status Exists – improve / New page needed / Merge
Current position 14 (Search Console)
Internal links from Commercial cleaning hub, home page, Brisbane location page

Mapping rules of thumb

  • One primary cluster per page. A page can rank for hundreds of related terms, but it should have one clear job.
  • Use the primary phrase naturally in the title, main heading and early copy, and cover secondary terms and questions in the body where they fit. Never force awkward phrasing.
  • Prefer improving existing pages that already rank for a cluster over creating new ones.
  • Map every important cluster to exactly one page. Empty rows (clusters without a page) become your content plan.

International and multi-location sites

If you serve several countries, the same cluster may need separate pages per country with genuinely localised content (prices, regulations, spelling, examples), connected with hreflang annotations. For multiple cities, location pages are justified when each has real local information, such as an address, team, projects or service details, not just the city name swapped.

Keyword cannibalisation

Cannibalisation happens when two or more of your pages target the same intent. Instead of one strong page, you have several weaker ones, and search engines may alternate between them.

Signs to look for

  • In Search Console's Performance report, filter by a query and open the Pages tab: several URLs receive impressions for the same query, and the ranking page switches over time.
  • Rankings for a term fluctuate unusually, with different URLs appearing.
  • You find two blog posts on essentially the same topic, often written years apart.
  • A blog post outranks the service page for a commercial query.

Note that two of your pages appearing for the same query isn't always a problem; it can be fine if they serve genuinely different intents and both rank well. It's a problem when they're near-duplicates or when the wrong page ranks.

Fixes

Choose the fix that fits the situation:

  1. Merge and redirect: combine the best content into one page and 301 redirect the others to it. This is the most common fix for duplicate blog posts.
  2. Differentiate: rewrite one page so it clearly serves a different intent (for example turn a general "office cleaning" post into "office cleaning checklist for managers" and link to the service page).
  3. Re-point internal links: make sure navigation and contextual links point to the page you want to rank for that topic, with descriptive anchor text.
  4. Canonicalise: where duplicates must exist (such as product variants or print versions), use rel="canonical" to the preferred URL.
  5. Noindex low-value duplicates that need to exist for users but shouldn't appear in search.

Worked example: an online pet store in the UK

The store had "best dog food for puppies" (2019 blog post), "puppy food guide" (2022 post) and a "Puppy food" category page, all appearing for "puppy food". Search Console showed impressions split across all three, none in the top five. The fix:

  • Merge both blog posts into one updated guide, "How to choose puppy food", with a 301 from the old URL.
  • Make the category page the target for the transactional query "puppy food", improving its intro copy and filters.
  • Link the guide to the category with "shop puppy food" and the category to the guide with "how to choose puppy food".

Within a few months, the category page ranked for the transactional queries and the guide for the informational ones.

Keep the map alive

A keyword map is a working document. Update it when you publish, merge or remove pages, and review it quarterly against Search Console data. New queries appearing for a page may justify a new section; a page attracting the wrong queries may need its focus clarified.

A step-by-step mapping workflow

If you're starting from a spreadsheet of a few hundred keywords, this sequence works well:

  1. Export current rankings from Search Console (queries and pages, last 3–6 months) so you know which pages already have visibility.
  2. Remove irrelevant terms: services you don't offer, places you don't serve, and competitor brand names you can't legitimately target.
  3. Cluster the remaining keywords by intent and SERP overlap.
  4. Assign each cluster to an existing page where Search Console shows that page already ranks or the page clearly fits.
  5. Mark unassigned clusters as new pages, noting the page type the SERP rewards.
  6. Check for conflicts: any cluster assigned to two pages, or any page with two unrelated primary clusters.
  7. Prioritise with relevance, value, opportunity and effort, then turn the top rows into briefs.

A sole trader's map may fit on one screen. A large retailer's map can span thousands of rows, usually managed by category templates rather than page by page. The principle is the same: one clear job per page.

Key takeaways

  • Cluster keywords that share intent so one page can target the whole group.
  • Use meaning, SERP overlap and topic hierarchy to cluster.
  • A keyword map gives every important cluster exactly one page; empty rows are your content plan.
  • Cannibalisation is several pages chasing the same intent; diagnose it in Search Console's Pages tab.
  • Fix with merge and 301, differentiate, re-point internal links, canonicalise or noindex.
  • Review the map regularly against real data.