Keyword clustering: how to group keywords so each group gets one page

What is keyword clustering?

Keyword clustering is the practice of grouping keywords that a single page can fully answer, so that each group, or cluster, is assigned to exactly one page. A cluster has a primary keyword that names the page and a set of variants, synonyms and sub-questions that the same page covers in its sections.

The test for whether two keywords belong together is simple: would someone who searched keyword A be happy landing on the page written for keyword B? “orthopedic dog bed” and “orthopedic dog beds” pass easily. “how to wash a dog bed” and “how to make a dog bed” share four words and fail, because one reader wants laundry instructions and the other wants a sewing project.

Clustering sits between keyword research and writing. Research gives you a long list of searches; clustering turns that list into a short list of pages. If you keep a master keyword list, the cluster is its “topic” column.

Why does keyword clustering matter for SEO?

Keyword clustering matters because it decides how many pages you write and which page each search belongs to, and both mistakes are expensive. Too few clusters and one page tries to answer questions with different intents. Too many and you publish near-duplicate pages that compete with each other.

• It prevents cannibalization before it starts. Two pages written for one cluster split clicks, links and Google's attention. Fixing keyword cannibalization after the fact means merging and redirecting; clustering first avoids it.
• It keeps you clear of doorway pages. Google's spam policies describe doorway abuse as pages created to rank for specific, similar search queries. One page per variant is exactly that pattern.
• It makes better pages. A page that covers a whole cluster answers the follow-up questions a reader has, instead of making them search again.
• It makes tracking meaningful. Tracking a cluster's primary keyword against one target URL tells you whether the page is working. Tracking 40 variants against no particular page only tells you that numbers moved.

What are the three ways to cluster keywords?

Keywords can be clustered by the words they share, by what they mean, or by whether the same pages rank for them in Google. Most tools combine the first two because they are cheap; the third is the most reliable but needs results-page data.

A note on the third method. Google's spam policies list sending automated queries to Google, including scraping results for rank-checking, as machine-generated traffic. MrKeyword does not collect Google's results, so it does not offer results-page clustering. Searching a handful of keywords yourself, in a private window, is how you get the same evidence for the clusters you care about most.

How do you group keywords by hand?

To group keywords by hand, strip the shared head term from each keyword, sort by what is left, merge rows that mean the same thing, split rows whose intent differs, and confirm close calls by searching both keywords. A spreadsheet is enough for a few hundred keywords.

1. Collect and clean the list

   Export your queries from Search Console and your ideas from a keyword tool into one column. Lower-case everything, collapse double spaces, and remove exact duplicates. Keep plurals for now; they are grouped, not deleted.

2. Strip the head term

   In a helper column, remove the words every keyword shares (“dog bed”). What remains, “orthopedic”, “for large dogs”, “how to wash a”, is what makes each keyword different, and it is what you cluster on. Without this step every keyword looks similar to every other.

3. Sort and merge by meaning

   Sort by the helper column so similar modifiers sit together, then give each row a cluster name. Merge synonyms the sort cannot see: price, cost and cheap; elevated and raised; wash, washing and washable.

4. Split by intent

   Label each keyword informational, commercial, transactional or navigational from its wording. A cluster that mixes “how to” questions with “best” or “price” keywords is two or three clusters.

5. Check the close calls in Google

   For any pair you are unsure about, search both. If most of the first-page results are the same pages, Google treats them as one topic and so should you. If the results are different, keep them apart.

6. Pick the primary keyword and the page

   Mark the keyword with the most measured demand as primary, and give the cluster one URL: an existing page if one already ranks, or a planned slug if not.

Worked example: clustering 30 “dog bed” keywords

Here is a real run of MrKeyword's grouping code on 30 dog-bed keywords, followed by the review a person still has to do. The keywords are realistic, but the volumes are made up for illustration; the groups are exactly what the code produced from them.

Read it the way you would read a colleague's first draft:

• Good: the plurals landed together, “price” and “cost” met through the related-terms table, and the “best” group correctly collected “top rated” and “reviews” as one comparison intent.
• Wrong: the “how to” group holds two topics. Washing and cleaning a dog bed is one page; making a dog bed (with “diy dog bed”, which was left ungrouped) is another.
• Missed: “cheap dog beds” stayed out of the price group, “washable dog bed” and “washing dog bed” stayed apart from washing, and “elevated” never met “raised”, because no word links them.
• A judgement call: “orthopedic dog bed for large dogs” could live on the orthopedic page or the large-dog page. Search it: if the results are mostly orthopedic beds, it belongs there.

After review, the 30 keywords become seven pages: orthopedic beds, large-dog beds, elevated beds, a buying guide (best, top rated, reviews), how to wash a dog bed, how to make a dog bed, and a price or shop page for the price, cost and cheap searches. “calming dog bed” and “dog bed near me” each need their own decision, and the bare “dog bed” goes to the main category page.

How do you cluster keywords for free with Google's tools?

Google's free tools do not cluster keywords for SEO, but Keyword Planner can group ideas into categories and Search Console shows which page already ranks for each query, which covers most of what you need.

• Keyword Planner categories. After you discover new keywords, Keyword Planner's Refine keywords panel sorts ideas into categories related to them (its help page uses shoe colour for running shoes as the example). Unticking a category removes those ideas, which is a quick first pass. It needs a Google Ads account with billing details entered; our guide to using Google Keyword Planner for free covers the setup.
• Search Console's Pages tab. In the Performance report, click a query and open the Pages tab. If one of your pages already ranks for several queries, those queries are a cluster Google has already formed for you, with that page as its URL.
• Your own searches. For the close calls, the results page is the evidence. Five manual searches settle most disagreements.

How does MrKeyword's Groups tab cluster keywords?

In MrKeyword's Keyword Research, the Groups tab sorts the Keyword Planner ideas for your search into topic groups by the words each idea adds to your phrase, so you see clusters instead of a flat list. Grouping runs on the ideas already on the page, so it does not use another lookup.

• When it appears. The tab shows when a search returns at least 20 ideas and they form at least 2 groups of 3 or more keywords. Smaller lists are easier to read as one list.
• What it compares. Your searched words and common filler words are removed from each idea first, so “orthopedic dog bed” is compared as “orthopedic”. The remainder is compared on whole words, three-letter fragments (which catch plurals and spelling variants) and a built-in table of related terms: price, cost and fee; near, nearby and local; best, top and review; vs and compare; course, class and training; how and step; buy and order; job and salary. The table also covers common Hindi and Hinglish equivalents.
• How groups form. Ideas are taken from the most searched down; each joins the group it is most similar to if the similarity is at least 0.42, otherwise it starts a new group. Up to 40 groups are shown; ideas that fit no group stay in the All tab and the page says how many.
• Names. A group is named after the word, or pair of words, that at least half of its keywords share beyond your phrase: “large”, “price”, “how to”.
• Searches. Each group's searches add up its distinct keywords. Keyword Planner gives close variants such as a plural the same number, so each variant set counts once, at its highest figure. Each group also shows its average difficulty and its mix of intents.
• Working with a group. Pick a group and every other tab, including Matching, Questions and Easy wins, shows only that group's ideas. Tick keywords to save them to a keyword list or track them, and Export CSV on the Groups tab gives one row per group.

Each search is one live Keyword Planner lookup. The Free plan includes 10 lookups a month and Starter includes 200; see pricing for the rest.

How do you cluster the keywords you already rank for?

Cluster your existing Search Console queries the same way, but add one column: the page Google ranks today for each query, because that page is usually the right home for the cluster. Clusters with two ranking pages are cannibalization candidates; clusters with no page that earns clicks are content gaps.

In MrKeyword this is the Keyword Strategy page. It takes up to the 2,000 busiest queries from the last 28 or 90 days of the website's own Search Console data, groups them with the same clustering, and for each cluster shows:

• The pillar page: the page that wins the cluster's clicks, and how many of your pages compete for it.
• A label: Quick win for an average position between 8 and 20 with at least 100 impressions, Defend for the top 3, Improve clicks between the top 3 and position 8, Needs a page beyond position 20, and Watch for everything else, such as a cluster in the quick-win band with too few impressions yet.
• One next step: merge competing pages when the queries split between pages hold at least 20% of the cluster's impressions, create a page when no page earns meaningful clicks and none ranks in the top 20, otherwise strengthen the pillar page.

Numbers there are impressions, clicks and positions from Search Console, never invented search volume. For a one-off review without the app, a keyword audit does the same job in a spreadsheet.

What do you do with keyword clusters once you have them?

Give each cluster one page, write that page to cover the whole cluster, and track the primary keyword against that page so you notice when Google picks a different one.

1. Map one URL per cluster

   Write the target URL next to every keyword in the cluster. If two existing pages already serve one cluster, pick one and fold the other into it; Google's canonicalization guide calls a redirect a strong signal of which URL should be canonical.

2. Cover the cluster on the page

   The primary keyword names the title and H1. The other keywords become sections, questions or table rows, written for the reader rather than repeated for their own sake.

3. Prioritize with difficulty and demand

   Rank clusters by total searches (or impressions) and by how hard the primary keyword is. Our keyword difficulty checker scores a keyword without a sign-up, and longer clusters often contain long-tail keywords that are easier to win first.

4. Track the primary keyword against its page

   In Rank Tracking, set the cluster's page as each keyword's target URL and use the cluster name as a tag. MrKeyword shows a Wrong page badge when Search Console reports a different URL for the keyword. The Free plan tracks 25 keywords, enough for the primary keyword of every cluster on a small site.

Is keyword clustering the same as topic clusters?

No. Keyword clustering decides which searches one page should answer; a topic cluster is a group of separate pages on one subject, usually a broad pillar page linked to narrower pages. The two work together: each page in a topic cluster is the home of one keyword cluster. In the dog-bed example, the buying guide could be a pillar page that links to the orthopedic, large-dog and elevated pages, each of which owns its own keyword cluster.

How many keywords should be in one cluster?

As many as one page can answer well, which is usually between a handful and a few dozen. There is no target size. A narrow transactional cluster (“dog bed price”, “dog bed cost”) may have three keywords; a broad guide may cover forty question variants. If you find yourself adding sections that a reader of the primary keyword would skip, the cluster is too big. Google's helpful content guidance also answers the related question directly: it has no preferred word count, so cluster size should follow the topic, not a length target.

What are the limits of automatic keyword clustering in MrKeyword?

• It never looks at Google's results. Groups come from words and a related-terms table, not results-page overlap, because MrKeyword does not scrape Google. Two keywords with the same words but different intent can land together, as the “how to” group showed.
• Its synonym table is short. Synonyms outside it, such as elevated and raised, stay apart until you merge them.
• Groups cover one search at a time. The Groups tab clusters the ideas for the phrase you searched. Combine several searches in a keyword list or a spreadsheet to cluster across topics.
• Searches are Keyword Planner estimates. Group totals inherit Keyword Planner's figures; where it reports none, the group shows no total rather than a guess.
• Keyword Strategy needs Search Console. It clusters only queries your site already appears for, and queries Search Console anonymizes never appear.

Questions and answers