Keyword clustering sorts your keyword list into groups that each stand for exactly one page. You clean the list, pre-sort by search intent, group by meaning and check the important groups against the search results. That shows you which keywords belong together and which need a page of their own.
After thorough research you often have hundreds or thousands of rows, many of them variants of the same question. After reading, you can turn them into a page plan in which every primary keyword has exactly one target page and no two pages compete for the same search.
What a keyword cluster is
A keyword cluster groups every query that the same page answers best. The searchers share the same intent, even if they phrase it differently. "Rank tracker", "monitor rankings" and "track keyword positions" are three ways of describing one task and belong on one page.
That makes a cluster a page decision, not just a word group. This is what sets it apart from simply sorting by shared words. "Keyword research tools" and "keyword research guide" share two words but call for completely different pages: a comparison and a how-to.
Several keyword clusters on a broader subject together form a content cluster. That is the level above, where you plan how the pages link to each other.
Why clustering matters
Without clustering, two typical mistakes happen. The first: a separate page gets planned for every keyword variant. The result is many thin pages competing with each other. The glossary describes this problem as keyword cannibalization. The second mistake is the opposite: completely different intents end up on one page that serves none of them properly.
Clean clustering solves both. It shows how many pages you really need, which terms a page should also cover and where an existing page just needs to be expanded.
Three ways to cluster
By words. Keywords with shared word stems are grouped together. This is fast and fine for a first sort, but it produces many errors because the same words do not mean the same intent.
By meaning. Semantic methods group keywords that are similar in content, even if they share no words. "Monitor rankings" and "track positions" end up together. This works well for large lists but can merge queries that Google treats differently. The glossary covers the background under semantic search.
By search results. Here you compare which pages Google shows for two keywords. If the top results overlap strongly, Google evidently treats both searches as the same task. This is the most reliable basis for a page decision, because it does not rest on assumptions but on what Google actually does.
In practice you combine the approaches: pre-sort semantically, then compare search results for important groups.
Step by step to a cluster
- Clean the list. Remove duplicates, other brands, job searches and anything that does not fit your offer. Every row you drop here saves work later.
- Pre-sort by intent. Roughly divide the keywords by search intent: informational, comparison, transactional, navigational. Keywords with different intents almost never belong in the same cluster.
- Group semantically. Within each intent group, combine phrasings that ask the same question. For small lists you can do this by hand, for large ones a tool helps.
- Check against search results. For each group, take the keyword with the most demand and compare its top results with those of the other members. A common rule of thumb: if two keywords share three or more results in the top ten, they usually belong on the same page. If they share hardly any, separate them. The glossary explains how to read results for this under SERP analysis.
- Set a primary keyword. Every cluster gets a primary keyword, usually the one with the most demand and the clearest intent. It determines the title and main message of the page.
- Assign a target page. Check whether you already have a suitable page. Expanding an existing page is often enough. A new page is only worth it if there really is none.
- Record the decision. Note the primary keyword, members, target page and reason for each cluster. Without this note, the discussion starts from scratch next time.
Running the SERP comparison in practice
You can approach the comparison of search results in three ways, depending on the size of the list.
By hand. For a handful of important clusters, it is enough to search the keywords one after another and note the top ten URLs. Use a private browser window and the right country setting. It only takes a few minutes per keyword and is time well spent for page decisions with a large impact.
With a spreadsheet. If you can export the top ten URLs from a tool, put them side by side in a spreadsheet and count the shared URLs per keyword pair. That lets you compare a few dozen keywords with reasonable effort.
With a clustering tool. For large lists there are tools that fetch search results automatically and group keywords by overlap. This saves a lot of time but costs money per results page fetched. So it pays to clean the list thoroughly first and only check the keywords that matter for a page decision.
Whichever path you choose, check the results of the most important clusters with your own eyes as well. A tool counts overlaps, but it does not recognize whether the ranking pages really serve the same purpose.
Handling edge cases
Not every group can be assigned clearly. These cases come up often.
The search results are mixed. Google shows how-to guides, comparisons and product pages for one keyword at the same time. The intent is evidently ambiguous. A page that serves several intents is often worth it here, such as a guide with a short comparison, or you deliberately choose the intent that fits your offer best.
A keyword fits two clusters. Assign it to the cluster whose search results match most closely. A keyword always belongs to exactly one target page. The other page can mention and link to the topic but should not treat it as its main subject.
The search results change. Search results are a snapshot. For important clusters it is worth repeating the check after a few months, especially if rankings change unexpectedly.
Singular and plural. Tools such as Google Keyword Planner group close variants together. The search results can still differ, for example when the plural shows mostly comparison pages and the singular mostly definitions. Check both for important terms.
An example
A company selling accounting software has about 400 keywords after research. After cleaning, 260 remain. Pre-sorting by intent produces four large groups: basics such as "what is an incoming invoice", how-tos such as "how to write an invoice", comparisons such as "accounting software comparison" and purchase terms such as "buy accounting software".
Within the how-tos, semantic grouping combines "how to write an invoice", "create an invoice" and "invoice template". Comparing the search results shows, however, that almost only downloadable templates rank for "template", while guides rank for the other two. The team therefore splits "template" into its own cluster for a template page. The 260 keywords become 34 clusters, 21 of which match existing pages and only 13 of which require new pages.
Maintaining clusters in practice
Clustering is not a one-off task. New keywords come in, existing pages get reworked and search results change. A few habits help you keep track:
- Maintain clusters in one fixed place, not in changing spreadsheets.
- Assign new keywords to existing clusters before creating new ones.
- Watch which URL actually ranks for a keyword. If it differs from the planned target page, either the assignment or the page is off.
- Link the pages of a topic to each other, for example through a central pillar page, so Google and readers can see how they relate.
In the Crawl Foundry keyword database you import your keyword list for free and organize it into lists and tags, for example one list per cluster and tags for intent and target page. You fetch search volume, search intent and SERP data only for the keywords you need for the decision and see the cost before you confirm. Later, a rank tracker shows whether the planned page ranks for each keyword.
From keyword list to page plan
Keyword clustering connects research and page planning. Clean the list, pre-sort by intent, group by meaning and check important groups against the search results. Every cluster gets a primary keyword and exactly one target page, and existing pages take priority over new ones.
Start with the cleanup and remove every keyword that is irrelevant to what you offer.