Keyword Clustering Tool

Keyword clustering involves grouping keywords that are similar to each other together. Essentially a “keyword cluster” is a group of keywords that mean the same topically and can be targeted together on a single page.

Upload your keyword list into Keyword Insights and we’ll analyse the search engine result pages to group any keywords together where the ranking URLs are the same or similar.

Watch Video Demo

Optimise for more than one keyword. Get better results.

It’s so much easier (and organically rewarding) to create content that can rank for many low-volume, low-difficulty keywords than creating a piece targeting a single high-volume keyword.

Cluster your keywords and aim to rank for groups of keywords.

Learn more about clustering here

Get 750 Credits for $1

Take the guesswork out of content production.

When creating new content, it can be very difficult to know when a certain piece should be broken out into more specific sub-topics. For example, if we had the keywords "architect fees" and "how much do architects cost?" would you quickly know whether you needed 2 different pages to target these, or if they could both be targeted on the same blog?

Keyword Insights makes this quick and painless. We use live search results pages and group keywords based on what's ranking. So you’ll know, in seconds, when a page should be broken out into sub-topics to stand the best chance of ranking.

Find gaps in your content quickly. Like, really quick.

With the flick of a switch, we allow you to pull in your current rank and the ranking URL for each cluster.

This will allow you to very quickly spot gaps in your content... It's just a case of looking for the clusters that don't rank (or rank poorly).

We cluster better. We really do.

We understand every niche is different so we give you maximum flexibility with your keyword clustering settings. You're able to adjust the URL overlap (default is 40%), the type of clustering algorithm used and how strict the NLP is to form your hub and spoke recommendations.

The Centroid Clustering Algorithm

We take the keyword with the largest search volume and then group all other keywords which share x number of URLs in common with it from the top 10 (x can be changed by you, but it is set at 4 by default).

All keywords in the group will have a common URL (the one with the highest search volume), but they won’t necessarily have common URLs with each other. The group will always be named after the keyword with the most volume. This method of clustering generally results in larger clusters but doesn’t eliminate the chance that irrelevant search queries will be clustered together. Recommended if your keyword list is small (<1000) or you need the results quickly.

The Agglomerative Clustering Algorithm

All keywords are compared against one another and are clustered into a group if they share the x number of URLs in common from the top 10 (x can be changed by you, but it is set at 4 by default).

This approach provides a higher level of accuracy as every keyword in the set shares x amount of URLs in common with each other and not just the keyword with the most volume (as in the centroid version of clustering). It eliminates the chance of irrelevant search queries being clustered. This method of grouping generally results in smaller, tighter clusters but takes a little longer to produce the report as every keyword is being compared against each other. Recommended if you need more accuracy and have time to wait for the report.

We make topical modelling simple.

Once you have your keyword clusters, it may be helpful to know how closely related those clusters are to other clusters. We apply NLP to find the semantic relationship between the clusters.

What this allows you to do is view how similar certain clusters are to one another so that you can easily plan and create a content calendar around a given "topic".

The hub and spoke insight is part of the “clustering” insight, so you won’t need to select this as an additional option.

Actionable Insights through Cluster Analysis with Our Algorithm

We understand that analyzing cluster outputs can be a laborious task. That's why we created an in-house algorithm that carefully analyzes your output clusters, selects the best keyword from each cluster, and even highlights it for content briefing and writing.

Our algorithm takes into account search volume and uses a machine learning model to determine the keyword that generates the most articles in Google SERPs. With just a click of a button, you can send the cluster for brief creation, saving you time and effort.

We operate in Google’s world

Unlike some clustering tools which attempt to use Natural Language Processing to group keywords together, we actually analyse Google search results and group them based on what's ranking. It means we're more accurate.

Please note: We do use NLP to "cluster the clusters" together as detailed in the section above, but not to cluster the keywords together.

Our clustering report uses live, country-specific SERP (Search Engine Results Page) data for each query.
To cluster keywords, we check every keyword in the top 10 results, if a keyword shares 40% or more URLs in common, we’ll determine that they are a cluster (although you are able to adjust this 40% as you see fit).
By flcking a switch in the settings, our clustering report can also be enhhanced with GPT-3 technology from OpenAI to generate content title ideas.

“Keyword insights has quickly become an essential part of our agency's tool stack. The clustering, intent classification and briefing features help us supercharge our SEO and content team's productivity levels, allowing them to spend more time delivering valuable insights for our clients.”

William Nye

Technical Director at Builtvisible

See our Keyword Clustering in action

The clustering report KWI generates can be used to easily build topic clusters around parent keywords. Our report will clearly show you what each page’s head keyword should be, whilst also listing other keywords that page can target. Combine our cluster insight with our rank tracking feature and spot gaps or cannibilisation issues in your content quickly.

Play Video

Frequently asked questions

We have put together some commonly asked questions.

What countries do you support for keyword clustering?

Keyword clustering is not language-specific but rather geo/country-specific. We support all the countries in the world.

Let's take a look at a few examples.

I have an SEO client in France, and I'm targeting users in France. When I prepare my keyword list, I will select all the keywords and search volume data for those keywords. I will choose France under the country dropdown when I run my order through Keyword Insights.

Keyword Insights will then scrape google.fr and build the clusters based on how keywords rank in google.fr.

You can choose to upload French keywords, English keywords, or any other language. Regardless of language, Keyword Insights will scrape google.fr for those keywords.

As you can see, language has no influence whatsoever on our clustering process.

Why do I have no_cluster in the output?

If the keyword it's not clustered with any keywords, it now either means one of two things:

1) I need to do more research on this word. It's interesting to me, so I'll set it aside and do more research on it in the future. I'm clearly not going into enough depth with my current keyword list.

2) This keyword is just incredibly specific and requires a deep dive on just one page.

Solution.

To improve your results, you could add more keywords to your upload. No_cluster simply means those keywords don't share at least 4 URLs with any other keyword so if there are any in there that you want to write about or are interesting to you or have a lot of volumes, our recommendation would be to do more research around those terms.

You'll also sometimes see some surprising results (i.e. keywords you THINK should be in a cluster as they're really similar to others, like pluralised versions of the same keyword) but if you actually test them, you'll see Google does in fact show very different results.

We recently came across one cluster "50mm skate wheels" and then a keyword in the no_cluster "50mm skate wheel" (singular version), and we thought it was a bug. We searched both, and, right enough, the results were very different.

Do special characters affect the output?

Special characters are common in many languages. Such as Norwegian, Thai, Swedish, German etc.

If you're using Microsoft Excel or Mac Numbers, please export your files with UTF-8 encoding. Else your special characters will not be exported correctly, and Keyword insights will show you an error.

When possible, we highly recommend you upload an excel XLXS file when dealing with special characters instead CSVs. The CSV encoding can convert special characters into random unreadable text, and it will cause our clustering and hub/spoke algorithm to produce incorrect output results.

How many keywords can you support?

We can support up to 4m keywords at a time.

However, please be aware that Google sheets won't work after 5 million cells, so if you want to use the Google sheets output (which is better), we recommend not uploading more than 150k keywords.

If you upload more than 150k keywords, you will still be able to use the excel report we provide. We just find it's a little more difficult to use. If this is a concern to you, and you want to upload more than 150k keywords at once, get in touch and we can help you with some workarounds.

How does Keyword clustering works?

Do I need other SEO tools?

Start your trial now for only $1

Try Keyword Insights. Get 750 credits for $1. Cluster upto 750 keywords, create upto 7 briefs or do 1500 keyword discovery searches.