Keyword clustering isn’t new but it is a concept that not everyone is using – though they should. Back in the day – like way way back in the day when mobile phones didn’t exist and people built websites using tables – search engine professionals would optimize each page of a website for a single keyword. In this ancient world, a website would have two separate pages optimized for cute kitten memes and memes of cute kittens … because they were two separate keywords. Yeah, silly.
The ghosts of this day still exist, though. The world’s most popular WordPress SEO Plugin – Yoast – still asks for a single focus keyword that it then uses robotically to tell you how well you’ve optimized your page for a single keyword. Keyword clustering – on the other hand – is a technique for creating groups of similar keywords that can be used to optimize website content in place of a single keyword.
This blog post covers the concepts of keyword clustering and gives detailed instruction for how to create keyword groups using our favorite keyword clustering tool by Serpstat. We’ll follow up with a second blog post about Serpstat’s Text Analysis tool that takes the process to the next level by using keyword clusters to analyze the on-page content of a single web page and make suggestions for content improvement based on other high ranking pages associated with that same keyword group.
Table of contents
- What is keyword clustering?
- Why bother?
- How are keywords grouped together?
- Clustering Type – Soft or Hard
- Clustering Level – Weak, Medium or Strong
- How to use Serptstat to create keyword clusters
- What to do with the list of Protoclusters, Superclusters and Clusters?
What is keyword clustering?
The basic concept is simple – keyword clustering is a technique for creating groups of similar keywords. The complexity comes in how the group is made and what you do with the group. When keyword grouping first came around, search engine optimization professionals used complex Excel sheets and a bazaar process called lemmatisation. These days, there’s awesome software to do the trick. Easy peasy.
Why should you use keyword clustering instead of a tried-and-true single keyword approach? Let’s think about it this way. If you were in charge of an $864.73 billion dollar company with 88,110 employees around the world that generated a staggeringly huge amount of money from users coming back to its main product – a search engine – would you run that search engine on a single keyword approach? Probably not. So why are you optimizing your website using a single keyword approach?
How are keywords grouped together?
Now that we’re agreed on using groups of keywords rather than single keywords, the question becomes – who makes the groups, and based on what strategy? Most people answer this with “I make the groups, and they’re grouped based on what keywords I think should go together.” While this makes some sense and takes very little energy, what makes even more sense is to base keyword groups on that $864.73 billion dollar company with 88,110 employees around the world that generated a staggeringly huge amount of money from users coming back to its main product – a search engine. Yes – let’s ask Google what keywords should go together. So most modern keyword grouping software uses Google search results to define the relationships between keywords.
There are a few levers you can pull to control how the groups are made. These generally work to balance two opposing forces:
|More Clusters||Fewer Clusters|
|Each cluster contains fewer keywords||Each cluster contains more keywords|
|Semantic relation among keywords will be greater||Semantic relation among keywords will be lower|
The two most common mechanisms for controlling how related the keywords within the groups are as well as how many groups of keywords are created are Clustering Type (Soft or Hard) and Clustering Level (Weak, Medium or Strong).
Clustering Type – Soft or Hard
Clustering type determines the how keywords must be related in order to be added to a group. Specifically, clustering type defines whether all keywords need to share common URLs or whether any two keywords can share common URLs.
With soft clustering, a cluster is created out of keywords pairs where any keywords share common URLs. Not all keywords in the cluster will share common URLs, but all keywords will be related to other keywords by sharing common URLs.
With hard clustering, a cluster is created only when all keywords share common URLs.
Clustering Level – Weak, Medium or Strong
Clustering level is easier to understand than clustering type. Clustering level defines the number of common URLs that are required in order to create a cluster. Each keyword clustering software varies, but with Serpstat:
Keywords must share at least 3 common URLs among the top 30 search results as a condition for grouping into a cluster
Keywords must share at least 5 common URLs among the top 30 search results as a condition for grouping into a cluster
Keywords must share at least 7 common URLs among the top 30 search results as a condition for grouping into a cluster
How to use Serptstat to create keyword clusters
Serpstat provides killer SEO tools – some even for free – and with a paid account, you get access to two really awesome tools that work together to help you optimize website pages for search engine traffic. One of these tools is their Keyword Clustering tool that does an amazing job of taking a long list of keywords and groups them into keyword clusters based on Google search results. The other is a Text Analysis tool that uses a group of keywords to analyze a page on a website and make specific suggestions for content and meta content updates. These two tools work amazing together. Here, we’ll be talking just about the Serpstat Keyword Clustering tool.
To start off, you can keep track of multiple projects and re-visit the data later, which is great for longer running projects where keyword grouping is used to plan website architecture or heading and subhead structure in a long form article.
Make a new project, import all your keywords (we’ve run up to 1,000 at a time) then select the search engine and even specific region you want to use when generating the keyword grouping. This is huge, especially for those working on Local SEO or geography specific content. Serpstat gives access to an number of search engines including Yandex and Google.
After selecting the search engine and region, select the Clustering Type and Clustering Level as described above.
Once Serpstat’s robots do all their voodoo math, you’ll get a list of:
- Protoclusters – a set of superclusters, which can be useful for thinking about broader website structure
- Superclusters – a set of clusters, useful for thinking about website structure or on-page headings and subhead structure
- Clusters – a cluster of the keywords you originally uploaded, grouped based on the Clustering Type and Clustering Level settings you chose and the search engine and region you selected
Each cluster is then given the following additional details:
- Homogeneity, showing the semantic consistency of a cluster of a scale from 0 to 1
- URL: If you specified a domain while creating the project, Serpstat will look at the domain and display the page which is the closest to the cluster’s topic
Additionally, each keyword is give the following additional detail:
- Connection Strength: how close the keyword is to the cluster’s topic on a scale from 0 to 1
What to do with the list of Protoclusters, Superclusters and Clusters?
This is $64,000 question. Generating lists doesn’t make anyone rich or famous (if it does, please let us know). So then what’s the benefit of all these groups?
Website structure analysis
Creating keyword clusters from a very long list of keywords – either keywords you hope your website will rank for or keywords your website already does rank for – can help you see patterns and categories that you may not have seen when you grouped the keywords together. Seeing things through the eyes of a search engine can open your eyes to different groupings and patterns.
One specific example of this comes from just looking at the keyword clusters created as examples for this blog post. We took 1,000 keywords that this website ranks for and ran them all through Serpstat’s tool for demonstration purposes. What we found, though, was that Google saw pattern in keywords for our portfolio pages. Normally, we make portfolio pages targeted around the name of the company we did the project for. But what we saw is that Google saw patterns in the type of project rather than the company’s name. So we plan to go back to target things like auto garage local seo rather than Marketing For Joe’s Auto Garage.
On-Site text analysis
This is where the real magic happens. Combining Serpstat’s Keyword Clustering with their Text Analytics tool is so awesome that we’ll need to write another blog post about just the Text Analytics tool. Serpstat’s Text Analysis tool allows you to take a keyword cluster – rather than just a single keyword – and use it to analyze the On-Site SEO content of a single page. Serpstat’s tool then actually compares the target page (the one you’re looking to improve) against other pages that rank well for the selected keyword cluster and makes specific suggestions for changes to
- Body content
- Body text length
What’s awesome about this is that it is based on real world websites that rank well, not a theoretical equation like the one Yoast uses. The tool also uses a group of keywords rather than a single keyword, giving more robust results.
Whew… that was a lot. How about you? Are you using keyword clustering for your search engine optimization? What’s working for you? Let us know in the comments.
Full disclosure – I pay Serpstat to use their services and like them so much that I blog about them. I have an affiliate link that I’ve included above, but it’s not likely to make me any cash. Besides, I spent way longer putting together this blog post than I’d ever in a million years make back through affiliate marketing. But hey, good to be honest.