Clustering Types

We have multiple keyword clustering algorithms.

You can change the clustering algorithm in our advanced settings. Go to Keyword clustering and click the advanced settings toggle.

Select the algorithm.

What are the keyword clustering algorithms and how do they work?

Centroids

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 3 by default). All keywords in the group will have a common URL (the one with the highest search volume), but they wonโ€™t necessarily have common URLs with each other.

This method of clustering generally results in larger clusters.

Agglomerative

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 3 by default).

This method of grouping generally results in smaller, tighter clusters and takes a little longer to produce the report as every keyword is being compared against each other.

Which algorithm should I choose?

Centroids Algorithm

Advantages:

  1. Simplicity and Speed: This algorithm can be faster as it only compares keywords against one key term (the one with the highest search volume).

  2. Larger Clusters: It generally creates larger clusters, which might be beneficial if a user is trying to create broad topics or themes.

  3. Easier to understand: This algorithm is less sensitive to keyword cluster fragmentation. Keyword cluster fragmentation is when 2 or more clusters seem really similar and would not actually require a separate page to target them. See the video here explaining this in more detail.

Disadvantages:

  1. Lack of Nuance: It may create clusters that are somewhat arbitrary or lack specificity because they're hinged on a single term.

  2. Missed Opportunities: Some potentially relevant keyword clusters may be overlooked if they do not have enough commonality with the high-volume keyword.

Choose the Centroids Algorithm if:

  • You're seeking larger, overarching themes or topics.

  • You need quicker results and are working with larger keyword sets.

  • You prefer a simpler, more straightforward approach.

Agglomerative Algorithm

Advantages:

  1. Specificity: It typically produces smaller, more precise clusters, which could be more relevant for targeted marketing or SEO campaigns.

  2. Holistic View: As all keywords are compared against each other, it may uncover unique or unexpected keyword groupings.

  3. Thoroughness: This algorithm can be more robust and reliable for forming tightly-knit, highly relevant clusters.

Disadvantages:

  1. Computational Intensity: This approach may take longer for you to get your report.

  2. Complexity: Depending on the implementation, it might be harder for users to understand or interpret the logic behind the clustering. The reports are sometimes more difficult to understand because they are more prone to keyword fragmentation. Keyword fragmentation is explained in the video below ๐Ÿ‘‡.

Recommendations for Customers

Choose the Agglomerative Algorithm if:

  • You're working with a relatively smaller set of keywords.

  • Uncovering unique and tightly-knit clusters is important to your strategy.

  • You're willing to allocate more time for thorough analysis.

For most customers, in most niches, we recommend sticking with the centroid algorithm.

Last updated