Keyword Clustering
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.
Keyword Insights uses live, country-specific SERP data for each query.
To cluster keywords, we check every keyword in the top 10 results and, if a keyword shares 40% or more URLs in common, we’ll determine that they are a cluster.
There is also a little NLP weaved in and some other clever logic, such as using GPT-3 technology from OpenAI to generate content title ideas.
Hint: You can now adjust this percentage yourself from the advanced settings.
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.
- 1.Which keywords can be targeted with a single page, or which need to be broken into multiple pages.
- 2.How to target that cluster - with a transactional page or editorial content.
Watch the walkthrough video or read the documentation below.
Launch the app and select Keyword clustering from the menu.

Now give your project a name. Select the country and language.

Next, we need to select the clustering settings.

- 1.Toggle cluster insights - Turn on this insight.
- 2.Selecting the clustering algorithm type - We offer two types of algorithms. Both are similar, but there are key differences. You can read about the differences and use cases here. For most use cases, we recommend the centroid approach.
- 3.Keyword grouping accuracy - The minimum number of URLs in the SERPs required to group keywords. The selected grouping accuracy will influence the number of keywords in a single group. The higher the accuracy, the fewer keywords will be put into a single group. The default and the recommended setting is 4.
- 4.Hub creation method - Once we create the clusters using the SERP data, we apply NLP (Natural language processing) to find the semantic relationship between various clusters. This enables us to build the hub/spoke model. Selecting soft will apply minimal relevance, and hard will apply maximum relevance. We found medium to be the sweet spot.
Now the settings are done, we can move on to uploading the keywords.

Click Upload file.
We allow multiple ways to upload your keywords.
- 1.Local file - If you have your own data you pulled from other tools or Google keyword planner, you can use this feature.
- 2.Keyword discovery - If you used our Keyword discovery module to generate keywords, you can bring the data from here.

Upload from a local file
Once you upload a CSV or XLXS file, you will be able to map the keywords and search volume columns in the next step.
The highlighted columns will turn green.
You can upload multiple files and combine everything to make one large file. It's useful when you have keyword data from multiple sources.

Upload from Keyword discovery
Click "Keyword discoveries", and you will be shown a list of all your previous keyword searches. You can click the + icon to add all your keywords.

Once the file(s) are uploaded, click summary.

The next and last stage will allow you to check your credit status and edit the project settings before you submit your order.

Click "Generate report".

The status will go from "Confirmed" to "Processing" and to "Completed"
You can go to projects and find your processed file. You will also get an email with the link to your Google sheets.
Last modified 4mo ago