Query Clustering and Grouping

Keyword Query Clusterer

This tool enables SEO specialists and marketers to swiftly organize key queries into thematic groups, transforming an extensive list of keywords into a structured semantic core in just a few minutes. With simple yet effective analysis and grouping algorithms, it ensures high accuracy without the risk of errors. Supporting over 70 languages, it is ideal for international projects.

Use Cases

• Creating a Semantic Core: Quickly convert a list of keywords into a well-structured semantic core for SEO and content marketing.
• Content Optimization: Create optimized content using clustered keywords to precisely match the target audience's queries.

Clustering Methods

Soft and Hard group keywords by search results: the top results are collected for every keyword, and keywords are grouped by the number of shared pages. This gives an exact split by site pages, but you have to wait for the results for all keywords.

• Hard: Keywords with the highest number of connections form unique groups. This method is suitable for creating well-defined thematic clusters.
• Soft: Groups are formed around the most frequent keywords. This method allows for creating more flexible and comprehensive groups.
• By meaning (fast): Groups keywords in a few seconds, without search results. Good for sketching the structure of a keyword set, sorting keywords into sections and finding junk.

How grouping by meaning works

1. A neural network turns every keyword into a vector, a set of numbers that describes the meaning of the query. Queries with similar meaning get close vectors even without shared words: “iphone repair” and “fix my iphone screen” end up next to each other.
2. Cosine similarity is calculated for every pair of keywords: the closer it is to 1, the closer the queries are in meaning.
3. Keywords are merged into groups until the average similarity inside a group drops below the threshold. The threshold is set by the Strictness slider: stricter gives small precise groups, looser gives large topical ones. Groups for every slider position are calculated at once, so the slider works without a new run.
4. The most general query becomes the main keyword of a group: its words appear in most of the other keywords of the group. The rest are sorted by similarity to the main keyword.
5. Keywords without a match in meaning go to a separate list.

Group density is the average cosine similarity of its keywords: the higher it is, the more uniform the group. Wordstat search volume is added with a checkbox, and groups are then sorted by demand.

Instructions

1. Upload your list of keywords.
2. Select clustering parameters, including language and region.
3. Start the clustering process to generate the semantic core.
4. Choose the level of precision for forming groups after the analysis.

Frequently Asked Questions

• How long does clustering take?

The time required depends on the data volume and can range from a few seconds to several minutes.

• Can keywords in different languages be clustered simultaneously?

No, the tool does not support simultaneous multilingual clustering.

• How is grouping by meaning different from Soft and Hard?

Soft and Hard compare search results: keywords go into one group when the search engine shows the same pages for them. This gives an exact split by site pages, but the top results have to be collected for every keyword. Grouping by meaning does not check search results and compares the queries themselves, so it takes seconds. A handy approach is to sort the keyword set by meaning first and then refine the split by pages with Soft or Hard.