Documentation
Topic Clusters
Group queries into topics with keyword lists or regex, then track each cluster's clicks, impressions, CTR, position and share, and the queries inside it.
A topic cluster is a named set of queries defined by keywords: every search that mentions pricing, every query about one product family, every comparison with a competitor. The Topic Clusters report adds up clicks, impressions, CTR and position for each cluster, compares them with the earlier period, follows each cluster day by day and shows the queries and pages inside it. Use it to judge whole topics rather than single keywords.
Creating a cluster
Choose Manage Clusters on the report, or Create Cluster under Property Setup → Content Groups & Topic Clusters, then New cluster.
- Name the cluster.
- Choose how it matches (see the table below).
- Enter the keywords, one per line or separated by commas, up to 200 per cluster.
- Check the preview, which counts the queries and clicks the cluster would capture in the current range and warns when some already belong to another cluster, then save.
| Matching | A query joins the cluster when |
|---|---|
| Contains any keyword | It contains at least one keyword as a whole word or phrase; a plural ending in "s" or "es" still counts |
| Contains all keywords | It contains every keyword |
| Matches regex | Any keyword, read as a regular expression, matches it |
A property holds up to 50 clusters. A query joins the first cluster it matches, so place narrow clusters before broad ones. Matching ignores case, and keywords written in scripts other than the Latin alphabet match anywhere in a query rather than as whole words.
Suggestions from your queries proposes clusters from words and two-word phrases that recur in at least three of your queries, leaving out common filler words and your brand terms. Import CSV creates many clusters at once: one cluster per line, its name followed by its keywords in quotes so their commas are not read as new columns, and one match mode for the whole import. Rows that cannot be used are listed after the import instead of vanishing.
Pricing,"price, pricing, cost, plans"
Integrations,"slack, teams, zapier, webhook"Reading the report
- Cards for Topic clusters, Clustered queries, Clustered clicks and Unclustered clicks, the queries no keyword matched.
- Clicks & impressions per cluster and Share of clicks for the selected period.
- Cluster Performance Over Time: daily clicks or impressions per cluster; click a cluster below the chart to hide it.
- The cluster table: Queries, Clicks, Impressions, CTR and Avg. position, compared with the earlier period, plus Share of clicks and the Keywords.
- Inside a cluster: choose one to list its Queries, each with the keyword that matched it, or its Pages, each with the number of the cluster's queries it ranks for. Up to 200 rows each.
The Filter panel works as on any other report, branded split included. Cluster totals leave out the rare queries Google withholds, so a cluster's real traffic is usually larger than shown (why).
Ways to use clusters
- Follow a content pillar. Cluster the topic, then watch whether new articles grow its clicks and impressions or only move them between pages. The Pages view shows when several pages share a cluster's queries, and Cannibalization examines that query by query.
- Find demand you do not yet serve. A cluster with many impressions but a weak position marks a topic that needs better pages. The content engine plans topics from Search Console searches on its own, and clusters let you check what it achieves.
- Keep the queries. Under Inside a cluster, Save to keyword library stores a cluster's queries for research and rank tracking.