Keyword Clustering: How to Group Keywords Into Content

Why cluster keywords at all

Google increasingly ranks pages by topic, not phrase. Ten near-identical pages targeting variations of the same query cannibalise each other; one strong page targeting the whole cluster wins.

Clustering answers: which of these keywords should live on the same page?

The two methods

SERP-based clustering

How: for every keyword, pull Google's top 10 results. Two keywords cluster together if they share ≥ 3 URLs in their top 10.

Pros: empirical — reflects what Google actually treats as the same intent. Cons: requires a SERP fetch per keyword (API cost). Best for: deciding what to build.

Semantic clustering

How: embed every keyword with a text model, group by cosine similarity above a threshold (usually 0.75+).

Pros: fast, cheap, doesn't need SERP fetches. Cons: can group queries with different intent that happen to share vocabulary. Best for: initial brainstorming, deduplicating giant keyword lists.

Use both: semantic for the big sweep, SERP-based for the final commit.

Step-by-step workflow

1. Dump your seed keywords into one file (500–5,000 is typical).
2. Run semantic clustering to shrink to ~50–200 rough clusters.
3. Run SERP-based clustering within each rough cluster to confirm groupings.
4. Name each cluster with its highest-volume phrase (that becomes the H1 candidate).
5. Assign one URL per cluster in your content plan.

Worked example: grouping 12 keywords by hand

Search Console shows people asking how to do keyword clustering and how to group keywords — so here is the manual version, no tooling required.

Take this list: rank tracker, rank tracking tool, best rank tracker, rank tracker pricing, how to track keyword rankings, how to track rankings automatically, check my google ranking, google rank checker, free rank checker, serp checker, rank tracker for agencies, white label rank tracker.

1. Search three of them and compare the top 10. rank tracker and rank tracking tool return 7 shared URLs — same cluster. how to track keyword rankings returns tutorials, not product pages — different cluster.
2. Split by intent, not vocabulary. Four groups fall out: a product cluster (rank tracker, rank tracking tool, best rank tracker), an instructional cluster (how to track keyword rankings, how to track rankings automatically), a tool-use cluster (check my google ranking, google rank checker, free rank checker, serp checker), and a segment cluster (rank tracker for agencies, white label rank tracker).
3. Leave rank tracker pricing on its own. Its SERP is comparison tables, not product pages — one shared URL with the product cluster is not enough to merge.
4. Name each cluster with its highest-volume phrase and assign one URL. Four clusters, four pages, twelve keywords covered — instead of twelve thin pages competing.

The rule that does the work: two keywords belong together only when Google already answers them with the same pages.

How LemRank clusters

The Keyword Research module runs semantic clustering automatically on every imported keyword list, then SERP-based verification on your priority list. Clusters show:

• Total volume across the cluster.
• Best-target head phrase (highest volume + reasonable KD).
• Suggested URL slug.
• Whether an existing page already targets the cluster (avoids cannibalization).

Common mistakes

• Over-clustering. Grouping 40 keywords onto one page dilutes intent — the page ends up ranking for none.
• Under-clustering. Publishing 10 similar pages creates cannibalization risk.
• Ignoring intent split. "Best CRM" and "CRM software" look similar semantically but often have distinct SERPs.

What this doesn't do

• Clustering doesn't decide what to write — it decides which queries go together.
• Clusters need re-verification quarterly because SERPs shift.

Related

• Long-tail keywords guide
• How to do a keyword gap analysis