Top 10 overlap
Only the first 10 results by position take part, and a pair must share at least 3 of them before it groups at all.
Keyword clustering
Group keywords by what Google already agrees about.
Two keywords belong together when their results overlap. Clustering reads the SERP observations you already stored, groups by shared URLs, and shows you exactly which ones.
Overlap, not guesswork
One page can serve both keywords — the results say so.
The grouping is arithmetic over stored results. AI may name a cluster; it can never change who is in it.
The evidence is the output
Every member shows the URLs it shares with its pivot and the exact date each observation was captured.
Optional AI labels
A cluster can be given a short suggested name, always marked as an AI suggestion and always safe to ignore.
Blocked keywords are listed
A keyword without a fresh observation is named with its reason — missing, stale, or empty — rather than quietly dropped from the run.
Product truth
A cluster is a count of shared URLs.
This illustrative frame mirrors normalized output. It is not a live search.
| Signal | What the product shows | Source |
|---|---|---|
| Shared URLs | The exact results two keywords have in common | Provider observation |
| Observed date | When each member's observation was captured | Provider observation |
| Cluster label | A suggested name, marked as AI and never used to group | AI interpretation |
How it runs
Deterministic, bounded, and free of vendor calls.
Cadence, source, and run bounds stay visible beside every run.
On demand
Runs happen when you ask. The same stored observations always produce the same clusters, so a re-run is reproducible rather than a fresh roll of the dice.
Stored observations only
Clustering reads SERP observations captured no more than 7 days ago. It issues no vendor request of any kind.
Bounded per run
A run considers at most 200 tracked keywords, which keeps the pairwise comparison predictable and the result readable.
What we don't doHonest limit Honest limit Honest limit
Three things this feature will never claim.
Every limit below is enforced in the product, not written on a disclaimer page.
Fresh observations required, and nothing is fetched
Only Google-engine keywords with an observation captured in the last 7 days take part. Keywords that are missing, stale, or stored with no results are named with their reason — this feature never runs a rank check to fill the gap.
A threshold, not a similarity score
A pair groups only when it shares at least 3 of the top 10 results. Two shared results is noise, not a cluster, and the product will not soften the threshold to produce a tidier picture.
AI names clusters, it never forms them
Membership is decided entirely by shared-URL arithmetic. A label is a suggestion shown behind an AI marker; if labelling fails, every cluster simply keeps no label and the run still completes.
Every limit below is enforced by the product, not a disclaimer.
Zero vendor spend
Find the keywords one page can win together.
Create an account to cluster your tracked keywords, or begin with a free site audit.