Keyword clustering tool
Paste the messy list. Get back groups, and each group is one page to write. Group instantly by shared words, or let AI group by meaning when synonyms hide the overlap.
How two keywords get linked
Filler words are dropped and endings trimmed (“cleaning” → “clean”, “skillets” → “skillet”). Then: shared words ÷ all distinct words. In a full list, words found in more than a third of the keywords (usually the seed, like “cast iron” or “pan”) are ignored, since they don’t tell pages apart.
0.60 similarity: same cluster at threshold 0.40. Try “skillet” vs “pan”: no shared word, so only AI grouping will join them.
Why clustering decides your content plan
A keyword list is not a list of articles. Google shows largely the same results for "how to clean cast iron skillet" and "cleaning cast iron pan", so writing two pages means splitting your effort and making your own pages compete. Clustering turns 300 keywords into perhaps 40 pages, each targeting a primary keyword with the rest as supporting phrases in headings, body copy and FAQs.
Three ways to cluster, from cheapest to most reliable
- Lexical (shared words). Fast and transparent: you can see why two keywords were joined. It misses synonyms and sometimes joins keywords that share words but not intent ("cast iron recipes" and "cast iron rust").
- Semantic (meaning). A language model groups by what the searcher means. Better at synonyms and intent, less transparent, and worth a sanity check.
- SERP overlap. Two keywords belong together if their top-10 Google results share several URLs. This is the gold standard, and it needs live results for every keyword. To check a borderline pair by hand, search both and compare page one.
Whichever method you use, name each cluster after the phrase you would put in the title, choose one page type per cluster (guide, list, product, tool), and check that the intent is consistent before you brief a writer.
Questions people ask
What is keyword clustering?
Grouping keywords that one page can rank for together. A cluster usually shares a meaning and a search intent: "how to clean a cast iron pan" and "cleaning cast iron skillet" belong together; "best cast iron pan" does not.
How does this clustering tool group keywords?
The instant mode compares the words in each pair of keywords after removing filler words and plural endings, and links pairs whose overlap (Jaccard similarity) is at or above your threshold. The AI mode asks a language model to group by meaning and intent, which catches synonyms that share no words.
Is SERP-based clustering better?
Grouping by overlapping Google results is the most reliable method because it reflects how Google itself groups queries, but it needs live SERP data for every keyword, which paid tools charge for. Word overlap and AI grouping are a good free first pass; spot-check the borderline clusters by searching them.
How many keywords can I cluster?
The instant mode handles a few thousand keywords in the browser. The AI mode takes up to 120 at a time to keep it fast and free.
What threshold should I use?
Start around 0.4. Lower thresholds merge more aggressively (fewer, broader pages); higher thresholds keep only near-duplicates together. Drag the slider and watch where the groups stop making sense.