How are keywords grouped?
The local algorithm removes common stop words, lightly folds English suffixes and compares token sets using Jaccard similarity. It is deterministic and does not use search results or embeddings.
Deduplicate up to 500 keywords, create deterministic topic clusters and names, inspect assignment similarity and download CSV.
The tool normalises Unicode word tokens, removes common stop words, applies light suffix folding and greedily groups keywords by Jaccard similarity at a fixed 0.40 threshold.
Input: running shoes; best running shoes; email marketing; email marketing guide
Output: Named running-shoes and email-marketing groups with assignment similarity
The local algorithm removes common stop words, lightly folds English suffixes and compares token sets using Jaccard similarity. It is deterministic and does not use search results or embeddings.
No. Token similarity is a topic heuristic, not SERP overlap or search-engine evidence. Review every group before changing a site structure.
TXT and CSV text files up to 256 KiB are supported, with at most 500 unique keywords and 120 characters per keyword.
Yes. The browser reads and clusters them locally; the list is not sent to Tool Teaze or an external provider.