Silhouette Score (Selecting k)

Silhouette Score is a clustering quality metric that measures how well each data point fits within its assigned cluster versus neighboring clusters. The score ranges from -1 to 1, where 1.0 indicates perfectly separated clusters, 0 means the point sits on cluster boundaries, and -1 suggests misassignment to the wrong cluster.

In SEO clustering workflows, practitioners calculate Silhouette Scores for different k values (typically 2-20) and choose the k that yields the highest score to reduce manual guesswork about cluster count. Good clustering in SEO typically shows scores of 0.4–0.6, while scores above 0.6 indicate excellent separation.

Scores below 0.3 suggest you should reconsider your embedding model or expand the keyword pool. When silhouette scores are consistently low, avoid proceeding with clustering, as this indicates excessive topical overlap between keywords.