Overview
Clustering fails when the segments it produces can't be acted on. A k-means solution with k=7 that produces segments labeled "Cluster 1" through "Cluster 7" with overlapping profiles is not a segmentation — it is a partition. The segments must be distinct enough to justify different treatment, stable enough to persist over time, and interpretable enough for non-technical stakeholders to understand and act on.
Most clustering implementations choose k arbitrarily, apply k-means regardless of data structure, and report cluster centroids without translating them into business language. The result is a technically correct analysis that produces no change in how the business operates.
The Clustering & Segmentation Prompt generates a complete segmentation framework: algorithm selection by data structure, optimal cluster count determination, cluster stability validation, segment profiling in business language, and an operationalization plan that connects segments to specific business actions.
What you get: - Algorithm selection matrix by data structure and cluster shape - Optimal cluster count determination with multiple criteria - Cluster stability validation across random seeds and time - Business-language segment profiling - Operationalization plan with segment-specific actions
Built for: data scientists and analysts building customer, product, or behavioral segmentation for marketing, product, and operations teams.