Customer Clustering Using Fuzzy K-Means and Cluster Evaluation via MADM to Improve Sales and Marketing Performance

This applied, exploratory study uses data-driven CRM to improve sales and marketing at Sina Fidar Kimia Company (۴۹ customers). Customers are segmented with K-Means and fuzzy C-Means, then the clusters are evaluated using Multi-Attribute Decision Making (MADM). The approach integrates three steps: clustering, fuzzification, and criterion weighting via the Best–Worst Method (BWM). The novelty lies in combining cluster-wise BWM weighting with higher-order (type-۳) fuzzification to reflect uncertainty in expert-based CRM indicators and to derive segment-specific managerial actions. Eleven indicators are analyzed, including purchase method, on-time payment, legal status (individual/legal entity), credit backing, reputation, brand, expert judgments (commercial manager and specialists), purchase share, and economic factors. Fuzzification is performed with a type-۳ fuzzy logic model in logistic form using composite sine–cosine membership functions (Sheffer type ۴) to capture nonlinear relationships more flexibly than triangular/trapezoidal functions. Based on inputs from ۱۴ experts, BWM weights are calculated within clusters. Results identify five distinct customer clusters with different priority indicators, enabling targeted sales and marketing actions and supporting personalized service recommendations to enhance profitability, satisfaction, and loyalty.

کلیدواژه ها:

Clustering ، Customers ، Validation ، Multi-criteria Decision Making