Customer Segmentation using Machine Learning for Strategic Supply Chain

Planning

Zhang Jun

Abstract

Amid increasing market competition and rising demand from clients, enterprises are under intense pressure due to increasing operating costs and limited growth prospects. With stagnating margins and changing business landscape, it is critical for companies to work on customer retention as it can be much more cost-effective than acquiring new customers. Successful customer retention requires a strong understanding of customers and building long-term relationships with customers. To achieve this, it is necessary for companies to adopt a strategic approach to manage customer interactions and relationships. One of the core functions of customer relationship management is customer segmentation, which is the process of grouping customers based on their characteristics. By segmenting existing customers and analyzing their behaviors, companies can identify customer profitability and understand customer buying behaviors, allowing companies to concentrate marketing and operating efforts by targeting the right customer group, to achieve optimized company resources and profits.

The purpose of this project is to develop a model using the concept of the RFM analysis and unsupervised machine learning algorithms to perform customer segmentation for better business management, including a demonstration of applying the segmentation results to devise tailored and differentiated supply chain strategies for different customer segments to maximize outcomes. The study is carried out using data set from a ship spare parts distribution company. This report first performs an RFM analysis and further extends to cluster the same using unsupervised machine learning methods including K-mean, Hierarchical clustering, DBSCAN, and Affinity propagation. The methodologies for each algorithm are carefully analyzed and a comparative analysis of the segmentation outcomes for different clustering approaches based on their experimental procedures, clustering speed, noise identification capabilities, and cluster compactness is presented. Furthermore, strategic supply chain planning for distinct customer segments is discussed, and the model that maps customer characteristics with company supply chain strategy is developed for simplified connections between experimental outcomes with real-world practices.

In this study, a novel approach to extending the topic of customer segmentation to differentiated supply chain strategies with an established framework to link the two concepts is proposed. This incorporation provides detailed guidelines for companies to design their supply chain systems to fully benefit from carrying out customer segmentation.