CUSTOMER SEGMENTATION WITH CLUSTERING METHODS IN THE RETAIL INDUSTRY

Abstract

The marketing world moves away from product-oriented work, understood customer importance, and shifts towards customer-centered practices. Today with tech development and increasing competition, company-customer relations become more important. Creating a customer profile is critical for businesses to recognize their customers and distinguish their most profitable customers. By understanding their customer behavior, companies can tailor their marketing and customer relationship management strategies to suit them and fulfill their customer needs, increasing their satisfaction and loyalty to their business, and encouraging them to shop from them again. Thus, this study aims to categorize customers based on RFM metrics and interpret the obtained clusters from a marketing perspective. At the segmentation phase, hierarchical and non-hierarchical clustering methods, namely k-means, AGNES, and DBSCAN, are used and the results are compared. First, data, which consist of the shopping information of 38975 customers who shopped from e-commerce in one year, are collected from a textile retail company in Istanbul. Then, the purchase amount spent by customers is additionally scored to reveal the most valuable customers. It is observed that better results are mined from the k-means algorithms. As a result, four different customer types are determined: loyal customer, potential customer, new customer, and lost customer types. In conclusion, profile-oriented marketing strategies are presented.