Segmentación de clientes mediante técnicas de clustering para optimizar las estrategias de contacto en la colocación de productos en el Banco de Loja

Abstract

Abstract: This paper addresses one of the most common problems within the banking environment, focusing on the low effectiveness of the contact strategies and actions of Banco de Loja, caused by massive and homogeneous actions. The main objective was to segment the customer base using clusters to optimize the placement of financial products. The methodology employed was quantitative and descriptive, using a sample of 156 historical records combined with customer surveys. For subsequent data analysis, the K-Means unsupervised learning model was applied. The results identified two independent clusters based on sociodemographic profile, level of engagement, and communication preference: Cluster 0 (77.5%) is characterized by grouping customers with a high affinity for asynchronous channels such as WhatsApp, while Cluster 1 (22.5%) encompasses young, mature users with greater engagement in policies, who prefer email as their communication channel. The results demonstrate that the current no-contact rate (40%) is explained by homogeneous and impersonal strategies, and an asynchronous omnichannel approach with personalized attention is proposed.