A model to improve management of banking customers

Purpose The purpose of this paper is to provide a customer lifetime value (CLV) model to carefully assess and classify banking customers using individual measures and covering customers’ relationships with a portfolio of products of the company. Design/methodology/approach The proposed model comprises two sub-models: (sub-model 1) modelling and prediction of CLV in a multiproduct context using Hierarchical Bayesian models as input to (sub-model 2) a value-based segmentation specially designed to manage customers and products using the latent class regression. The model is tested using real transaction data of 1,357 customers of a bank. Findings This research demonstrates which drivers of customer value better predict the contribution margin and product usage for each of the products considered in order to get the CLV measure. Using this measure, the model implements a value-based segmentation, which helps banks to facilitate the process of customer management. Originality/value Previous CLV models are mostly conceptual, generalisation is one of their main concerns, are usually focussed on single product categories using aggregated customer data, and they are not design with a special emphasis on their application as support for managerial decisions. In response to these drawbacks, the proposed model will enable decision makers to improve the understanding of the value of each customer and their behaviour towards different financial products.