Evaluation of Digital Banking Business Models Based on Customer Complaint Topic Analysis Using Latent Dirichlet Allocation

Abstract

The rapid growth of digital banking services has increased the volume and complexity of customer complaints, making it essential for organizations to understand them from a strategic perspective. This study aims to evaluate the digital business model of banking services by mapping customer complaint topics—identified through Latent Dirichlet Allocation (LDA) topic modeling—to the nine elements of the Business Model Canvas (BMC). Using complaint data from ConsumerFinance.gov consisting of 6.3 million entries (2011–2024), and LDA results with an optimized coherence score of 0.56, ten major complaint topics were mapped to corresponding BMC elements. Results show that the most critical elements affected are Customer Relationships (33.19%), Value Proposition (28.33%), Key Activities (12.20%), and Revenue Streams (10.77%). These findings suggest that banking institutions need to prioritize improving their complaint resolution systems, transaction processes, and data authorization mechanisms. This study contributes a novel framework for evaluating digital business models using topic modeling, offering actionable insights for strategic decision-making in digital banking.