From Data to Improvement: Analyzing Performance and Identifying Bottlenecks in the Lending Process with a Data-Driven Process Mining Approach

The loan approval process, characterized by its complexity, path diversity, and time sensitivity, has long been a critical domain for enhancing operational efficiency and customer satisfaction in the banking sector. This study aims to analyze performance and identify bottlenecks within the loan approval process of a financial institution using a process mining approach. The research utilizes data from the BPI Challenge 2012 dataset, which contains over 13,000 loan applications. A quantitative, data-driven methodology was employed, consisting of seven sequential stages: data preparation, process discovery, performance analysis, bottleneck identification, resource analysis, conformance checking, and, finally, the formulation of improvement strategies and continuous monitoring. The results reveal that the loan approval process exhibits a relatively long cycle time, a high rate of rejected or canceled applications, substantial workload concentration among a limited number of users, and low conformance with the reference model. The main bottlenecks were detected in the Application Completion and Credit Assessment stages, while many process deviations were attributed to repetitive or non-standard task sequences. Based on these insights, several improvement strategies were proposed, including the adoption of digital checklists, workload redistribution, standardization of main process paths, automation of time-consuming activities, and deployment of real-time monitoring tools. These actions are expected to improve productivity, reduce operational costs, and enhance customer satisfaction in the banking industry.