Enhancing Credit Card Risk Assessment with Alternate Data: A Success Story for a Leading Bank

Our client, a prominent private sector bank, sought to revolutionize its credit card application approval process. They aimed to leverage alternate data to gain deeper insights into the credit risk profiles of potential credit card customers.

• Development of Alternate Data-Based Score: We crafted a unique credit assessment score by harnessing alternate data sources.
• Smartphone Data Utilization: We collected valuable data from customers’ smartphones, including information on the types of apps installed, average credit card usage, and the total number of credit cards held by customers.
• App Classification and Risk Assessment: We categorized the different types of apps on customers’ phones and analyzed the default rates associated with these app categories. This information was used to identify risk patterns.
• Tree-Based Model Integration: We incorporated a tree-based model to pinpoint the riskiest customers. This model’s output served as a key variable in the subsequent XGBoost model.
• Crosstab Analysis: To create more refined customer segments, we conducted a crosstab analysis comparing the traditional scorecard (based on inquiry and Credit Bureau data) with the score derived from alternate data.

• Enhanced Customer Risk Profiling: The bank gained a comprehensive understanding of customer risk profiles, thanks to the utilization of alternate data.
• Risk Reduction and Revenue Increase: By implementing our innovative approach, the bank managed to reduce risk while simultaneously boosting revenue.
• Reclassification of Medium Risk Customers: Notably, some customers originally classified as medium risk by the traditional scoring model were reclassified as low risk or high risk when assessed using our alternate data model. These reclassified groups exhibited default rates consistent with low-risk and high-risk categories, leading to more accurate risk assessment.

In summary, our collaboration with this leading bank showcases how leveraging alternate data can transform credit risk assessment, resulting in more precise customer segmentation and, ultimately, improved risk management and revenue generation.