Fragmented Decisioning Process

Client had a fragmented credit decisioning process with multiple siloed systems, manual data aggregation, and offline policy checks. The process also had high levels of “non-decision” outcomes that required manual expert intervention for investigation leading to lengthy decisioning and loan disbursal times and subsequently lower customer satisfaction and lost potential revenue to competition.

Enhanced Decisioning Process

• Built a centralized data repository combining the multiple data sources into a single database in an standardized, central database to automate data aggregation process.
• Built a custom ML model based on the client’s consumer durable application data and underwriter decision along with its reason codes to improve model performance, leading to lower decisioning time and lower number of cases requiring investigation
• Further enhanced the model performance using feature generation process by introducing relevant alternative variables to create multiple benchmarking models
• Integrated the final deployed ML model with existing client systems through APIs

Improved Decisioning and Optimized Alerts

Solytics solution allowed the client a singular, more relevant view of the credit worthiness of its customers by utilizing relevant customer attributes like utility bills, bank statements, statement of assets. The ML model also improved the effectiveness of the decisioning process by optimizing the alert ratio, and reducing the overall decisioning time. The final impact was in terms of improved user experience and lower total cost of operations.

‍

‍