Fresh Unassisted Newbies (FUN) Qorus Banking Innovation Awards 2023

In response to the dynamic changes in modern banking, the bank aimed to enhance its digital services, particularly with a strong emphasis on its workforce. Our bank has traditionally incurred significant costs in manned channels, involving payments to relationship managers, vendors, and third-party partners who engage with customers to sell products. In contrast, digital and unassisted channels are cost-effective, yielding higher profitability. The Chief Marketing Officer (CMO) set an ambitious goal of achieving 30% of business through unassisted channels, double the existing level. This target served as the inspiration for the development of the unassisted AI/ML framework. Approach and cutting-edge technologies leveraged The approach involved harnessing data from over 15 sources to gather various transactional and digital footprints of the customers. ~45 crore transactions each month were processed to create model features. A Python server with 512 GB RAM, a 80 TB PostgreSQL database was leveraged to ensure smooth execution. Multiple significant features were engineered - transaction volumes, frequency and diversity of digital transactions. The XGBoost modeling technique was used for predictive modelling to successfully capture 85% digital business penetration within the recommended base. Further, leveraging the product recommendations, the propensity scores and the digital savviness, each customer was provided a digital index score and consequently assigned a stage between Stage 1 (assisted) to Stage 6 (unassisted).