Discovery Bank combines ML and GenAI on Databricks to power a hyper-personalized next-best-action banking model

Overview

Discovery Bank built a next-best-action (NBA) framework on the Databricks Data + AI Platform, using Delta Lake, MLflow and Unity Catalog, that combines actuarial models, machine learning and generative AI to create a behavioral fingerprint for each client, driving segmentation, pricing, risk management and AI-generated communication templates for client servicing. The NBA model produced a 40% uplift in the impact of customer engagement activities, with data processing 20x faster and data product creation 5x faster than the bank's legacy environment.

The challenge

Discovery Bank's legacy on-premises environment struggled to keep up with the volume and velocity of data needed for its shared value banking model, lacking the flexibility to rapidly iterate on data products due to the complexity of integrating multiple components in its ecosystem that were often not geared for advanced analytics.

The solution

Discovery Bank standardized and consolidated data, ML and AI on the Databricks Data + AI Platform, using the medallion architecture in Delta Lake for automated ETL workflows, MLflow to build model requirements for governance and quality, and Unity Catalog for lineage, table history and access controls, integrating data processors, advanced ML models, complex actuarial models and generative AI across the client lifecycle to power a next-best-action framework.

Reported business value

Discovery Bank achieved a return on investment of more than 500%, data processing times are 20x faster and data product creation and implementation times are 5x faster, model-building capacity increased to more than 300 models per day, and the next-best-action model produced a 40% uplift in the impact of customer engagement initiatives.