Case Study: BNP Paribas Personal Finance reduces fraud by 20% with Neo4j

BNP Paribas Personal Finance Reduces Fraud by 20%

BNP Paribas Personal Finance needed a stronger way to detect fraud across more than 800,000 credit applications a year, as fraudsters reused and altered information to bypass traditional blacklist and rules-based checks. The company turned to Neo4j and its graph database to uncover hidden relationships in connected data and support real-time scoring decisions.

Neo4j worked closely with BNP Paribas Personal Finance to build and pilot a new graph-powered fraud detection model, including data modeling and query development. The result was a system with a maximum query latency of 2 seconds that helped reduce total fraud by 20% while still approving valid applications efficiently.

BNP Paribas Personal Finance

Julie Cavarroc

Data Scientist in Scoring Center, Central Risk