Fortune 500 Financial Services Company Enhances Fraud Detection with Neo4j and Amazon SageMaker on AWS
A Fortune 500 financial services company faced challenges in fraud detection due to slow, inefficient queries with their previous Microsoft SQL Server relational database, impacting analysts' ability to meet SLAs and detect fraudulent activity efficiently. To solve this, the company deployed Neo4j Enterprise Edition on AWS to enable real-time graph data analysis and visualization, providing faster, more expansive fraud pattern recognition with scalability for billions of nodes and relationships. They also integrated Neo4j with Amazon SageMaker machine learning to enrich fraud detection models, improving suspicious pattern identification. The deployment included a multi-region disaster recovery setup on AWS ensuring high availability and security with role-based access control combined with AWS security features. The solution cut manual review times by half, doubled reviewed transactions daily, uncovered previously undetected suspicious clusters, and enabled instant stopping of fraudulent transactions saving the company thousands daily.
- Organization
- Fortune 500 Financial Services Company
- Industry
- Finance
- Location
- United States
- Published
- November 2024
Reported outcomes
+20%
quantified impactRevenue & growth
Strategic outcomes
Catalog median for revenue & growth deployments: +34% across 150 reported metrics. Compare benchmarks →
Primary read
Use case focus
Showing 3 of 3
- 1Fraud Detection
- 2Real-Time Data Analysis
- 3Graph Database Analytics
- The legacy relational database system caused slow, complex queries requiring multiple JOINs, taking over five minutes per query, which was unsustainable given the volume of up to 10,000 daily transactions per analyst.
- Inefficient manual fraud reviews and delayed data availability compromised fraud detection speed and accuracy, risking missed fraudulent transactions and poor customer experience.
- Implemented Neo4j Enterprise Edition graph database on AWS for scalable real-time graph analysis and data visualization, supporting billions of nodes and relationships.
- Integrated Neo4j with Amazon SageMaker to enrich and enhance fraud detection machine learning models for better suspicious pattern identification.
- Established a multi-region disaster recovery architecture on AWS for high system availability and data security using Neo4j's role-based access control combined with AWS security features.
- Exploring use of Amazon Bedrock to further ground AI models in knowledge graphs for improved fraud scheme identification.
- Manual review times were cut in half, enabling analysts to review twice as many transactions daily, improving fraud detection throughput.
- Uncovered new, previously unnoticed clusters and relationships indicating potential fraud, enhancing detection capabilities.
- Detected and prevented fraudulent transactions worth millions annually, saving thousands of dollars per day via integration with the company's decision platform for instant fraud blocking.
- Achieved high scalability supporting over 4.8 billion nodes and 14.2 billion relationships with 20% annual data growth, ensuring future-proof fraud detection operations.
Architecture
Neo4j Enterprise Edition is deployed on AWS to handle large graph datasets (~2+ TB), enabling real-time fraud detection analysis. Integration with Amazon SageMaker enhances fraud models. Multi-region disaster recovery architecture on AWS ensures high availability. Role-based access control combined with AWS security protects sensitive financial data.
Implementation partners1
Sources & evidence1
AI-generated summary. Verify important details with the linked sources before relying on this case.
Explore related AI use cases
Was this useful?