Case Study: Global Payment Solutions Provider achieves 20% uplift in transaction fraud detection and 5x faster model builds with DataVisor

Global Card Network Relies on DataVisor’s AI Solutions to Boost Transaction Fraud Detection

Global Payment Solutions Provider, which handles several trillion dollars in payments annually, was struggling with rapidly evolving fraud tactics, fast model decay and lengthy model-build cycles, incomplete digital data and a lack of modern infrastructure for real‑time, large‑scale fraud detection. To address these gaps the company engaged DataVisor and adopted DataVisor’s open machine learning modeling platform, combining unsupervised and supervised learning plus feature-engineering and big‑data capabilities.

DataVisor implemented unsupervised ML to augment the client’s supervised models, a feature‑engineering layer that works with partially filled data, and a high‑throughput platform for real‑time scoring and indexing. As a result, the Global Payment Solutions Provider saw a 20% uplift in transaction fraud detection, 94% detection accuracy, and a 5x faster model build time (from 4–6 months to weeks), while achieving high QPS processing with 50–100 ms latency.