How Capitec Built a Graph-Powered Fraud Scoring Pipeline for 3.5M+ Daily Cases

Capitec set out to improve how it detects authorized push payment fraud, where scams often look legitimate in a single transaction but reveal themselves across connected accounts, payment paths, and repeated network patterns. What the team faced was a fraud problem that could not be solved well through isolated transaction review alone, especially at banking scale.

In this succes story, learn how Capitec used Memgraph to move from manual investigation maps to production-grade fraud scoring. The pipeine is able to process more than 3.5M daily records in about two hours on average. By combining built-in graph algorithms, graph visualization and graph-powered feature engineering, the team built a system that helped reveal hidden fraud connections and scale graph adoption from one live graph to seven. Here is what we will cover in this whitepaper:

What you’ll learn:

• Why authorized push payment fraud requires a connected view of payment behavior
• How graph analysis exposed deeper fraud patterns beyond isolated transactions
• How Memgraph supported production-grade fraud scoring
• How Capitec scaled graph adoption from one live graph to seven