Payment Fraud Prevention

Our Client

UnionPay handles over $17 trillion worth of transactions annually.  It is the world’s largest card processing company and it operates in 174 countries.

The Challenge

UnionPay wanted to prevent fraudulent transactions at the time that a card was presented for payment. Traditionally, card fraud is often detected only after the payment has been made.

  • While the card company can remedy the situation for the legitimate card holder, the bank must work to recoup the loss.
  • As ever-more card transactions take place daily, the ability to analyze those transactions is putting the company under increasing pressure.
  • UnionPay manages millions of payments each second and needed a real-time processing solution that could identify fraudulent activity instantly.

The Desired Outcome

UnionPay wanted to significantly improve their anti-fraud capabilities by preventing fraudulent activity immediately with a real-time detection and prevention solution.

  • UnionPay needed a very sophisticated, high-speed processing data streaming capability, with specific rules applied to stop fraudulent activity.
  • UnionPay wanted a solution that could analyze millions of concurrent transactions in real time to detect suspicious activity and intervene to prevent fraud before the crime had been committed.
  • The solution can incorporate machine learning models to dynamically adapt to changing behavior, evolve to changing patterns and produce higher accuracy with fewer false positives.

The Solution & Key Benefits

Horizon8 leveraged its real-time data streaming analytics platform, known for its high-capacity, high-speed, and low latency capabilities, to deliver the solution.

The solution uses a rules engine to analyze data in real-time, allowing anomalies and outliers to be detected instantaneously.  These business rules are designed and applied in line with client requirements.

  • We set up fraud detection models in the rules engine, which observed spending behaviors and patterns of cardholders.
  • We identified scenarios and use cases and created over 200 dynamic models to detect fraudulent activity.
  • The solution was easy to integrate within UnionPay’s infrastructure and operating environment.
  • In the first year of using the solution, UnionPay was able to prevent $30 million globally in fraudulent transactions.