Multinational transaction processor realizes 7-digit cost savings by detecting fraudulent activities

Download Case Study

Industry:

Financial Services

The customer is a public multinational transaction processor for retail B2B and B2C industry, with subsidiaries and partners in over 25 countries. Facing an increase in fraudulent activities and dispute calls, the customer turned to Lore IO to investigate root cause and perfect its internal machine learning model.

The Solution

Lore IO blended order, transaction, demographic, merchant location, and purchase catalog data and detected patterns of fraudulent credit cards in just five days. It developed a profile with unique attributes of reported fraud activities.

The Results

The customer realized cost savings in the tens of millions of dollars by singling out all gift cards that match fraud patterns and their potential dollar impact. The customer also identified historical transactions that matched its optimized machine learning model. The customer aggregated card behavior over time and verified a list of hypotheses that indicated fraud. Machine learning teams used virtualization to build complex set of features. And business users built out complex journey patterns for repeated use.

The Bottom Line

"While we had a great data science team building a good fraud detection service, we suddenly saw a spike in fraud and needed something to understand where it was coming from the Journey feature is the most powerful concept in Analytics and BI I have seen in a while. It allows rapid creation of complex scenarios. Combined with the ability to apply them on top of raw data allowed our team to very quickly iterate on and understand the patterns in the data We were also able to leverage this to improve our models significantly. It's a critical part of fraud monitoring infrastructure now. "