Visualizing Fraud Detection
Retail Database Compromise
As someone who has located and resolved multiple credit card fraud rings across the United States I wanted to share some of the methods for how these types of compromises are detected using fraud data visualization techniques.
These types of compromises are rarely found by retailers. Instead they are found by banks who are able to trace the full activity of a compromised account. In the recent example of the Target data theft, full magnetic stripe information was compromised. This makes is difficult (almost impossible) for a bank to differentiate between a valid card and a compromised card.
In the past, fraudsters would use a valid credit card number and input that card number into a credit card generator program. This program would generate all other cards matching the card sequence. Some accounts would be valid and some wouldn't be valid. Fraudsters would go out to the stores and test the cards to determine if they were valid or not. Banks would have an easy time determining fraud because the only matching element would be the card number.
In this case, all of the data on a card was stolen and "Cloned" onto a different card. To the bank, it looks and acts exactly the same as a valid account. I will show you using visual analytics how this compromise could have been detected by an issuing bank.
Detection Method by Terminal
The 61 cards below have been identified using advanced detection algorithms or customer calls as fraudulent. The visualization below shows how a group of cards would typically be analyzed for a common point of compromise. Compromises in the past have typically been gas pumps, restaurants, etc.
Detection Method by Location and Company
The 61 cards below have been identified using advanced detection algorithms or customer calls as fraudulent. The visualization below shows an analysis by card geography and by company. The analysis below shows that all 61 cards were used at Target. This is a 100% match by company.
0 comments:
Post a Comment