One of the top priorities for financial institutions today is fraud prevention and detection. It’s a topic all over the news and our clients are constantly looking for new ways to better manage fraud across the board. Due to the constantly changing nature of fraud, banks need to be able to adapt quickly to stay ahead of the sophisticated tactics fraudsters use.
There are so many different types of fraud and variables to consider, but one of the most perplexing is catching fraud in realtime in the payment stream and stopping it before it actually happens. The majority of the time fraud is caught on the back end and then steps are taken to stop it from happening again. This is not ideal, as fraud losses are costly and nobody wants to be a victim of fraud. Imagine being able to cut losses and protect customers from a negative experience by preventing fraud before it occurs.
The data and analytics are available to accomplish realtime fraud detection; the difficulty is maintaining the speed of the transaction. Because the majority of realtime transactions are not fraudulent, a bank does not want to slow down the 95 percent plus transactions that are low risk. They only want to apply extra time to the fewer than 5 percent that may be of concern. Where does a financial institution start? The use of customer and transaction segmentation helps identify those situations that require flagging and establishing specific rules will determine the type of data financial institutions need to apply to individual transactions. The rules banks need to apply are common sense based on well known fraud prevention strategies.
Incorporating the use of behavioral modeling to determine whether a specific transaction for a specific customer is fraudulent is a key component for catching fraud in realtime. Applying a standard set of rules across all customers will not work and may even cause false positives. It is important to know how individual customers use their cards to avoid stopping valid transactions and creating a poor customer experience. If it is known that a particular customer never charges more than a couple of hundred dollars on their card in any given month and suddenly charges for thousands of dollars appear, that is likely going to be flagged as potentially fraudulent. On the flip side if it is known that a particular customer travels frequently and it is not unusual for them to use their card in New York and London on the same day, those transactions should not be flagged. Unique individual behavior tied to a payment type can save a lot of difficult conversations and perhaps a customer.
Banks need to apply initial rules to all transactions that run extremely fast and do not impede processing time. Optimally these rules take into account individual customer behavior. Additionally, a second set of rules need to be applied to those transactions that have a high risk profile. This logic also needs to be executed quickly, but is more intensive and perhaps includes external data for a more comprehensive evaluation of the transaction. Using this process banks can catch more fraud in the transaction approval stream without creating a bad customer experience for their cardholders. This is where the industry is headed. It’s a positive step toward fraud prevention and eradication.