Three separate machine learning fraud solutions targeting credit card issuers, auto loans and online groceries were released last week by Fiserv, PointPredictive and Ocado Technology, respectively.
Fiserv teams up with Mastercard
Financial services technology provider Fiserv made the biggest buzz when it announced that it had begun offering Mastercard’s Decision Intelligence fraud detection service to its clients. Decision Intelligence, a fraud detection service is expected to increase the accuracy of real-time approvals of genuine transactions, reduce the number of false declines and improve the overall cardholder experience.
Decision Intelligence uses artificial intelligence modeling techniques and thousands of data points to help card issuers decide whether to approve or deny a transaction in real-time. Some of the data points included in the decision-making process include location, merchant, device data, time of day and type of purchase made.
“Minimizing the risk of financial losses while avoiding declining genuine consumer transactions requires delicate balance,” said Fiserv VP of Card Services Patrick Davie. “False declines can damage the relationship with cardholders, so evaluating multiple factors, including information about the consumer, merchant and issuer, throughout the shopping experience can enhance that cardholder experience and approve more genuine transactions without increasing risk.”
Fiserv research shows that the number of active cards after two or more false positive denials drops by an average of 20% over a six-month period following the last false positive denial. This suggests that around 20% of cardholders may stop using the card altogether after two or more false declines.
Machine learning sythentic ID solution for auto-loans
On the same day last week, fraud solution provider PointPredictive launched its new Synthetic ID Alert to help auto-lenders detect fraudulent accounts created using synthetic identities. Synthetic ID Alert produces alerts on auto-loan applications that exhibit patterns consistent with synthetic ID fraud. The machine learning-based solution was trained on over 40 million historical auto-loan applications. The solution is a complement to Point Predictive’s comprehensive application fraud scoring solution Auto Fraud Manager and can easily be installed within a lender’s technology system to be used in real-time application review.
“Auto lenders participating in our consortium meetings have identified that one of their top three issues this year is solving synthetic identity fraud, said PointPredictive’s VP of Fraud and Product Strategy Eric Werab. “Our analysis shows that synthetic identity fraud accounts for 15-20% of fraud and misrepresentation losses across the industry, which equates to more than $1 billion in synthetic identity originations this year.”
Online groceries means online fraud
Meanwhile, the technology arm of the U.K.-based online grocery Ocado rolled out a machine learning fraud detection system for online grocery orders. Historical data from the more than 260,000 grocery orders Ocado receives every week were used to train the model. Among the data points included for analysis were the cost of online carts and the number of past deliveries.
The solution is part of Ocado Technology’s Ocado Smart Platform (OSP), an end-to-end e-commerce solution for grocers. According to Ocado Technology, the machine learning model is currently being used to shadow the Ocado’s pre-existing rules-based fraud prevention system and has already registered a 15-fold improvement in the number of failed payments predicted. Ocado has already licensed OSP to major grocers in Canada, the U.K. and France.
Company spokesman Alex Voica said that the deploymeny of the machine learning solution in the OSP would ensure that both Ocado and its licensed partners would benefit from the unique insights delivered by the retail grocery operations to improve the model. “This leads to far better solutions that are tailored specifically for the online grocery market compared to off-the-shelf generic [fraud] solutions,” Voica said.
