Battling fraud is not a battle that should be fought single-handedly. CA Technologies is debuting a consortium-driven, real-time behavioral analytics model that seeks to leverage real-time learning and instant risk scoring into real-time victories against fraudsters, scammers and bad guys.
Fighting online payments fraud is a tough battle if done singularly. Teamwork might make the Herculean task a bit less, well, onerous.
To that end, CA Technologies is debuting the CA Risk Analytics Network, which the company is billing as a real-time behavioral analytics fraud network that ties in with card issuers who opt into that broader network. The premise, in part, is that shared data and knowledge is valuable to the community at large when combatting fraud. The upshot is measurable, with a 25 percent reduction in fraud losses or a 35 percent reduction in false positives.
In an interview with James Rendell, vice president of Payment Security Strategy at CA Technologies, PYMNTS’ Karen Webster delved into the effectiveness of behavioral analytics and machine learning deployed on a network scale for financial services.
CA Risk Analytics Network incorporates a new advanced neural network model, backed by unique real-time machine learning algorithms and networked device and entity reputation data, to protect 3-D Secure card-not-present transactions. It learns from, and adapts to, suspected fraudulent transactions in an average of five milliseconds, instantly closing the gap for potential fraud, using the same card or device, across all members of the network.
Clearly, effectively using vast amounts of data in combination with risk-based models becomes increasingly important in payments and commerce. But “in terms of how we get there … First, CA Risk Analytics Network is a product that we’ve launched that is specific to 3-D Secure card-not-present transactions,” said Rendell.
Classical authorization messaging data flow depends on data fields that are not specific to internet transactions, said Rendell, whereas the 3-D Secure protocol provides a wealth of data points unique to an internet shopping transaction, ranging from IP addresses to browser information, he said.
Rendell maintained that additional value lies in the sheer volume of data that has been processed by CA’s Payment Security Suite over the years. “This is the real edge in terms of the quality of the predictive model we can build, and there are just years and years of data available” that can help to build this consortium model, he said.
“Due to vast amounts of data we process, we believe that we process for the larger part of the market overall … with real-time updates to the model itself,” he added. The CA Risk Analytics Network model, he said, learns in milliseconds, and “it learns from and adapts to fraudulent behavior on cards and devices in real time. If deemed fraudulent, the model stops the very next transaction that comes in. Ultimately, the model will have learned from and adapted to the previous transaction in real time, and that is a really big deal.”
Other companies and systems, he said, while they may store transaction results in real time for reporting and similar purposes, do not update their neural network predictive models in real time.
“To complement the cardholder behavior profiling that we’ve always done,” CA Risk Analytics Network leverages CA’s global consortium of device data, said the executive, which has been built up and is continuing to be built up over the years. He said that the core product is also poised to take full advantage of the forthcoming EMV 3-D Secure 2.0 protocol, which gathers a lot of metadata tied to mobile devices, augmenting traditional browser fingerprinting.
CA Technologies is heavily involved as a “Technical Associate” of EMVCo, the organization which is developing the new 3-D Secure 2.0 protocol. The reason why the new specification is so important in the predictive risk modeling space is that it defines even richer transaction metadata, including a wealth of device data, perfectly complementing CA Risk Analytics Network’s fraud prevention capabilities, which leverage consortium device data.
CA Risk Analytics Network will benefit customers using the current 3-D Secure 1.0.2 protocol, seamlessly integrating and taking advantage of the new EMV 3-D Secure 2.0 protocol as merchants and issuers adopt it worldwide.
The benefit accrues to customers using both the data and the model without having to share anything confidential or putting data at risk.
To learn more about CA Risk Analytics Network, join us for our May 25 webcast.