Amidst the massive explosion of consumer, payments and transaction data, the question remains: how are we going to safeguard this stuff?
According to the Nilson Report, credit card transactions rose 28 percent, debit card transactions 46 percent and electronic transactions 45 percent between 2010 and 2015, for a collective increase of 34.2 billion transactions annually. That was before it was possible to use your Fitbit to buy a kale juice. In addition to volume, the variety of transactions is also exploding. Mobile wallets, the IoT, P2P networks, digital banking – all of these contribute to greater complexity.
The good news? Credit unions have more member data than ever before. The bad news: as we have all witnessed, keeping massive stores of fast-moving, complicated data safe isn’t easy.
Detecting fraud is a significant challenge. Why? Existing fraud detection has trouble “seeing” the transactions coming in. Data is also disjointed, lowering the level of detail and making it difficult to perceive trends.
At the same time, tolerance for “nuisance” declines is at an all-time low. Members may expect you to keep them and their data safe, but they do not wish to be inconvenienced or falsely pegged as bad transactors. According to Javelin, nearly 40 percent of people who experience a false decline stop using their credit card altogether.
Feeding the machine
“Machine learning is just one of the technologies CO-OP is investing in to combat fraud,” says CO-OP’s chief information officer Nick Calcanes. “Machine learning allows us to access and process a great deal more data than traditional methods do. Data scientists have to build models and teach the machine, but once they do, the program learns at a pace that is far beyond what we humans can match. Fraudsters are constantly evolving. Machine learning gives us the speed and flexibility to keep up.”
While data can put a strain on traditional fraud-fighting resources, it feeds the effectiveness of machine learning: “With machine learning, the massive flow of data we’re seeing is a huge opportunity,” Calcanes says. “You can take data, roll it into a machine learning environment and look for trends related specifically to our client base.” The computational power and self-learning capabilities that make machine learning different enable this type of fraud detection to turn a data avalanche into an advantage.
Venturing into machine learning isn’t without challenges. Just getting the data into one place is a hefty undertaking. “When we began, the data was completely disaggregated,” Calcanes says. “It was spread across our many platforms and within the credit unions themselves. The strategy that we’re executing on is to build a unified platform that will aggregate the data within our span of control, then apply machine learning across the system.”
Though this project is still underway, significant process has already been made. “We expect to have our machine learning platform in place by mid-2018,” says Calcanes. “We then will continue to deploy and tune our models.” Initially, machine learning will work side by side with CO-OP’s longstanding neural network technology. Over time, CO-OP may switch to machine learning entirely or keep both systems in place as the ultimate safeguard.
Speed and Scale
New security technology is not quite an unmitigated blessing in a chaotic fraud environment. Two factors – speed and scale – are critical. We’re already experiencing an avalanche of data and evolving forms of fraud that include account takeover and ransomware. Waiting years for technology to deploy simply isn’t a workable option. Scale matters to, as the broad application of new security technology is necessary for machine learning to work.
CO-OP is facilitating both speed and scale. With the acquisition of TMG, CO-OP processed more than 4 billion transactions in 2017. CO-OP credit unions will pool their data, “We won’t share real data across credit unions, but we will aggregate it for modeling,” says Calcanes. The result: effective fraud detection that benefits the industry and individual credit unions. “Whether you’re a $3 billion credit union or a $300 million credit union, our teams will work with you to make sure we have aligned the full leverage of their technology to your specific needs.”
This machine learning initiative not only marks the beginnings of a new approach to fraud detection at CO-OP: it also heralds a new data strategy that will help credit unions with business intelligence, member experience, marketing and more. “Fraud, of course, is a hot topic and, because the cost of fraud is so high, it’s a great rationale for investing in this technology,” says Calcanes. “We’re using this initiative to stand the platform up and leverage the capability across all we do.”
As we’ve mentioned, harnessing data is one of the most important issues facing credit unions right now. If they fail to learn how, data – and its accompanying security risk – can easily overwhelm. If they succeed, though, new doors can open and the possibilities can be down-right uplifting.
“That’s the beautiful thing about being part of these data-driven initiatives at CO-OP,” Calcanes says. “We can bring horsepower, innovation and scale to credit unions that would not have access to these technologies on their own.”