Screening out bad guys is a constant challenge for PayPal, says Mike Vergara, vice president of consumer risk at the online payment service.
And screening efforts don't always go right. "Sometimes we say no to good people and yes to bad people, but the goal is to be as accurate as you can," he says.
Knowing that effective fraud detection is a necessity for a company that handles people's money, Vergara and other PayPal executives continually work to improve the technologies the company uses for that task.
As part of the effort, PayPal created an advanced end-to-end enterprise fraud management (EFM) system to replace an existing one. Hui Wang, senior director of global risk sciences, says the team had to build something that would work in a global market where patterns vary from country to country. They also set a very high bar, because mistakes can be costly.
Wang says PayPal's huge cache of data, combined with information from other sources, forms the foundation of the new system. Machine learning and a combination of other technologies strengthen the system.
Vergara says PayPal gained key technology for the new system in 2008, when it bought Fraud Sciences, an Israeli company specializing in detecting online fraud. PayPal fully deployed the new EFM system in 2013.