As more transactions become digitized every month, either through the traditional swiping of credit cards of through more advanced payment methods like ApplePay, the risks of online fraud and damaged credit ratings become more pronounced. Currently, online fraud costs the economy over $190 billion a year, and although there are many safeguards to rectify fraud after the fact, not much is being done to predict and prevent fraud before it happens, even with people’s private lives increasingly becoming less private thanks to social media. As a result, online fraud prevention is an unusually fertile marketplace for innovation and creativity in fraud detection.
In steps British startup HelloSoda, a fraud detection company founded by Paul Shepard and James Blake and staffed by a team of data analysts and advanced computer coders poached from top universities and from businesses like Skype. The company’s specialty is to use Bayesian Belief Network principles to detect fraud risk, a process which relies on analyzing structured and unstructured data from social media, Internet blogs, and various interactions in cyberspace. These “psycholinguistic” techniques analyze aspects of people’s lives that a lot of lending agencies don’t necessarily look at, like life-changing events, frequency in honest reporting of personal statistics on the Internet, and various spending metrics on online commodities.
These analyses are sold to clients not just in lending and insurance market, but also gaming companies that use the data for specific marketing practices as well as monitoring gaming habits and spotting bot and fake accounts. These and other conventional metrics, such as residence and job description, seek to help lenders and insurers assess the true costs of lending and insurance, as well as spotting suspicious behavior even before the individual fills out his or her first loan document. While this kind of predictive modeling is still in its early stages, it’s a first foray in what may soon be a vibrant, competitive market for online fraud prevention.