Let’s take a trip with the Ghost of Christmas Future. The year is 2016, and George Bailey, a former banker, now a part-time consultant, is looking for a 30-year fixed-rate mortgage for a co-op in the super-hot neighborhood of Bedford Falls (BeFa). He has never missed a loan payment and has zero credit card debt. He submits his information to the online-only PotterBank.com, but halfway through the application process, the website asks for his Facebook login. Then his Twitter. Then LinkedIn. The cartoon loan officer avatar begins to frown as the algorithm discovers Mr. Bailey’s taxi-driving buddy Ernie was once turned down by PotterBank for a loan; then it starts browsing his daughter Zuzu’s photo album, “Saturday Nite!” And what was this tweet from a few years back: “FML, about to jump off a goddamn bridge”?
A new wave of startups is working on algorithms gathering data for banks from the web of associations on the internet known as “the social graph,” in which people are “nodes” connected to each other by “edges.” Banks are already using social media to befriend their customers, and increasingly, their customers’ friends. The specifics are still shaking out, but the gist is that eventually, social media will account for at least the tippy-top of the mountain of data banks keep on their customers.
“There is this concept of ‘birds of a feather flock together,’” said Ken Lin, CEO of the San Francisco-based credit scoring startup Credit Karma. “If you are a profitable customer for a bank, it suggests that a lot of your friends are going to be the same credit profile. So they’ll look through the social network and see if they can identify your friends online and then maybe they send more marketing to them. That definitely exists today.”
And in the last year or so, financial institutions have started exploring ways to use data from Facebook, Twitter and other networks to round out an individual borrower’s risk profile—although most entrepreneurs working on the problem say the technology is three to five years away from mainstream adoption.
“Credit score is a lagging indicator,” said Brett King, a tall, puffy Australian with white blond hair who is the founder of the online-only bank Movenbank and author of BANK 2.0: How Customer Behavior and Technology Will Change the Future of Financial Services. “At best, your credit score is about 60 days behind. What we’re trying to do is look for things that reflect the likelihood of a future default, rather than what’s happened in the past.”
Movenbank is an online bank in private alpha release that replaces plastic credit and debit cards with a mobile device such as an iPad or smartphone. Mr. King is a major proponent of the questionable young science of using social media to evaluate creditworthiness.
When it comes to online privacy, Mr. King subscribes to the Mark Zuckerberg school of thought: standards are evolving, and the world will be better for it. (As long as you’re connecting and sharing, only good things can happen to you!) “Our view of what ‘private’ is, is changing,” Mr. King said. “We make friends with people we barely know!”
He predicts that banks will soon start asking customers to verify their social media profiles. Not everyone has a social media presence, of course, so submitting your Twitter handle will first be pitched as a way to provide customer support or account alerts, which will later open the door for “more complex products,” Mr. King said.
Employers have already started using social media to evaluate potential candidates, and in 2009 a woman in Quebec stopped receiving disability payments for major depression after Manulife decided, based on beach vacation photos on Facebook, that she seemed happy enough to work after all. “I’m sure that insurers now are looking at Facebook profiles and saying, ‘You’ve said you’re not a smoker? Well how come in three of these ten photos where you’re out with friends, you’re smoking?’” Mr. King said.
That means that tweet, “Just got fired, man. Spending my severance at the bar!” may have been ill-considered.
Mr. King is especially interested in identifying customers who can evangelize the service to a sizable crowd of cloud-friends. Movenbank requires users to connect their Facebook accounts upon registering, data from which will be baked into a proprietary “CRED” score, a number that determines which rates and products are available. The exact recipe is still being written, but eventually Movenbank will boost your CRED as you hook it up to your accounts on Twitter, LinkedIn and even eBay, which calculates a reputation score based on buyer feedback. It’s not the only metric, Mr. King said, but a strong Twitter presence could tip the scale in favor of a marginally risky borrower.
Much of this is driven by enterprising techies looking for the next big sector of the economy to disrupt with a social twist. Back in July, the 34-year-old internet pundit, angel investor and startup entrepreneur Kevin Rose, best known as the founder of Digg, sat down in front of his webcam in a t-shirt and baseball cap to talk to the internet about credit cards. “This might be potentially the dumbest, least-vetted idea I’ve ever put out there,” he said. “What if we could make credit cards a little more social?”
Mr. Rose was just spitballing, and his idea seemed innocuous enough. But there’s a nightmare scenario: if banks learn how to use social media, they could gather information they aren’t allowed to ask for on a credit application—including race, marital status and receipt of public assistance—or worse, to redline segments of the social graph.
In other words: choose your online friends wisely, for they may one day determine your APR.
Lenddo, a Hong Kong-based microlending startup incubated in New York’s FinTech Innovation Lab, calls itself “the first credit scoring service that uses your online social network to assess credit.” The first thing Lenddo asks for is a Facebook account; then it wants access to Gmail, Twitter, Yahoo, and Windows Live. The Observer was given a respectable score of 470. But when we tried to apply for a loan, we were told “you need at least 3 connections with scores above 400 in your Lenddo trusted network.” (We wouldn’t have been able to get a loan anyway: Lenddo is only available in the Philippines, although it recently hired an ex-Googler to head up the Americas.)
The company’s algorithm is proprietary and secret, said CEO Jeff Stewart, but the primary metric is what Lenddo knows about the people you’re friends with. “We think that in the age of the internet you should be able to establish your reputation and your identity through your social graph, through your on- and offline community, and use that to get access to financial products and information,” he said.
If Lenddo sees one of your best Facebook buddies took out a loan and paid it back, there’s a good chance you will too. “Our backgrounds are in machine learning and pattern recognition,” Mr. Stewart said. “It’s some serious math.
“There’s no reason there shouldn’t be thousands of engineers working to assess creditworthiness.”
In another nifty but nefarious innovation, Lenddo reserves the right to broadcast your loan status if you fall into default. As the site warns: “Failure to repay will negatively impact your Lenddo score, as well as the score of your Lenddo friends. Lenddo MAINTAINS THE RIGHT TO NOTIFY YOUR FRIENDS, FAMILY AND COMMUNITY if the borrower fails to repay, however, this is only done after several notifications to the borrower and an attempt to work out a payment plan.”
“I think Mark Zuckerberg said it best,” Mr. Stewart said. “Every industry will be in fact impacted by social.”
Banks have been curious about using social media to gauge risk for at least a year, said Matt Thomson, VP of platform at Klout, which calculates “influence” based on a user’s social media activity. Determining creditworthiness is not a core product of Klout’s, he said, but banks have approached the startup to ask about it. He wouldn’t name names. “It’s really like the who’s who of banking,” he said.
(Mr. Stewart of Lenddo also said his startup is approached “regularly” by major banks curious about the algorithm.)
Klout, arguably the leader in developing a metric for social media power users, has taken a beating from bloggers for being spammy and potentially insecure. The New York Times wrote about shocked parents who discovered Klout had autogenerated skeleton profiles for their children, based on what it had gathered from their connections to others; the science fiction writer Charles Stross called the service “the internet equivalent of herpes.” R. Ethan Smith, who blogs as The Startupist, recently wrote a critique of Movenbank’s projected partnership with Klout. “Klout claims that I am influential about New Jersey, coffee, and iPads,” he wrote, noting that he has no real expertise in any of the three and doesn’t even own an iPad. “Now, let’s assume that King is completely serious about using online social profile data to determine a Movenbank user’s influences, which will essentially determine their ability access a line of credit… To stake tangible dollars on what seems to be a relatively easily manipulable algorithm is not something I would characterize as ‘good business sense.’”
Media theorist Douglas Rushkoff dismissed the idea that social media credit scoring is a serious erosion of privacy, mostly because there’s nothing left to hide. “We’re already in the nightmare scenario,” he wrote in an email. “They already know everything about you—more than most of us realize. If anything, the addition of social networking information to this data mining will help us come to some understanding of how much more these companies know about us than we know about ourselves.”
The precise formula for FICO, the most widely used credit score, is secret and proprietary to the Fair Isaac Corporation, a publicly traded company. Experian and TransUnion, two of the three national credit bureaus, did not respond to requests for comment on this story; Equifax, the third, did respond. “Our corporate development professionals are very aware of the opportunities to enhance our proprietary data and partner with companies who add value to the accuracy of our reporting, which helps our customers make better decisions prior to lending,” a company rep said in an email, adding that Equifax can’t comment on future strategies because it’s a public company.
This new use for social media data could turn out to be empowering, Mr. Rushkoff pointed out, if it leads to people lending to one another. A reputation score based on the social graph could lower the barrier to entry for peer-to-peer lending startups. “Instead of everyone outsourcing their savings, investments, and borrowing to truly evil institutions who use what information they about us simply as an excuse to drain more money from us,” Mr. Rushkoff said, “we would invest in one another.”
Snow is falling lightly outside as Mr. Bailey logs off of PotterBank.com. Suddenly, a bell rings. It’s his iPhone: a text message from Lending Club, a peer-to-peer lending startup based in San Francisco. His friends saw his Tumblr post with photos of the coveted apartment, and forwarded it to friends of friends. Collectively, they’ve pledged to invest over and above the needed deposit. He looks up, smiles; looks back at his phone, and taps out a tweet: “No man is a failure who has friends!”
CORRECTION: The original post incorrectly said Lenddo is based in New York. The company is officially headquartered in Hong Kong.