

Even by lofty fintech standards, OakNorth is soaring. The London-based startup bank’s $440 million funding round, led by SoftBank’s Vision Fund, was the biggest ever for a European firm of its kind. Less than four years since it launched, the company is profitable and valued at $2.8 billion. OakNorth says not one of its customers has ever defaulted, or even had a late payment.
OakNorth’s founders, who sold their analytics startup to Moody’s in 2014, say the company is designed for entrepreneurs. Its customers are small- and medium-sized businesses such as Leon, a restaurant chain, as well as bars, hotels, property developers, and private equity firms. The company plans to use the money it recently raised for expansion in the US and abroad.
But what about the perfect lending record? It sounds almost too good to be true. The bank has lent out more than £2 billion ($2.6 billion), which it funds mainly through equity and deposits from more than 30,000 savings customers.
Sean Hunter, CIO of OakNorth’s analytics unit, said about one in 10 companies makes it through an initial screening before getting signaled to apply for a loan. The prospective borrower then goes to a credit committee, where about 90% of applicants get approved. Roughly speaking, the approval rate is about in line with broader small-business lending in the UK, according to BBA trade association data.
Hunter says the bank uses artificial intelligence to automate as much of its credit analysis as possible. Machines are adept at sorting through mountains of data to identify patterns. And so one example is finding statistics for comparable borrowers. For a hotel company, it might identify property values for similar businesses. “There are lots of little sub-jobs you can get AI to do,” said Hunter, who previously worked at Silicon Valley data science company Palantir $PLTR.
The company also uses outside data like TripAdvisor ratings and sentiment analysis based on Yelp reviews. Its OakNorth Analytical Intelligence business licenses its tech platform’s services to other banks.
But not every borrower behaves as expected. In a “handful” of cases, Hunter said OakNorth has modified a customer’s loans. Another strategy is to simply jettison the weakest credits. Co-founder Joel Perlman told a London conference recently that the company has on occasion urged a customer to refinance elsewhere. “We’ve said, ‘go renegotiate with another bank and refinance,'” Perlman said. “And they’ve gone and refinanced and then a few months later they’ve gone into default.”
OakNorth says its loans are specifically designed for each customer, similar to the treatment a large company would expect from a top Wall Street lender. Humans make the lending decisions, but they’re able to improve the analysis and speed up the process using AI. Hunter said the technology helps it bring the cost of customized service down to a level where it can be used for relatively small loans.
The company also continues monitoring the customer, either by plugging directly into accounts to access operating data, or by having customers provide documentation that gets uploaded. The idea is to detect problems before the customer runs out of options. “Getting all that data in different formats from different businesses is complicated,” Hunter said. “That’s a technology problem that we’ve taken on.”
Big banks with vast portfolios aren’t always able to inspect each every loan on a regular basis, said Francesc Rodriguez Tous, a finance professor at Cass Business School in London. Instead, they might focus on their most important accounts unless there are obvious red flags coming from the smaller clients.
But if every company that seeks to refinance an OakNorth loan is already in trouble, other banks will eventually notice, he said. They’ll be less likely offer refinancing, making it harder for OakNorth to unload problem debts in the future (and besmirching its perfect lending record). Without specifically pointing to OakNorth, he noted that loan modifications are fine unless they’re simply reclassifying a borrower who still can’t afford to service a debt. And a key question, of course, for every startup lender is how well they will withstand a downturn, following a long stretch of relatively benign economic conditions.
The company points out that some of its clients have already been through a slump. For Perlman, profitability—it reported a £10.6 million profit in 2017—is a key to withstanding a potential slowdown. “We don’t believe in that story about building revenues and eventually we’ll make money,” Perlman said at the conference. “When the downturn comes, being profitable is a big thing.”


podcast that to write a finance program for Mother’s Day: “So, the very first program I wrote was to basically take all of the kind of utility bills in her house and I produced a program that would look, you know, basically automate them for my mom. And when I look back at that, she was a bit kind of bewildered.”

Feb. 1: Apple should issue a debit card
