NYC Startup Uses Big Data To Disrupt Small Business Lending

For the vast majority of small businesses — the ones away from New York and Silicon Valley and in industries like dentistry and auto repair — getting financing is a massive pain.

It involves reams of documents, weeks of waiting, and frequent rejection.

New York startup On Deck Capital hopes to appeal to these millions of businesses with a loan application process that takes minutes to complete and can be approved within a day.

On Deck's secret is software that evaluates loans within minutes based on analysis of data including cash flow, past credit use, and vendor payment history.

While banks look closely at a business owner's personal history, the startup makes decisions based on data about the business itself.

At the same time, the startup is efficient enough to deal with small clients.

"Banks typically don't love lending to smaller businesses," On Deck CEO Noah Breslow told Business Insider. "It's much more efficient for a bank to make a million dollar loan than a $50,000 loan."

This has been the status quo for years, and banks have been slow to innovate in the space due to tight regulatory scrutiny, their sheer size, and issues with their legacy business.

Other startups in the space include Capital Access Network and Kabbage, and even Amazon is getting into the game.

For On Deck to succeed, it will have to prove that its rapid loan evaluations are effective and gain a market foothold before competition moves in.

All about data

Using online data and relatively little input from the business owner, On Deck's algorithms asses risk, build credit profiles, and make lending decisions quickly.

There are about 15 fields in the application for a $30,000 loan.

"When they apply, to get a smaller loan, it's about the same as applying for a credit card," Breslow says.

For loans up to $150,000, On Deck collects one-to-three months of cash flow data via transaction-based data tracker or uploaded bank statements.

Compared to regular banks, that's nothing.

"There are 14 different pieces of information you have to bring in," Breslow says. "You have to have two or three years of personal tax returns and two or three years of business tax returns. Usually you have to build a business plan around the loan and create pro-forma financial projections going forward and have that as a part of your package as well."

Breslow recently visited a "decent-sized" bank in the Midwest that does a 16-page individual writeup for a $100,000 loan. That's a week of a person's life running financials and doing sensitivity analysis, he says. "It's just not cost effective."

"We're never going to see these people."

"A lot of the banks' process was about me looking you in the eye, and then a bunch of self-reported data," Breslow says.

On Deck works because it uses third-party verified data. It's about sophisticated data analysis, not a gut feeling nor what small business owners hope their business will be in the future.

Everything from cash flow to past credit use to vendor payment history to legal history goes into the model. It even takes into account a presence on sites like Yelp.

"It's not that we make a core underwriting decision based on it," Breslow said, "but it helps validate a business we've never seen before. If you see 80 Yelp reviews, you know this guy exists, he's got a flow of customers, and the food's not half bad."

The big data method lets them approve people who have been conditioned to rejection from big banks yet who are perfectly capable of growing their business and repaying a loan.

The firm has loaned half a billion dollars since starting in 2007, and — after raising a lot of money — is set to loan half a billion dollars in this year alone.

Changing the risk management game

A classic bank loan might be three to five years. On Deck's are three to 18 months, and rather than monthly payments, they're collected in daily "micropayments" from the business.

Rather than loaning to, say, a contractor who builds one home a quarter and won't be able to pay if a deal falls through, the company focuses on businesses with steady, small cash flow, like restaurants, dentists, and repair shops.

While technology, data, and the credit model reduce risk before a loan is granted, daily payment helps reduce risk afterward.

The long-term vision is to make things even more fluid.

"Eventually we're going to be at a place where the data from a business continually unlocks credit for them. They don't have to apply; applying is superfluous because their data has already enabled the credit to be there," Breslow says. "In time, the business owner is going to just log into their bank account and be like "oh, I need $50,000 for this,' and bang it's there. We're continually iterating to get closer to that."

After raising money from Thiel and Google Ventures, the company will put some muscle into marketing and increasing name recognition.

There certainly is a demand for better small business lending in America.