Dan O’Malley became a believer in the power and speed of automated loan approvals a decade ago, while working as a senior director for Capital One Financial Corp., the big credit-card issuer. “We had a model for every single person in the United States, so that if you came onto the website, we could instant-decision you for a credit card,” he recalls.
As chief digital officer at $9.9 billion asset Eastern Bank since 2014, O’Malley has brought the same taste for technology-driven speed to small-business lending. O’Malley, also head of Eastern Labs, the Boston-based bank’s innovation unit, says he can approve and close a loan of up to $100,000—from application through funding—in as little as three minutes.
How? Every night, Eastern’s software assesses the loan eligibility of virtually every business in Massachusetts, using information from public databases, court records and media reports, among other sources, to give a preliminary thumbs-up or down. It then makes conditional offers, via email and digital channels.
When an existing customer enters Eastern’s website, the business owner is prompted to answer three questions—how much they want to borrow, where the money should be deposited and a Social Security number. The rest of the application—names, addresses, business revenues—prefills with data the bank already has. The process for noncustomers takes about 10 minutes, a bit more time.
After a quick check of the prospective borrower’s FICO score, cash flows and other loan commitments, the customer digitally signs the loan documents and the money is deposited, almost instantly, into the borrower’s account. “It takes as much time as it takes to draw down a line of credit,” O’Malley says. “People are shocked by how fast it is.”
Until recently, banks shied away from rushing through the lending process and giving software such a big role in small-business lending decisions. Personal interaction and analysis has always been considered fundamental to good credit underwriting. But times are changing.
Marketplace lenders such as Kabbage, Lending Club and Prosper have stolen some market share and changed borrower expectations with intuitive online applications and the use of data algorithms that can assess creditworthiness and underwrite loans in minutes instead of weeks. If banks can’t provide the same speed and experience, the thinking goes, they risk losing a significant chunk of business that otherwise might come their way.
“The marketplace is demanding quicker decisions through technology,” says Pierre Naude, CEO of nCino, a maker of bank operating systems. Bank customers, he says, are clamoring for special products and specialized coding that enable greater automation of the small-business lending process. “Bankers are waking up to the fact that speed and convenience will trump price. You can lose a customer to an alternative lender if you don’t have it.”
Automating the small-business lending process requires some deep thinking from boards and management about how much faith they’re willing to place in technology, and their ability to embrace the cultural change implicit in basing lending decisions more on data than judgment.
The concept remains new enough that credit quality of loans underwritten by software programs has yet to be tested by a recession. (Lending Club made news in October with warnings about rising charge-offs.) Regulators like the audit trails that automated processes generate, but don’t like cookie-cutter loans devoid of human input, bankers say.
“The big question we get [from the board] is, ‘Can you be fast and accurate at the same time?’” says Hugh Connelly, president of small-business lending at $2.9 billion asset Univest Bank & Trust Co. in Souderton, Pennsylvania, which has special credit-decision programming built into its nCino operating system. “We don’t want to find out a year from now that in our quest for speed we made several hundred imperfect loans.”
The benefits can be significant. David O’Connell, a senior analyst with Aite Group, a research and advisory firm, says banks that use automation right can grow their small-business loan books faster and cut costs. Significantly, it also can provide a better handle on credit metrics, while leaving a more-complete audit trail for the compliance staff and regulators to review.
“Automation is a way for banks to achieve the scale that’s been elusive to them,” largely because the traditional lending process is too slow and cumbersome, O’Connell says. Streamlining the process by automating at least the drafting of documents “can takes days or weeks off the loan cycle,” and make loan officers more efficient.
“It’s beneficial to everyone in the same way CRM was beneficial,” O’Connell adds. “People spend less time on the clutter of underwriting—manually comparing a company to peers or looking at guarantor financial statements—and more time assessing risks and structuring deals.”
Outside of time, the best way to gain comfort with automation is to understand how it is used. Many banks employ automation primarily to simplify the collection of data and provide an initial screen on applications.
At Wilmington, North Carolina-based Live Oak Bank, a $1.4 billion asset, Small Business Administration lender, prospective borrowers enter their basic information in a web portal, where the bank’s proprietary lending screen, dubbed “the gauntlet,” pulls data from 19 different services to assess credit, fraud, identity and other borrower characteristics. It also looks at factors like location or social-media usage before generating an initial score.
“We know within seconds that there’s a strong likelihood we’re going to approve the loan,” says Steve Smits, Live Oak’s chief credit officer. If it clears, the application gets passed on to a loan officer who analyzes cash-flows, examines the project and interviews the potential borrower via phone or Skype. (Live Oak, which spun-off nCino and continues to use its technology, is a branchless bank with a national client roster in niches such as small-town pharmacies and veterinary clinics.)
The process, which accommodates loans of up to $350,000, takes longer than some—about 10 days to close—but still far less than the 60 days to 90 days it used to take. Machines are not making the decisions, Smits says. “But if it’s already passed the gauntlet,” he says, “then we’re in the mindset that it’s a good opportunity. We just want to make sure they can afford the debt.”
Univest takes things a step or two further by automating credit decisions and the order fulfillment process. Customers seeking a secured loan of up to $200,000 can come into a branch for a quick interview with a loan officer, but various sections of applications prefill with data, and a decision is made in as little as seven hours.
The bank’s proprietary software pulls credit reports, information from data providers Dun & Bradstreet and PayNet, and it collects and downloads personal guarantor information, tax returns, financial statements and other important data. It then performs an analysis of collateral and debt-payment projections, and generates a credit memo that is electronically signed. Usually, such analysis doesn’t require independent appraisals of property value, and the bank uses a first blanket lien on all business collateral.
Credit decisions are rooted in the same underwriting standards that helped Univest survive the financial crisis, and “we still have people with intuition and judgment involved,” says Connelly.
“In banking, speed has traditionally meant errors, omissions and other mistakes,” he adds. “But today’s technology is so robust, speed goes hand-in-hand with accuracy and precision. They’re not mutually exclusive concepts.”