Every other bank in the region runs loan applications through an automated scoring model before a human ever sees the file. This one doesn't — every application, big or small, gets read start to finish by one of four loan officers before any decision gets made.
The bank's president doesn't dispute that the manual process is slower. An application that a scoring model could clear in minutes takes her staff most of a day. What she disputes is the assumption that slower is worse.
What a model misses that a person catches
Loan officers here can point to specific files a scoring model would likely have rejected — a seasonal business with uneven monthly revenue, a first-time borrower with thin credit history but a steady work record — that turned into some of the bank's most reliable accounts.
None of this means the manual process catches everything a model would. It's slower, it doesn't scale past a certain branch size, and the bank has no plans to grow beyond the handful of counties it already serves. Its president says that's the trade she'd make every time.