Underwriting from the processor: what settlement data says that a P&L does not

For most of the history of commercial credit, a lender learned about a business from documents the business prepared: a profit and loss statement, a balance sheet, a tax return, a stack of bank statements printed and handed over. Each was a summary, each was prepared after the fact, and each passed through the hands of the party with the most to gain from a favorable reading.

That is changing, and the change is structural rather than cosmetic. A growing share of small business revenue arrives through payment processors, and nearly all of it passes through bank accounts that can be read through permissioned data connections. A lender no longer has to ask what the business earned. With the owner’s consent it can look.

Three views of the same business

Consider a restaurant group applying for working capital. Its tax return shows one revenue figure, prepared to minimize taxable income. Its internally prepared profit and loss shows another, prepared to support the application. Its processor settlement data shows a third: every card transaction, by day, gross and net of fees, with refunds and chargebacks broken out. The three rarely agree, and the gaps between them are the underwriting.

Settlement data answers questions a statement cannot. How much of the reported revenue is card and how much is cash, which affects how much of it a lender can actually see and control? What is the chargeback rate, and is it rising? What is the average ticket, and is the business growing by serving more customers or by charging the same customers more? Does revenue arrive evenly through the week and the month, or in a pattern that suggests a few large events? Processor data is the business’s revenue as its customers actually paid it.

What bank data adds

Bank transaction data, retrieved through a permissioned connection to the borrower’s bank rather than as a PDF, gives the other half of the picture. Data aggregators such as Plaid document lender-facing products that return account balances, categorized transaction histories and cash-flow analytics directly from the institution, in formats built for underwriting. The immediate benefit is authenticity: a transaction feed pulled from the bank cannot be edited in a PDF tool. The larger benefit is completeness. The connection reveals every account the business holds at that institution, not only the one it chose to show, and it shows the recurring debits of every other lender already taking a share of the same revenue.

From that feed a lender reads the things that predict repayment in short-term commercial credit: deposit frequency and consistency, average daily balance, the count of overdrafts and returned items, the ratio of existing debt service to deposits, and seasonality across a full cycle. A business whose deposits arrive daily and whose balance never goes negative is a different credit from one with the same monthly total arriving in two lumps and three overdraft fees, even though a monthly statement would summarize them identically.

Reconciling the record to the receivable

For a receivables lender the data does one more job. An invoice is a claim that a customer owes money. Settlement and deposit data show whether customers in fact pay, how long they take, and how much they pay short. Reconciling the receivables aging to cash receipts, invoice by invoice where the systems allow, turns dilution from an estimate into a measurement and turns verification from a sample into a census. An invoice that never becomes a deposit is either slow, disputed or fictitious, and the data says which faster than a phone call.

Round-tripping and the limits of a deposit

Direct data also closes a specific fraud. Round-tripping is the practice of moving the owner’s own money through customer or affiliate accounts so that it lands in the operating account looking like revenue. On a bank statement it is a deposit like any other. In a transaction feed with counterparty detail, it is a transfer from an account with the same beneficial owner, or a deposit that never has a matching invoice or processor settlement. Tracing deposits to their origin, and matching them to sales records, is how underwriters distinguish revenue from recycling.

Control follows data

The same connection that lets a lender read the processor lets it participate in the settlement. Where a processor supports split funding, the lender’s share of each day’s card receipts is settled to the lender’s account before the remainder reaches the merchant, so repayment is taken from revenue rather than requested from the borrower. A lockbox does the same for businesses paid by invoice. Both are forms of cash dominion, and both are more reliable than an automated debit against an account the borrower controls, which can be blocked, moved or emptied.

What data does not tell you

Direct data is a better record, not a complete one. It says nothing about a pending lawsuit, a lease that ends next quarter, a key employee who has resigned, or a change of ownership. It cannot see cash revenue that never reaches a bank. And a feed can be gamed for a time by a determined applicant who funds a business for a few months to build a pattern. The Federal Reserve’s Small Business Credit Survey continues to document how many small firms turn to non-bank and online lenders, and speed of decision is a large part of why. The lenders that hold up over a cycle are the ones that use the data to underwrite faster without using it to underwrite less.