The Most Expensive Outcome in Your Book Is Not a Decline. It Is a Partial Approval.

Every funder tracks approval rate. It is the wrong instrument at the small end of the market, and the reason is arithmetic rather than credit.

Among applicants for a loan, line of credit or merchant cash advance with 1 to 4 employees, 47 percent were approved in full, 30 percent were approved in part, and 23 percent were denied. Among applicants with 50 to 499 employees: 72 percent, 20 percent, 8 percent.

The reading everyone reaches for is the first and last columns. Small firms are riskier, so more of them get declined. That is true, and the denial gradient is steep: a firm with fewer than five employees is nearly three times as likely to be turned down as one with fifty or more.

But the denial column is not where the money goes.

Look at the middle

Partial approval runs at 30 percent for the smallest applicants and 33 percent for firms with 10 to 19 employees, against 20 percent at the top of the range. Roughly a third of applications from the small end resolve as “some of what you asked for.”

Consider what that outcome costs a funder. You ran the pulls. You analyzed the bank statements. You verified the deposits, checked for existing positions, assembled the file, produced the offer, delivered the disclosure and funded the transaction. Every fixed cost in your origination stack was incurred in full, because a partial approval is a complete underwrite. What differs is only the number at the end.

A decline is cheap by comparison. It stops consuming cost at the decision point, and a good decline stops early. A full approval earns the whole ticket against the whole cost. A partial approval is the only outcome that pays the full price of a yes and collects a fraction of the revenue.

Cost per booked dollar, not cost per file, is the number that governs whether the small end of the market is worth serving. On that measure a 30 percent partial rate is a larger problem than a 23 percent denial rate, and the industry has no metric that surfaces it, because approval rate counts a partial as a win.

Why this lands on the non-bank channel specifically

Partial approval would be a tidy problem if it were spread evenly. It is not.

Among applicants with 1 to 4 employees, 32 percent applied to an online lender. Among those with 5 to 9 employees, 35 percent. At 50 to 499 employees, 8 percent. The non-bank channel is concentrated in precisely the size bands where partial approval is most common and where the fixed cost of underwriting is spread across the smallest tickets.

The concentration in small tickets is not a market the channel drifted into. It is the market it exists to serve. Which means the partial-approval economics are not an edge case in an alternative lender’s book. They are the book.

The conclusion the data does not state, and the one it does

The survey reports the gradient. It does not say why it exists, and honesty requires separating the two.

What it does say is that outcomes improve monotonically with borrower size across every column. What I am arguing is the mechanism: the fixed cost of underwriting does not scale down with ticket size, so as files get smaller, the cost of reaching a defensible yes converges on the value of that yes. Somewhere on that curve, the rational answer stops being a full approval and becomes a partial one, sized to the amount the file can carry rather than to the amount the borrower needs. That decision is booked as a credit judgment. Part of it is an operations constraint wearing a credit label.

The test of that claim is straightforward and every funder can run it internally. Take your partial approvals from the last twelve months. Compare the amount requested against the amount funded, and set the difference against your fully loaded cost per file. If the ratio degrades as ticket size falls, your partial-approval rate is being set by your cost structure and not only by your credit box.

How CXO Solves This

The lever is the fixed cost, because it is the only term in that arithmetic a funder controls directly.

Client Onboarding Automation carries intake, document collection, eligibility screening and status communication as agentic steps rather than as analyst hours. That moves the fixed cost per file down, and moving it down changes the ticket size at which a full approval is the rational answer.

Custom Agentic Workflow handles what a standard tool will not: multi-system verification, existing-position checks, and the file assembly that sits between a submission and a decision, which is where most of the unrecoverable cost actually accumulates.

Neither widens the credit box. They change the economics that quietly narrow it, which is a different and considerably cheaper thing to fix.

In most operations, far more work can be automated than leadership realizes. One discovery call is enough to size what automating it would return to your bottom line. Book it at https://cxocorporation.com/contact.