For the people who run outsourced underwriting for credit unions

Your loans arrive on Saturday. Your people were hired for Wednesday.

You sell your clients’ rulebooks, applied exactly, around the clock, at a variable cost. You deliver it with people, by the hour. This page is about that gap.

Five questions, two minutes, your score on the spot. No form to see the result.

A typical week on an outsourced desk Share of the week, illustrative
Applications received versus underwriter hours staffed, by day of week Applications peak on Saturday at about a third of the week. Staffed hours are about the same every day, slightly higher on Saturday, so Saturday demand runs well past Saturday capacity.
View as a table
DayApplicationsStaffed hours
Monday10%14%
Tuesday9%14%
Wednesday9%14%
Thursday9%14%
Friday13%14%
Saturday35%16%
Sunday15%14%

The problem

The promise you sell is the problem you carry

Every outsourced underwriting desk makes the same four promises to its credit unions. They are good promises. They are also, one for one, the hardest things about running the desk.

You sell

“Decisions around the clock. Nights, weekends, holidays.”

You carry

A roster that is about the same size every day, and a demand curve built by car dealers. Saturday is the biggest day of the week for indirect applications, and it can run several times a midweek day. Your people already work Saturdays and take a weekday off, and it is still not enough. You either staff every day to the Saturday peak and pay for idle Tuesdays, or you staff to a typical day and every client sees it on Saturday.

You sell

“Decisions made to your written lending policy.”

You carry

Every client’s written policy. Dozens or hundreds of hotsheets, rate sheets, and addenda, each different in the details that matter, each living in a shared drive and in the heads of whichever underwriters happen to know it best.

You sell

“Turn your fixed cost into a variable cost.”

You carry

The fixed cost. Your clients’ volume moves up and down. Your payroll does not. When a spike lands, you call the temp agencies, and the least-trained people on the floor take the most rule-heavy work on the busiest day.

You sell

“We work in your systems, under your name.”

You carry

Systems you cannot configure. Whether applications land in your own platform or in each client’s, the policy is theirs, the decision engine is theirs, and the right to change either is theirs. You get the work the engine could not clear.

None of this is a failure of management. It is the shape of the business.

Why it is still manual

Every tool built for this problem was built for someone else

The credit union owns the policy and the loan origination system. You own the labor. So you absorb the full cost of variety, and you have no right to simplify its source.

  • Decision engines inside the LOS are sold to credit unions. Many of your clients outsourced precisely so they would not have to configure and maintain one.
  • Robotic process automation moves keystrokes between screens. It does not read a hotsheet or weigh a compensating factor.
  • Rules engines execute rules. Someone still has to turn a hundred prose policies into rules, keep them current as the emails arrive, and put the result on the underwriter’s desk. Those are the hard parts.
  • Automated underwriting products assume one policy. Yours. Your problem is many policies, none of them yours.

Nobody built for the desk that applies many clients’ policies, with its own people, on systems it does not control. So the knowledge stayed where it always was. In people.

On the floor

What it looks like on a Saturday

Walk the queue with one of your underwriters on the busiest day of the week.

  1. The unfamiliar client

    An application lands from a credit union she works twice a month. She knows the policy exists. She does not know it. Turn time across comparable clients can vary about twofold, and complexity does not explain the spread. Familiarity does. Some of the most complex guideline sets are the fastest served, because the people who know them cold are on shift.

  2. The page-flip

    The hotsheet is a PDF. There is no search that understands it. Checking one rule means reading a few pages to find it, then a few more to confirm nothing later overrides it. Multiply by the number of rules that matter on this deal.

  3. The first decline

    More than half of an outsourced desk’s decisions are declines, and they take more than half of the time. They are the most rule-driven, most documentation-heavy work on the floor. Under pressure, an underwriter finds the first failing rule, writes it up, and moves on. The adverse-action reasons come out thinner than the policy, and the approvable deal behind it in the queue waits.

  4. The counter

    The deal misses on loan-to-value by a little. The right answer is a counter: the cash down or the reduced amount financed that brings it inside the cap, at the tier’s rate. That is calculator math done by hand, against a rate sheet that may have changed last month.

  5. The resubmission

    A meaningful share of applications are resubmissions of a deal already worked. The dealer changed the structure, or added a co-borrower. The underwriter starts from the top, because there is no easy way to see what changed.

  6. The email

    A client tweaks a rule. The change arrives as an email and a new PDF. Someone updates the shared drive. Nobody can say afterward which version applied to a decision made last Tuesday.

On a desk like this, a meaningful share of decisions are recorded after the application has already left the queue. The dealer has moved on. The client sees the turn time. You see it later.

What it costs

What it costs you

In the terms your P&L uses.

  • Peak labor at agency rates. Every spike is bought from a temp agency at the worst price, delivered by people who do not know the policies.
  • Idle capacity midweek. The roster that covers Saturday sits partly idle on Tuesday. You pay for both.
  • Turn time your clients see before you do. In indirect lending the fastest sound decision wins the contract. When volume grows and turn time stretches, your clients count the deals they lost before you count the hours you spent.
  • Veterans who are the policy. A minority of your underwriters carry most of the volume, because they carry most of the knowledge. Newer underwriters run about a third slower. When a veteran leaves, a credit union’s rulebook leaves with them.
  • A long tail you cannot afford to serve. A small credit union costs the same to learn as a large one and brings a fraction of the volume. The clients who need you most are the ones you can least afford to serve well.
  • Versions you cannot prove. When an examiner or a client asks which policy applied to a decision, the honest answer is a reconstruction.

Growth on this desk is linear in people. Each new client, each new program, each new percent of volume costs a matching unit of payroll. Margin cannot expand. So the mandate to automate a meaningful share of inbound sits on your desk with no headcount behind it.

Funding

It does not end at the decision

The same rulebooks get executed a second time at funding and document stacking, by a different team, from the same PDFs. Same problem, later stage. Underwriting is the front door.

The other side

What the other side looks like

Not a product. A day.

  • Capacity is present at the Saturday peak without a Saturday roster. Minutes saved on every decision are the one capacity lever that follows the demand curve on its own.
  • The same answer from any underwriter on any client. Familiarity stops being the variable.
  • A new client is live in days. A hotsheet and a review cycle, not a training program.
  • Every decision is traceable to a rule and a version. Adverse-action reasons come from the policy, not from the first failing line.
  • Growth is absorbed at flat headcount. The desk’s ceiling becomes its pipeline, not its roster.
  • The underwriter still decides. Every time.

Two seats

You will hand this to two other people

They should find themselves here too.

If you run the underwriters

Your problem is accuracy and your people’s day. A client with a hundred rules checked against thirty is your nightmare, and silence about the other seventy is worse than a wrong answer. You will not accept anything that slows a good underwriter down, takes judgment away from her, or fails on a phone at two in the morning. You feel resubmissions as pure rework.

Your test is simple: check every rule, every time, show the work, and let my people decide.

If you own the platform

Your problem is boundary and proof. Anything that writes to the LOS, polls the queue, moves member data where it should not go, or crosses a vendor’s terms of service is dead on arrival. You will ask for read-only access by application number, no write-back, no PII leaving your control, an audit trail with versions, a SOC 2 report, and a result measured in your own data against your own baseline.

Those are the right questions. This page was written expecting them.

Diagnostic

Where does your knowledge live?

Five questions. Two minutes. Your score on the spot, no form required. If you tell us who you are, we will send you how comparable desks answered.

Question 1 of 5
How many distinct client guideline sets do your underwriters apply?
When an underwriter needs a rule, where do they find it?
What share of your applications arrive outside staffed hours or on weekends?
How long before a new underwriter is productive across your full client mix?
A client asks which version of their policy applied to a decision three weeks ago. How do you answer?

Sidecar

Fifteen minutes on your workflow

Sidecar helps you field more business without additional headcount. Bring one client’s hotsheet and one recent application. We walk your workflow, not ours. No deck.