What Exam Season Costs an Operations Team That Nobody Budgets For

There is a line in Fannie Mae's Selling Guide that most people outside a QC function have never read. It says that if the lender's QC cycle is in arrears more than one 30-day cycle, the lender must provide written notice to its Fannie Mae customer account team or QC Specialist. For most leading operations, falling behind on quality control is not an internal scheduling problem to be quietly caught up over a long weekend. It is a reportable condition. The clock is not advisory.

Most operations leaders we talk to know the cycle is tight. Fewer have connected that tightness to what actually happens inside their own department three or four times a year, when loan volume climbs or an examination date lands on the calendar and the review queue has to be cleared regardless. What follows is not a compliance problem. It is a capacity problem wearing a compliance costume, and almost nobody measures it.

The clock does not care how many loans you closed

The requirement is specific. Loans must be selected for post-closing QC review on at least a monthly basis, and the entire cycle has to be finished inside 90 days of the month the loan disbursed or was acquired. The part that catches teams out is what the cycle includes. It is not 90 days of review. It is selection, review, rebuttal, and reporting, all four, inside the same window. Rebuttal in particular is unpredictable, because it depends on how quickly other people answer questions. A reviewer can finish the file work in week six and still be chasing a response in week eleven.

None of that flexes with volume. A month where the institution closed sixty loans and a month where it closed a hundred and sixty are given the same calendar. The requirement was written to make defects surface sooner, which is a reasonable thing to want. The practical effect at a smaller institution is that a busy month arrives with a fixed deadline already attached to it.

The sample is a floor, not a target

The second fixed quantity is the size of the review itself. A lender choosing the straightforward path must select a minimum of 10% of the loans that it originates or acquires using a random selection methodology. The alternative is a statistically valid sample calculated at a 95% confidence level with a 2% precision rate, which most community institutions pass on for the sensible reason that ten percent is easier to explain to a board.

Two details matter here and both tend to get lost.

First, that random sample is doing more work than it appears to. The Selling Guide describes it as a critical measure of loan quality that serves as the basis for calculating the lender's defect rate. The sample is not just a spot check. It is the measurement instrument. Which is precisely why it cannot be quietly trimmed in a heavy month without changing the number the institution reports about itself.

Second, discretionary selections do not help with this. The guide is direct that discretionary samples supplement, and do not replace, the random sample. Targeting higher-risk loans is good practice and it is additional work, not a substitute for the baseline.

So the institution enters a busy month with a fixed window and a minimum volume of review that scales up with originations. Everything in the equation is fixed except one thing.

Which leaves people as the only variable

This is the part that never appears in a policy document. When the numbers do not work, somebody gets borrowed. A processor who knows loan files well gets pulled into QC review for three weeks. A deposit operations specialist who has done QC before gets asked to help clear the queue. At a $740M community bank with two dedicated reviewers, adding a third body for most of a month is not a dramatic intervention. It is Tuesday.

From our experience, nobody in that conversation thinks the process is wrong. They think the process is what the resources allow. The borrowing is not a failure of planning, it is the planning. It happens often enough that it has stopped registering as an event worth recording, which is exactly why it never shows up in any report about what quality control costs the institution. And because it never shows up, it never gets fixed. Work that is invisible does not get a line item, a project sponsor, or a place on next year's priorities. It just recurs.

What a borrowed reviewer actually costs

The first real cost is consistency. Two reviewers looking at the same missing document will describe it differently. One writes a short note referencing the loan file, another records it against the borrower, a third flags it in a spreadsheet column that only she uses. None of them is wrong. But when the findings are assembled for reporting, somebody has to reconcile three vocabularies into one defect summary, and that reconciliation happens inside the same 90 days as everything else. Rework at the reporting stage is the most expensive rework there is, because it arrives when the calendar has the least room left.

The second cost is the work that stopped. The processor who spent three weeks on QC did not spend three weeks on the thing she was hired to do. Whatever that was, it slipped. A vendor review that moves to next quarter. An exception queue that grows while nobody is clearing it. A new hire whose training gets postponed. These are real costs with real consequences, and they are almost never attributed back to the QC cycle that caused them, because by the time the consequence shows up the cycle is closed and everyone has moved on.

Coverage is a capacity question before it is a compliance question

It is worth being clear about something, because the industry conversation around sampling tends to carry an accusation in it. Nobody samples because they believe sampling is sufficient. They sample because reviewing everything by hand was never possible with the people available, and ten percent is the number the rules settled on. An institution testing at the minimum is doing what is required of it. The constraint has never been judgment or diligence. It has been arithmetic.

That framing matters because it changes the question. If the limit is capacity rather than conviction, then the interesting question is not whether the sample is big enough. It is what would become possible if coverage stopped being the thing that has to be rationed. We have written separately about why a 10% QC sample is not a quality control program, and the argument there is about risk. This one is about the calendar and the people on it, which is a different problem with the same root.

Three things worth measuring before the next cycle

If any of the above is familiar, there are three numbers worth collecting during the next review cycle. None of them requires a project. Count how many people touched the cycle who do not normally work in QC. Not hours, headcount. It is the fastest proxy for how much the department is borrowing, and most leaders are surprised by it.

Count how many of the ninety days went to rebuttal and reporting rather than file review. If that share is large, the bottleneck is not review capacity, it is the handoffs around it, and adding reviewers will not fix it. Write down what stopped. One line per person borrowed, recorded at the time rather than reconstructed afterward. This is the hardest of the three because it requires somebody to notice in the moment, and it is the only one that makes the real cost visible to anyone outside the department.

Those three numbers will tell an operations leader more about their quality control program than the defect rate will, because the defect rate measures the loans and these measure the institution's ability to keep looking at them.

Have a Problem to Solve?

The institutions that make progress here rarely start with a plan to overhaul quality control. They start by picking the single most annoying recurring piece of it and getting that one thing off people's desks, which is also the approach we wrote about for automating deposit account QC without adding headcount.

We handle automated QC and data quality review for community banks and credit unions, with the institution's own written procedures as the ruleset and its own team making every judgment call. If the next cycle is already looking tight, reach out to Shore Group and we can compare notes on how yours runs today.