Flux & Variance

Why your flux commentary keeps getting rewritten

The control tests that an explanation exists. It doesn't test whether the explanation is any good — and the bar that decides that is your reviewer's judgment, held individually and never written down anywhere you can read it.

If your commentary gets deleted and rewritten by a reviewer, the natural reading is that you did the work badly. Sometimes that's what happened. Often it isn't. Often the reviewer has a view of the account — what drives it, what matters about it, how the business thinks about that activity — and your explanation is defensible but not the one they would have written.

There is a lot of grey area in that, and it helps to know where the grey starts.

What the control actually tests

Flux analysis is an internal control over financial reporting, though almost nobody preparing one is thinking about it that way. You know it has to be done, and you do it.

The control itself tests something narrow: for accounts whose movement exceeds the defined threshold, is there commentary, and did someone other than the preparer review and approve it?

That is a presence test. It asks whether an explanation is there. It does not ask whether the explanation is correct, whether it names the driver that actually moved the account, or whether it would tell a reader anything useful about next month. Those are quality questions and the control isn't designed to answer them.

This isn't a defect in how your company wrote the control. It's common, because quality is hard to test in a way that produces evidence. Presence produces a clean artifact: the commentary is in the file or it isn't, the sign-off is there or it isn't. A quality standard would require a reviewer to document why an explanation was judged sufficient — which is judgment, and judgment doesn't reduce to a checkbox without becoming one.

So the standard stops at presence. Everything past presence still matters — it's the entire reason the control exists — but it's enforced informally.

What "informally" means in practice

It means the bar is whatever the reviewer thinks a good explanation of that account looks like, applied across thirty accounts, under a submission deadline.

That bar is real. Reviewers generally have a clear sense of what drives the accounts they own. But it's held individually rather than written down, so a preparer learns it by producing commentary and seeing what comes back. Two reviewers can hold different bars without either being wrong, because neither was ever documented to conflict with the other.

Under time pressure there's a fast option: rather than send commentary back with an explanation of what's missing, the reviewer deletes it, writes it the way they would have written it, and notes that this isn't what's driving the account. That closes the item immediately, and it's often the right use of their time.

What it doesn't do is transfer the standard.

The asymmetry that shapes review

There's a structural feature that determines what flux review catches, and it runs in the direction of catching less.

Challenging an item that turns out to be correct has an immediate cost. Someone prepared that account, someone else likely looked at it, and questioning it implies neither of them understood what they were doing. The response is often mildly pointed, and both parties have now spent time on nothing. Missing something, by contrast, costs nothing today. It surfaces later, if at all, and by then it's rarely traced back to the review that let it through.

Under deadline, that asymmetry pushes reviewers toward explaining what they're seeing rather than challenging whether it should be there. Explaining is the safer position to defend. Challenging is exposed.

Reversals are the clearest case. A reversal that looks unusual is almost always correct — it reverses a prior accrual, and someone booked it deliberately. The productive work isn't flagging it, it's spending enough time to understand why it's reversing, because raising it without that understanding is how you spend credibility on nothing. That understanding takes time, and time is the thing in shortest supply.

Where the reviewer's seat ends

Some of what looks like a variance is a methodology decision rather than an error.

Accrual cutoff timing is the common example. How many days of activity get accrued, and where the line falls each month, is often a judgment made by the person ultimately responsible for the books — balancing accuracy against a department that has to close on a schedule and produce the same result repeatably. A reviewer who treats the resulting variance as a mistake isn't identifying an error. They're questioning a design decision made deliberately by someone with the authority to make it.

Where this actually surfaces is in front of management. You're walking through the explanation, and it becomes visible that the story doesn't line up with what makes sense for the business. At that point the honest move is to say so — looking back at last month, that wasn't completely right — rather than defend it.

High-volume accounts are harder for a reason that isn't skill

The accounts that resist clean explanation are often the ones with heavy transaction activity you don't personally follow.

It isn't that the detail is unavailable. It's that sifting through a large volume of activity you're unfamiliar with, to find which items actually moved the account, is slow — and being unfamiliar with the activity also means you can't quickly tell what normal looks like for it. Both problems land on the same accounts.

So where does that leave the standard?

Roughly here: the control tests that an explanation exists. Whether it's a good explanation is a judgment your reviewer makes, informed by what they know about the account and the business, and it isn't written down anywhere you can read it.

Which means the fastest way through isn't guessing at a universal quality bar. It's finding out what the reviewer of that specific account expects an explanation to contain — what they think drives it, and what they'd want to see named.

Some of it does reduce to arithmetic, and that part is worth getting out of the way before review rather than during it: whether the drivers you named actually account for the movement, whether they point the same direction the balance moved, whether each one says where to look. More on that here.

The rest is the conversation.

If you work this in Excel: the free flux template flags what needs explaining and shows the residual your drivers don't cover. The CloseOps Flux & Variance System ($79) starts from your trial balance instead: confirm each account's classification and it builds the income statement and balance sheet flux statements, then ranks what to investigate.

Found something wrong or missing? Tell me here — anonymous, thirty seconds.