It is the third day of close, it is late, and one account will not tie. The difference is small — small enough that the fastest path to a clean reconciliation and a finished day is to book it to a clearing account and move on. Tick, tie, initial, next. Everyone in finance has sat in that chair. The reconciliation gets treated as a thing to complete, and the break gets treated as the thing standing between you and completing it.

That framing is the whole problem. A reconciliation is not a form you fill out. It is a question you put to the system: do these two records of the same reality agree? When they don't, the difference is not an obstacle to clear. It is the answer to the question — and often the most useful thing the close will tell you all month.

A break is not what's keeping you from a clean reconciliation. It is the reconciliation, doing its job.

What a difference actually says

Take the most common one: a sub-ledger that doesn't agree with its GL control account. The AR sub-ledger is the detailed record of what every customer owes; the control account is the single general-ledger balance meant to mirror their sum. When the two diverge, the system is telling you something specific and true — a transaction posted to one and not the other, a manual journal hit the control account directly, an interface dropped a record, a timing window caught a transaction mid-flight. Each of those is a finding with a cause. None of them is fixed by forcing the two numbers to match.

The plug — the entry to suspense that makes the difference disappear — does exactly the wrong thing. It removes the symptom and keeps the disease. The accounts now tie, the box is ticked, and the actual defect is still live, still mis-stating something, now harder to see because the one signal that would have surfaced it has been booked away.

The small ones are the dangerous ones

The instinct is to chase large variances and wave through small ones on materiality. But a small net difference is often two large errors that nearly cancel — an overstatement and an understatement that happen to land close together this month and won't next month. The size of the break tells you almost nothing about the size of the cause. We have seen a few-hundred-dollar variance turn out to be a mapping error moving real money to the wrong account, netted down to a rounding-sized residual that a team had been clearing, uninvestigated, for a year.

Treat the break as a finding
  • Name the cause, not just the amount. "$1,240 to suspense" is a plug. "An interface dropped three invoices" is a finding you can fix.
  • Be suspicious of small net differences. A tiny residual can be two large errors cancelling out. Decompose it before you dismiss it.
  • Read what the clearing account collects. A suspense account that only ever grows is a list of defects nobody went back to read.

The discipline

The practice is unglamorous, and it is the whole thing: every difference gets a cause before it gets an entry, even the immaterial ones — because the pattern across the immaterial ones is itself a signal. You will still clear most of them in a minute, once you know why. The point is never to plug something you haven't understood. A clearing account should be a waystation with a known reason and an exit, not a graveyard.

Done that way, reconciliation stops being a formality and becomes what it was always for: the place where the system gets to tell you the truth before anyone downstream relies on the number. The reconciliations you can trust are the ones where you chased the breaks you were allowed to ignore.