Reconciling a bank statement against your books

The purpose of a bank reconciliation is routinely misdescribed as finding mistakes. Most of the time there are no mistakes. Two records disagree because they record the same events at different moments and because each knows about some events the other does not, and reconciliation is the exercise of accounting for that difference until nothing unexplained is left.

Your books record a payment when you make it — when the cheque is written, the transfer is instructed, the obligation is discharged from your side. The bank records it when it happens to the bank. Those are genuinely different dates, and neither record is wrong.

Separately, the bank knows things your books have not been told. It has levied charges, credited interest, executed a standing instruction, or returned a cheque you believed had cleared. Until the statement arrives, those events exist only on one side.

So the output of a good reconciliation is not a corrected figure. It is a list: every item explaining the difference, each classified as either a timing difference that will resolve itself or an error that will not. The first category needs no action beyond being understood. The second is the reason the exercise is worth doing at all, and it is a small minority of the items.

The conventional statement starts from one balance and works to the other, and the direction matters less than doing it consistently. Starting from the bank's closing balance:

Add deposits recorded in your books that the bank has not yet credited — receipts in transit. Subtract payments recorded in your books that the bank has not yet paid — principally cheques you have issued that remain unpresented. Then adjust for everything the bank has recorded and your books have not: charges, interest, direct credits from customers you were unaware of, standing instructions and mandates, and returned items. Finally, correct any actual errors on either side.

What you should arrive at is the balance in your own books. If you do not, the difference is unexplained, and an unexplained difference is the whole finding — it is not a rounding to be absorbed or a small amount to be written off, because the size of a discrepancy has no relationship to the size of what caused it. A missing zero and a duplicated small charge can produce the same gap.

One discipline makes this tractable: reconcile a defined period against a statement whose opening balance equals the previous period's closing balance. A reconciliation performed on statements with a gap in them cannot balance, and a great deal of time gets lost to arithmetic that was never going to work.

Unpresented cheques have always been the largest timing item for businesses that pay by cheque. You issued it, your books show the payment, and the bank has not seen it — because it is in the post, or on a desk, or the payee has not banked it.

That category has recently become smaller and shorter-lived at one end. Clearing has moved from batch processing with next-working-day settlement to continuous clearing with settlement on realisation, phased in by the Reserve Bank from October 2025 and January 2026, so once a cheque is presented it is now confirmed or returned far more quickly than the old cycle allowed. The clearing lag has largely gone.

What has not gone, and what remains the real cause of the item, is the interval before presentation — the time between your writing the cheque and the payee depositing it, which no clearing reform touches. So the shape of this reconciling item has changed: it is now almost entirely about payee behaviour rather than about system delay.

The other timing items are receipts deposited near a cut-off and appearing in the next period, and transfers instructed late in a day. Each of these resolves on its own in the following period, which is the test for whether something is a timing difference at all: if it will not clear itself next month, it is not one.

This category causes more unreconciled balances in small businesses than any other, because the entries are small, numerous and require no action from you to occur.

Bank charges lead the list: non-maintenance charges, cash-handling charges beyond a free allowance, transaction charges, cheque return charges and service charges on facilities. Individually they are minor and collectively they are the reason a book balance drifts steadily above a bank balance over months.

Interest works the same way in the other direction, whether credited on a deposit or debited on an overdraft, and overdraft interest computed on daily balances will rarely be a figure your books predicted.

Standing instructions and electronic mandates execute silently. The ones to look for are those you have stopped needing: a mandate given for a service since discontinued keeps debiting until it is cancelled with the bank, and the statement is where you will discover it is still running.

Direct credits are the pleasant version of the same problem — a customer paid by transfer, quoted no invoice reference, and told nobody. The receipt is in the bank and unrecorded in your books, and it will sit there as an unexplained credit until someone matches it, which is much easier in the month it arrives than a year later.

Errors are the minority of items and the reason for the exercise, and they have recognisable shapes.

On your side, the common ones are transposition — digits swapped, which produces a difference divisible by nine and is worth knowing as a quick test; an entry posted twice; an entry posted to the wrong side; and a payment recorded at the invoice amount when a different amount was actually paid, which is frequent where a discount or a deduction was applied at payment.

On the bank's side, errors are less common but not absent: a charge levied that does not match the published schedule for your account variant, a transaction debited twice, a reversal for a failed transaction that never arrived, or an entry belonging to another account.

The category that deserves separate treatment is a debit you did not authorise. That is not a bookkeeping error and it should not be quietly journalised. The Reserve Bank's framework on limiting customer liability for unauthorised electronic transactions ties the outcome to where the fault lay and to how quickly the transaction is reported, so the reconciliation is where it is found and the bank's complaint channel is where it goes.

The practical argument for reconciling monthly rather than annually is precisely this: reporting speed matters for some of what you will find, and a year-old debit forecloses options a month-old one leaves open.

The frequency question has a clear answer for reasons that have little to do with bookkeeping tidiness.

The first is evidence. Matching an unreferenced credit to a customer, or reconstructing what a terse charge narration referred to, is straightforward within weeks and genuinely difficult a year later. The people who would remember have moved on and the context has gone.

The second is the reporting-speed point above: some of what a reconciliation surfaces is time-sensitive in a way that a bookkeeping error is not.

The third is compounding. An unexplained difference carried forward contaminates every subsequent period, because each month's reconciliation starts from a balance that already contains something unaccounted for. A single unresolved item makes twelve months of work harder rather than one.

What makes it easy is unglamorous and effective. Keep the transaction reference against your own record of each payment, so a query is a lookup rather than a search. Record charges as they appear rather than in a year-end sweep. Cancel mandates at the bank when a service ends rather than only with the provider. Ask customers to quote an invoice reference on transfers, since the narration will carry whatever they put there. And close each period rather than leaving it open, because an open period is where unexplained items go to be forgotten.