What Reconciliation Means, and Why Skipping It Breaks Everything

Reconciliation sounds like an accounting formality. It is closer to the opposite — it is the only step in bookkeeping that tests your records against something you did not create.

Everything else in the books is internal. You decide how to categorise a transaction, you decide when to record it. Reconciliation takes the resulting balance and asks a third party — your bank — whether it is true.

What actually happens

You take the closing balance on the statement. You take the closing balance in the books. You identify every transaction that explains the difference: cheques not yet cashed, deposits still in transit, payments the bank has not processed.

If the adjusted figures agree, the account reconciles. If they do not, something is wrong, and the difference is a lead rather than a nuisance.

The discipline is in that last sentence. A difference is information. Treating it as an obstacle to get past is where books go bad.

What it catches

Reconciliation is the control that finds problems nothing else will:

• Missing transactions. A bank feed silently drops a few days. Nothing in the file indicates a gap; the reconciliation simply will not balance.
• Duplicates. The same expense imported twice, or entered manually and then imported.
• Wrong amounts. A transposition — 5,481 entered as 5,148 — which is invisible in a category total but shows up here immediately.
• Unauthorised activity. A charge nobody recognises. This is the check that catches card fraud and internal theft, and in small businesses it is frequently the only one.

That last point deserves weight. In a business where one person records transactions and reconciles the accounts they recorded, the control is compromised by design. That is precisely the structure most small businesses have.

The forced reconciliation

Every accounting package will let you post an adjusting entry to make a stubborn difference disappear. It is occasionally legitimate — a small bank fee nobody recorded — and mostly not.

A forced reconciliation is worse than an unfinished one, for a specific reason: it looks complete. Nobody revisits it. The underlying error stays in the file, now wearing the costume of a legitimate entry, and it will surface at the least convenient moment.

If you inherit a set of books, searching for adjusting entries with vague descriptions and non-round amounts is one of the fastest ways to assess how much you can trust them.

Not just the bank

The same logic applies to every balance sheet account that has an external counterpart:

Credit cards to card statements. Loans to lender statements — this one drifts constantly, because payments get expensed rather than split. Payroll liabilities to the provider's filings. Sales tax liability to what was actually collected and remitted. Merchant deposits to processor settlements.

Each of these is a place where the books can quietly diverge from reality, and each is checkable against a document somebody else produced.

The frequency question

Monthly. Not quarterly, not annually.

The reason is arithmetic rather than diligence. An error found within thirty days is bounded by one statement and one person's memory. The same error found eleven months later has to be located somewhere in a year of activity, and whoever could have explained it has forgotten.

Reconciling a clean month takes minutes. Reconciling twelve at once takes days, and produces worse answers.

It is the least glamorous part of the job and the one everything else rests on — which is why it is a defined step in every monthly close we do, and why a cleanup runs oldest-period-first rather than starting with the month that feels most urgent.