Why Ledgers Shouldn't Have an Undo Button

Every spreadsheet has an undo button. Every notebook has an eraser, or a line scratched out and rewritten. It feels natural, then, to expect the same from software: type the wrong number, fix it, move on. A ledger that actually needs to be trusted months later can't work that way, and understanding why is worth more than it first seems.

The difference between an edit and a reversal

An edit replaces what a record used to say. Once it's done, there's no trace that the record ever said something else, unless a separate change-log happens to have caught it. A reversal is different: it's a new entry that cancels out the effect of an old one, while leaving both the original and the reversal fully visible. The balance ends up the same either way. The history doesn't.

Why the history is the part that matters

A single wrong entry, corrected the same day, rarely matters either way. The case for reversals over edits shows up months later, when a number looks off and someone needs to work out why. "This customer's outstanding is ₹18,000 lower than expected" is a question with an answer if every correction that happened along the way is still visible as its own entry. It's not a question with an answer if corrections silently overwrote the numbers that were there before.

What this looks like for a specific mistake

Say a payment was entered against the wrong customer. Editing it would change the amount, or the customer, on the original entry, and the mistake disappears entirely, along with any record that it happened. Reversing it instead means: the original payment stays exactly as it was entered, a reversal entry cancels its balance effect, and a new, correct payment is entered separately. Three entries instead of one, but all three explain themselves.

The same logic applies past payments

A purchase, a sale, a production run, any posted transaction, follows the same rule. Cancelling a sale reverses its stock and balance impact and stays in the record's history with who cancelled it and, ideally, why. That "who and why" is what turns a correction into something explainable rather than something suspicious.

This costs something, on purpose

It's slightly slower to reverse-and-reenter than to just fix a number in place. That's the actual trade being made: a small amount of friction on every correction, in exchange for a ledger that can still answer "why does this say what it says" a year later. For a book that exists specifically to be trusted after the fact, that trade is usually worth making.

More on what changes when purchase, production and sales records connect