What is reconciliation, really?
Reconciliation isn't matching two columns of numbers. It's proving two records describe the same economic event. Riya walks through it in 3 panels.
“Books in one column. Bank in the other. The story should be the same.”
Transcript›
Books show ₹12,300. Bank shows ₹12,600. Which one is right?
Both. The gap is timing, not error. Reconciliation finds which is which.
“Three reconciling items. Each one is timing, not error.”
Transcript›
Three items explain the ₹300. None of them are fraud.
Most reconciliation gaps are timing. The math is simple — the finding is the work.
“What reconciliation actually proves.”
Transcript›
Adjusted, both sides land at ₹12,450. The books didn't lie. The bank didn't lie. The gap was timing.
Reconciliation isn't balancing totals. It's proving the story behind them.
The longer take
The first audit I worked on, I matched 12,000 rows in Excel and felt like I had reconciled something. My senior partner walked over, looked at the file, and said: you matched nothing. You proved that two columns of numbers happen to sum to the same total. You did not prove that they describe the same economic events.
That distinction is the difference between bookkeeping and reconciliation. Bookkeeping records transactions; reconciliation proves the records hold together as a coherent story. A ₹300 gap between a book balance and a statement balance can be three things: a true error that needs correcting, a timing difference that will resolve in the next cycle, or evidence of fraud. Each requires a different response. The job is to find which one it is, not to make the columns tie.
The pattern in the panels above is the simplest version of the standard month-end shape: the General Ledger says one balance, the bank statement says another, the difference is explained by a small set of timing items — a customer credit the bank received but we haven't posted, a payment we issued that hasn't cleared, bank charges the bank deducted that we haven't recorded — and once each is accounted for, the gap closes to zero and both sides agree on an adjusted balance. The discipline isn't in solving any single mismatch. It's in doing the same investigation every cycle and leaving an audit trail that proves you did it.
What goes wrong in practice is volume. A small business with 20 transactions a month can do this in a notebook. The same business at 2,000 transactions a month cannot, because the human attention budget required to investigate each mismatch grows linearly with transaction count and the human attention available grows not at all. The middle ground — semi-automated reconciliation with spreadsheets — works until format drift, split settlements, or counterparty timing breaks the underlying assumption that each row pairs with exactly one other row. When that breaks, the reconciliation stops being a discipline and starts being a Friday afternoon ritual of writing off the residual to make the books tie.
The honest version of the trade-off is that reconciliation is not optional, and the question is not whether to do it but how to scale it. A tool that does this well doesn't replace the human judgement at the disposition step; it removes the row-by-row matching from the human's plate so the human can spend time on the variances that actually require thought. That's the architecture. Everything else is implementation.