How to read a day-wise report and catch the mistakes the next morning

On the Saturday before Diwali your report says ₹24,600 came in and ₹19,150 went out. Both figures are ₹1,000 too high. A ₹1,000 payment to the flour mill got saved twice, and a ₹900 credit was entered as ₹1,900. The balance between them is right to the rupee. A total is a sum, and two errors of opposite sign cancel inside it perfectly, which is why the balance is the last thing on the page worth reading.

Start by counting the lines. The bill book settles it: last serial of the day minus the first is thirty-one, and the ledger has nineteen entries in it. Twelve sales never got written down, and the figure you were studying belongs to a day that did not happen. No arithmetic catches this, because a sum cannot miss what it was never given.

Then look at the times. A shop open from nine to nine whose nineteen entries are all stamped between 3.05 and 3.20 did not record the day; it reconstructed the morning in the afternoon lull. On paper the same thing shows as one hand, one pen and an unbroken run of neat figures. Those amounts were written from memory an hour or three after the money moved, and the drift in them is not visible anywhere in the report.

The box holds ₹8,800 at closing and the day's cash lines say ₹9,300. Counting the notes a second time will not tell you anything the first count did not; the ₹500 is a reason to open the report, not a problem the drawer can answer. A gap that size has only a few possible homes in the day, and the report shows all of them.

Start with the method on each line, because the commonest of them leaves every total correct. A ₹500 payment that came in notes and got saved as UPI puts the day's takings right to the rupee and the drawer exactly ₹500 short, so no amount of adding up will ever point at it. Read the day with the transfers filtered out and look at the cash lines on their own, against the figure you counted: the line that does not belong is usually obvious once it is the only kind of line on the page. It runs the other way too, and a ₹500 transfer marked cash leaves the box looking heavy.

If no cash line matches, read the tail of the previous evening and the first entries of the next morning, because a slipped date is where a missing ₹500 most often sits. Then read the pending lines: a promise that was actually settled in cash and never confirmed is listed and left out of the arithmetic, which makes the box heavier than the report rather than lighter. What survives all of that is money that left the drawer with nobody writing it — ₹500 to the helper at six o'clock, or the tempo driver's slip still in your shirt pocket — and that is a question for whoever was standing at the counter, asked the same evening. Four unexplained ₹500s in a month is ₹2,000 that nobody stole and nobody can account for.

Scan each day for the same amount appearing twice close together, and look hardest at round figures: ₹500, ₹1,000, ₹2,500. Genuine repeats are ordinary; two customers paying ₹1,000 each is a Tuesday. One payment entered twice looks identical inside a total. The shape to hold in your head is two entries pointing the SAME way — two credits, or two debits — separated by nothing but the clock and the note; a credit and a debit of the same amount a day apart is a different mistake altogether, one movement of money written down from both ends, and that one is found by sorting on the amount rather than by reading the times.

Of the two signals here, the clock is the weaker. 11.42 and 11.44 is one payment saved twice by somebody who thought the first attempt had failed; 11.42 and 12.50 usually is not. The note is what settles it. Two ₹2,500 credits an hour apart, both reading Rajubhai stock payment, are one payment written down twice by two people who each assumed the other had missed it.

The expensive version involves a promise. A customer says he will pay ₹2,000 on Monday and you mark the entry pending. On Wednesday the money arrives, somebody adds a fresh credit of ₹2,000, and somebody also goes back and marks Monday's entry received. The books now show ₹4,000 from a man who paid ₹2,000. It is hard to see coming, because a pending entry is not counted in a balance while it is pending. The gap opens at the moment it is confirmed. Tell everybody who touches the ledger the same sentence: when the money arrives, settle the promise you already recorded, and never write it a second time.

A wrong name survives every check so far. The count is right, the cash matches, nothing is doubled, and the quarrel is still coming. A committee of twelve pays ₹5,000 each. Ramesh pays on the 3rd, Rajesh on the 4th, and the 4th's instalment is filed against Ramesh. The month's collection is ₹60,000, correct to the rupee. Ramesh now appears to have paid twice and Rajesh appears not to have paid at all, and you find out when you ask Rajesh for ₹5,000 he counted into your hand on the 4th.

So read the report by person as well as by day. Filter to one member at a time across the month and read that column against your collection sheet, or against the receipt the member himself is holding. Not against memory: if memory could be trusted here, the instalment would not have gone to the wrong name in the first place. Do it before you tell anybody they are behind.

When something is misfiled, who filed it starts to matter. Khatamel records who added each entry and when, so a correction is a change somebody made and can be traced to them. A paper ledger does the same with initials in the margin and corrections struck through, never erased.

A day-wise report tells you what moved. Tags tell you where it went. Six or eight labels will cover a shop: Stock, Transport, Labour, Rent, Electricity, Festival credit, Home. A short list gets used the same way every time. Thirty labels means every entry gets a slightly different one and none of them add up to anything by month end.

Tempo charges of ₹1,200 feel like nothing on the day you pay them. ₹1,200 is also the only figure you ever see, so ₹1,200 is the number in your head. Tag them for a month and read that label's total over the range: twenty-six delivery days is ₹31,200. Nothing has gone wrong, the money went where you sent it. But nobody negotiates a rate, or decides to put two trips into one, over a figure they have never seen written down.

The entry you cannot label is the one to look at hardest. Six labels cover a shop because a shop does six kinds of thing with money, so an entry that fits none of them usually means the note was too thin to say what happened — not that you need a seventh label. ₹2,000 out on a Tuesday with “paid” against it is not a Transport entry and it is not a Stock entry. It is an entry nobody can read. Sort it out tonight, while the day is still in your head, and the label follows by itself.

Open yesterday's report before the shop fills up.

• Count the lines and compare that against the bill book.
• Match the cash figure you wrote at closing against cash in minus cash out.
• Look for anything still marked pending that was promised more than a week ago.

Five minutes, standing up, with yesterday still close enough to argue with. The same three checks attempted on the 30th take an hour and end in your word against a customer's, because an ₹8,000 udhaar entry with no note, no label and no name on it is not evidence of anything. An error found the next morning is one phone call. The same error found at month end is a negotiation.

If your ledger is paper, keep yesterday's page loose and tick it once it has been read. If it is on a phone, filtering the day-wise report to yesterday is the whole review; a PDF of the month is for the settlement conversation, not the morning one. Start with yesterday. Today's page is not finished yet.