Find the lowest cash balance in the month, not the closing one

Run the balance forward one week at a time — what is in the account, plus what lands that week, minus what leaves — and take the smallest number in the column. That is the lowest point, and it is the only figure that says whether the money runs out. The closing balance cannot say it, because a balance that goes to minus 2,100 in week seven and recovers by week nine closes the month in credit.

Why the closing balance hides it

Profit is an opinion about a period; cash is a fact about a date. A business does not fail because it is unprofitable. It fails because on a particular day there is less in the account than the payroll needs, and that day can fall in a month that ends in credit, inside a year that ends profitably. The month-end figure is the balance on one date, and nothing happened to bounce on that date. The bank cared about a different one.

Take a quarter that opens with 42,000 in the account. Two contracts pay 18,000 every four weeks and 9,500 every two; one sale of 24,000 lands in week nine. Payroll of 14,800 and rent of 3,100 go out every four weeks, suppliers take 6,200 a week, and a VAT bill of 11,400 is due in week six. Over thirteen weeks 153,000 comes in and 163,600 goes out. It closes at 31,400 and its lowest point is 15,900, in week seven. Nothing wrong yet.

Now start the same quarter with 24,000 instead. Same receipts, same payments, same 10,600 more out than in. It closes at 13,400, in credit, and in week seven it is at minus 2,100:

The reason is timing, not totals. The VAT bill lands in week six and the sale that covers it arrives in week nine, so for one week in between there is less in the account than the suppliers' payment needs. Read only the last row and the quarter looks fine. Read the column and it is a forecast of a bounced payment with a date on it.

How to find it

1. Start from what is in the account today, cleared, not from last month's closing figure and not from the ledger balance.
2. List every receipt and payment by the week it actually lands. Payroll on the day it leaves the account; VAT on the day it is due, not the quarter it accrued in; a customer at their terms plus however late they usually are, not the month the invoice was raised.
3. Run the balance forward. Each week's opening plus its receipts minus its payments is the next week's opening.
4. Take the smallest number in the balance column, and the week it falls in. That is the lowest point. The week is the date to plan around.

Weeks rather than months, because a month is long enough to hide the crossing: payroll on the 25th and a receipt on the 30th net to nothing over the month and to a bounced payment on the 25th. Thirteen weeks rather than a year, because a quarter is far enough ahead to see a quarter-end tax bill and near enough that the entries are dates rather than hopes.

The second number: how much has to go wrong

A forecast that clears zero by 15,900 looks safe until you ask how. The way to ask is to re-run it with everything a little worse and see where it breaks: every receipt one week late, then two, until the low point crosses zero; and separately with one per cent of every receipt failing to arrive, then two, until the same thing happens. Those are the two things that actually go wrong — customers pay late, and some of them do not pay — and they are separated because they have different remedies.

For the quarter that opens at 42,000, the answers are these. Receipts can be late by nought weeks. One week late on everything, and week nine fails at minus 4,900: the sale that was going to carry week nine's payroll arrives in week ten instead. And a fifth of every receipt can fail to arrive before the low point reaches zero; at 21 per cent it does not. So a forecast that never goes below its buffer has no tolerance at all for lateness, and that, not the 15,900, is what a conversation with a bank or a board is actually about.

Nobody finds those two numbers by hand, because each one takes a dozen recalculations. CASHFLOW finds them by search: it re-runs the forecast until it breaks and reports the boundary, rather than asking you to imagine a bad quarter.

Where the low point usually hides

The buffer

Zero is the wrong line to watch, because by the time the balance is at zero there is nothing left to pay the thing that arrives unannounced. Decide the least you are willing to see in the account — a payroll's worth is the usual answer — and treat the first week under it as the warning, not the first week under nought. In the quarter above with a buffer of 10,000, that week is week six, one week before the money runs out, which is the week you would want to have known about.

What CASHFLOW does and does not do

CASHFLOW takes the opening balance, the horizon, the buffer and the lines — what each one is, in or out, how much, when it starts, how often it repeats and when it stops — and runs the balance. It reports the lowest point and when it comes, the first week under the buffer or under zero, and the two boundaries above; draws the shape; lays out the week-by-week table; and hands back the working as text or as a CSV. Nothing is uploaded, which for a cash position is not a small point, and it works with the connection off.

It does not read a bank statement or an accounts package. You type the lines, so it is a forecast of what you told it, and the customer you put down at thirty days is in the forecast at thirty days. It does not model interest, an overdraft facility or exchange rates. A week that would go below zero is shown below zero, which is the point of it.