The 13-week cash flow, rebuilt every Monday
A 13-week cash forecast is the single most useful document a small business can keep, and most of the ones that get built are dead within a month. Someone spends a careful afternoon on it, it looks good, and then reality diverges and nobody updates it.
The model was never the point. The forecast is a habit, and the habit is what tells you things — because the interesting information is not in the projection, it is in the gap between the projection and what actually happened.
Why thirteen weeks
It is a quarter, in weeks. Long enough that you can still act — chase receivables, delay a purchase, arrange credit — and short enough that the numbers are grounded in real invoices and real bills rather than assumptions.
Weekly matters as much as thirteen. Monthly buckets hide the problem: a month can end comfortably while the second week of it was two days from an overdraft. Payroll, rent and tax dates do not spread themselves evenly across a month, and neither do customer payments.
The Monday routine, twenty minutes
- Enter last week's actuals. Real opening balance, real receipts, real payments. Five minutes.
- Compare to what you forecast. This is the step everyone skips and the only one that teaches you anything.
- Roll the window forward. Drop the week that closed, add a new week 13.
- Update what you now know. A large invoice went out, a customer promised Thursday, an equipment deposit is due. Adjust only what has genuinely changed.
- Read the low point. The lowest projected balance and the week it lands. That is the output.
| Forecast | Actual | Variance | |
|---|---|---|---|
| Opening balance | $96,000 | $96,000 | — |
| Receipts | $74,000 | $56,000 | −$18,000 |
| Payments | $68,000 | $72,000 | −$4,000 |
| Closing balance | $102,000 | $80,000 | −$22,000 |
The variance is the whole point
Nobody forecasts cash accurately, and accuracy is not the goal. What you are building is calibration — a sense of how wrong you tend to be and in which direction.
After two months of comparing, patterns appear. Customers who say the 15th pay on the 22nd. Card receipts land two days after the sale. Your supplier payments always run a little heavier than planned. Those adjustments make week three onward dramatically better, and you cannot learn any of them without recording what you expected before you found out.
Weekly time
20min
including the variance review
Useful after
6wks
when calibration starts
The output
1
the low point, and its date
Free · no account · no card
Get your 2027 budget built from your QuickBooks — P&L, balance sheet and cash flow, in Excel.
Keep it deliberately rough
The temptation is to make it more precise: every customer modelled, every expense category split out. Resist it. A forecast that takes ninety minutes gets abandoned by week four, and precision buys you very little when the underlying uncertainty is when a customer feels like paying.
Six or eight lines of receipts and eight or ten of payments is plenty. Group the small stuff. The goal is a number you trust to within a few thousand dollars, produced fast enough that you will still be doing it in March.
Act on the low point, not the average
The only number that matters is the trough: the lowest balance in the window and the week it occurs. Compare it to the floor you refuse to go below and you have a binary answer to whether anything needs to change.
If it breaches, you have weeks of warning and cheap options — chase specific invoices, move a payment, delay a purchase, draw on a facility you already arranged. The same problem found in the week it happens has only expensive options left.
Need your financials for a lender, a buyer or your accountant? Get the free financials pack →
See what an AI CFO says about your own numbers.
$99/mo, everything included. Free for 14 days, no card.