Every finance leader has built a 13-week cash forecast at least once, usually during a crisis. The format survives because it answers the only question a board really asks about cash: will we be fine next quarter, and how sure are we?
Why thirteen weeks
A quarter is long enough to see payroll cycles, tax dates and large vendor runs, and short enough that most inflows are already contracted. Beyond that horizon you are forecasting the business; inside it you are forecasting cash.
Where spreadsheets break
The model itself is simple. What breaks is the input:
- Balances are copied from bank portals on Monday and are stale by Tuesday.
- Intercompany transfers are counted twice, once leaving and once arriving.
- FX is applied at whatever rate the analyst used last month.
With three entities that is an annoyance. With thirty it is a different number every time someone opens the file.
A better loop
Start from live balances, not a copy. Classify flows by rule (payroll, tax, AP runs, customer receipts) so actuals roll into the forecast automatically. Eliminate intercompany flows at the group level and revalue foreign balances daily.
The goal isn’t a perfect forecast. It’s a forecast whose errors you understand.
Review variance weekly: which category missed, by how much, and why. After a quarter of this, most teams find their forecast error drops below 3% at the four-week horizon.