Loading
Burning. The runway figure assumes this month repeats, which is the assumption worth stress-testing first.
Collections rather than invoiced value, because a forecast built on invoices raised is a forecast of what you are owed rather than what you can spend.
A thirteen-week forecast with receipts and payments by week, built from your open invoices and bills rather than from a monthly average.
Thirteen weeks is deliberate: long enough to see a problem, short enough to be worth maintaining.
Enter open invoices by due date, then adjust each by how that customer actually pays rather than by the terms. A customer whose pattern is terms plus twenty days belongs twenty days later.
Add committed payments: bills, payroll, rent, tax. Payroll is the one people forget because it feels automatic.
It will not predict what a customer will do. Adjusting for behaviour is better than using terms, and it is still an estimate.
It also cannot see a cost nobody has committed yet.
Look for the weeks where the balance dips rather than at the ending figure. A forecast that ends comfortably can pass through a week that cannot be funded.
Run a pessimistic version where your three largest receipts arrive a fortnight late. If that version fails, the plan needs changing now.
Long enough to act, short enough to maintain.
Behaviour. Terms are what was agreed, not what happens.
Yes — largest receipts late is the useful one.
Half an hour on your own figures, and an honest answer about the parts Treepie does not improve.