The examples and architectures in this series are my personal educational, non-commercial experiments; they are not an offer of IT or consulting services.
Every forecast is wrong. The useful ones are wrong early enough to do something about it. Post #14 priced the days a gap stays open, and the cheapest way to shorten them is to see the gap before it arrives. The question is how far before.
The 13 weeks described below are a composite built from patterns I've seen repeat across finance teams over the years. The weeks, the gap and the confidence levels are illustrative; they don't describe a specific company or forecast.
One gap, three ways of looking
Lead time is what the forecast is for
A one-week view shows the gap in the week it arrives, which is the day-of picture from Post #14. A four-week view shows it three weeks ahead. A thirteen-week view shows it five weeks ahead, long enough to renegotiate a payment, chase a receipt, or arrange financing on terms rather than in a hurry.
| View | Gap first visible | Warning |
|---|---|---|
| 1 week | Week 6 | 0 weeks |
| 4 weeks | Week 3 | 3 weeks |
| 13 weeks | Week 1 | 5 weeks |
Illustrative composite from Post #15. The weeks, balances and the gap are made-up units and do not describe a specific company or forecast.
What the 13 columns have to contain
It's the same set of signals from the last five posts, laid out by week. One row per signal, thirteen columns, and one new line at the bottom.
| Row | What goes in it | How sure it is |
|---|---|---|
| Opening balance | Last week's closing balance | A fact |
| Receipts: confirmed | Invoices with an agreed payment date or a lifted credit hold | High |
| Receipts: expected | Invoices due, discounted by how often that customer pays late (Post #12) | Medium |
| Payments: committed | Payment run, payroll, tax, debt service | High |
| Payments: movable | What can shift within its terms (Post #11) | Medium |
| Minimum buffer | The line you don't want to fall below (Post #10) | Set by policy |
| Closing balance, first week below buffer | The result, and the week a decision is needed (Posts #13 and #14) | Follows from the rows above |
What "rolling" actually means
When a week ends, it drops off the left and a new week 14 enters on the right. The horizon never shrinks. You always see thirteen weeks ahead, instead of twelve, then eleven, then a scramble at quarter end.
| What it looks like | What's actually true |
|---|---|
| A longer horizon means a more accurate forecast | A longer horizon means an earlier signal and a less accurate number. The point is to make that trade on purpose |
| The forecast should be right | It should be readable about what it knows and what it assumes: weeks 1 to 2 come from payment runs and receivables, weeks 3 to 6 from patterns, weeks 7 to 13 mostly from assumptions |
| Rolling means rebuilding the sheet every week | Rolling means one week drops off and one is added. The structure stays |
| The forecast missed, so it's broken | A forecast that missed by two weeks and warned you five weeks ahead has done its job. The sheet that never misses is often the one nobody updates |
A forecast isn't judged by whether it hit the number. It's judged by how much earlier it showed the gap.
A test you can run this week
Take the last three gaps you only noticed on the day a payment was due. For each one, ask what a thirteen-week view built four weeks earlier would have shown, and with how much confidence. The answer often sits in the structure of the sheet, not in better arithmetic.
I haven't put a script behind this one. A 13-week forecast is a layout rather than an algorithm, and the layout above is the part worth copying into your own sheet.
Follow the series
Post #16 looks at the payment that didn't look like the others, and what a normal week has to look like before an odd one stands out.