Building a 13-week cash forecast your board will trust
A direct cash forecast is only as good as the drivers underneath it. A practical guide to choosing drivers, setting the weekly rhythm, and measuring accuracy honestly.
The 13-week forecast has become the default horizon for mid-market treasury, and for good reason: it is long enough to see a quarter-end and short enough that the assumptions are still defensible. But most of the ones we see fail the same test — nobody can explain why last week was wrong.
Build it direct, not indirect
An indirect forecast, derived from the P&L and balance sheet, is fine for annual planning. For a 13-week horizon it is the wrong tool. You need a direct forecast: expected receipts and payments, week by week, from the underlying operational data.
That means five categories in most groups.
- Customer receipts, driven by the receivables ledger and actual payment behaviour rather than stated terms.
- Supplier payments, driven by approved invoices and your payment run calendar.
- Payroll and related taxes, which are the most predictable line you have and should never be estimated.
- Debt service — interest, amortisation, facility fees — taken from the loan schedule.
- Everything else, which should be small. If it is not, something in the first four is missing.
Drivers, not spreadsheet rows
The difference between a forecast that survives and one that does not is whether the numbers are typed or derived. If a week's receipts figure was typed, nobody can revisit the logic. If it was derived from ageing plus a collection curve, you can adjust the curve and watch the whole forecast move.
Use actual behaviour for the collection curve. If your 30-day terms are paid on average at 47 days, forecast 47. Optimism about terms is the single largest source of forecast error we see.
Lock the weekly rhythm
Pick a day and never move it. Ours is Monday morning, with actuals for the week just closed replacing forecast, a new week 13 added at the far end, and variances explained before anything else is touched.
The discipline matters more than the day. A forecast rebuilt on an irregular cadence cannot be compared to itself.
Measure accuracy, and publish it
Most teams never measure forecast accuracy, which means nobody knows whether to believe it. Track the absolute percentage error on the week-one closing balance, rolling over the last eight weeks. Publish that number on the front of the report.
Two things happen. The forecast improves, because errors become visible. And the board's confidence in it goes up, because a forecast that admits to a 4% error band is more credible than one that implies none.
Common failure modes
- Netting off intercompany movements too early, which hides real timing risk inside a single entity.
- Forecasting VAT and corporation tax as a smooth monthly line rather than on their actual payment dates.
- Ignoring the facility. Available headroom on a revolving credit facility is part of your liquidity picture and belongs in the same view.
Get those right and a 13-week forecast stops being a compliance exercise and starts being the thing you make decisions with.