The cash forecast a contractor actually needs
Profitable firms fail on timing, not margin. A short method for a rolling cash view built from data you already record.
A contractor can be profitable on every job and still run out of money, because the money leaves on subcontractor and payroll dates and arrives on certification dates, and nobody arranged for those to line up.
Why the annual budget does not help
An annual budget is an average. Cash is a sequence. The question is never whether the year works, it is whether the third week of next month works, and that answer is invisible in any monthly view.
Weekly buckets, rolling forward, out about 13 weeks. Far enough to act, near enough that the numbers are real.
The five lines
| Line | Source | Common error |
|---|---|---|
| Certified income | Valuations submitted and their contractual payment dates | Using the submission date instead of the payment date |
| Retention released | Contract terms per project | Forgetting it entirely until it is late |
| Subcontractor payments | Signed subcontracts and certified work | Only counting invoices already received |
| Payroll and labour | Attendance and rates | Averaging a month when the site had a push week |
| Materials and overheads | Open purchase orders plus fixed monthly costs | Missing committed orders not yet invoiced |
The two lines that catch people
The certification lag
Between submitting a valuation and receiving the money there is an assessment period, a payment period and, honestly, a habit. Model the habit, not the contract: if a client has paid at day 52 for the last four valuations, forecast day 52.
Retention
A few percent held on every certificate accumulates into a serious number that arrives long after the site closes. Firms forecast the release optimistically and then borrow to cover the gap. Put each project retention release in the week it will genuinely land, including the second moiety.
Retention plus a slow final account is the classic pattern behind a profitable firm with an empty account. Both are known in advance, which means both are forecastable.
Building it from data you already have
If costs are already posted against projects and valuations are recorded with dates, the forecast is a query rather than a modelling exercise. Committed costs give you the outflow, certified and pending valuations give you the inflow, and the payment habit per client gives you the timing.
- 1
Start from the bank balance
Actual, today, across all accounts. Not the ledger balance if they differ, and if they differ, find out why first.
- 2
Lay in the committed outflows
Payroll dates, subcontractor certificates, open purchase orders with expected delivery and terms.
- 3
Lay in the inflows at habitual dates
Per client and per project, using the observed lag rather than the contractual one.
- 4
Mark the worst week
The lowest point in the 13 weeks is the number that matters. Everything else is context.
- 5
Roll it weekly
Same day each week, ten minutes. A forecast rebuilt monthly is a report, not a control.
What to do with the worst week
- Move a valuation earlier if the work supports it. This is the cheapest lever and the most often ignored.
- Sequence a large purchase after a known receipt rather than before it.
- Agree staged payment with a subcontractor before you need to, not during the week in question.
- Arrange the facility while the forecast still shows the gap in eight weeks rather than eight days.
Questions this raises
Is 13 weeks a rule?+
It is a convention that works because it covers a full monthly billing cycle plus the payment lag. Firms with 60 day terms often extend to 16.
Do we need an accountant to run this?+
No. It is arithmetic on data you already record. An accountant should review the assumptions, not assemble the sheet.