Cash flow visibility for small contractors: seeing the gap before it hits
Contractors do not fail because the jobs lose money. They fail because the money arrives later than the wages do. A firm can be profitable on paper and insolvent in practice, and the gap between those two states is usually visible six weeks out, if anyone is looking. Most small contractors are not looking, because their cash flow picture lives in a spreadsheet updated when someone has an evening free.
The two raw materials for a proper picture already exist in every contracting business: the payment calendar the Construction Act imposes on the contract, and the committed costs sitting in orders, subcontracts and timesheets. Neither changes daily on a whim. Both are exactly the kind of structured, deadline-driven data that automation handles better than people do.
Why do profitable contractors run out of cash?
Because construction payment is asymmetric by design. You pay labour weekly and merchants on 30 days, but you are paid monthly in arrears, after an application, a notice period and a final date that can sit 45 days or more from the work being done. Add one under-certified application or one client paying late, and the timing gap swallows the margin.
The failure is rarely a surprise in the data. It is a surprise to the director, because the data was never assembled in one place. The application went in late, the payment notice was never checked against it, the pay less notice arrived on the last permitted day, and the first anyone senior heard was the bank balance. Every one of those steps had a date attached, known weeks in advance. This is the real cost of doing nothing: not a dramatic loss, just a slow bleed of position that was visible the whole time.
What does the Construction Act payment calendar actually say?
The Housing Grants, Construction and Regeneration Act 1996, as amended, gives every construction contract a payment skeleton. Contracts set their own periods within it, and where they are silent the Scheme for Construction Contracts fills the gaps. The shape is always the same:
| Event | Typical timing | What it means for you |
|---|---|---|
| Application for payment | Contract date each cycle, often monthly | Miss it and you can wait a whole extra cycle for the money |
| Payment due date | Fixed by the contract, or by the Scheme where silent | The anchor every other deadline hangs off |
| Payment notice | Not later than 5 days after the due date | The payer states what they intend to pay; check it against your application |
| Pay less notice | A prescribed period before the final date (7 days under the Scheme) | The last legitimate moment the number can drop; after this, silence means the notified sum is payable |
| Final date for payment | Fixed by the contract (17 days after the due date under the Scheme) | Money in the bank, or a statutory right to suspend and to interest |
Every row is a date. None of them requires judgement to compute; they require someone to know the contract terms and count days. That is a calendar problem, and calendar problems are what machines are for. A system that knows each contract's periods can tell you every Monday which applications are due this week, which payment notices should have landed, and which pay less deadlines pass in the next seven days. The Small Business Commissioner exists in part because small firms tolerate late payment they were entitled to challenge; most of that tolerance is really just nobody watching the dates.
Why is a spreadsheet updated monthly not enough?
Because the update cycle is longer than the damage cycle. A monthly spreadsheet shows you the iceberg after the hull is open.
The information that moves your cash position moves weekly: an order placed, a valuation certified short, a subcontractor's application arriving bigger than expected, a client slipping the due date. A monthly snapshot averages all of that into a number that was true two weeks ago. Zero-click automation is the right pattern here: the forecast should rebuild itself when the underlying data changes, not when someone remembers to open the file.
There is also a quieter input: what gets said in meetings. The client mentioning a funding drawdown date, the QS flagging a contra charge, the PM agreeing to bring a package forward. Those are cash events, and they evaporate unless meeting notes are captured and mined rather than scribbled and lost.
What does forecasting from committed costs look like?
Simpler than the phrase sounds. You are combining three feeds you already have.
First, the receivables side: each live contract's application cycle and notice dates, plus a realistic lag assumption per client based on how they have actually paid, not how the contract says they will. Second, the committed cost side: purchase orders, subcontract orders and payroll, each with its own payment terms, which together describe money already promised out of the door. Third, the pipeline: awarded work not yet started, at a sensible start date.
Line those three up week by week and the gap, if there is one, appears on the screen weeks before it appears at the bank. Illustrative arithmetic makes the point: a firm turning over £2m a year is moving roughly £40,000 a week; a client slipping one monthly payment by three weeks puts a £100,000-plus hole in a specific week you can name in advance. Knowing the week is the difference between arranging cover calmly and ringing the bank on a Friday. RICS guidance treats cash forecasting as a core discipline of commercial management for exactly this reason; it is not an accounting nicety.
Where do you start?
Not with software. With a list: every live contract, its application date, its due date, its notice periods and its final date, in one place, owned by one person. That alone catches the missed application, which is the single most expensive routine failure in small contracting.
Then automate the calendar, because humans are bad at deadlines that repeat forever, and machines are not. AI Metric builds this kind of pipeline for contractors, quietly, from the systems already in use. Cash stops being a monthly fright and becomes a number you manage, which is all it ever needed to be.