Anonymised, illustrative composite. A British Columbia civil/earthworks subcontractor watched a healthy-looking job slide toward break-even and traced the whole slide to two cost codes.
At a glance
The job looked profitable at bid stage and looked worse every month after. Job costing showed cost climbing steadily, but progress billings stayed almost flat against the original estimate. Nobody had gone back to remeasure the two cost codes actually driving the site’s volume.
The contract was a CCDC 4 – 2023 Unit Price Contract, under which CCDC states the work is priced for a “pre-determined, fixed amount for each specified unit of work performed,” with the total price determined by multiplying the unit price by the “actual, measured quantity” of work performed. The crew’s cost tracking had been reconciling actual cost against the estimated quantities from the bid takeoff, never against a field-verified re-measurement — so as actual volume ran ahead of the estimate, cost rose with it but billing did not.
Common excavation was budgeted at 4,200 m³ at $38/m³ ($159,600). A field survey showed 4,950 m³ actually excavated. At $38/m³, that is $188,100 — $28,500 more than what the estimate had been billing against.
Granular backfill was budgeted at 1,800 t at $22/t ($39,600). The scale tickets totalled 2,150 t actually placed. At $22/t, that is $47,300 — $7,700 more than the estimate.
Together the two codes account for $36,200 of quantity the crew had genuinely placed and genuinely paid crew and material cost for, but had never billed, because nobody had submitted the re-measurement the contract entitles — and requires — the subcontractor to bill against.
CCDC 4 does not pay on the estimate; it pays on actual, measured quantity. That single clause is the whole rule, and it cuts both ways — a unit-price subcontractor who underperforms an estimated quantity is only owed for what was actually done, but one who exceeds it is owed for the excess too, provided the quantity is measured and submitted. The margin fade was not a pricing problem; it was a measurement-and-billing problem sitting on top of an otherwise correctly priced contract.
The subcontractor submitted survey-verified re-measurement on both codes and recovered the $36,200 as a progress billing adjustment, with the field survey and scale tickets as backup. Every subsequent unit-price code on the job now gets re-measured monthly rather than left against the original takeoff. On tracking the codes themselves, see unit price contracts and quantity risk. The related overhead-allocation problem is in how a job costing rebuild exposed unbilled shop overhead, and the same firm’s progress-billing reconciliation is in what an eleven-point WIP gap turned up on another contract.
Had the crew never re-measured, the $36,200 of genuinely performed, genuinely paid-for work would simply have been lost — a CCDC 4 unit-price contract pays on submitted, measured quantity, not on what a crew can later prove it did. An unsubmitted re-measurement is not a claim waiting to be made; it is revenue that was never billed and, past a certain point, becomes difficult to substantiate at all once site conditions change.
Track the ratio of actual-to-budgeted quantity on every unit-price code monthly, not only when a job's overall margin looks wrong. Excavation running 4,950 ÷ 4,200 — 18% over budgeted volume — and backfill running 2,150 ÷ 1,800 — 19% over — were both a visible, calculable early warning; the margin fade this case caught only became visible months after the quantities themselves already told the story.
Every other line item on the job’s cost sheet was tracking close to its original estimate; the fade traced entirely to the two codes where actual site conditions produced meaningfully more volume than the bid takeoff assumed. That narrowness is itself useful information: a margin problem spread evenly across every cost code usually points to a pricing error at bid stage, while a margin problem concentrated in one or two codes, as here, usually points to a measurement-and-billing gap on a contract that was priced correctly in the first place. The subcontractor's monthly re-measurement schedule now flags any code running more than 10% over its budgeted quantity for review, rather than waiting for the job's overall numbers to look wrong.
The original $38/m³ and $22/t unit prices were not the problem — both held up as fair pricing once the correct quantities were billed against them. That is worth stating plainly, because a margin-fade investigation that starts by re-checking unit prices, rather than the quantities behind them, spends time on the wrong half of the contract. A unit-price job's margin depends on two independent numbers, price and quantity, and this one confirmed the price was never in question.
A 30-minute call is enough to tell you whether AI pays for itself here.
