Your Most Expensive Cost Line Doesn't Appear in Your Accounts
Ask most operators in a project-based business what their biggest controllable cost is and they'll say labour, or subcontractors, or materials. All defensible answers. But sitting inside those numbers, unnamed and unmeasured, is a cost that in many businesses runs into the hundreds of thousands of dollars a year.
Rework.
The cost is real, the record isn't
When something has to be redone — a wall set out incorrectly, a service installed to a superseded drawing, a finish that didn't meet spec — the response is operational. Get someone back on site, put it right, move on.
The financial trail follows the same path as the original work. Hours go to the job's labour code. Materials go to the job's materials code. Plant hire, travel and supervision all follow. Every dollar is captured accurately against the job.
And every dollar is invisible as rework.
The information exists in people's heads. The site supervisor knows the crew went back twice. The project manager knows why. But by the time it reaches a monthly report, it has been absorbed into a single line that reads "labour: over budget".
What that does to your decision-making
This is where the cost compounds, because the wrong conversation happens next.
A job comes in under margin. The obvious explanations are examined: the estimate was too tight, the labour rate has moved, productivity was poor, the client was difficult. One of those gets accepted and something gets adjusted — usually the estimating allowance, which gets nudged up on the next similar job.
So the business prices in the cost of its own rework and calls it a market rate. It becomes less competitive on tender, wins less work, and the underlying problem is now funded rather than fixed.
Meanwhile the actual cause — a detailing issue that recurs on a particular type of job, or a supervision gap on a particular crew, or a supplier whose material consistently arrives out of tolerance — remains unaddressed, because nobody has the data to isolate it.
The estimating feedback loop breaks
There's a second-order effect worth naming. Estimating relies on historical actuals. If your actuals include rework you don't know about, your cost base is inflated by an amount that varies unpredictably from job to job.
That means your estimating variance widens for reasons that look random. Two similar jobs return different margins and nobody can explain it convincingly. Confidence in the numbers erodes, and estimators start applying judgement over data — which is how a cost base drifts.
Clean rework out of your actuals and your estimating accuracy improves immediately, before you've fixed a single quality issue.
Coding rework is a small change
This is not a call for a quality management system. Those have their place, but they usually arrive as documentation exercises and rarely produce cost data anyone uses.
What produces useful data is a single additional cost code, applied at the point of timesheet entry, with a mandatory cause selected from a short list. Five options is enough:
Design or documentation error
Supervision or setout
Materials or supply
Workmanship
Client-directed change
The last one matters, because a proportion of what people call rework is actually unpriced variation work. Separating those two is worth doing on its own — one is a cost problem, the other is a revenue problem.
What you'll learn in a quarter
Within three months you'll have a number: rework as a percentage of labour cost. That number is your baseline, and it is almost always higher than management expects.
Within six months you'll have concentration. Rework rarely distributes evenly. It clusters — around a job type, a client, a design consultant, a crew, a supplier. Once you can see the cluster, the intervention is usually obvious and often inexpensive.
The businesses that reduce rework meaningfully aren't the ones with the best quality manuals. They're the ones that can see where it's happening while it's still happening.
You can't manage a cost you can't name.