BAMML Executive Insights
Your Job Costing Is Telling You the Wrong Jobs Are Profitable
Most engineering and job-shop businesses can tell you which jobs were profitable last quarter. Very few can tell you whether that report is true.
The report is produced by a costing model. The model contains a labour rate, an overhead absorption basis and a set of assumptions about what a job consumes. If any of those has drifted away from reality — and they drift quietly — then the report is an accurate statement about the model and an inaccurate statement about the business.
The difficulty is that a wrong costing model does not produce obviously wrong answers. It produces plausible ones, in the wrong order.
What is actually happening
Three things drift, usually at different speeds.
The labour rate. The rate used in quoting was calculated at some point from wages, plus an allowance for non-productive time, plus employer costs. Since then wages have moved, the premium mix has changed, absence may be higher, and productive hours per employee may be lower than the model assumes. The rate in the quoting spreadsheet is rarely rebuilt — it gets escalated by a percentage, which preserves whatever error was in it.
The absorption basis. Most job costing absorbs overhead on labour hours, because that was the sensible basis when the business was labour-intensive. As equipment, compliance cost, supervision and systems have grown, overhead has become less connected to labour hours. Absorbing it on that basis then systematically misallocates: jobs with many labour hours carry overhead they did not cause, and jobs that are equipment-heavy or setup-heavy carry too little.
What is excluded entirely. Setup and changeover. Rework. Warranty returns. Idle time between jobs. Time spent on quotes that did not convert. These are real costs of doing the work, and they usually sit in overhead rather than against the jobs that caused them.
Why it matters
The consequence is not random error. It is systematic bias, and it bites in a specific direction.
Complex jobs — more setup, more changeover, more rework risk, more engineering input — tend to absorb less cost than they cause. They look more profitable than they are. Simple, repetitive, high-labour-hour jobs absorb more overhead than they cause. They look less profitable than they are.
So the business does exactly what the numbers tell it to do. It pursues the complex work, prices it competitively because the model says there is margin in it, and wins more of it. It treats the simple work as marginal, prices it defensively, and loses it to competitors.
Over a few years a business can restructure its entire order book in the wrong direction, with every individual decision fully supported by its own management accounts.
The Integrated View
| Discipline | What it should be testing | What usually happens |
|---|---|---|
| Financial Governance | Whether the labour rate and absorption basis still reflect reality | Escalated annually by a percentage |
| Operational Performance | Setup, changeover, rework and idle time, costed to the jobs that cause them | Absorbed into overhead, invisible per job |
| Workforce Governance | Actual productive hours per employee, including absence and premiums | An assumption inside the rate, rarely revisited |
| Risk & Compliance | Exposure on fixed-price work quoted off a drifting model | Not framed as a risk at all |
| Executive Decision Intelligence | Quoted cost compared to actual cost, after completion | Almost never done |
That last row is the one that matters most, and it is the cheapest to fix. Very few job shops systematically compare what a job was quoted at to what it actually cost once finished. Without that comparison, the costing model never receives feedback — and a model that never receives feedback does not improve. It only drifts further.
What good looks like
Rebuild the rate rather than escalating it. Start from actual payroll cost and actual productive hours, not from last year's rate plus a percentage. The number that comes out is often uncomfortable, which is usually a sign it is closer to correct.
Test whether labour hours are still the right absorption basis. If a meaningful share of overhead now relates to equipment, setup, compliance or supervision rather than to labour, absorbing all of it on labour hours is distorting every job in the book.
Cost setup, rework and changeover to jobs. Not perfectly — approximately is a very large improvement over not at all.
Then close the loop. Compare quoted to actual on every completed job above a threshold you set. The variance report is the single most valuable document in a job shop, and most do not produce it.
What to do on Monday
- Take ten recently completed jobs. For each, compare quoted cost to actual cost, including labour at real rates and any rework.
- Look at the direction of the variance. If the complex jobs are the ones that overran, your absorption basis is the likely cause.
- Ask when the labour rate was last rebuilt from first principles — not escalated. If nobody can say, that is your answer.
- Ask where setup, rework and idle time are recorded. If the answer is overhead, they are invisible to every job decision you make.
- Instruct that quoted-versus-actual becomes a standing report. It does not need a system. It needs somebody to do it monthly.
The objective is not a perfect costing model. It is a model that gets corrected by reality often enough that it cannot drift far.
