Job costing on the factory floor: measured versus guessed
Ask a job shop how it prices and you will hear a rate per hour plus material plus a margin. Ask where the rate came from and the answer is usually a calculation someone did once, before the machine list changed, the power tariff went up and the product mix moved.
2 min readOadbox
Four inputs, two of which are guessed
A job costs material, machine time, labour and overhead. Material is known — it was issued. Labour is roughly known. Machine time and overhead allocation are where the guessing happens, and they are usually the larger half.
The consequence is not that quotes are wrong on average. It is that they are wrong per job — the simple work is overpriced and lost, the difficult work is underpriced and won.
Measure setup separately from run
Setup time is largely fixed per job; run time scales with quantity. A shop that costs them together will systematically misprice small batches, which is exactly the work that fills the schedule when demand softens.
Separating the two is a small change in how operations are reported and a large change in what the quote looks like at quantity ten versus quantity a thousand.
Rejection has a cost and an owner
Scrap and rework are the difference between the theoretical cost and the actual one, and they are the input most often left out. A five per cent rejection rate at the final operation carries the full accumulated cost of every prior operation.
- Record rejection at the operation where it was detected and where it was caused.
- Cost rework as an operation, not as an adjustment.
- Track rejection by part, machine and reason — the three axes that lead anywhere.
- Feed the rate back into the quote, so pricing reflects the process you have.
Close the loop back to the quote
Actual costing is only useful if the actuals reach the person quoting the next job. In most shops that feedback loop does not exist — costing is a finance exercise done monthly, and quoting is a sales exercise done daily.
A monthly review of quoted versus actual on the ten largest jobs, with the estimator in the room, is a low-tech version of the same loop and it corrects more error than any amount of software.
Written by the Oadbox team. Something here not match how it works in your business? We would genuinely like to hear it — connect@oadbox.com.