Ask a plant manager how a particular job performed and, in most factories, the honest answer is: we will know at month-end. By then the job has shipped, the margin is spent, and the costing is a post-mortem. The information arrives precisely too late to be useful.
Estimated versus actual, reconstructed weeks later
The standard pattern is that a job is quoted against an estimate, run on the floor, and reconciled long afterward when the accounts close. Material, labour, scrap and rework are gathered by hand and compared to the estimate. If the job lost money, you learn why after there is nothing to be done about it. The lesson is expensive and it is always retrospective.
What "live" job costing changes
The alternative is to post actual costs against the work order as the job moves. Material drawn, labour booked, scrap recorded, rework attributed — each landing on the job in near real time rather than at reconciliation. The moment the picture becomes live, two things follow.
Drift becomes visible mid-job
A job trending over its estimated cost shows up while it is still open. That is the window in which you can act — tighten a process, escalate a material issue, renegotiate a change — instead of absorbing the loss and discovering it later.
Predicted landed margin
With enough of the job posted, you can project where it will land. A job that is going to close below margin at, say, 40% completion is a job you can still influence. The same fact at invoice is merely a record.
Attributing scrap and rework where it belongs
Scrap and rework are usually absorbed into a general overhead and forgotten. Attributed instead to the machine, shift, operator or process that produced them, they stop being an anonymous cost and start being a solvable problem. Patterns emerge — a particular setup, a particular material, a particular time of day — that no month-end summary would ever reveal.
From post-mortem to early warning
The shift is not really about accounting. It is about when you learn things. Costing that arrives at month-end can only explain the past. Costing that posts as work happens can change the outcome. The same data, delivered at a different time, becomes a different kind of tool entirely.
Building it over what you already run
None of this requires replacing your shop-floor systems. It requires reading from them — the ERP, the machine logs, the material issues — and capturing directly where no system exists, such as scrap and rework at the point they happen. The goal is one live picture of every open job, so that "how is this job doing?" has an answer today rather than next month.