Build the Variance Report by Operation
Start with a closed-job report that lists, for each work order, the estimated and actual hours for each operation: kit cut, layup, bag, cure, demold, trim, inspect, finish and pack. Add material estimated versus issued, scrap quantity and reason, and outside processing cost. Sum to a job margin, but use the operation view to find the cause.
A job that is ten percent over in total might be thirty percent over on layup and under on trim, which points to a quoting method rather than a shop efficiency problem. If your ERP stores only total hours per job, correcting how operators clock labor is the first project.
- Estimated versus actual hours by operation
- Quoted versus actual material yield and scrap
- Cure load fill percent and cure hours charged
- Rework and repair hours with reason code
- Tooling repair and replacement cost by program
Typical Causes, Ranked by How Often They Hide
Layup variance is the most common culprit because the estimate is built from ideal minutes per ply. Real parts have wrinkles to rework, debulk waits, and operators learning a new program. Next comes material: yield worse than quoted, or prepreg expired in the freezer or at the table.
Scrap from vacuum leaks and cure errors shows up as extra material and repeated cure time. Underfilled autoclave loads inflate the real cost per part. Finally, engineering changes mid-program change the ply book without a price change.
Each cause calls for a different fix, which is why a single margin figure per job is not enough.
Separate Learning Curve From Estimating Error
New programs cost more on the first parts because operators learn the drape, the tooling is new and the process is still tuned. If you treat these hours as pure variance, you will blame the estimator for a normal curve. Plot hours per part against cumulative units for each program.
A falling curve that levels near the estimate means the quote was reasonable but ignored learning. A flat curve well above the estimate points to a wrong labor standard. Build learning allowances into quotes for new programs, and use steady-state hours for repeat work.
Try it here, free
Job Costing Worksheet for Global Shop Solutions Users
Compare estimated and actual labor and material for each job and see the cost variance and variance percent. Good for monthly job reviews and quote feedback.
Feed Actuals Back Into Standards
A report nobody uses is just overhead. Set a monthly review where estimators, planners and the floor lead look at the five jobs with the worst variance and the five best. Update labor standards per ply by family, yield assumptions by roll width and the standard allowance for scrap.
Record changes and the effective date so later comparisons are fair. 35. This habit alone often matters more than any software purchase.
Where AI Fits and Where It Does Not
Questions like which part families run over on layup hours this year, or what is our real yield on 60 inch carbon fabric are hard to answer without a report project. A private natural-language layer over your ERP can answer these from live data, show the underlying query and respect user roles. It will not fix poor data, so check that labor postings and material issues are accurate before trusting any answer.
If you run only a few jobs a month, a manual review of each job with the foreman may be quicker than any tooling.
Frequently asked questions
How do you calculate job margin for composite parts?
Subtract actual material, direct labor by operation, a share of cure cycle cost, outside processing, scrap and tooling charges from revenue, then compare with the estimate by line. Use operation-level actuals so you can tell whether the shortfall came from quoting assumptions or shop execution.
Why do composite jobs run over estimated hours?
Common reasons are layup minutes per ply set too low, rework from wrinkles or bag leaks, learning curve on new programs, underfilled cure loads and design changes that were not repriced. Operation-level tracking reveals which one is dominant, so fixes target the actual cause.
Can a mid-size ERP report actual versus estimated by operation?
Yes, if the estimate is stored as a routing with operation standards and labor is clocked per operation. Systems like SyteLine, Epicor, JobBOSS and E2 can report this. Where operators clock only at job level, fix data capture first, because any report built on it will mislead.
Netray's ERPray lets estimators and plant managers ask margin and variance questions of their own ERP in plain English, with the query shown and ERP roles enforced.