Where the Margin Actually Goes
Most optics shops can name the jobs that went badly, but not the pattern. Common leaks are polishing cycles beyond the quoted number, scrap discovered late at final inspection after coating, coating runs repeated for spectral drift, interferometer time that was never charged to a job and new tooling bought for one order. Each is small in a single job and large across a year.
Without data by operation, the shop responds by raising prices across the board, which loses easy work and does not fix the hard jobs.
- Extra polish and test cycles beyond the plan
- Scrap found after value was added, especially post-coating
- Rerun or stripped coatings
- Unbilled metrology and engineering time
- One-off tooling and test plates absorbed into overhead
Capture Scrap by Operation and Reason
A single scrap quantity at the end of the job tells you nothing. Record scrap at each operation with a reason code, such as edge chip, scratch, figure out of tolerance, centration error, coating defect or breakage. Because cost accumulates, scrap after coating costs far more than scrap after grinding.
Use these records to set yield assumptions per process route, material and tolerance class, and feed them into the next quote.
Charge Time Where It Is Spent
Shop-floor time capture often rounds work into a few operations. For optics, record time by operation including test and rework, and make sure shared resources such as the interferometer and the coating chamber are charged to the jobs that used them. If everything goes into overhead, jobs with heavy metrology look cheaper than they are.
Even simple time stamps at start and stop of each operation are better than end-of-day estimates.
Try it here, free
Quote Margin Analyzer
Pressure-test a quote by amortizing setup across the order quantity, adding commission, and showing the price required to hit a 30% target margin.
Compare Quote to Actual by Job Class
After each job, compare quoted hours, yield and cost to actual, then group results by customer, material, tolerance class and process route. A pattern like tight-figure fused silica running a quarter over every time is a pricing rule waiting to be written. Review the report monthly with the estimator and the production lead, not just finance.
The goal is to change the next quote, not to assign blame.
How Far to Take It
A small shop with mostly repeat work can get most of the benefit from a reason-coded scrap sheet and a monthly review of the five worst jobs. Larger shops with many custom jobs benefit from ERP-level job costing and natural-language access to the results, so an estimator can ask what similar parts actually cost without asking IT for a report.
Frequently asked questions
Why do optics shops lose money on jobs that were profitable on paper?
Usually because rework, late scrap and extra metrology time were not in the quote and not visible in the cost report. A job that needed extra polish cycles or lost pieces after coating can erase the margin while standard cost still looks acceptable.
What yield should I assume in an optics quote?
Use your own history by process route, material and tolerance class rather than a generic percentage. Yield varies widely with surface figure, scratch-dig grade and substrate, and for small lots it is better to quote a start quantity that covers expected loss.
How do I cost shared resources like the interferometer?
Charge time to jobs through operation-level time capture, using a machine rate that reflects depreciation, calibration and operator cost. If metrology is rolled into general overhead, jobs that demand heavy measurement are underpriced and simple jobs are overpriced.
Netray's ERPray layer can answer quote-versus-actual questions over your live ERP data, read-only and on-prem, so estimators see real history while they quote.