A real case
A run of stainless sensor bodies came off the floor three hours over the quoted machine time, and the repeat order is already in. Read the ledger line by line, see what each line's confidence rests on, correct the one that was wrong with what actually happened, and quote the repeat on your own shop's numbers.
The situation
The 316L sensor bodies came off the floor in March: a hundred and twenty off, quoted at eleven hours of machine time, run in a shade over fourteen. Everybody in the building knows the number was wrong. Nobody in the building can say which part of it was wrong — whether the second setup took longer than anyone allowed for, whether the finishing pass on the bore was quoted off a cycle time nothing had ever measured, or whether the outside passivation simply sat in a vendor's queue for four days. Now the same customer has sent the June repeat: same family, two hundred and fifty off, one extra port and a tighter bore tolerance. This page is about what happens after the quote — the loop between what you sent in March and what you send in June — and all of it is entered by hand on the job you already ran. Nothing here reads hours out of your job-costing system, and nothing infers what happened behind your back.
Step by step
A quote here is not one number with a margin on it. Every cost line traces to a named rate, a posture setting or a rule in the engine, so the three hours have somewhere to land: the setup line, the machining line, the handling and queue time between departments, or the outside finishing step. Drift that had no address gets one, which is the whole difference between correcting an estimate and adding a percentage to it.
Every result carries a confidence band that names its basis: a published prior, your own shop-calibrated actuals, a deterministic rule, or a flag that a human needs to look at it. On a job your shop has never run before, the lines resting on a published prior are the ones borrowing somebody else's numbers, and they are the first place to look when quoted and actual pull apart the same way twice.
Contour-parallel roughing, pocket finishing and adaptive or trochoidal clearing are planned from the same voxel removal simulation that reconciles cycle time, flags unreachable features and calls out leftover stock and gouge depth — so the machining line reflects an actual toolpath rather than an average taken from parts that were not this part. On the sensor body that is where an eleven-hour figure either survives contact or does not.
Suggesting a correction on a cost line — what that step really took when the job ran — feeds a calibration record, and confidence bands tighten as your own shop's actuals come in. The correction goes against the line it belongs to, not against the total, so the next quote does not inherit a shop-wide fudge factor that nobody can test or explain — a blanket uplift overprices the jobs that were always fine while still underpricing the ones that were not, and leaves the lines that were right alone only by accident. Whoever closes the job enters it; no reader pulls hours out of your job-costing system, and nothing is reconciled between systems for the correction to count.
The same family comes back at two hundred and fifty off, and the line that was wrong in March now rests on your recorded outcome and says so. A line that still has no history behind it still names the published prior it is standing on rather than dressing a borrowed figure up as yours. A should-cost curve runs at every quantity, so the jump from a hundred and twenty to two hundred and fifty is priced rather than extrapolated by hand.
What you get
What you end up holding is a June quote whose machining line rests on a March job you actually ran, plus a plain statement of which lines still do not — and a customer conversation you can have, because a price that moved between two quotes moved somewhere specific: a rate, a route step, a corrected cycle time. The three hours stop being a story about a bad month and become a line item somebody fixed. Partimation is decision support, not a guarantee — every result names the rule behind it, and the calibration is only as good as the corrections your shop actually enters. The one thing to be clear about before you start: this loop runs on corrections typed against the line by the person closing the job, not on an automatic feed from wherever your actual hours are recorded today.
Already shipped
Built into CostDFM.
Built into CostDFM.
Built into CostDFM.
Built into CostDFM.
Built into MachDFM.
Built into MillDFM.
Built into RouteDFM.
FAQ
Corrections go against the line they belong to — the setup, the machining time, the handling and queue time between departments, or the outside finishing step — because every line already names the rate, posture setting or engine rule that produced it. Three hours that used to be absorbed into a total has an address, and correcting that address is what feeds the next quote.
Corrections are typed against the cost line by whoever closes the job. No feed reads hours out of the system where your actuals are recorded today, and nothing has to be reconciled between two systems for a correction to count — which also means nothing is inferred behind your back.
Contour-parallel roughing, pocket finishing and adaptive or trochoidal clearing are planned from the same voxel removal simulation that reconciles cycle time, flags unreachable features and calls out leftover stock and gouge depth. The quoted machining time therefore comes off an actual toolpath on the geometry in front of you.
Confidence on every result names one of four bases: a published prior, your own shop-calibrated actuals, a deterministic rule, or a flag that a human needs to look at it. A line resting on a published prior is borrowed until your shop's own outcomes replace it, and it says so before the job runs rather than after.
A blanket uplift overprices the jobs that were always fine while still underpricing the ones that were not, and nobody can test it because it is attached to nothing. A correction entered against the cost line that was actually wrong tightens the bands on the jobs after it and leaves the lines that were right alone.
Where this fits
You run a CNC or contract machine shop, twenty to fifty RFQs land a week, and one senior estimator gates all of them. Upload the model or the drawing you were already sent, get the route built and ranked, and send a price where every line names the rate or rule behind it.
The quote said eleven hours, the floor ran fourteen, and the next quote for the same family said eleven again. Every cost line names the rate or rule that produced it and states what its confidence rests on, so a correction lands on the line that was wrong and tightens the bands on the jobs after it.
Get started
Send a model or a drawing and book a walkthrough — or start free and run one through yourself.
Book a demo
Tell us about your shop and we'll set up a walkthrough on a real part.