Your numbers
This job
Worth reading before this quote goes out
Where the money goes
| Cost line | This lot | Per part | Scales with |
|---|---|---|---|
| Direct cost | |||
| Overhead | |||
| Total cost |
The same part at every lot size
Recomputed from scratch at each quantity, so the material steps and the amortised setup are both real rather than interpolated.
| Lot | Starts | Stock units | Flat work per part | Cost per part | Price at target | Margin at your quote |
|---|
How this is worked out
Material is a step, not a rate
A bar that yields 53 parts costs the same whether you cut 31 or 53 out of it. So the material figure here is the number of whole bars or sheets you have to buy, not the weight of the finished parts — and when the stock you are buying holds more parts than you are starting, the tool says so, because that material is already paid for.
Scrap is an integer
To ship 10 good parts at a 3% fallout you start 11, which is a 10% hit and not a 3% one. The shortcut of multiplying by one plus the scrap rate understates it at every quantity: at 10% fallout, 100 good parts need 112 starts. This page uses the yield form and rounds up to a whole part.
Setup, programming and first article are per lot
They cost the same at 5 pieces as at 500, which is the entire reason a short run is expensive per part. Programming is spread over the parts you expect off the program rather than only this lot, because a reorder owes no new CAM time. Whether to bill that block as its own line or fold it into the part price is a real decision with consequences either way — setup fee or in the part price works through what each choice does to your quote and your reorders.
An outside-process minimum behaves like a second setup
Anodising at $2.10 a part with a $75 lot minimum is $15.00 a part on a 5-piece run. The minimum, not the rate, is what prices a short run, and the tool charges it on the parts you send out rather than the parts you ship — a plater does not refund the ones you scrap afterwards.
Already quoting out of a spreadsheet?
Most job shops are, and it is a reasonable thing to do — a sheet holds the history, the customers and the estimates you built over years, and this page holds none of that. What a sheet makes easy is the four costing errors above, all of which move the cost the same way on short runs. This calculator vs. quoting in a spreadsheet sets out what each tool is for, and shows the same $38 part price earning 57.4% margin at 50 pieces and losing 19.3% at 10.
What it does not do
It does not know what your market pays, what your material is worth today, what a reasonable cycle time is for your machine, or what your competitor quoted. Every number above comes from your own invoices, payroll, tooling records and machine monitor. It is arithmetic on your figures, not a benchmark.
Questions shops ask
How do you calculate a CNC machining price per part?
Add the material actually consumed, the machine and operator time for the cycle plus handling, the perishable tooling worn making the part, and any outside process. Then add the work that happens once per lot no matter how many parts you make — setup, program proveout, first-article inspection — and divide that by the lot size. Add your overhead as a percentage of that direct cost, and the result is your cost per part. Price above it by the margin you need. This calculator does that arithmetic on your own numbers; what the customer will pay is a separate question it cannot answer.
Why is a 5-piece order so much more per part than 500?
Because setup, programming and the first-article report cost the same at either quantity. Four hours of dialling in a fixture and proving out a program is four hours whether you then cut 5 parts or 500. Spread over 5 pieces that can exceed the machining itself; over 500 it rounds to cents. Quoting both off one price list is the most common way a job shop loses money on the short run, which is why the break table on this page costs the same part at every standard lot size.
How much material do I actually pay for on a machining job?
Whole stock units, not the weight of the parts. A 12 ft bar that yields 53 pieces costs the same whether you cut 31 out of it or 53, so material cost per part is a step function rather than a rate: the 32nd part costs nothing in material and the 54th costs a whole new bar. This calculator computes yield from the bar length less your chuck grip and end loss, divided by the part length plus the parting cut — or from a plain rectangular nest for plate, tried both ways round — and then charges you for the bars or sheets you actually have to buy.
How should I handle scrap allowance when quoting?
As an integer, because that is what it is. To ship 10 good parts at a 3% fallout you do not start 10.3 parts, you start 11 — which is a 10% hit on material, machine time and outside-process cost, not 3%. The common shortcut of multiplying the quantity by one plus the scrap rate also understates it at any quantity: at a 10% fallout, 100 good parts need 112 starts, not 110. This calculator uses the yield form and rounds up, and tells you the effective rate when the rounding makes it materially worse than what you entered.
Should I bill setup as a separate line or bury it in the part price?
Either works, and the calculator shows the consequence of each. A separate setup line keeps the per-part price comparable between quantities and matters most on repeats: a second order off the same program owes no new programming and often a shorter setup, so a shop that buried those in the per-part price has to explain why the reorder is cheaper. Enter what you actually bill as a setup fee and the break-even figure adjusts to only the flat work the per-part price still has to carry.
Does anything I type here get uploaded?
No. Every calculation runs in this browser, and nothing you enter is sent anywhere — there are no accounts, no cookies, and no analytics vendor. Your shop rates and a customer's part dimensions never leave the device. If you use the save button, your numbers are written to this browser's own local storage on this device only, and the forget button deletes them. Anonymous feature counts are stored the same local way and can be switched off.