How to calculate a machine shop rate that covers your costs
Ask 3 people in a machine shop what the rate on the vertical mill is and you will get 3 answers, all of them round numbers, none of them derived. A machine shop rate calculation is arithmetic, and doing it once properly settles arguments that otherwise run for years.
What belongs in a burdened rate covers which costs are legitimately in scope. This post does the sum. One work center, 7 layers, then the division that decides everything. Every figure is illustrative. The rate this produces is what a machined part build-up multiplies against every operation.
The 7 layers that make up an hourly cost
Start with annual cost for the whole work center, not with an hourly figure. Hourly is the output of this exercise and treating it as the input is how floors end up with a rate nobody can defend.
For an illustrative 3 axis vertical machining centre, the layers run as follows. Machine ownership, taking a $180,000 purchase over a 7 year life, gives $25,714 a year. Floor space, at 120 square feet and $9 a foot, is $1,080.
Power at an average 15 kilowatt draw across 2,080 hours at $0.11 a kilowatt hour is $3,432. Maintenance, service contracts and consumables run $6,000. Workholding, fixture plates and the durable tooling that belongs to the machine rather than to a job is $3,000.
The operator at $52,000 with 28 percent burden is $66,560. Allocated overhead, meaning supervision, quality, the building and the office, is $34,000.
Adding it up before dividing it
Those 7 layers total $139,786 a year for one machine and the person on it. That number is the honest annual cost of having this work center available, and it is the numerator for everything that follows.
The reason to total it before dividing is that the layers are auditable and the rate is not. A finance person can check $180,000 over 7 years. Nobody can check $95 an hour without being shown the sum behind it.
Annual cost first, hourly second.
Two of the layers deserve a second look because they are the ones most often missing. Allocated overhead is frequently left out entirely, which produces a rate that covers the machine and not the company. Durable tooling and workholding are usually expensed and forgotten, which quietly under-recovers a few thousand a year on every machine.
Depreciation is worth a decision rather than an accident. Book depreciation serves a tax purpose. For quoting, the useful life is the number of years you will actually run the machine before replacing it, and using the tax figure because it is available is how a fully depreciated machine ends up appearing to cost nothing.
The utilisation assumption that decides everything
Now divide, and this is where most rates go wrong. The machine is available for 2,080 hours on one shift. It is not sold for 2,080 hours.
Setup that you do not charge, maintenance, the gap between jobs, the morning the program was not ready, the afternoon the material had not arrived. All of it is time the machine was available and not earning.
If you divide $139,786 by 2,080 you get $67.20 an hour, and that rate is wrong in a specific and expensive direction. It assumes every available hour is sold, so any hour that is not sold is a loss you have already promised away in your prices.
Divide by hours sold instead. At 60 percent utilisation the machine sells 1,248 hours and the rate is $112.01. At 75 percent it sells 1,560 hours and the rate is $89.61. At 90 percent it sells 1,872 hours and the rate is $74.67.
Why honest utilisation is the hardest number here
The spread between those 3 rates is 50 percent, which is larger than any argument you will ever have about overhead allocation. Everything else on this page is arithmetic. This one is a judgement about your own floor and people are optimistic about it.
Get the real number from your records rather than from memory. Total spindle hours booked to jobs over a year, divided by hours the machine was staffed and available. Most floors that have never measured it find a number lower than they expected.
Ninety percent utilisation on a single machine in a job environment is unusual. It requires work waiting at every changeover and nothing going wrong. Quoting at a rate derived from 90 percent when you run at 65 means under-recovering on every hour, on every job, permanently.
The failure is invisible because it does not show up per job. It shows up as a year where the floor was busy, every quote looked profitable, and the company made nothing.
Divide by hours sold.
There is a second reason to measure it rather than assume it. Utilisation is the one input on this page you can change. Ownership, space and overhead are largely fixed for the year, so the fastest route to a competitive rate is selling more of the hours you are already paying for, which is a scheduling problem rather than an estimating one.
What to do with a machine that is genuinely underused
A low utilisation machine produces a high rate, and a high rate can lose you the work that would have filled it. That circularity is real and it needs a decision rather than a formula.
The honest options are 3. Quote at the true rate and accept the machine stays quiet. Quote at a lower rate as a deliberate contribution decision, taking work that covers marginal cost and something toward the fixed layers. Or move the work to a machine that is already busy and price it there.
What breaks floors is doing the second one without naming it. Marginal pricing is a legitimate tactic on a specific job with spare capacity. It is not a rate, and once it becomes the rate, the fixed layers stop being recovered anywhere.
Write down which of the 3 you chose and why, on the quote. A year later somebody will ask why this part was priced at $78 an hour when the standard is $90, and the answer needs to exist.
Marginal pricing is not a rate.
The same discipline applies to a brand new machine. A press bought for one program carries its whole ownership layer against very few sold hours in year one, which produces a rate that looks absurd. Quoting that rate loses the work that would fill it. Spreading the ownership across the life you expect, rather than the first year, is the defensible answer, and it belongs in writing beside the calculation.
Feeding the rate into a quote without losing it
A rate is only useful when it meets a cycle time, and the pairing has to be consistent. If the rate was derived from hours sold, then the hours you multiply it by must be the hours you will book to the job, including setup if setup is charged.
Keep the rate per work center rather than one floor average. A grinder, a rotary transfer and a manual lathe have different ownership costs, utilisation and operators, and a blended rate prices every job slightly wrong in a direction that depends on which machines it touches. That is why a machined part build-up carries a separate rate against every operation rather than one number across the routing.
Review the layers annually and the utilisation quarterly. Machine ownership changes when you buy, overhead changes when you hire, and utilisation moves with the order book, which is the fastest moving input on the page.
The last discipline is to keep the calculation somewhere other than one person’s file. A burdened rate nobody can reconstruct is a number the floor will eventually stop trusting, and once that happens estimators start adding their own private margin on top of it.