Low volume CNC machining is the band between “one part to see if it fits” and “a quantity large enough to justify a mould”. In that band the arithmetic is different: the one-time work of programming, fixturing and proving the first part is spread over very few pieces, so unit cost is dominated by something that has nothing to do with how long the machine runs.
This guide sets out how that trade-off actually behaves, with a worked amortisation calculation you can repeat with your own numbers, the levers that move speed and the ones that move cost, and an honest account of when low volume machining is the wrong answer entirely.
Key takeaways
- Unit cost in a small batch follows C(n) = S/n + v, where S is the one-time setup and v is the per-part cost. Everything about low-volume pricing follows from that single equation.
- If setup equals the machining cost of eight parts, unit cost falls from 9v at one piece to 1.8v at ten — about 90% of the total achievable saving is captured by the tenth part. Going from 50 to 500 recovers very little.
- Time to first part is usually set by material availability, programming and fixture design, not spindle hours. MW+ stocks 70+ materials, which removes the most common delay.
- Buying speed compresses queue time and overlap; it does not compress cure times, heat treatment, third-party plating or inspection. Know which part of your schedule you are actually paying to shorten.
- Loosening a tolerance that has no function is the cheapest lever available. General machining at MW+ is held to ±0.01 mm against ISO 2768-m, with ±0.005 mm and ±0.001 mm reserved for features that need them.
- MW+ takes single-piece orders with no minimum order quantity: 48-hour express, 3–5 business days standard prototype, 10–15 business days volume production.
- What counts as low volume, and why does the band matter?
- What actually drives cost in a small batch?
- Worked example: how fast does setup amortise?
- What actually sets the time to first part?
- When should you pay for speed, and when for cost?
- Which design decisions move both cost and lead time?
- When is low volume CNC the wrong answer?
- How do you buy a small batch without paying a penalty?
- Frequently asked questions
What counts as low volume, and why does the band matter?
There is no standards body definition. In practice the useful boundary is economic: a batch is low volume when the one-time cost of preparing to make the part is comparable to, or larger than, the cost of making it. Below that line the decisions that matter are about setup and scheduling. Above it, they are about cycle time and capacity.
| Band | Typical quantity | What it is for | What dominates the price |
|---|---|---|---|
| Proving | 1–10 | Form, fit and function checks; a physical model of a CAD assumption | Programming, fixturing, first-article inspection |
| Pilot | 11–50 | Field trials, certification builds, early customer samples | Setup still dominant, but amortising fast |
| Bridge | 51–200 | Supplying real demand while tooling for another process is made | Cycle time and material begin to dominate |
| Pre-production | 201–500 | Proving the process at something close to production rhythm | Cycle time, material, secondary operations |
The band matters because advice that is correct in one is wrong in the next. Consolidating two orders of five into one order of ten is transformative. Consolidating two orders of 250 into one of 500 usually is not — by then the setup is already spread thin and you are buying inventory rather than saving money.
What actually drives cost in a small batch?
Buyers tend to ask for a machine-hour rate. At low volume that is the wrong question, because the machine is not the largest line. Cost separates into work done once and work done per part.
| Cost element | Once per job, or per part? | What moves it |
|---|---|---|
| CAM programming and toolpath proving | Once | Number of distinct features, multi-axis interpolation, number of orientations |
| Fixture design and workholding | Once | Whether the part can be held in a vice or needs a soft jaw, a plate or a bespoke fixture |
| First article verification | Once | Number of characteristics on the drawing, not the number of parts |
| Material stock | Per part, plus a purchase minimum | Billet envelope, alloy availability, minimum mill quantity under specifications such as ASTM B221 |
| Cycle time | Per part | Volume of material removed, tolerance, surface finish, number of tool changes |
| Deburring and finishing | Per part | Edge count, cosmetic requirements, coatings |
| Inspection after the first article | Per part or per sample | Sampling plan, whether a full dimensional report is required per piece |
| Packing and logistics | Mostly per shipment | Consolidation, incoterms, whether parts ship as they finish |
Two of those rows explain most of the surprises in low-volume quotations. The first is material minimums: a mill will not sell 80 mm of a rare bar, so a single part can carry the cost of a full stock length. The second is first-article inspection, which is priced by the drawing rather than by the batch — a heavily toleranced part costs the same to verify whether you order one or twenty.
Worked example: how fast does setup amortise?
Unit cost in a small batch is C(n) = S/n + v, where S is everything done once and v is everything done per part. Work in multiples of v rather than currency, and the result transfers to any shop and any part.
Take a part where the one-time work — programming, fixture, proving, first article — equals the per-part cost of eight pieces. So S = 8v.
| Quantity n | Setup per part, S/n | Unit cost, C(n) = S/n + v | Share of the achievable saving captured |
|---|---|---|---|
| 1 | 8.00v | 9.00v | 0% |
| 5 | 1.60v | 2.60v | 80.0% |
| 10 | 0.80v | 1.80v | 90.0% |
| 25 | 0.32v | 1.32v | 96.0% |
| 50 | 0.16v | 1.16v | 98.0% |
| 100 | 0.08v | 1.08v | 99.0% |
| 500 | 0.016v | 1.016v | 99.8% |
Step by step at n = 10: setup per part is 8v ÷ 10 = 0.80v; unit cost is 0.80v + v = 1.80v; the reduction achieved is 9.00v − 1.80v = 7.20v out of a maximum possible 9.00v − 1.00v = 8.00v, which is 7.20 ÷ 8.00 = 90%.
The practical reading: nine-tenths of the available reduction has happened by the tenth part. Ordering 100 instead of 10 improves unit cost by a further 0.72v — real, but far smaller than most buyers expect, and it commits capital against a design that may still change.
The equation also tells you where to push. If setup is large relative to per-part cost, attack S: fewer orientations, simpler workholding, fewer toleranced features. If setup is small, attack v: material envelope, cycle time, finishing. Asking a supplier which of the two dominates your quote is a fair and useful question, and the answer directs the next design revision.
What actually sets the time to first part?
Buyers often assume lead time is machining time. At low volume, machining is frequently the shortest element in the chain.
| Stage | What it depends on | How to compress it |
|---|---|---|
| Quotation and DFM review | Completeness of the file set | Send a 3D model plus a drawing carrying tolerances and datums |
| Material availability | Whether the alloy and section are stocked | Choose a stocked grade and a standard section where function allows |
| CAM programming | Feature count, orientations, multi-axis strategy | Reduce distinct setups; avoid features reachable from only one awkward angle |
| Fixturing | Whether standard workholding reaches the geometry | Leave a clampable datum face; avoid parts that must be held on a finished surface |
| Machining | Removal volume, tolerance, tool changes | Larger corner radii, fewer deep narrow pockets |
| Deburring and finishing | Edge count, cosmetic class | Specify finish only where it is functional |
| Inspection | Number of characteristics, whether FAI applies | Mark critical characteristics rather than tolerancing everything tightly |
| Coating or heat treatment | Third-party queue, batch minimums | Decide the finish before quotation, not after |
Two of these sit outside the machinist’s control and routinely dominate the schedule: raw material procurement and external coating. Broad stock removes the first. The second is why finish belongs in the RFQ — discovering after machining that the part needs anodising in a specific colour can add more calendar time than the machining took. MW+ stocks 70+ materials in common bar, plate and tube sections, and quotes return within 24 hours with a DFM review included.
When should you pay for speed, and when for cost?
The choice is not a preference. It depends on what the part is blocking.
| Situation | Optimise for | Reasoning |
|---|---|---|
| Part blocks a test that gates the next design decision | Speed | Idle engineering time costs more than the expedite |
| Regulatory or certification submission with a fixed date | Speed | The deadline is external and unrecoverable |
| Bridge supply while a mould or die is being cut | Speed, then cost | Missing the bridge means a line stop; once bridged, re-quote |
| Design is still moving | Speed, small batch | Do not buy quantity against a geometry that will change |
| Design frozen, demand forecast credible | Cost | Consolidate releases and amortise properly |
| Non-critical brackets, covers, spacers | Cost | Nothing downstream is waiting on them |
| Repeat annual requirement | Cost | Schedule the release; reuse the existing programme and fixture |
The most expensive pattern is paying an expedite premium on a part nothing is waiting for, because “urgent” was the default on the requisition. The second is the reverse: taking the economical lead time on the one part that gates a validation test.
What buying speed can and cannot compress
An expedite buys queue priority and process overlap: the job jumps ahead, and downstream steps start before the batch finishes. It does not shorten heat treatment soak, coating cure, third-party plating queues or the calendar an approval takes. Ask which segment is actually being shortened; if it is not on your critical path, the premium buys nothing.
Which design decisions move both cost and lead time?
Some changes reduce cost at the expense of time, and vice versa. A smaller set moves both in the right direction, and those are the ones worth a design revision.
- Reduce the number of orientations. Every additional setup adds fixturing, programming and a tolerance stack between faces. A geometry reachable in two setups instead of four is cheaper and faster, and more accurate. Where the geometry genuinely needs it, multi-axis machining collapses setups rather than adding them.
- Use tolerances that mean something. A blanket tight tolerance on every dimension multiplies inspection and scrap risk. Tolerance mating features to a fit class from ISO 286 and let the rest sit at the general band.
- Match internal corner radii to a real cutter. A 1 mm internal radius forces a small tool, shallow depths of cut and long cycle times. Increasing it to 3 mm can transform the cycle without changing function.
- Avoid deep, narrow pockets. Depth-to-width ratio drives tool deflection, chatter and passes. If the depth is not functional, reduce it.
- Keep a clampable datum. A part with no plain face to hold needs a bespoke fixture, which is pure S in the equation above.
- Choose a stocked alloy and a standard section. Exotic material is not only more expensive per kilogram, it is a procurement lead time.
- Specify finish per surface. Ra 0.4 µm on two functional faces is a different job from Ra 0.4 µm everywhere; state the parameter against ISO 21920-2, which supersedes ISO 4287.
These are the same levers examined in detail in our guide to design for manufacturability in CNC machining, which goes through how a review is run and what to send.
When is low volume CNC the wrong answer?
Machining is the default at small quantities because it needs no hard tooling. That does not make it correct in every case, and a supplier who never says so is not being useful.
| Situation | Better process | Where the crossover usually sits |
|---|---|---|
| Flat parts from sheet, simple profiles, bends | Laser cutting and forming | Almost immediately, once the part is essentially 2D |
| Thousands of identical small polymer parts | Injection moulding | Once quantity justifies the tool, typically thousands |
| Large, thick-walled, complex castable shapes | Casting plus finish machining | When removal volume from billet becomes the dominant cost |
| Non-functional visual models | Additive manufacturing | Where dimensional accuracy and material properties are not required |
| Very deep narrow slots, square internal corners, hardened material | wire EDM services | Where a rotating tool physically cannot produce the geometry |
| Very small diameters and long slender turned parts | Swiss machining | Where part slenderness defeats conventional turning |
Two honest caveats. At a quantity of one, unit price looks shocking next to any catalogue part, because you are buying an engineering set-up rather than a commodity; if an off-the-shelf component meets the requirement, buy it. And if the design is genuinely unstable, even a small machined batch may be waste — one part to check the concept, then a pause, often beats ten parts that are obsolete on arrival.
How do you buy a small batch without paying a penalty?
Most of the avoidable premium in low-volume work comes from incomplete RFQs and from splitting orders that could have been combined.
- Send a 3D model and a drawing. The model gives nominal geometry; the drawing gives tolerances, datums, finish and material. A model alone forces the supplier to assume, and assumptions are priced conservatively.
- State the real quantity and the real repeat expectation. A supplier who knows a second release is coming can retain the fixture and the programme.
- Mark the critical characteristics. Two or three features flagged on the drawing, using ASME Y14.5 callouts, cost far less to control than forty tight ones.
- Decide the finish before you quote. Adding a coating afterwards changes both schedule and price.
- Consolidate releases inside a band, not across bands. Combining 5 + 5 into 10 is worth doing; combining 250 + 250 usually is not.
- Ask which dominates, setup or per-part. The answer tells you whether the next revision should simplify workholding or reduce machining.
MW+ accepts single-piece orders with no minimum order quantity, run on the same machines and the same inspection regime as production work: 60+ CNC machining centres, ±0.01 mm general tolerance to ISO 2768-m, and a certificate of conformity, CMM inspection report and material certificates with every order. For a proving batch rather than production, CNC prototyping is the entry point; the process envelope sits under CNC machining services and the equipment list under CNC machining capabilities.
Frequently asked questions
Is there a minimum order quantity for low volume CNC machining?
Not at MW+ — single-piece orders are accepted. The economic point differs from the contractual one: at a quantity of one you pay the whole setup, so unit price is at its maximum. If a second unit is plausible, ordering two at once costs far less than ordering one twice.
Why is my quote for ten parts not ten times the quote for one?
Because the one-time work does not repeat. Using C(n) = S/n + v with S = 8v, one part costs 9v and ten parts cost 18v in total, not 90v. That is the whole economics of low-volume machining in one line, and it is why asking for a price break at quantity is a reasonable request rather than a negotiation tactic.
Does ordering a larger batch always reduce the unit price?
It reduces it, but with sharply diminishing returns. The table above shows 90% of the achievable reduction captured by the tenth part and 98% by the fiftieth. Beyond that you are trading cash and storage for very small unit savings, against a design that may still change. Order to the demand you can actually forecast.
What lead time should I expect for a small batch?
MW+ runs 48-hour express for urgent prototypes, 3–5 business days for standard prototypes and 10–15 business days for volume production, with quotes returned within 24 hours. What most often moves those numbers is material availability and external coating, which is why both belong in the RFQ.
Is it cheaper to prototype in aluminium and switch material later?
Often, but not always safely. Aluminium cuts faster and costs less to buy, so the proving batch is cheaper; however different thermal expansion, springback and residual stress mean some features will behave differently in the production material. Validate geometry in aluminium if you like, but plan a first article on the production material before release — see our comparison of aluminium and stainless steel machining.
How does low volume differ from full production pricing?
In low volume, unit cost is dominated by amortised setup; in production it is dominated by cycle time, material and yield, and the levers change accordingly. The transition and where the curve flattens are covered in our guides to prototype versus mass production and scaling across volume tiers.
Can I hold production tolerances on a batch of five?
Yes. Tolerance capability is a property of the machine, the fixture and the metrology, not of batch size. What changes is the cost of proving it: verifying a heavily toleranced drawing costs the same whether the batch is five or fifty. Mark the characteristics that matter, using ISO 1101 geometric controls where form or position is what counts, and the burden falls sharply.
What should I send to get an accurate small-batch quote?
A 3D model in STEP, IGES or native SolidWorks, a 2D drawing in PDF or DXF with tolerances, datums and finish callouts, the material and condition, the quantity, the expected repeat and the required date. With that, request a CNC machining quote and an engineer responds within 24 hours with a DFM review attached.
