Home  /  Resources  /  Small-batch CNC

Small-batch CNC in China: keeping the unit cost sane

Part of the process guide series — see all →

Fifty CNC parts almost never cost twice twenty-five. They cost four times — and the quote is honest when the shop can tell you exactly which fixed costs it's spreading over how many parts.

Somewhere between "send me three prototypes" and "run five thousand a month" sits the small batch — ten to five hundred pieces, ordered again and often. It's the volume range where unit prices look irrational and buyers feel extorted. They aren't, usually. The price is just telling the truth about setup: programming, fixturing, first articles and machine time all get amortized over whatever quantity you bring. The good news is that most of that fixed cost is design-controlled. Here are the six levers, in the order I pull them.

Why the unit price is what it is

A small-batch CNC quote has two stacked halves. The fixed half: CAM programming, fixture building, first-article inspection, machine setup and teardown — identical whether the order is 40 parts or 400. The variable half: cycle time × machine rate + material + finish, scaling with quantity. On a 40-part order the fixed half can be 40–60% of the unit price; on a 400-part order it's single digits. That's the whole arithmetic. A supplier who can't show you the split is either hiding margin or hasn't done the engineering.

Lever 1 — Cut operations, not corners

Every additional setup is a full repeat of fix, indicate, prove, run — plus a tolerance stack between setups. The biggest single cost lever in small-batch machining is operations count:

I've seen a bracket drop 35% in unit price purely from reorienting two features onto the primary face — same part, same function, one fewer setup.

Lever 2 — Start from the right stock

Material is where small batches bleed invisibly. Three checks before the drawing goes out:

Lever 3 — Tolerances must pay rent

Every tight tolerance on a small batch is paid for three times: slower cycle (spring passes, finishing passes), inspection time, and scrap allowance. The discipline is the one in the DFM checklist — tight only where function lives — but it bites harder at 50 pieces, because there's no volume to dilute a CMM cycle. A blanket ±0.01 mm on a 50-part order can add double-digit percentage to the quote. Keep ±0.02 mm and tighter for bores that locate something, sealing faces, and press fits; let the rest float at ±0.1 mm or ISO 2768-m.

Lever 4 — Standardize features

One M5 thread, not three flavors of "M5-ish". One drill diameter wherever a Ø5 hole appears. One chamfer callout. Each unique feature is a tool change and its setup offsets; on a small batch, toolchanges are a bigger share of cycle time than they'd be at volume. This lever costs nothing in function and routinely shaves 5–10% off cycle-heavy parts.

Lever 5 — Pick the finish for the batch size

For sheet-metal assemblies the same logic applies with different players — the trade-offs are laid out in the anodize vs powder vs plating guide.

Lever 6 — Batch strategy: buy the curve, not the point

Small batches have a habit of repeating. Two structures keep the repeat orders cheap:

And always ask for the price curve — unit price at 50 / 100 / 250 / 500. It tells you where the fixed cost actually sits and what one more batch buys you.

Reading a small-batch quote

A quotable small-batch quote shows: setup/tooling as a separate line, unit price by quantity (at least two breakpoints), material grade and stock form named, finish priced separately, and lead time anchored to a defined date. One flat "USD 38.5/pc, 2 weeks" tells you nothing except that the number will move after your deposit. This mirrors the RFQ discipline in the sourcing process guide: quotes only compare when they're built the same way.

A quick worked example: two shops quote a 100-piece aluminum bracket. Shop A: USD 41/pc flat. Shop B: USD 1,100 setup + USD 29.50/pc. At 100 pieces, Shop B's math lands at USD 40.50 — nearly identical. But the repeat order tells them apart: Shop B's second batch prices near USD 29.50, because the setup is already paid and on the shelf; Shop A quotes 41 again, because you were always paying the setup, just invisibly. The itemized quote isn't pedantry — it's the only form where batch strategy (Lever 6) can actually work.

When machining isn't the right small batch at all

Below roughly 300–500 units for a typical aluminum housing, machining wins — a die-cast tool takes 25–35 days and a five-figure investment that never amortizes. Past that, the crossover depends on the part: heavier, thinner-walled, more organic geometry crosses over earlier. The decision framework is in die casting in China, and a real example of switching mid-life — turned part redesigned for die casting at −50% unit cost — is in this case study. Price both routes at your true annual volume before committing; the answer changes with the answer to "how many, for how long".

Where I fit in

Small-batch CNC is where my engineering background earns its keep: I run the DFM pass that pulls the six levers on the drawing side, match the part to shops whose machine list fits the batch size (not every CNC shop wants your 80-piece order — some are structured for it), and normalize the quotes so the price curve comparison is real. One engineer, one accountable quote comparison, before anything goes to production.

See the CNC machining floor →

CNC machining in China: the full capability guide →

Send a drawing — small-batch quotes compared →

Common questions

The shop those small batches run in is described on the CNC machining factory page — equipment, tolerance bands and what counts as a small order there.

Small batch, big quote? DFM pass free

Send the drawing and quantities — I'll pull the six levers and compare shops that actually want the batch.

Send your part →
Send Your Idea →