For a functional sheet-metal prototype, the route is settled first by tolerance and geometry: formed sheet metal is reported around ±0.1–0.5 mm, while CNC machining is reported at ±0.05 mm or better. So a feature tighter than that points to machining.

Cost at 1–100 pcs cannot decide it, because the two sources that address low volume disagree on which route is cheaper, so you quote both for the actual part. 3D printing is outside this comparison: no tolerance, cost, lead-time or material data for it appears in these sources.

Rule out the route that cannot hold your tightest feature

Start with the single tightest functional feature on the drawing, not with the price. Formed sheet metal is reported in a band of roughly ±0.1–0.5 mm: one supplier's prototype service quotes ±0.2 mm on laser cutting, ±1.0° on bend angle and ±0.2 mm bend-to-hole9 6.

And a general comparison puts bent features at ±0.2–0.5 mm with laser-cut holes at ±0.1 mm20. A separate prototyping guide gives ±0.1–0.2 mm for laser-cut and bent features, noting the figure moves with material thickness, bend complexity and part size6.

CNC machining is reported tighter — ±0.05 mm or better on critical features20, and ±0.001 in in another comparison24. That gap is what decides the route.

If a mating, sealing or locating feature needs to sit below about ±0.05 mm, the formed route is reported not to reach it and the part goes to machining regardless of what the quote says20,24.

One fabricator states the boundary in imperial terms: anything under ±0.005″ typically needs CNC machining rather than forming or casting alone14. Treat that as a screening line to check against your drawing, not as a certified grade — no source here ties either route to a standard tolerance class.

Check whether the geometry is flat-plus-bends or true 3D

Tolerance rules out one route; geometry can rule out the other. Sheet metal is described as a 2.5D process — flat blanks plus bends — with welding adding cost, while CNC machining covers true 3D forms, undercuts and internal cavities20.

Another comparison reaches the same split: sheet metal suits flat panels, bends and boxes, and cannot easily produce complex curves or internal cavities25. If your prototype is an enclosure, bracket, chassis or panel, the formed route is geometrically open to you.

If it needs a pocketed internal cavity or an undercut, it is not. Sheet thickness sets the second boundary.

Standard bending is reported best suited to 0.5–6 mm sheet, with thicker material needing special equipment20, and thickness governs minimum bend radius, hole sizing and flange length20.

Check your gauge against that band before you assume the formed route is available; the sources give no numeric bend radius or hole-to-thickness rule, so those come from the fabricator's DFM review.

Quote both routes instead of trusting a cost ranking

At prototype quantity the published cost rankings contradict each other, so neither route can be assumed cheaper. One comparison says CNC is usually lower per part at 1–100 pcs, because setup is limited to fixturing and toolpaths, while sheet metal carries blank programming, bend sequencing and handling20.

Another says sheet metal has the lower single-unit cost and CNC the higher, and recommends CNC for prototypes and low-volume custom enclosures25. Both statements are about the same quantity band and they point opposite ways.

The mechanism behind the disagreement is how fixed effort is spread. Programming, fixture planning and prove-out are project costs that get divided across however many parts you order, and that division differs between prototypes, small batches and repeat production17.

Material use pulls the other way: sheet metal is reported at 70–85% utilization with nesting against 30–50% for CNC because of chip waste20. Which effect dominates for your part is not something the sources settle — send the same drawing to both routes and compare the quotes line by line.

Do not expect lead time to break the tie

Lead-time ranges for the two routes overlap, so delivery date rarely picks the process for you. One supplier publishes 2–5 business days for sheet-metal prototypes, splitting into 1–2 days for laser cutting only, 2–3 days for cutting plus bending, and 4–7 days for full fabrication with welding and finishing9.

A general comparison gives CNC at 1–5 working days and sheet metal at 3–1020. A third shop advertises 3–7 days standard for sheet metal, as little as 3 days19.

What moves the date is complexity and material availability, not the route label6. A simple flat part and a simple machined part can both land in days; a multi-bend assembly with welding and finishing, or a machined part with deep cavities and long toolpaths, will not.

Plan the schedule around the hardest feature on the part, and ask each supplier which step in their process sets their quoted date.

Weigh how much the design will still change

If the design is still moving, the two routes carry different revision risk. CNC is rated high on design-change flexibility because toolpaths are edited quickly; sheet metal is rated medium because a flat pattern and bend sequence need re-validation20.

The sources do not explain what that re-validation involves or what it costs in time or money, so treat the rating as a direction, not a penalty figure. For a design you expect to revise more than once, that rating favours holding the machined route for the features most likely to move.

And asking the fabricator how they handle a revision before you place the first order. One fabricator advises bringing engineering review in before the design is locked14; the sources do not quantify what that saves, so ask what a second iteration would involve on their line.

Lock inspection and documentation into the RFQ

Inspection and documentation are priced work, not a free afterthought, and a late request can move the delivery date because inspection programming, sample selection and certificate review all take time30. Define acceptance at RFQ stage so the quote separates machining cost from evidence cost30.

  • ✓State which features need a CMM report and which only need dimensional checks against the drawing.
  • ✓Ask whether material certificates, first-article inspection or PPAP records are required, and get them priced in the quote.
  • ✓Confirm the shop's quality certification (ISO 9001, AS9100 where applicable) and the inspection equipment it actually runs.
  • ✓Ask whether inspection happens before or after finishing, and whether records ship with each batch.
  • ✓Request DFM feedback with the quote rather than accepting a print-as-quoted price.

Budget post-processing before you order

Finishing is a separate line in the prototype budget and it can disturb the fit you just validated. Powder coating is reported to offer limited control over coating thickness, which can produce uneven finish levels and affect surface texture, and reuse of coating material carries a cross-contamination risk that affects colour7.

On a prototype where a coated panel meets a mating part, that thickness variation sits on top of the formed tolerance band. Decide the finish before the parts are cut, and tell the supplier which surfaces must stay bare for fit or grounding.

The sources here do not give a coating thickness figure or a fit allowance, so ask the finisher what build they expect on your specified coating.

Where the two cost sources disagree

The low-volume cost question is the one place the sources directly contradict each other, and the buyer needs to see both positions rather than a merged verdict.

Disputed item (unit)One source reportsAnother reportsWhat the buyer should do
Cost per part at 1–100 pcs ($/part)CNC usually lower; sheet metal higher on blank programming and bend sequencingSheet metal lower single-unit cost; CNC higher, CNC preferred for prototypesQuote the same drawing on both routes and compare totals, not unit rates alone

Before you request a quote

  • Ask the supplier for model-specific evidence and record the condition that would change your buying decision.
  • Request a written quote with price basis, MOQ, tooling, payment terms and production lead time.
  • Ask the supplier for model-specific evidence and record the condition that would change your buying decision.
  • Ask the supplier for model-specific evidence and record the condition that would change your buying decision.
  • Ask the supplier for model-specific evidence and record the condition that would change your buying decision.
  • Ask the supplier for model-specific evidence and record the condition that would change your buying decision.
Sources · 11

Technical references cited for verifiability — not supplier recommendations. Browse the research library.