In this article
This guide shows which design decisions move the number, by how much, and how to read your own print for cost before you release it. Every figure cites a source.
The cheapest lever you have on a machined part is not the shop rate. It is the drawing.
By the time you send a STEP file out for quote, most of what the part will cost is already locked in — set by the geometry, the tolerances, the material, and the finish you specified. The shop can optimize toolpaths and trim a few points off the cycle. They cannot un-specify a ±0.0005″ tolerance you didn't need, or un-draw a sharp internal corner that forces a tiny end mill.
This is the central finding of Design for Manufacturability (DFM): across the cost-engineering literature, 70–80% of a product's total manufacturing cost is committed during design — before a supplier ever quotes it. That figure traces to the founding work of Geoffrey Boothroyd and Peter Dewhurst (creators of DFMA, recipients of the U.S. National Medal of Technology) and is corroborated by peer-reviewed cost research: a literature review titled, literally, "Design determines 70% of cost," and CIRP Annals work finding that roughly 75% of manufacturing cost is committed by the end of conceptual design.
For a procurement or engineering lead, that has a blunt implication: the biggest cost lever is the model, not the negotiating table.
What "Design for Manufacturability" Actually Means
Design for manufacturability (DFM) is the practice of designing a part so it can be machined reliably, at the lowest cost, without sacrificing the function it has to perform. In CNC machining, DFM means matching the part's geometry, tolerances, material, and surface finish to how a cutting tool actually removes material — round tools, finite reach, work-holding limits, and cycle time. A DFM review reads a model or drawing for the features that quietly drive cost or scrap, then proposes changes that keep function intact while making the part faster and cheaper to cut.
DFM is not "make the part cheaper by making it worse." It is the opposite: it protects the function you need while removing the cost you don't. A wall that has to be 2 mm thick for stiffness stays 2 mm. A bore that has to be ±0.0002″ for a bearing fit stays ±0.0002″. Everything else — the features specified tight out of habit, the corners drawn sharp out of CAD convenience, the finish called out fine "to be safe" — is where the money leaks.
The reason DFM works is timing. A design decision takes minutes to change in CAD and produces cost consequences that take hours — and tooling, and scrap — to change later. The same finding caught at concept costs an order of magnitude less than the one caught at first article. That is the whole game: move the decision earlier.
Why Design Decides 70–80% of the Cost
There is a well-known curve in cost engineering. Early in a project, very little money has actually been spent — but most of it has already been committed. By the time the design is frozen, the major levers (material, process, geometry, tolerance) are set, and everything downstream is just executing on choices already made.
Roughly 70–80% of a part's total manufacturing cost is determined by design decisions, and most of it is committed before the part is ever quoted or prototyped. Sourcing, negotiation, and shop-floor efficiency compete over the remaining ~20%. This is the founding premise of Design for Manufacture and Assembly (DFMA) and is consistent across the cost-engineering literature.
This is why chasing only the shop rate is a losing game. Move a part from a U.S. shop to a lower-cost region and you act on the labor slice of cost — real, but bounded. Redesign the part so it needs one setup instead of three, holds a sensible tolerance instead of a heroic one, and starts from less stock — and you act on the 70–80% that design controls. The two are not mutually exclusive. The point is that the design lever is bigger, and it is the one most buyers skip.
A practical way to think about it: sourcing changes what each machine-hour costs; DFM changes how many machine-hours the part needs.
The CNC Cost Stack: Where the Money Actually Goes
Before you can design for cost, you have to know what you're designing against. A typical CNC machined part breaks down roughly like this (GPW 2026 cost model — directional, varies by part):
| Cost element | Share of part cost | What design controls |
|---|---|---|
| Material | 25–35% | Alloy choice, starting stock size, material removed |
| Direct labor / cycle time | 15–55% | Setups, tool changes, feeds limited by geometry |
| Machine burden (tooling, wear, depreciation) | 20–30% | Special tools forced by features; tool wear from hard materials |
| Overhead (QC, setup prep, handling) | 10–20% | Inspection forced by tight tolerances; first-article scope |
| Margin | 10–20% | — |
The single most important column is the right one. Almost every line in that stack has a design knob behind it. Labor share alone swings from ~15% on a simple, material-heavy bracket to ~55% on a complex 5-axis part — which is exactly why the same shop quotes wildly different "savings" depending on what part you hand it.
Two of those elements — tolerance (which drives labor, machine burden, and QC together) and geometry (which drives setups and tool selection) — are where DFM finds the most money. We'll take them in turn.
Tolerance Budgeting: The Single Biggest Design Lever
Tolerance is the most expensive thing on most drawings, and the most over-specified. Tightening a tolerance does not nudge cost up — it bends an exponential curve.
Tolerance budgeting is the practice of assigning each dimension only the tolerance its function actually requires, instead of applying one tight tolerance across the whole part. Tight tolerances cost money — through slower cutting, extra finishing passes, more inspection, and higher scrap — so a tolerance is a budget to be spent deliberately on the few features that need it (bearing fits, sealing faces, mating interfaces) and kept loose everywhere else.
Start from the default. The standard machining tolerance — used by the large majority of shops worldwide when no tolerance is called out — is ±0.005″ (0.127 mm), which maps to roughly ISO 2768-medium. At that level there is no premium; it's just how the machine cuts. The cost only starts climbing when you ask for tighter.
Here is roughly how the curve behaves (representative figures from machining cost-engineering sources; exact numbers vary by feature and material):
| Tolerance band | Relative cost vs. standard | Typical use |
|---|---|---|
| ±0.005″ (0.13 mm) — standard | 1× (baseline, no premium) | General features, non-critical dimensions |
| ±0.001″ (0.025 mm) — precision | ~2–4× | Locating features, light press fits |
| ±0.0005″ (0.013 mm) — tight | ~5× and up | Bearing seats, sealing faces |
| ±0.0001″ (0.0025 mm) — ultra | up to ~24× | Optical, aerospace, gauge work |
The mechanism is not mysterious: tighter tolerances mean slower feeds, more finishing passes, more frequent tool changes, controlled-temperature inspection, and a higher scrap rate when a part drifts out of band. One consensus point worth internalizing — moving from ±0.05 mm to ±0.02 mm might add ~50%, but going from ±0.02 mm to ±0.005 mm can add 300–500%. The curve is non-linear, and you live on the steep part of it the moment you specify tight tolerances you don't need.
The DFM move: default the whole part to the standard tolerance, then deliberately tighten only the handful of features that mate, seal, or locate. On a typical part that is 3–6 dimensions out of dozens. Everything else rides the default and costs nothing extra.
The Geometry That Quietly Drives Cost
After tolerance, the biggest cost driver is geometry that fights the tool. CNC end mills are round, finite in length, and held in a spindle that can deflect. Features that ignore those facts force smaller tools, slower feeds, extra setups, or special work-holding — each of which shows up on the quote. None of these require a citation; they are how metal cutting works.
| Feature as drawn | Why it drives cost | Lower-cost design move |
|---|---|---|
| Sharp internal corners | A round end mill can't cut a sharp inside corner; a tiny tool must, slowly | Add a corner radius ≥ the tool radius (e.g., R ≥ 1/3 of pocket depth) |
| Deep, narrow pockets | High depth-to-diameter means tool deflection, chatter, special long tools, slow feeds | Limit cavity depth to ~3–4× tool diameter; relax depth where possible |
| Thin walls | Vibration and deflection cause chatter and scrap; needs slow finishing | Keep walls ≥ ~0.8 mm (metal); add ribs for stiffness |
| Features on many faces | Each new orientation is another setup — re-fixturing, re-indicating, stack-up risk | Group features onto fewer faces; design for 1–2 setups where function allows |
| Tiny holes, deep holes | Small drills are fragile and slow; deep holes need peck cycles | Use standard drill sizes; keep depth ≤ ~5× diameter |
| Custom / non-standard threads | Special taps or single-point threading add tooling and time | Specify standard thread series; tap only as deep as the joint needs |
| Engraved text, fine cosmetic detail | Fine detail is a separate, slow operation | Keep it minimal, or move it to a secondary process if cosmetic |
The recurring theme: setups and tool size are where labor cost is born. Every additional setup adds fixturing time, machine idle time, and a tolerance stack-up across datums. Geometry that lets a part be cut in one or two orientations, with standard-size tools, is geometry that quotes well — whether it runs on a 3-axis mill or a lathe.
Material: The Choice You Make Before the First Chip
Material is 25–35% of part cost on its own — but its bigger effect is on cycle time, because how fast you can cut depends on what you're cutting. Machinability is a published, relative metric. The faster a material cuts, the lower the labor cost behind the same geometry.
| Material | Relative machinability | DFM note |
|---|---|---|
| Free-cutting brass (C360) | ~100% (industry baseline) | Cuts fast and clean; ideal where strength allows |
| Aluminum 6061-T6 | High (~90% of brass; far above steels) | The default cost-effective choice for most parts |
| Steel 1018 / 4140 | Moderate | Predictable; cost climbs with hardness and alloying |
| Stainless 304 / 316 | Lower (work-hardens) | Specify only where corrosion resistance is required |
| Titanium (Ti-6Al-4V) | Low (slow, hard on tools) | Cost-effective only where strength-to-weight is non-negotiable |
(Relative machinability indices vary by reference baseline; figures above are representative of published machinability rating charts.)
The DFM lesson is not "always use aluminum." It's "don't pay for titanium-grade machining time on a part that 6061 would serve." Material selection is a function call, not a default — and it's one of the cheapest reviews to run, because changing it on the model costs nothing and changing it after tooling costs everything. GPW's engineering support and our metals and plastics guidance exist for exactly this conversation.
A 10-Point DFM Checklist You Can Run on Your Own Print
Before you release a part for quote, run it against these. Each one is a place we routinely find cost in a DFM review:
- Default the tolerances. Is the whole part at ±0.005″ / ISO 2768-m, with only the functional features called tighter?
- Count the tight callouts. Can you justify each sub-±0.001″ dimension by a fit, seal, or location? If not, loosen it.
- Radius the internal corners. Are inside corners radiused to fit a reasonable tool, not drawn sharp?
- Check pocket depth. Is any cavity deeper than ~3–4× the tool diameter that has to reach the bottom?
- Count the setups. How many orientations does the part need? Can features move to fewer faces?
- Check wall thickness and spacing. Any wall — or web between a hole and a nearby edge or feature — thin enough to chatter, deflect, or break out under cutting load?
- Standardize holes and threads. Standard drill sizes? Standard thread series? Tapped no deeper than the thread engagement the bolted joint requires?
- Right-size the material. Is the alloy the cheapest one that meets the requirement — and is the starting stock close to the finished envelope?
- Specify finish only where it matters. Is a fine surface finish called out only on functional/cosmetic faces, not blanket across the part?
- Confirm datums and GD&T are defined by function. Do datums and tolerances reflect how the part mounts and functions — and reference features the shop can actually fixture and inspect from?
If your print clears all ten, it's a well-designed part for machining. If it doesn't, that's not a problem — it's the 5–15%.
Keep it handy: Download the 10-point DFM checklist as a one-page PDF — with a tolerance cheat-sheet — and run it on your next drawing before it goes out for quote.
How GPW Approaches DFM
We don't quote first and find the problems later. Every part gets a free DFM review before we send a number.
Before any shop in our coordinated Monterrey machine-shop network cuts the first chip, an engineer reads your model the way the tool will: where the tight tolerances are, how many setups it needs, which corners force a small tool, whether the material is right, what the finish actually requires. This is not automated software feedback — it's a person who has watched these parts run, telling you where the cost is and what would change it.
That review feeds three things:
- A part that quotes better. Most parts carry an additional 5–15% in savings that only surface under part-specific DFM — the kind a generic instant-quote engine can't see (GPW observation across quoted work).
- A quote you can audit. We itemize line by line — material, labor, machine time, finishing, freight, USMCA filing — so you can compare against your current supplier without back-calculating what's included.
- One point of contact. The same engineer from review to shipment. No handoff to an account manager who never saw the part.
DFM is also where the nearshore math compounds. The landed-cost savings of sourcing in Monterrey are real on their own; layered on top of a part that's been designed (or redesigned) for manufacturability, the two stack. Sourcing acts on what the machine-hour costs. DFM acts on how many you need.
Send Us Your Print. Get a Free DFM Review Before We Quote.
Upload a STEP file or PDF drawing. An engineer reviews it for tolerance risk, setup count, material fit, and cost-driving geometry — then returns a line-item quote you can compare directly against your current supplier.
Frequently Asked Questions
What is design for manufacturability (DFM) in CNC machining?
DFM is designing a part so it can be machined reliably and at the lowest cost without losing the function it has to perform. In practice it means matching geometry, tolerances, material, and finish to how a cutting tool actually removes material — round tools, finite reach, work-holding limits, and cycle time. A DFM review reads a model for the features that quietly drive cost, then proposes changes that keep function intact.
How much of a part's cost is determined by its design?
Roughly 70–80% of a part's total manufacturing cost is committed by design decisions, most of it before the part is ever quoted or prototyped. The figure comes from the Design for Manufacture and Assembly (DFMA) work of Boothroyd and Dewhurst and is corroborated by peer-reviewed cost-engineering research. Sourcing and shop-floor efficiency compete over the remaining ~20%.
What design features make a CNC part more expensive?
The biggest drivers are over-tight tolerances, sharp internal corners, deep narrow pockets, thin walls, features spread across many faces (forcing multiple setups), non-standard holes and threads, and hard-to-machine materials. Each forces smaller tools, slower feeds, extra setups, or more inspection — all of which add cycle time and cost.
What is the standard CNC machining tolerance, and does tightening it cost more?
The standard default is ±0.005″ (0.127 mm), roughly ISO 2768-medium, applied at no premium when no tolerance is specified. Tightening costs more on an exponential curve: cutting from ±0.02 mm to ±0.005 mm can raise cost 300–500%, and ultra-tight ±0.0001″ work can reach roughly 24× standard. Specify tight tolerances only where function requires them.
What is tolerance budgeting?
Tolerance budgeting is assigning each dimension only the tolerance its function requires, rather than applying one tight tolerance across the whole part. Because tight tolerances cost money, you spend your budget on the few features that mate, seal, or locate, and keep everything else at the standard default.
How do I reduce the cost of a machined part without changing what it does?
Default the whole part to standard tolerance and tighten only functional features; radius internal corners to fit a reasonable tool; reduce the number of setups by grouping features onto fewer faces; use standard hole sizes and thread series; choose the cheapest material that meets the requirement; and call out fine surface finish only where it's needed. A DFM review systematizes all of this before you release the part.
When should a DFM review happen?
As early as possible — ideally before the design is frozen and certainly before quoting. A change caught in CAD costs minutes; the same change after tooling or first article costs an order of magnitude more. GPW runs a free DFM review before quoting every part for exactly this reason.
Sources
All figures in this guide trace to one of the sources below. Accessed May 2026.
- Boothroyd, G., Dewhurst, P. & Knight, W. — Product Design for Manufacture and Assembly (DFMA methodology; origin of the "design determines 70–80% of cost" premise). Boothroyd Dewhurst, Inc. / DFMA
- Love, D. et al. — "Design determines 70% of cost: A review of implications for design evaluation" (academic literature review of the claim). ResearchGate
- Part cost estimation at early design phase — CIRP Annals (~75% of manufacturing cost committed by end of conceptual design). ScienceDirect
- ISO 2768 general tolerances & standard machining default (±0.005″/0.127 mm). Fictiv — What is ISO 2768
- Tolerance-vs-cost curve (exponential cost of tight tolerances). Tormach — The High Cost of Tight Tolerances
- Machinability rating charts (relative machinability of aluminum, brass, steel, stainless, titanium). SAM — Machinability Rating Charts
- GPW 2026 cost model (cost-stack shares, labor-share-by-complexity, 5–15% DFM savings) — GPW internal model; presented as our own, not third-party data. See the Mexico vs US CNC Cost Study for the full landed-cost methodology.
Published May 18, 2026. The cost-stack and DFM-savings figures are GPW's own model and are labeled as such; the design-cost, tolerance, and machinability figures are sourced to the public engineering references above.