7 Costly Mistakes in CNC Machining: Are You Making Them?
Seven failure patterns we see on real drawings, quoted jobs, and first-article inspections. For design engineers and buyers who need to know what went wrong and what to change in the next revision.

In this article
- 1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
Seven costly mistakes, mapped to their fixes
Start with the symptom you actually see on the part or the quote.
| Symptom | Likely cause | What to do |
|---|---|---|
| Sharp internal corner won't clean up | End mill radius larger than the corner | Add corner relief or accept the tool radius |
| Quote 3× the target price | Tolerance applied to every feature | Tighten only the 3–5 features that matter |
| 0.8 mm wall chatters and rings | Thin wall, no support during cutting | Thicken to 1.2 mm or add a support rib |
| Cost jumps when another process is added | Each vendor quotes its own setup and freight | Keep grinding and finishing under one roof |
| Cheapest hourly rate wins, part still late | Rate hides setup, programming, and inspection | Compare landed cost per good part |
| 304 part galls and tears on the finish pass | Material and cutter geometry mismatch | Switch grade or change feed and coating |
| Parts ship without a report | Verification treated as paperwork | Request dimensional and material reports |
| Supplier changes every order | No process owner, no stable program | Assign one team and freeze the setup |
Designing parts that fight the cutter
A CAD model shows the shape. It does not show how a tool reaches that shape. Sharp internal corners, 8:1 deep pockets, and walls under 1 mm are the usual fingerprints of a part that was never reviewed for machining. The cutter is round. It leaves a radius equal to its own diameter in every internal corner.
This is the first of the seven costly mistakes because it is the cheapest to avoid. A corner relief notch, a slightly larger fillet, or a small change to pocket depth can remove an entire EDM operation from the routing. That operation might have added a week and a separate fixture.
Before release, look at the three hardest features on the part and ask how a tool gets there. If you cannot answer, the shop will either add cost or send the drawing back. Free DFM feedback at the quoting stage is where most of this gets caught.
Thin walls deserve a separate look. A 0.8 mm wall in aluminium will deflect under cutting force and sing. Move it to 1.2 mm, or keep 0.8 mm and accept a slower, more expensive pass with light depths of cut.
- 1Corner reliefAdd a notch so a smaller tool can clear the corner
- 2Wall thickness1.2 mm minimum for aluminium, more for steel
- 3Pocket depthKeep depth under 4× tool diameter where possible
Over-specifying tolerances, then accepting worse
A drawing that calls ±0.005 mm on every dimension is not a quality statement. It is a price multiplier. Each tight feature needs a controlled process, a specific cutter, and often a separate inspection step. Put that on forty features and the quote reflects forty tight features.
Then the parts arrive, and three features are out. The buyer accepts them because the schedule is gone. That is the real cost: you paid for tolerance you never received, and you have no margin left to negotiate.
The useful approach is to name the critical features. On a shaft that spins at 15,000 RPM, the bearing journals and the shoulder faces matter. Bolt clearance holes do not. Apply ±0.005 mm there and let the rest follow ISO 2768 medium or fine, which is a documented general grade rather than a guess.
Ask the shop what metrology backs the number. Our inspection runs raw material checks, in-process monitoring, and final inspection, with reports on request. A tight callout without that backup is a promise nobody can verify.
- 1Critical featuresIdentify 3–5 dimensions that control function
- 2General gradeLet the rest follow ISO 2768 medium or fine
- 3VerificationConfirm the shop can measure what it promises
Underestimating mixed-process parts
A housing might start on a 5-axis mill, need wire EDM for two corners, then bead blasting, anodizing, and laser marking. Four vendors, four setups, four chances for a tolerance stack to drift or a box to arrive late. Each handoff also re-clamps the part, and every re-clamp adds variation.
The cost is not just freight. It is the queue time at each shop, the re-qualification of the first article, and the paperwork that follows the part between suppliers. On a 10,000-part run, one extra day per handoff is a visible number.
Keeping the chain short is the practical fix. We run 127 high-precision CNC machines across 3 wholly-owned plants, covering milling, turning, mill-turn, and surface finishing in one flow. Parts that need anodizing or laser marking do not leave the building.
If you must split the work, define who owns the final dimension. Usually that is the last operation, and that vendor needs the drawing plus the inspection data from the previous step.
- 1Handoff countEach vendor adds queue time and stack-up risk
- 2Single roofMachining and finishing in one flow shortens the chain
- 3Final ownerName the vendor responsible for the last dimension
Choosing a supplier on hourly rate alone
The hourly rate is the most visible number and the least useful one. It excludes programming, fixturing, tooling, inspection, and the cost of a late part. Two shops can quote the same rate and land 40 percent apart on total cost.
Compare landed cost per good part instead. That means the quoted price plus the cost of any rework, plus the schedule risk. A shop that ships in 3–5 days and holds a historical late-delivery probability below 2 percent is often cheaper than a low-rate shop that misses dates.
Ask two questions before you commit. How is the setup amortised across the run, and what happens to scrap. A supplier with 99.99% qualification rate has already paid for the process control that keeps scrap low.
Volume matters too. No minimum order quantity means a single prototype and a 10,000+ part run use the same route. That removes the usual re-qualification jump between the prototype shop and the production shop.
- 1Landed costPrice plus rework plus schedule risk
- 2Setup amortisationAsk how the fixture and program are spread over the run
- 3Scrap ownershipConfirm who pays for parts that fail inspection
Ignoring how material and cutter interact
Material choice changes the cutting strategy, not just the price. 304 stainless work-hardens, so a light pass with a dull cutter raises surface hardness and the next pass tears. 6061-T6 cuts clean and fast, which is why so many prototypes start there. 7075 machines well but moves more after stress relief.
Titanium and Inconel are heat problems. TC4 (Ti-6Al-4V) and Inconel push heat into the tool rather than the chip, so feed and speed need to be paired with the right coating and coolant. Run them like aluminium and you burn cutters and scrap parts.
Plastics have their own rules. POM and PEEK hold tight tolerances but clamp marks easily. ABS and PP need sharp cutters and air blast, because coolant can stain or stress-crack them.
The fix is to state the material grade on the drawing and let the shop choose the parameters. Vague callouts like 'stainless steel' leave 303 and 316L on the table, and those two machine very differently.
- 1Work hardening303 and 304 need a sharp cutter and steady feed
- 2Heat-resistant alloysTC4 and Inconel need matched coating and coolant
- 3PlasticsAir blast instead of flood coolant for ABS and PP
Cutting verification to save a day
Skipping inspection to hit a ship date is a loan against the next order. The parts leave, the problem arrives at the customer's line, and the rework costs more than the original job. It also burns the schedule you were trying to protect.
Verification is not one gate. It starts with the raw material certificate, continues with in-process checks on critical features, and ends with a final dimensional inspection. Our standard is 100% inspection before shipment, with reports on request.
The practical question is what to measure on the floor. First article on every setup, then sampling on the features that the process tends to drift. Bearing bores and sealing faces get measured more often than bolt holes.
If a job is genuinely urgent, cut scope rather than inspection. Ship the parts that are complete and verified, and hold the rest. That keeps the quality data honest.
- 1Raw materialCheck the certificate against the stated grade
- 2In-processMeasure critical features while the setup is still live
- 3Final100% inspection before shipment, reports on request
Treating the supplier as a one-off transaction
A shop that keeps the same program, fixture, and operator on your part will hold tolerance better than one that rebuilds the setup each time. That continuity is the seventh item on the list because it is the one that quietly compounds. Re-qualifying a process every order reintroduces variation you already paid to remove.
Long-term stability shows up in small ways. The same engineer answers your DFM questions. The inspection data from last year is still on file, so a drift is visible. A revision gets compared against the previous first article instead of starting from zero.
This matters most for regulated work. Our plants hold ISO 9001:2015, IATF 16949:2016, ISO 13485:2016, and ISO 27001:2022. Those systems only pay off when the same team runs the same process across orders.
You do not need a contract to get this. Assign one process owner, keep the drawing revision under control, and share inspection feedback. Confidentiality is handled by default, and an NDA is available on request.
- 1Process ownerOne team, one program, one fixture per part number
- 2Data continuityKeep first-article and inspection records across orders
- 3ConfidentialitySecure uploads by default, NDA available on request
Step by step: review your next drawing before release
Run these checks in order. Most cost comes out at steps 1 and 2.
- 1Mark the critical featuresCircle 3–5 dimensions that control function. Apply ±0.005 mm only there and let the rest follow ISO 2768 medium.
- 2Check tool accessFor every internal corner, confirm the radius is at least half the smallest practical cutter. Add corner relief where it is not.
- 3Review wall and floor thicknessKeep aluminium walls at 1.2 mm or more, steel at 1.5 mm or more. Below that, expect chatter and a slower pass.
- 4Count the process handoffsList every operation and the vendor that performs it. If more than two vendors touch the part, look for a shop that covers finishing in house.
- 5Set the material gradeWrite the exact grade on the drawing, such as 6061-T6 or 316L. A general 'stainless' callout can swing the quote by a wide margin.
- 6Confirm the inspection planAgree on first article, in-process checks on critical features, and 100% inspection before shipment. Ask for reports up front.
- 7Name the process ownerGive the shop one contact and one revision. Keep the same program and fixture across repeat orders.
Questions engineers ask after the first quote
How tight a tolerance can CNC machining actually hold?
We work to ±0.005 mm (±0.0002 in) on critical features, with the right cutter and a controlled setup. That number only makes sense on features that need it.
Applied across a whole drawing, it raises cost sharply and leaves no margin for the features that matter. Name the critical dimensions and let the rest follow a general grade.
What surface finish should I call out?
As-machined surfaces sit around Ra 1.6–3.2 μm. A high-quality finish is Ra 0.8–1.6 μm, and fine finishing reaches Ra 0.2–0.8 μm.
Call out the finish only on faces that need it. Sealing faces and bearing surfaces usually do. Bracket sides rarely do.
When is 5-axis machining worth the extra cost?
When the part has features on multiple faces and the alternative is several setups. Each re-clamp adds variation, so one 5-axis setup often holds position better than three 3-axis operations.
For simple prismatic parts with features on one face, 3-axis is the cheaper and faster route.
Can you machine a prototype and then the production run?
Yes. There is no minimum order quantity, so a single prototype and a 10,000+ part run use the same process route. That avoids re-qualifying the part when it moves to production.
Quotation and free DFM analysis come back within 12 hours, and production can start within 24 hours.
How do you handle confidential drawings?
Uploads are secure and confidential. We can sign an NDA before you send the files, and access is limited to the team on your job.
Our quality systems include ISO 27001:2022 for information security, alongside ISO 9001:2015, IATF 16949:2016, and ISO 13485:2016.
What should I send for an accurate quote?
A STEP or native CAD file, the 2D drawing with tolerances and finish callouts, the material grade, the quantity, and any inspection or certification requirements.
If the drawing is still open, send it anyway. DFM feedback at that stage is where most of the seven mistakes get removed.
Send your drawing and get DFM feedback before you commit
Quotation and free DFM analysis within 12 hours. No minimum order quantity, from one prototype to 10,000+ parts.
12-hour quote100% inspectionNo MOQ