I've spent six years placing custom manufacturing orders across CNC machining, injection molding, and 3D printing. In that time, I've personally made-and documented-4 significant mistakes that totaled roughly $9,000 in wasted budget.
This article is a direct result of those failures. I'm sharing the exact pre-submission checklist I now use before uploading any CAD file to a digital manufacturing platform. It's not theoretical advice. It's a list born from real, expensive screw-ups.
When to use this checklist
Use this before you hit Request Quote on Fictiv or any similar manufacturing platform. It's designed for:
- Engineers submitting a design for the first time
- Teams who are moving from traditional machine shops to a digital platform
- Anyone ordering multiple parts or using multiple manufacturing processes
It takes 15 minutes. It will save you from 1-2 week delays and hundreds (or thousands) of dollars in rework.
The 5-step checklist
1. Check your CAD file's export format and units
This sounds ridiculously basic, but it's where my first mistake happened. In 2019, I uploaded a STEP file to a platform, didn't notice the units were set to inches, and got a part that was 25.4 times larger than it should have been. The entire order was scrap.
Action: Explicitly state the units in the order notes. Don't assume it's obvious. Checkpoint: Go into your CAD model, check the physical properties, and confirm the units match your intent.
Honestly, the golden rule is: your file should be exactly what you'd hand to a machinist in person. If you'd say, "It's in inches," then put that in the notes.
2NCheck your material and finish spec against the process
The most common error in this phase is not the material itself, it's the finish. People assume that specifying the material is enough. But a 3D printed part in gray resin will need different post-processing than a CNC part in polycarbonate.
Action: For each part, define both the material and the surface finish requirements separately. Checkpoint: If you need a smooth, paint-ready surface, say so. If you need a machined finish only, say that too.
I once ordered 300 CNC parts with a 2B finish that I didn't actually need. It looked great, but it added $0.42 per part and just delayed the project. That's $126 I'll never see again.
3. Redefine your tolerances beyond the basic dimensions
This is a subtle one. You might have a callout for a critical bore diameter of ±0.005mm, but what about the distance between that bore and a mounting hole? Or the flatness of the surface that it mounts to?
Action: Create a tolerance map for your part. Don't just rely on the default tolerances your software applies. Checkpoint: If you wouldn't check it with a micrometer, don't expect the vendor to nail it first try without a clear tolerance.
This gets into GD&T territory, which isn't my expertise. What I can tell you from a design-engineering perspective is that being specific about datums and surface profile requirements prevents a whole class of misunderstandings.
4. Count your critical-to-quality (CTQ) features
Here's the step that most people ignore. Not the dimensions, but the features that absolutely must be right for the part to function at all. For example, a snap-fit tab is a CTQ. The aesthetic surface next to it is not.
Action: List your CTQs in the notes. For example: "CTQ: The boss height 5.1mm must be exact. The slot width is reference only." Checkpoint: If your part doesn't work when this dimension is off by a hair, it's a CTQ.
On Fictiv, you can add notes to specific features in the DFM feedback. Use that space. It tells the engineer (and the software) where to focus their inspection.
People assume the vendor will just know what's important. What they don't see is that for a high-volume order, the platform's automated quoting system may prioritize faster turnaround over a nuanced finish that's only in your head.
5. Review the DFM feedback literally, not aspirationally
When the platform gives you Design for Manufacturing feedback, it's tempting to read the green "Proceed to Checkout" button and assume everything is fine. The reality is that DFM feedback often contains warnings that won't prevent you from ordering, but will affect the outcome.
Action: Read every line of DFM feedback. If it says "wall thickness below recommended for injection molding," your part may have cosmetic sink marks or structural issues. Don't proceed unless you've accepted that risk. Checkpoint: Print the DFM report, mark it up, and file it with your project documentation. Yes, even for digital orders.
In Q1 2024, after the third rejection on a nylon 12 3D-printed part, I finally realized the issue wasn't the printer. It was that my design had a latticed section that the platform's automated support could not generate correctly with the default settings. The DFM feedback had actually flagged this as a "possible quality risk." I missed it because I was in a rush. I should add that we've caught 47 potential errors using this checklist in the past 18 months. 47.
Notes on selecting materials for the main processes
This gets back to the common question I get about materials and process capability. It's tempting to think you can just ask for a carbon-fiber-reinforced part and it will behave like a solid carbon sheet. But it depends entirely on the process:
- CNC machining: You're cutting it from a solid block. Here, a "flying fiber laser coder customized" (a fancy way to say you need a custom-coded, laser-marked texture) is actually a valid secondary operation for marking part numbers, logos, or QR codes. It's not an additive process; it's a subtractive marking process.
- Injection molding origins: The whole reason we have injection molding is for high-volume, repeatable plastic parts. If you want a carbon-fiber look without the cost, sometimes a molded texture or paint is the smart play. But that adds a tooling texture spec to your drawing (that's a critical checkpoint in my Step 2).
- 3D printing: Regarding the constant question, "can 3d printers print carbon fiber?" Yes, but it's usually a carbon-fiber-filled polymer (like nylon), not pure carbon. It's stiffer and more dimensionally stable than unfilled nylon, but it's not isotropic like a machined block of aluminum. If you need a truly structural carbon fiber part, that's generally a layup process, not typical 3D printing.
The over-rating problem
From the outside, it looks like more capability is always better. The reality is that a skilled engineer will pick the right tool, not the most expensive one. For example, for a quick prototype of a bracket, a white resin print might be 80% as good as a machined aluminum part, but at 30% of the cost. And honestly, the lead time is way faster.
Pricing realities (not a marketing promise)
Based on publicly listed quotes I've seen on Fictiv's platform in early 2025 (and other platforms I won't name, since I'm not going to bash them), CNC machining is your premium option, injection molding is your cost-per-part winner at scale, and 3D printing is the flexibility champion for low volumes.
A sample price point (not a guarantee): a simple 1-inch aluminum 6061 block, 3D-printed, might be $15-25 at low volume. The same part CNC'd might be $35-60, based on quotes from January 2025. Those are just reference numbers, not a price promise. Verify current rates before your order.
The cheapest insurance is a phone call
One thing the digital checklist can't replace? Talking to a human. When I submitted my "flying fiber laser coder customized" logo spec for a transparent acrylic part, I got a DFM note back saying the laser would leave a dark ablation mark that's normal. I called, they offered a test piece for $20 before running the full 100. That test saved me $780 in wasted parts (plus the embarrassment of shipping something with ugly, unreliable marks).
Final common mistakes in the ordering process
To wrap up, here are three things I see people (including my past self) get wrong:
- Ignoring the DFM note about "cosmetic finish acceptable." If it's not acceptable to you, you need to talk to a person. Or change the finish spec.
- Assuming your team member's CAD file is right just because they gave it to you. You must verify the export format matches the process. A STEP file is universal, but if you're using a specific slicer or 5-axis CNC, an STL may not be enough.
- Not defining the inspection report level. A basic CMM report might just say "in spec." If you need a full First Article Inspection (FAI) report, that should be in the order notes before you click buy, not after you get the parts.
Use this list. It's the cheapest 15-minute insurance you'll ever buy. (Mental note: I really should add a note about packaging for fragile parts. *Adds to list.*)
