Last quarter, I rejected 1,500 injection-molded housings because the wall thickness wandered about 0.015" outside tolerance. The vendor said it was “within industry standard.” It wasn't in our spec. We fought for a week, they re-molded everything, and we still shipped late. This wasn't a one-off, and it wasn't bad luck.
The Surface Problem: Parts That Don't Meet Spec
When a part shows up out of tolerance, the obvious suspect is the machine. Maybe the tool is worn. Maybe the operator rushed. For a CNC and lathe machine job, you can chase spindle accuracy and tool wear for days. But most of the time, the machine is fine. The real issue is the gap between what you specified and what the supplier thought they were selling.
I'm not a design engineer, so I won't lecture you on draft angles or feed rates. What I can tell you from a quality perspective is this: tolerance disputes are rarely about capability. They're about process control and communication. A vendor quotes your drawing, but they're practically promising to try. Samples run under ideal conditions. Production is a different story.
What I've learned is that most suppliers aren't trying to fool you. They genuinely think they can hit the tolerance. But their quality control only measures a few features, or they inspect the first article and assume the rest of the run will behave. That assumption is where the risk lives.
Terminology doesn't help. In a machine shop, VMC means vertical machining center. If you're searching for “vmc stock dividend” or “who made vmc drinks,” you're in a different universe entirely. I once watched a new buyer spend an entire afternoon looking at the wrong VMC because they didn't realize the acronym had multiple meanings. The same confusion happens with Fictiv: people assume it's just a 3D printing website, so they never look at Fictiv injection molding or CNC capabilities. They make decisions based on a partial picture.
Why This Keeps Happening: You're Evaluating the Wrong Signals
After four years of reviewing first articles and vendor audits, I've got a biased view. But the pattern is consistent. Teams compare quotes, delivery dates, and website polish. They rarely compare how a supplier handles specification risk.
Here's something vendors won't tell you: the person answering your RFQ is often a salesperson, not a process engineer. They want the order. So they say yes to tolerances, quantities, and lead times that their production floor may not actually hold. Later, when parts fail, the conversation becomes “what is industry standard” instead of “what did we agree to.” That's the deeper problem.
The other signal I see ignored is the quote's assumptions. Does it include a material certificate? Does it specify inspection points? Does it say what happens if the part drifts mid-run? If a quote is just a number with a lead time, the vendor hasn't told you how they'll prove the part is right. And if you're not buying proof, you're buying hope.
Also, capability and capacity are not the same thing. A shop may have a five-axis machine and still not have the right inspection process for your part. It may have capacity to take your order and not the experience to make that specific feature. To be fair, smaller job shops can be incredibly good. But you can't verify that from a quote alone. And if you don't verify it, you're gambling with your lead time, not just your unit price.
The Real Cost of a Cheap Quote
Let's talk about what “cheap” actually costs. The first article looks good. The unit price is 12% below the next vendor. Then the first production run fails inspection, and you're stuck with non-conforming inventory and a launch date that doesn't move.
Last year, we approved a 50,000-unit annual order based on a supplier's sample. The first delivery was 8,000 units. They looked fine out of the box, but after two weeks in the warehouse they warped. The supplier had swapped a material grade to protect their margin. The redo cost $22,000, the launch slipped three weeks, and our supply chain team lost credibility. The cheap quote wasn't cheap.
That $22,000 doesn't count the hours spent on email threads, the emergency meetings, or the fact that engineering had to redesign the mounting boss to accommodate the replacement material. It doesn't count the opportunity cost of not launching on time. This is why I now think in total cost, not per-unit cost.
Looking back, I should have audited that supplier before production, not after samples. At the time, the audit felt like overhead. It wasn't. The same way you wouldn't bet a production line on an unverified machine, you shouldn't bet it on an unverified supplier.
What I Actually Do Now (and When I Don't Use Fictiv)
I'm not going to pretend there's one recommendation for everyone. The right supplier depends on the part, the volume, and the lifecycle stage. But for prototyping, short-run production, and projects where design changes are still coming, I've shifted a lot of work to Fictiv.
Full disclosure: Fictiv doesn't pay me. I'm sharing this because visibility is what separates a good quote from a gamble.
On the Fictiv website, you upload your part and get instant quotes across CNC machining, injection molding, and 3D printing. More importantly, you get design-for-manufacturing feedback before you spend money on tooling. That early feedback has caught more bad assumptions than any post-production inspection ever did.
Fictiv injection molding, for example, is not a black box. The platform reviews the part for draft, wall thickness, and gate placement before the mold is cut. Compare that with a typical quote cycle where those issues only surface after steel is machined and parts are sampled. On the machining side, if you need a CNC and lathe machine operation on the same part, the platform handles the handoff without you acting as the messenger between two shops.
But there are cases where I'd choose a specialized local shop instead. If you're running millions of parts with custom automation, or if you need an engineer physically present for every production run, a local partner is usually the better fit. Fictiv is not the best answer for every supply chain. No manufacturing platform is. The honest answer is to match the platform—or shop—to the actual problem.
So the next time you get a bad batch, don't only blame the machine. Look at how you selected the partner. Did you compare qualifications or just quotes? Did you understand the acronyms and the process limits? Did you look at the full manufacturing picture, or just the easiest search result?
Most quality problems are purchased before a single chip is cut. You can reduce them with better supplier selection. Fictiv helps because it makes manufacturing capabilities visible. But the visibility only helps if you use it.
