Engineering note

A 200-Part Order, 3 Rejections, and 9 Days: A Custom CNC Machining Service Sourcing Story

2026-11-17 Evelyn Chen
Precision manufacturing engineering article visual

The Call at 7:40 AM

Tuesday, September 10, 2024. 7:40 in the morning, and my phone buzzed before I'd finished my first coffee. A mechanical engineer at a consumer hardware startup was on the other end. If you've spent any time in custom CNC machining service sourcing, you already know roughly how these calls go.

Their product launch was set for September 18. They needed 200 custom aluminum brackets, anodized to a specific RAL color from their brand palette, delivered to their San Francisco office. Normal lead time for that kind of job? Three to four weeks. They had eight working days.

What made it worse: 200 pieces is a weird size. Too small for a real CNC machining factory — those shops usually dump the RFQ into an automated quote system and send an auto-reject a few hours later. Too custom for a low-volume job shop that wants to treat every order as a prototype. So you end up in the dead zone.

I took the call.

Four Hours, Three No's

The first thing I did was not send the drawings. It was make phone calls. Nine of them in the first hour. That's basically the job — you want to verify the shop will even touch the order before you spend time quoting it.

Three different custom CNC machining manufacturers said no outright.

The first one: "We're booked through the week. Three weeks minimum." No mention of a rush option. Nothing.

The second: "Two hundred pieces? That's prototyping, not production." Meaning, we'll price you out. And they did — $34 a bracket. The part itself is worth maybe $8 to $12.

The third was the most honest: "Nine days, we can't fit it. It's not that we don't want to do it, we literally can't slot it in."

By 1 PM we were down to five candidate shops. Two were small outfits that do great prototype work but couldn't handle the anodizing. One had the capacity but outsourced finishing, which added four days on its own. That math put us at 13 — nine days went out the window.

Honestly, that's when I started to think maybe we couldn't pull it off.

The Cheaper Option That Almost Cost Us the Launch

Here's where the story gets uncomfortable for me.

A broker I work with regularly tipped me off to a CNC machining factory I hadn't used before. They quoted roughly 30% less than the two finalists on my list. And they promised five days.

Someone warned me about them. Told me the last time they used that shop, the part surfaces came back wrong and they lost a week. I didn't listen.

Here's what was going through my head: the client was a small team. Budgets are tight at that stage. If I could get them a 30% saving on their first order, maybe that earns a long-term relationship. My logic was: small customer, big priority. In my role coordinating sourcing for smaller orders, that logic applies most of the time.

It was a dumb call in this specific case. I'm not going to pretend otherwise.

Five days later, the parts were finished. Ninety minutes after that, we caught the problem: 12 of the 200 brackets had contaminated surfaces. The anodizing bath hadn't been cleaned properly before the parts went in. Friday morning, six working days from the launch.

I called the shop. Could they redo 12 pieces? They said: "Five days." Not going to work.

The Turnaround: Back to the Known List

So I went back to a shop I'd worked with in 2022 — a small custom CNC machining service outfit in the Bay Area that specializes in prototypes and low volumes, and runs a small in-house anodizing tank. I'd left them off the first-round shortlist because their capacity ceiling worried me for 200 pieces. But now I only needed 12.

The operations manager on the phone had been doing this for over twenty years — the kind of person who has heard every panicked Friday call. I asked for a 72-hour turnaround on 12 pieces, same spec, same RAL color. She asked three questions: alloy, tolerance, and Milspec color code. Then she said: "Send the drawing. I'll have a quote to you by 3 PM."

They delivered in 71 hours. And this time — no surprises, no drama. Just a clean batch of brackets, packed individually, dropped at our shipping partner's dock on Monday morning, September 16.

The client had built in a two-day buffer. The launch went ahead on September 18. Nobody outside our shop knew the difference.

What It Actually Cost

I want to walk through the numbers, because this is where the lesson lives. I'm not going to pretend I remember every line item — some of these I'm rounding, some I've reconstructed from my own notes.

The initial "cheap" quote: about $1,500 for 200 pieces, or roughly $7.50 each. That's roughly $780 less than the two finalists we'd shortlisted earlier in the day.

The defect cost: 12 pieces from the small Bay Area shop, plus 72-hour rush, plus air freight to get them to the client by Monday. Together, somewhere around $1,100.

Total damage: roughly $2,600 — about $320 more than if we'd just gone with our original shortlist pick at $11.40 a piece, with zero drama, first time.

Not catastrophic. But that's not the part that bothers me. What bothers me is the two days I spent making phone calls, chasing photos, and explaining to the client that their launch might slip. The $320 was a line item. The two days were the actual cost.

What I Actually Took Away From This

I don't want to turn this into a lecture. Here's just what I've changed since.

Total cost of ownership includes: base part price, setup fees, shipping, rush fees, potential rework, and time lost. The lowest quoted number is often not the lowest total number.

That quote has been on my wall for a while. I followed it in theory and broke it in practice. September 10 reminded me why it's on the wall.

The other thing — and this is where I'll probably get disagreement — is that small orders get a raw deal in this industry. 200 pieces isn't a big contract for a CNC machining manufacturer. It gets auto-rejected or marked up three times to make it go away. But small orders have their own logic: pilot runs, seed-stage shipments, market validation, design iterations. Vendors who won't spare a second thought for them are leaving qualified buyers on the table for someone else.

The shop that came through for us didn't do it through charm. They did it by saying what they could do, saying what they couldn't, and not promising things outside their control. That's the whole thing. That's the whole job, honestly.

Three rules I now apply to every rush order

  1. Call first, send drawings second. A two-minute phone call saves an hour of email ping-pong, and it filters out the shops that were going to auto-reject you anyway.
  2. Price the rework, not just the quote. The cheapest number on the sheet is almost never the final cost once you factor in finishing, tolerances, and the chance you'll have to redo 5% of the batch.
  3. Ask: "If something goes wrong, what's the process?" An answer that sounds vague when things are calm will sound worse when they aren't. The shop that answers that question clearly is usually the shop worth keeping.

One Last Thing

The client sent a photo after the launch. All 200 brackets installed on the new device, lights on the staging table, the RAL color matched perfectly against their brand spec. Her note read: "Thanks for saying yes to this one."

That order made us maybe $1,800 in gross profit. The note is worth more than that. That's not a sales line — that's the reason I still pick up the phone at 7:40 in the morning.

Ask About This Topic Back to Resources
Evelyn Chen

Evelyn Chen

Evelyn Chen is an independent manufacturing quality and sourcing analyst covering DFM reviews, supplier qualification, FAI, PPAP, CMM reporting, traceability, and low-volume-to-production transitions. She relates ISO 9001 controls to drawing revision, calibration status, process capability, material certificates, sampling plans, change management, lead time, and total landed cost. Her procurement guides help OEM teams compare quotations, define evidence packages, manage supplier risk, and avoid hidden assumptions before purchase orders or production releases.