The problem we were actually trying to solve
Every shop knows the message. A buyer asks for "just a rough price." You send one. Then silence — not because the number was wrong, but because they were never serious about placing an order. They wanted a free ballpark to benchmark somewhere else.
We wanted to stop spending engineer time on those dead-end first passes. So instead of quoting manually, we built a tool that lets a buyer estimate a part themselves, straight from a 2D PDF drawing. No sign-up, no waiting, no human on the other end.
Day one: it was wrong by a mile
We shipped it. Within hours, a real order landed on the desk — 40 aluminium spacers, roughly 25 × 25 × 13 mm, real parts cost about $180. Our brand-new tool estimated $97.
That gap isn't a rounding error. At that distance, the tool isn't a convenience — it's misinformation. We could have quietly tuned it. Instead, we posted the screenshot on LinkedIn and asked peers to tear it apart.
What the peers told us (and what we changed)
The feedback was specific and useful, not polite. Three points kept coming up:
- "Your machine-time per part is too low." Small parts aren't cheaper in proportion — they still need setup, fixturing, and handling. We added a base minutes-per-part term so a 25 mm part isn't priced like a scaled-down 250 mm part.
- "You can't scale small parts linearly." Our dimension parser now reads diameters and x-separated dimensions (mm and inch), and estimates mass from the actual envelope instead of a fixed default.
- "Don't fake freight." Air-express cost for a 0.3 kg parcel and a 30 kg pallet are completely different animals. Rather than show a number we couldn't stand behind, the tool now asks the buyer to confirm freight with us.
The result of those changes: tested against the same real drawings, the estimate moved from "wrong by multiples" to "close enough to be useful." Not a substitute for a real quote — a fast first pass that respects the buyer's time.
Why we're telling this story
We're not a software company. We're a CNC shop that got tired of guessing and decided to build the thing we wished existed. Every improvement above came from someone who does this job every day telling us where we were wrong.
That's the model we want to keep. We're not going to keep the tool in a drawer. After one more round of polishing, we'll open it as a standalone page that any shop's customer can use — not just ours.
Two ways to help shape the next version
If you run CNC quoting daily, we'd rather hear from you than guess:
→ Try the estimator now and tell us what it gets wrong on your typical parts.
→ Or send us a drawing and we'll show you the real number next to the estimate.
FSCPRECISION is a Shenzhen-based CNC machining shop serving automotive-aftermarket and aerospace supply chains in the EU and US. This article is part of an open build log — we publish what we learn as we build tools for our own shop floor.