Why Your Tolerance Stack-Up Keeps Failing

And how to avoid costly mistakes when sourcing custom parts.

Designing a part that works on paper but fails once assembled usually comes down to one thing: an impaired tolerance stack-up.

We’ve seen it across CNC, sheet metal, and even additive manufacturing. The design meets all functional requirements. However, one assembly dimension was off by 0.3 mm, and now the entire part is going to scrap.

Here’s why it happens and how to get it right, especially when ordering parts online.

The Most Common Mistakes in Tolerance Stack-Ups

1. Assuming zero deviation

It’s tempting to believe that parts will be machined or printed exactly to spec. In reality, every process introduces variation. The tighter your tolerances, the more complex (and more expensive) it is to guarantee them, and the more your stack-up matters.

Tip: Always treat tolerances as ranges, not targets. Worst-case scenarios matter more than average ones.

2. Overconstraining multiple features

We often see customers assigning tolerances to every hole or surface as if each one were critical. However, not all features require ±0.01 mm accuracy. Overconstraining adds cost, limits suppliers, and increases the risk of parts being rejected, even if they would have worked fine.

Instead: Use functional Technical Drawings or GD&T. Focus on what must fit, slide, rotate, or align. Leave the rest looser.

3. Using linear stack-ups only

If your stack-up calculation looks like a sum of worst-case linear values, you’re missing the real-world interplay of tolerances. Angular misalignments, non-rigid assemblies, and 3D variations all matter.

Better: Use statistical analysis or tolerance simulation tools if possible or talk to our team to sense-check complex assemblies.