Xometry Tolerances: A Quality Inspector on CNC, Sheet Metal, and Resin Reality

There is no honest single-number answer to 'what are Xometry tolerances?', and anyone who gives you one without asking what you are making is saving you from a decision they should not be making for you. I'm a quality and brand compliance manager at a small manufacturing company. I review roughly 200+ unique components every year, from first articles to full production batches. I also buy parts through Xometry when our internal capacity is tight. The failures I see are rarely because the vendor was sloppy. More often, the part was quoted against a drawing that treated tolerance as a marketing number instead of a manufacturing plan.

In practical terms, I separate the work into four situations: CNC machined metal parts that need to fit together, sheet metal parts with bends, resin 3D printed parts that are still at the prototype stage, and the bigger question of whether your own shop should become a Xometry supplier. Each one has a different tolerance conversation, and mixing them up is the fastest way to overpay or under-specify.

Xometry tolerances are not one number

Searching for 'Xometry tolerances' usually happens right after someone opens the quoting engine and sees standard and precision options. The phrase sounds like a fixed rule, but it is actually a per-process decision. A machined bracket, a sheet metal enclosure, and a resin SLA prototype have completely different physics behind them.

According to Xometry's own CNC machining page, standard CNC tolerances typically run around plus or minus 0.005 inches, with precision available on critical dimensions depending on material and geometry. That is a useful starting point, not a contract to put on every feature. If you copy the same tight tolerance across the entire part because it feels safer, you are paying for capability you will never use.

My general rule before any quote: fit, form, and function decide tolerance, not how impressive the decimal looks. If the feature controls a bearing fit or a mating surface, hold it. If it is a cosmetic edge on a bracket that will never touch another component, let it breathe. That is not lazy engineering. It is TCO thinking: unit price, lead time, inspection cost, and rework risk are all connected to tolerance.

Fit, form, and function decide tolerance, not how impressive the decimal looks.

Scenario A: CNC machined metal parts and the axis of a CNC milling machine

Let me start with the scenario I see most often: you need CNC machined metal parts that will be assembled with other components. Before you add precision tolerances to every dimension, think about what happens inside the machine making the part.

The axis of a CNC milling machine is where this story lives. On a standard 3-axis machine, the X and Y axes control horizontal position, the Z axis controls depth, and every move carries a small amount of positioning error. Backlash in a ball screw, thermal growth in the spindle, tool deflection, and even a worn guide way can show up as a feature that is slightly out of place. Skilled machinists compensate for these variables, but they cannot compensate for a drawing that demands a tight tolerance on a non-critical feature and a loose tolerance on the one hole that actually matters.

My advice for this scenario is straightforward: use Xometry's standard tolerance as the default across the whole part, then select precision only for the features that determine function. In Q1 2024, our engineers asked for a quote on an aluminum bracket with three precision tolerance callouts across the entire model. The quoted price was noticeably higher than the same part with standard tolerances and precision only on the two bores that located the assembly. The part functioned identically, and the supplier had an easier time holding the features that actually mattered.

This is also where asking for a first article report pays for itself. A first article inspection tells you whether the axis of the CNC milling machine used for your part was holding position on your critical dimensions. It is cheaper to catch an issue on one part before a production run of 200 parts than after delivery.

Scenario B: Sheet metal parts, press brake bends, and Adira press brake problems

Sheet metal tolerances behave differently because the geometry is created by bending, not by cutting away material. A CNC mill removes metal with a spindle that holds position. A press brake bends metal by forcing it into a die, and springback means the final angle rarely equals the programmed angle. If you are ordering bent parts, the important dimensions are usually hole-to-bend distances, flatness across the part, and the bend angle itself.

If you run a press brake in your own shop, you already know that older machines can create strange intermittent quality issues. We run an Adira press brake, and it is a solid machine, but it has taught me to think carefully about how bend angles drift. When people search for 'Adira press brake problems', the most common pattern I see is not a sudden catastrophic failure. It is a gradual loss of parallelism between the upper beam and the lower die. The ram comes down slightly unevenly across the length of the bed, so one end of the part bends a little more than the other. Operators then try to correct the angle by adjusting bend allowance in the control, but the actual correction should come from checking the ram level, the tooling, and the back gauge repeatability.

One counterintuitive lesson: before you blame the machine, measure the material. Steel thickness can vary from one coil to another, and a few thousandths of an inch difference in material thickness will change the bend angle, especially if you are air bending. I have seen shops spend days chasing an Adira press brake problem that turned out to be a batch of material with inconsistent thickness. The machine was fine. The input was the problem.

For tolerance purposes, I recommend treating sheet metal bends as a process to qualify, not a number to guess. Run a test piece, measure the angle at both ends and the center, and confirm the hole-to-bend dimensions before you quote a tight tolerance on a production drawing. If you are sourcing bent parts through Xometry, do the same thing: make sure the vendor has clear bend reliefs and reasonable tolerances on the drawing, and always inspect the first article before authorizing the full quantity. The lowest quote is only the lowest total cost if the parts arrive flat, square, and within spec.

Scenario C: Resin 3D printed parts and how to change resin in a 3D printer

Resin 3D printing is the scenario where engineers most often over-specify tolerances. I understand why. Resin parts look smooth and detailed, so it is tempting to hold them to machined tolerances. But resin shrinks during printing, shifts slightly during post-curing, and can absorb moisture afterward. A resin prototype is usually meant to validate fit, ergonomics, or visual appearance, not to replace a machined part. If you need machined tolerance, machine the part. If you need a prototype, treat the resin part as a prototype.

That said, there is one resin workflow issue that causes more quality failures than tolerance ever will: changing the resin incorrectly. People often type 'how to change resin in 3d printer' into a search engine because they assume it is like swapping filament on an FDM printer. It is not. You cannot just pull out one material and push in another without cleaning the vat.

The safest process I use in our shop is this:

1. Drain the vat through a filter back into a labeled bottle, and keep the bottle away from direct light.
2. Clean the vat with a plastic spatula and lint-free wipes. Do not use paper towels, because they can leave fibers that show up as defects in the next print.
3. Inspect the FEP film for clouding, scratches, or dents. A damaged FEP will ruin prints no matter how carefully you level the build plate.
4. Clean the build plate and remove any cured residue from previous prints.
5. Refill with the new resin, and run a small calibration print before committing to a production batch.

The counterintuitive part is that most sudden quality drops after a resin change are not caused by bad calibration. They are caused by leftover old resin, settled pigment, or tiny cured particles stuck to the FEP. I have rejected parts that looked like they had a layer shift, only to discover that the real issue was contamination from a sloppy resin change. So glad we caught that one early. A failed batch of 200 print jobs would have been an expensive lesson.

Should your shop become a Xometry supplier?

If you run a machine shop or a sheet metal shop, you have probably searched 'Xometry become a supplier' at some point. The idea is attractive because the platform brings work to you instead of you chasing RFQs. But from my side of the quality process, there is an important distinction: becoming a supplier means you are agreeing to hold someone else's tolerances consistently, not just when conditions are perfect.

The signup process is straightforward, and Xometry's supplier application page lists the capabilities and certifications they need from your shop. But before you apply, be honest about your process controls. Can you produce first article documentation? Do you know how your press brake behaves across a full production run? Do you have a way to verify that the axis of your CNC milling machine is still within its original positioning specification? If the answer is no, no amount of quoting volume will fix that gap.

You also do not need the newest machine in the world. An older Adira press brake can absolutely produce conforming parts if it is maintained and verified. The platform is not paying for machine cosmetics. It is paying for parts that pass inspection. The shops that succeed are the ones that treat every order as if a quality inspector is going to measure every critical dimension on arrival, because that is exactly what I do.

How to decide which scenario applies to you

Instead of making you sort through general advice, here is the classification I use when someone asks me about tolerances:

• Your part is machined metal and will be assembled with other components: start with Xometry standard tolerance, then apply precision to the features that control fit. Ask for a first article report on critical dimensions.
• Your part is sheet metal with bends: focus on material thickness, bend radius, and hole-to-bend dimensions. Qualify the process with a test piece before production.
• Your part is a resin 3D print: treat it as a prototype or visual model. Do not demand machined tolerances on a part that will shrink and cure differently. And change your resin properly.
• Your own shop makes machined or sheet metal parts: look into becoming a Xometry supplier, but only after your process controls and inspection practices are ready for outside scrutiny.

If you still are not sure which bucket you are in, ask a different question: what happens if this feature is 0.002 inches off? If the answer is nothing, loosen it. If the answer is that the part will not assemble, tighten it. If you honestly do not know, get one part made, measure it, and learn from the first article. That data is worth more than any generic tolerance chart.

And keep the total cost view in mind. A quote that looks cheap on the surface can become expensive after inspection, rework, and delayed assembly. A slightly higher quote from a supplier who understands which tolerances matter is often the real bargain. Xometry's instant quoting engine makes it easy to compare standard and precision options in real time, but the engine cannot tell your story. Only you know how the part will be used. That is why tolerance is not just a number on a drawing. It is the language between the person who designs the part and the person who has to make it right.