From ISO 17025 to the Shop Floor: When Lab Standards Meet Manufacturing

There’s a conversation happening in the quality management world that most machinists and shop owners haven’t tuned into yet. It concerns ISO 17025 — the international standard for testing and calibration laboratories — and why its principles are increasingly showing up in manufacturing quality requirements.

I run a testing laboratory. My world is ISO 17025. But I’m writing this for a manufacturing audience because the boundary between “laboratory” and “shop floor” is dissolving faster than most people realize, and the shops that understand this early will have a significant competitive advantage.

What ISO 17025 actually requires (in plain language)

ISO 17025 is the global standard for laboratory competence. When a laboratory is accredited to ISO 17025, it means an independent assessment body has verified that the lab can produce technically valid test and calibration results.

The standard covers the entire measurement process:

Personnel competence. The people performing tests must be trained, qualified, and evaluated on an ongoing basis. Not just “they went to a training seminar once” — there must be documented evidence of initial qualification and periodic reassessment.

Method validation. Before a laboratory uses a test method on customer samples, it must demonstrate that the method produces reliable results for the specific material and measurement range. This involves running reference materials, assessing precision and bias, and documenting the conditions under which the method is valid.

Measurement uncertainty. Every test result must be accompanied by an estimated uncertainty — a quantified statement of how much the measured value might differ from the true value. This is calculated from all sources of variability: instrument resolution, environmental conditions, operator technique, sample preparation, and the method itself.

Traceability. Every measurement must be traceable to a recognized reference through an unbroken chain of calibrations. Your digital caliper’s calibration is traceable to gage blocks, which are traceable to a national metrology institute’s primary length standard.

Document control. Procedures, calibration records, personnel records, and test reports are all controlled documents. You can’t just print a procedure, stick it in a binder, and forget about it. There’s a formal process for review, revision, and distribution.

If you’re reading this list and thinking, “That sounds like it should apply to our quality lab too” — you’re right. And that’s exactly where the industry is heading.

The manufacturing measurement problem

Here’s the disconnect I see when I work with manufacturing companies:

A machinist produces a part to a tolerance of ±0.001”. He measures it with a digital micrometer that reads 1.2505”. The part is within tolerance. Ship it.

But how confident is he in that 1.2505” reading?

The micrometer was calibrated six months ago — is the calibration still valid? What’s the measurement uncertainty of the instrument? Was the part at thermal equilibrium when it was measured, or did it just come off the machine? Is the measuring force consistent with the calibration conditions? Has the micrometer been dropped, damaged, or stored improperly since calibration?

In a laboratory operating under ISO 17025, all of these factors are addressed before a measurement result is reported. In most machine shops, they’re not even considered.

I’m not saying this to be critical. Manufacturing and testing have historically been different disciplines with different quality frameworks. ISO 9001 covers manufacturing quality management. ISO 17025 covers testing and calibration. They coexist but rarely overlap in practice.

That’s changing.

Where the two worlds collide

Three forces are pushing ISO 17025 principles into the manufacturing environment:

Customer requirements are escalating. Aerospace (AS9100), medical (ISO 13485), and automotive (IATF 16949) quality standards increasingly reference measurement system analysis, measurement uncertainty, and calibration traceability. Some aerospace primes now require that their suppliers’ in-process inspection results include measurement uncertainty estimates — which is a direct import from ISO 17025.

Tolerance-to-uncertainty ratios are tightening. As parts get more precise, the ratio between the part tolerance and the measurement uncertainty becomes critically important. A widely accepted rule of thumb says your measurement uncertainty should be no more than 10–25% of the part tolerance. When you’re holding ±0.0005” on a bore, your measurement system needs to be capable of ±0.00005” to ±0.000125” uncertainty. That’s laboratory-grade metrology, not shop-floor estimation.

Regulatory scrutiny is increasing. The FDA’s current Good Manufacturing Practice (cGMP) requirements for medical devices explicitly require measurement system validation. The European Union’s Medical Device Regulation (MDR) has similar requirements. Shops that supply medical device manufacturers are finding that their customers’ regulatory obligations flow down to them through quality agreements.

Practical steps for shops (from a lab director)

I’m not suggesting that every machine shop needs to achieve ISO 17025 accreditation. That would be overkill for most operations. But there are ISO 17025 principles that every shop can and should adopt:

Calculate measurement uncertainty for your critical measurements. Start with your tightest tolerance features and the instruments you use to measure them. A basic uncertainty budget accounts for instrument resolution, calibration uncertainty, repeatability (measure the same feature 10 times and see the spread), reproducibility (have different operators measure and compare), and environmental effects (primarily temperature).

This doesn’t require a PhD in metrology. NIST has published free guides on measurement uncertainty estimation (GUM — Guide to the Expression of Uncertainty in Measurement). Your CMM manufacturer likely offers uncertainty analysis as part of their calibration service. Start there.

Maintain calibration records that actually mean something. “Calibration due: 06/2026” on a sticker is the minimum. Good calibration records include the as-found condition (what the instrument read before adjustment), the as-left condition (what it read after), the reference standards used, the environmental conditions during calibration, and the calculated uncertainty of the calibration itself.

If your calibration provider doesn’t give you this information, find one that does. And if you’re doing calibrations in-house, make sure your reference standards are traceable and current.

Control your measurement environment. Temperature is the single biggest source of measurement error in dimensional inspection, and most shops ignore it completely. Steel expands at approximately 11.7 µm/m/°C. A 100mm steel part measured at 25°C instead of the standard 20°C will read approximately 5.85 µm (0.00023”) larger than its true dimension at 20°C. That’s significant when you’re working to tight tolerances.

You don’t need a temperature-controlled clean room. But you should know the temperature of your measurement area, the temperature of the part when you measure it, and have a process for allowing parts to reach thermal equilibrium before critical measurements.

Document your measurement procedures. Which instrument do you use for which features? What technique do you use (single reading? average of three? where exactly do you place the contacts?)? What are the acceptance criteria? Write it down. If two operators measure the same feature differently because they have different techniques, your measurement system is adding variability to your process that doesn’t need to be there.

Participate in measurement comparison exercises. Some calibration labs and industry organizations offer interlaboratory comparison programs where participants measure the same artifact and compare results. If your customer’s incoming inspection disagrees with your final inspection, you have a measurement system problem that needs resolution before it becomes a quality problem.

The cultural shift

The hardest part of bringing laboratory discipline to the shop floor isn’t technical — it’s cultural. Machinists are skilled craftspeople who have been measuring parts their entire careers. Telling them that their measurement process needs improvement can feel like questioning their competence.

The key is framing it as a process improvement, not a personal criticism. Measurement uncertainty isn’t about whether the machinist is good at their job. It’s about understanding the limits of the measurement system so that everyone — the machinist, the quality inspector, and the customer — is working from the same set of assumptions.

In my laboratory, every technician understands that measurement uncertainty is not a weakness to hide — it’s a strength to quantify. The same mindset needs to take hold in manufacturing. Knowing that your measurement has ±0.0002” uncertainty isn’t a problem; not knowing your uncertainty is the problem.

Where this is heading

Five years from now, I expect that measurement uncertainty reporting will be a standard requirement in most precision manufacturing supply chains. The aerospace industry is already there. Medical is close behind. Automotive is moving in that direction.

The shops that start building this capability now — even informally, even imperfectly — will be better positioned than the shops that wait until a customer mandates it and scramble to comply.

ISO 17025 was written for laboratories. But its core principles — competence, traceability, uncertainty, and documentation — are universal principles of measurement excellence. They belong on the shop floor just as much as they belong in the lab.

Nour Abochama has directed ISO 17025 accredited laboratories for 17 years and holds a Master’s in Biomedical Engineering from Grenoble INP – Ense3.