ISO 17025 Method Validation: What the Standard Actually Requires — and the Questions Every Brand Should Be Asking Their Lab

Most supplement and cosmetic brands we speak with have the same assumption: if a lab is ISO 17025 accredited, its test results are trustworthy by definition. That’s not entirely wrong — but it’s incomplete in a way that causes real problems.

ISO/IEC 17025:2017 accreditation means a laboratory has demonstrated competence in specific testing activities, within a specific scope. That scope matters enormously. A lab can hold valid accreditation for a particular method on a particular matrix and simultaneously run other methods — methods outside its accreditation scope — with little to no formal validation behind them. You’d never know unless you asked.

Method validation is the mechanism that gives test results their meaning. Understanding what ISO 17025 actually requires — and where labs commonly fall short — is one of the more practically useful things a brand can know about the testing process.

What ISO/IEC 17025:2017 Clause 7.2 Actually Demands

The 2017 revision of ISO/IEC 17025 addresses method validation in Clause 7.2. The standard draws a clear line between two activities that often get conflated: validation and verification.

Validation applies to non-standard methods, laboratory-developed methods, or standard methods used outside their intended scope. When a lab adapts a published USP method for a different matrix, or develops an in-house LC-MS/MS method for a novel analyte, full validation is required. That means generating experimental data demonstrating the method is fit for its specific purpose — not just running a few spikes and calling it done.

Verification applies when a lab uses a recognized standard method exactly as published — say, AOAC 2016.12 for total metals in food, run on the matrix and concentration range the method was designed for. Here, the lab confirms it can reproduce the published performance characteristics. Less work than a full validation, but still a formal documented exercise. A printout of the AOAC procedure stapled to an instrument report doesn’t satisfy Clause 7.2.

The performance characteristics ISO 17025 expects labs to establish — through validation or verification, as appropriate — include:

• Accuracy (trueness): expressed as percent recovery from certified reference materials or spiked samples. For food and supplement matrices, 80–110% recovery is the commonly cited benchmark, though method-specific guidance may tighten or widen that window.
• Precision: both repeatability (within a single run, same day) and intermediate precision (across different days, analysts, or instruments). A %RSD of ≤5% is a reasonable expectation for most chromatographic or spectrometric methods, though complex botanical matrices can push that higher.
• Specificity/selectivity: the method’s ability to measure the target analyte in the presence of everything else in the sample. This is especially critical in botanical supplement matrices where co-eluting compounds are common.
• Linearity: the correlation between analyte concentration and instrument response, typically verified by r² ≥ 0.999 across the calibration range for instrumental methods.
• LOD and LOQ: the lowest concentrations the method can detect and reliably quantify. The LOQ, in particular, must sit below your specification limit — otherwise a “non-detect” result is scientifically empty.
• Range: the concentration interval over which all the above parameters hold simultaneously.
• Robustness: the method’s sensitivity to small, deliberate variations in operating conditions — column lot, mobile phase pH, extraction time. Even a basic ruggedness assessment is better than none.

One requirement that consistently surprises clients: ISO/IEC 17025:2017 Clause 7.6 mandates that labs evaluate and report measurement uncertainty for every result where it’s relevant to validity or compliance with a specification. This isn’t optional. And it’s not the same as a standard deviation. A proper uncertainty budget accounts for every source of variability in the measurement process — calibration standard uncertainty, sample preparation steps, instrument drift, matrix effects, analyst-to-analyst variability — and combines them into a single estimate of the range within which the true value is expected to fall.

Where Method Validation Most Commonly Falls Short

Even labs that take accreditation seriously can have gaps in their method validation programs. These are the failure modes we encounter most often when reviewing inter-laboratory discrepancies or subcontractor documentation.

Matrix-specific validation is absent. This is the most consequential gap. A method validated on aqueous calibration standards may perform very differently when applied to a gummy matrix, a protein powder, or an oil-based botanical extract. Phospholipids, pigments, sugars, and surfactants all suppress or enhance ionization in mass spectrometry; they interfere with extraction efficiency; they shift recovery curves. If a lab validated heavy metals by ICP-MS using diluted aqueous standards and then applies those calibration factors to a turmeric-curcumin softgel, the recovery data doesn’t support that application. The accreditation scope document should confirm the method was validated on a matrix comparable to yours — and if it doesn’t, you should ask.

Intermediate precision is underreported. Repeatability — running the same sample six times in one analytical sequence — is easy to generate and easy to look good on. Intermediate precision, measured across different days, different analysts, or different instruments, tells you how the method will actually perform in routine operation over time. Method validation summaries worth trusting will show precision data spanning at least three non-consecutive days, not a single validation event.

Measurement uncertainty isn’t propagated to results. Despite the explicit requirement in Clause 7.6, a significant share of COAs circulating in the industry carry no uncertainty estimate. That omission isn’t always a sign of a bad lab — some clients haven’t requested it, and labs don’t always volunteer it. But when a result sits near a specification limit — arsenic at 9.7 µg/g against a 10 µg/g spec, for instance — the expanded uncertainty is the only number that tells you whether “passing” is analytically defensible.

LOQ vs. specification limit misalignment. We’ve reviewed COAs where the reported LOQ for lead was 0.5 mg/kg on a product with a specification of exactly 0.5 mg/kg. A result of ”< LOQ” on that product tells you nothing useful. The LOQ must sit below the action level — typically by a factor of at least 5, and preferably 10 — to make the result interpretable for compliance purposes.

Scope creep beyond accreditation boundaries. A lab may hold accreditation for Method X on matrices A through C. If a client sends a matrix D sample and the lab runs the same method without revalidating for that matrix, they’re operating outside their accredited scope. The American Association for Laboratory Accreditation (A2LA) and Perry Johnson Laboratory Accreditation (PJLA) both publish searchable scope documents online — it takes about five minutes to cross-reference your sample type against what’s listed.

Four Questions Every Brand Should Ask Before Accepting a COA

The point isn’t that labs are being careless. Most ISO 17025 accredited labs we work alongside take their quality systems seriously. But method validation is technically demanding, documentation-heavy work, and brands are well within their rights to understand what sits behind the results they’re paying for.

1. Is this method in your accreditation scope for my specific product matrix? Ask the lab to point you to the relevant line in their A2LA or PJLA scope document. If they can’t, or if the listed matrix doesn’t match your product type, ask whether the method has been formally verified or validated for your matrix — and request the supporting data.

2. What is the LOQ for each analyte, and how does it compare to my specification limits? This is a quick calculation. If the LOQ is within a factor of 2 or 3 of your spec limit, you don’t have meaningful analytical headroom. Any result near that boundary is ambiguous.

3. Can I see the method validation summary or verification protocol? Reputable labs keep these on file and should be willing to share a summary — sometimes under a mutual NDA if the method is proprietary. Look for: percent recovery data by matrix type, precision data across multiple days, the calibration range with linearity coefficients, and a list of which validation parameters were assessed versus assumed.

4. What is the reported measurement uncertainty for this analysis, and can it appear on my COA? If the answer is “we don’t calculate uncertainty,” that’s a significant gap for an ISO 17025 accredited lab. If the answer is “we calculate it but don’t report it by default,” ask them to add it. When your results are ever reviewed by a retailer, an Amazon compliance team, or an FDA investigator, having documented uncertainty on file demonstrates exactly the kind of analytical rigor those audiences are looking for.

Accreditation Is a Floor, Not a Guarantee

ISO/IEC 17025:2017 accreditation is genuinely meaningful. The assessment process — conducted by A2LA, PJLA, or another ILAC-recognized accreditation body — involves rigorous on-site technical evaluation, equipment calibration record review, quality management system audits, and inter-laboratory proficiency performance data. Labs that maintain accreditation over multiple reassessment cycles are doing something substantively right.

But the standard’s value is only realized when labs apply it faithfully — including on method validation and uncertainty reporting. An accreditation certificate tells you that an independent technical body evaluated the lab’s systems at a point in time. It doesn’t tell you that every result you receive was generated under the full protection of those systems for your specific sample and matrix.

The scope document, the validation package, and the measurement uncertainty are what connect your COA to the analytical confidence the standard was designed to provide.

When you’re choosing a testing partner for a product destined for retail shelves, Amazon’s marketplace, or regulatory review, the right question isn’t “are you ISO 17025 accredited?” It’s “can you show me the validation data that backs up this specific method on this specific matrix?”

The answer to that second question tells you a great deal more than the certificate ever could.

---

Written by Nour Abochama, Vice President of Operations, Qalitex Laboratories. Learn more about our team

Talk to our team about your testing needs. Contact us

Related from our network

• FDA Software Validation and 21 CFR Part 11 Compliance for Quality Systems — Aurora TIC covers regulatory consulting for GAMP5, software validation, and FDA audit readiness.
• COA Verification and Supplier Qualification for Raw Material Testing — Ayah Labs helps global buyers verify certificates of analysis and qualify raw material suppliers against USP, Ph.Eur., and BP monographs.
• ISO 17025 Accredited NHP and Pharmaceutical Testing for Canadian Brands — Androxa provides Health Canada–compliant testing for natural health products and pharmaceuticals sold in Canada.