Method Validation vs Verification in ISO 17025

Every method on your scope needs performance evidence, and the amount depends on where the method came from. ISO/IEC 17025 splits that evidence into two routes: verification and validation. Lab managers often mix them up when sorting out method validation vs verification in ISO 17025, and the confusion shows at assessment, when a method file holds either too little evidence or the wrong kind.

This walkthrough sorts each method on your scope into the right route, using the logic of clause 7.2. It also covers what the assessor expects to see in each file, which performance characteristics to cover, and the mistakes that most often turn into nonconformities. The question of method validation vs verification in ISO 17025 can be settled method by method, before anyone external asks.

Assessors from accreditation bodies work under the same international framework, and accreditation to this standard is recognised across borders through the ILAC Mutual Recognition Arrangement and its signatories, which is one more reason the method evidence has to stand up.

The one question that decides it: is the method yours, modified, or used exactly as published?

Start with the origin of the method and how you apply it. Every method on your scope falls into one of three groups:

Method validation vs verification in ISO 17025: three lab benches, one for each method route
  1. A standard method used as published. The method comes from an issuing body (an international, regional or national standard, or a recognised technical organisation) and you run it as written, within its stated scope.
  2. A standard method that you modified or used outside its scope. You changed a step, a sample size, a matrix, a range or an instrument condition, or you apply the method to something it was not written for.
  3. A non-standard or lab-developed method. You designed it, adapted it from literature, took it from a manufacturer’s instructions, or built it for a customer need that no published method covers.

The routes follow directly:

  • Group 1 requires verification: evidence that your lab can achieve the method’s stated performance.
  • Groups 2 and 3 require validation: evidence that the method itself is fit for its intended use.

Verification asks, “Can we do this method properly?” Validation asks, “Does this method do what we claim it does?” The first question assumes someone else has already answered the second.

The test is whether anyone has already established the method’s performance, and whether your use matches what they established. If you changed something, the earlier work no longer covers you.

When verification is enough: standard methods used as written

Clause 7.2.1.5 requires the lab to verify that it can properly perform a method before introducing it, by confirming it can achieve the required performance. Records of verification must be kept. If the issuing body revises the method, verification is repeated to the extent necessary.

Three conditions have to hold for verification alone to be enough:

  • The method is current. Clause 7.2.1.3 expects you to use the latest valid version unless that is inappropriate or impossible, so a superseded edition on the bench is a problem before any performance data comes into play.
  • You follow it as written, with no changed steps, volumes, temperatures or acceptance rules.
  • The sample type and measuring range fall within the scope the standard states.

Verification is the lighter route, but it is not a paperwork exercise. The standard has already described the performance you should expect, often through precision or recovery data from a collaborative study. Your job is to show your own staff, equipment and environment reproduce it. That usually means running the method on reference materials, spiked samples, certified materials or a proficiency-style sample, and comparing the results against the method’s stated performance or your own acceptance criteria.

Analyst loading certified reference material into an autosampler for method verification

Where the standard gives no performance data, you set your own criteria based on the customer’s needs and demonstrate that you meet them.

Verification in a microbiology lab: ISO 4833-1. A food microbiology lab adds ISO 4833-1, the horizontal method for enumerating microorganisms by colony count at 30 °C using the pour plate technique, to its scope. The lab follows the method as published, on the food types the standard covers. Before reporting results, it verifies the method: analysts run the pour plate procedure on reference material or spiked samples with a known level of target organisms, check that recovery and repeatability meet the criteria the lab set, confirm media and incubator performance, and record each analyst’s competence. No new method is designed, so verification is the right route.

Verification in a chemistry lab: ISO 17294-2. A chemistry lab adds ISO 17294-2, the ICP-MS method for determining selected elements in water, to its scope. It runs the method as published, on the water types the standard covers. Before reporting, it verifies: analysts run certified reference material and spiked samples, check recovery and repeatability against the method’s stated performance, confirm detection limits are low enough for the customer’s needs, and record each analyst’s competence. Nothing about the method is changed, so verification is enough.

When you must validate: non-standard, lab-developed and modified methods

Clause 7.2.2.1 lists four cases where the lab must validate:

  • non-standard methods
  • laboratory-designed or developed methods
  • standard methods used outside their intended scope
  • amplifications or modifications of standard methods

Nobody else has established the performance of these methods under your conditions, so the evidence has to come from you. Validation must be as extensive as needed to meet the needs of the application. A method that supports a high-risk decision, or one that a customer or regulator depends on, needs deeper evidence than a method used for internal screening.

Two practical points trip labs up here:

  • A small modification still counts. Clause 7.2.2 treats modifications of standard methods as validation territory. Some modifications have little effect on performance, but the lab has to determine that and record it, since the original performance claims no longer cover you automatically. The depth of work can be proportionate to the change.
  • Validation is not a one-time event. When you change a validated method, you determine how the change affects the original validation. If the effect is meaningful, you validate again.

Validation in a chemistry lab: a lab-developed ICP-MS method. A testing lab is asked to report lead, cadmium, arsenic and mercury in a herbal tea, a matrix that no published standard method it holds covers. The lab builds its own ICP-MS method, with its own microwave digestion and internal standards. Because it is lab-developed and applied to a new matrix, the lab validates it. The plan, written before any data is collected, covers selectivity and interferences, linearity, limits of detection and quantification, trueness through certified reference material and spike recovery, precision, and matrix effects, and it ties the results to the measurement uncertainty estimate. The same lab running a published ICP-MS method for elements in water as written, on its listed matrix, would only verify.

Validation in a microbiology lab: a modified ISO 4833-1. The food microbiology lab from the verification example wants to replace the pour plate step with spiral plating to speed up its throughput, and to apply it to a food type outside the standard’s stated scope. That is a modification and an extension of scope, so the lab validates. Its plan, written first, compares the modified method against the reference method on naturally contaminated and spiked samples across the food type, and covers relative trueness, repeatability, the working range, selectivity against background flora, and the limit of quantification, with results tied to measurement uncertainty.

For a deeper treatment of how to plan and run a validation study, see the site’s article on ISO 17025 method validation and laboratory competence. This article stays with the decision about which route applies.

What an assessor expects to find in each method file

An assessor sampling your scope will usually pull the file for a method and look for a straight line from the method’s origin to the evidence. Build each file so that line is easy to follow.

Assessor reviewing a ring binder of method validation records

Verification file

  • The method identity: the standard’s number and the edition in use, with confirmation it is the current valid version.
  • A statement that the method is used as published, and that the sample type and range are within its scope.
  • The performance requirements you verified against: either the standard’s stated values or your own criteria, with the reason for them.
  • The verification data: raw results, who ran them, when, on which equipment, and with which reference materials or samples.
  • A comparison of results to the acceptance criteria, and a conclusion signed off by an authorised person.
  • Evidence of analyst competence on this method, since verification is partly about the lab’s ability to perform it.
  • A record of re-verification when the standard was revised.

Validation file

  • A description of the method and its intended use, including the customer or regulatory requirements it must meet.
  • The validation plan or procedure, written before the work began, with the performance characteristics selected and the acceptance criteria for each.
  • The determination of each performance characteristic, with the raw data and calculations.
  • The results obtained and a statement on whether the method is valid for its intended use.
  • The link to measurement uncertainty estimation (clause 7.6), since the validation data typically feeds it.
  • A change history showing what happened when the method was later modified.

In both cases, assessors check that the data is genuine, traceable to the instruments used, and signed off. A file that holds only a conclusion without the data behind it will not pass.

Performance characteristics to cover, and how much is enough

Clause 7.2.2 points to the characteristics a validation may need to address. Typical examples include:

  • measurement range
  • accuracy of results, expressed through trueness and precision
  • selectivity
  • linearity
  • sensitivity (including limits of detection and quantification where relevant)
  • robustness
  • cross-sensitivity or interference

You do not need every item for every method. Select the characteristics that matter for the intended use, and write down why you left others out. A qualitative test has no use for linearity. A method that reports a result at a fixed decision limit needs evidence around that limit more than across a wide range.

How much is enough depends on the application. Use these questions to decide:

  • What decision does the customer make with this result?
  • How close to a limit or tolerance will results typically fall?
  • How variable is the matrix or sample type?
  • Is the method new to the lab, or has it been performed for years with stable quality control data?

The answers set the number of replicates, the number of concentration levels, and how many analysts, days or instruments you include. Whatever you choose, record the reasoning next to the data. An assessor accepts a proportionate plan with a clear rationale far more readily than an exhaustive study with no explanation.

For verification, the scope is narrower. You normally confirm the characteristics the standard claims and that matter to your use, typically precision, trueness or recovery, and the working range or reporting limit you will rely on.

Common mistakes that raise nonconformities on method evidence

Several patterns come up repeatedly in method files:

  • Treating a modified standard method as a standard method. Labs change a sample prep step, shorten a run time or extend a matrix, then file the method under verification. The modification moves it into validation.
  • Verifying against a superseded edition. The data is real, but it supports the wrong version of the method.
  • Verification with no acceptance criteria. Results were collected, yet the file never states what counted as acceptable, so no conclusion can be checked.
  • Validation plan written after the data. If the criteria appear after the results, they look fitted to the outcome. Write the plan first.
  • Characteristics chosen without a reason. A list of tests with no link to the intended use suggests a template, not a decision.
  • No evidence of analyst competence. The method was verified once by one person, and other staff now perform it with nothing on file.
  • No re-verification after a revision. The standard changed, the lab kept running the old verification, and nobody assessed what the change affected.
  • Validation data that never reaches the uncertainty budget. Precision and bias figures sit in one file while the uncertainty estimate uses different numbers.
  • Scope drift. A method verified for one matrix is now being used for another, and the file does not cover it.

Most of these are easy to catch with a periodic review of the method list against the files.

A worked example: sorting five methods on a lab’s scope

Consider a hypothetical testing lab with five methods on its scope. The examples are illustrative, not a real lab’s records.

Five method folders with flags sorting each method into verification or validation

Method 1: a published international standard, run exactly as written, on the sample type it names. Route: verification. The file needs the current edition, a statement of conformity to the stated scope, and the lab’s own precision and recovery data against the standard’s performance claims or the lab’s criteria.

Method 2: a published standard where the lab shortened the extraction step to speed turnaround. Route: validation of the modification. The lab must show the changed step does not harm performance. The plan compares results from the original and modified procedures and records the acceptance criteria in advance.

Method 3: a published standard applied to a sample matrix the standard does not list. Route: validation. Use outside the intended scope is one of the four cases in clause 7.2.2.1. Matrix effects and recovery in the new matrix are the main characteristics to examine.

Method 4: an in-house method developed for a customer request that no standard covers. Route: full validation. The file needs the customer’s requirements, a plan, and data for the characteristics that matter to the use: probably range, trueness, precision, selectivity and robustness, along with the link to uncertainty.

Method 5: a standard method the lab has run for years, where the issuing body has just released a revised edition. Route: re-verification to the extent necessary. The lab compares the revised text against the old one, identifies what changed, and repeats the verification work that those changes touch. The file records the comparison and the decision about what to repeat.

The sorting result is two verifications (Methods 1 and 5), and three validations (Methods 2, 3 and 4), with the depth of each validation set by the risk of the decision behind the result. A simple table with columns for method, origin, route, performance characteristics required, file location and review date makes this exercise repeatable across the whole scope.

Verification is not calibration: where to go for the instrument side

Method verification and method validation deal with whether a method performs properly. Calibration deals with whether an instrument gives results traceable to reference standards. Both are required, and one does not substitute for the other. A balance with a current calibration certificate can still be part of a method nobody verified, and a verified method run on an instrument with an overdue calibration has an equally weak foundation.

If your lab is deciding how verification activities relate to calibration and where each applies, the site’s article Verification vs Calibration Lab: Which Do You Need? covers the instrument side in detail.

Sorting your scope this way takes a few hours, and it tells you which files need work before the assessor opens them. Pair the sorting table with a method-by-method checklist, and look through the lab quality tools and training resources recommended on this site, which can help you build validation and verification records before your next assessment.

🕒 Book Your Free 45-Minute Consultation

Have questions about ISO/IEC 17025 or ISO 9001 implementation or accreditation? Schedule a free 45-minute consultation with me to discuss your Company or laboratory’s needs and how we can achieve compliance together.

Schedule Your Consultation

Similar Posts

Leave a Reply

Your email address will not be published. Required fields are marked *