A CAPA effectiveness check template is the structured form you use to prove a corrective or preventive action actually eliminated the problem it was designed to fix. You build it around measurable acceptance criteria tied to the original failure, run it after the fix has had time to take hold, and record raw evidence that a reviewer can verify. Filling it out well is less about the form’s layout and more about the discipline of defining success before you start measuring.
The regulatory anchor changed on February 2, 2026. The FDA’s Quality Management System Regulation (QMSR) replaced the older Quality System Regulation, so 21 CFR 820.100 no longer exists. CAPA obligations now flow through ISO 13485 Clauses 8.5.2 and 8.5.3, which 21 CFR 820.10 incorporates by reference, and both clauses require that corrective actions be verified as effective and that the results be documented.1U.S. Food and Drug Administration. Quality Management System Regulation – Frequently Asked Questions2eCFR. 21 CFR 820.10 – Requirements for a Quality Management System If your template still cites §820.100, update the reference before your next audit.
Pull the Inputs From the CAPA File First
Every field on the template depends on data that already exists in the CAPA record. Gathering it up front prevents the check from drifting toward the wrong metric.
- CAPA tracking number. The unique identifier that links the effectiveness check back to the original investigation. Every document in the chain should carry it.
- Failure description. A plain account of the non-conformance or complaint, taken from the NCR or the initial complaint record rather than from memory.
- Confirmed root cause. The specific mechanism established during the investigation. The effectiveness check exists to verify this cause was eliminated, so a vague or disputed root cause makes the entire exercise unreliable.
- Implemented actions. The full list of what was done — procedural changes, retraining, hardware modifications, supplier changes.
- Baseline data. The problem metric as it stood before implementation. Without it, you have nothing to compare post-fix results against.
If any of these items is missing or ambiguous, resolve the gap before you start filling out the template. An effectiveness check built on an incomplete file is one of the fastest ways to draw scrutiny.
The Fields That Do the Work
Acceptance Criteria
This is the most important field on the form. It defines the specific, measurable standard the process has to hit for the CAPA to be considered successful. Wording like “monitor for improvement” or “ensure no recurrence” is functionally useless because it leaves the pass/fail call to subjective judgment. Write criteria that produce a binary outcome.
Good criteria look like “reject rate below 1.5 percent over ten production batches” or “zero repeat incidents of the same failure mode within three months of implementation.” The standard should relate directly to the root cause and the baseline you collected. If the original problem was a 4 percent seal failure rate, a criterion of “seal failure rate below 1 percent over the next 50 lots” gives you something concrete to measure against.
Sample Size
The number of units, records, or events you examine needs a documented rationale on the template. For attribute data (pass/fail, conforming/nonconforming), ANSI/ASQ Z1.4 tables match sample sizes to lot sizes and acceptable quality levels. For variable data (continuous measurements like pressure, weight, or temperature), ANSI/ASQ Z1.9 provides analogous plans built for measured values.
Risk level drives the confidence and reliability parameters. A Class III implantable device with a life-threatening failure mode demands higher confidence than a cosmetic defect on a Class I device. One common approach uses the success-run theorem: for 95 percent confidence and 95 percent reliability, you need 59 units with zero failures. Whichever method you use, record the justification on the template so an auditor can reconstruct your reasoning.
Verification Period
The timeframe has to be long enough for the process to generate meaningful volume, but short enough that a failing CAPA does not run uncorrected for months. Typical periods fall between three and six months. Some organizations count batches instead of days, which can be more meaningful when production runs are infrequent — “ten production batches after implementation” tells you more than “ninety days” if you only run a batch a month.
Verification Method
State the method on the form up front. Deciding how you will collect evidence after the fact is exactly the pattern auditors treat with justified suspicion.
Match the Method to the Corrective Action
A hardware fix calls for different evidence than a retraining effort. Pick the approach that actually tests whether the specific action worked.
Trend Analysis
Monitoring the problem metric over time is the most common method for high-volume processes. Compare post-implementation data against the pre-CAPA baseline and look for sustained improvement, not a brief dip. For attribute data, p-charts and np-charts track the proportion or count of nonconforming units across successive lots. For variable data, X-bar and R charts track process mean and variation. Using the wrong chart type can mask a shift or amplify noise, so the choice matters.
Focused Audits
When the corrective action changed a procedure or workflow, a focused audit checks whether people are actually following the new process. Review specific records, training logs, and operator behaviors tied to the change. This catches the failure mode where the fix exists on paper but nobody on the floor knows about it, a gap that trend data will not reveal until defects come back.
Physical Testing
Hardware changes, material substitutions, and design modifications call for direct testing. If the original failure was a seal leak, pressure-test a sample from post-implementation production. If a supplier change addressed a raw material contamination issue, run the incoming material through the relevant analytical tests. This produces the most concrete evidence, but only when the corrective action involved a tangible change to the product.
Behavioral Observation
For CAPAs rooted in human error, the check has to verify that people changed their behavior, not just that they sat through training. Quality staff can observe operators on the floor to confirm they follow the revised SOP. A more data-driven approach tracks deviations to the specific SOP step that was revised; if the training worked, deviation frequency at that step should drop to zero or near-zero over the verification period. Confirming that training records exist verifies the action was taken, not that it worked.
Running the Check and Signing It Off
Open the verification window only after the corrective action is fully implemented and enough production has occurred to generate representative data. Launching the check the week after a procedural change went live rarely produces meaningful results, and the waiting period also gives intermittent failure modes a realistic chance to resurface.
The person running the check should not be the person who implemented the corrective action. That separation provides objectivity, and industry practice calls for independent verification and sign-off. In a small operation where one person wears several hats, at minimum ensure final sign-off comes from someone outside the implementation team.
Record the actual data collected against each acceptance criterion. Not a summary, not a conclusion — the raw results. Any deviation from expected results gets documented with an explanation. If the data meets the criteria, the completed template moves to quality assurance, which confirms the evidence is complete and the methodology was followed as planned before approving closure.
Electronic Signatures
Most organizations now manage CAPA templates within a digital QMS. Electronic signatures used to approve the check must comply with 21 CFR Part 11: each signed record must display the signer’s printed name, the date and time the signature was executed, and the meaning of the signature (review, approval, responsibility, or authorship). The signature must be linked to the record so it cannot be copied or transferred to a different document, and it must be unique to one individual and never reassigned.3eCFR. 21 CFR Part 11 – Electronic Records; Electronic Signatures
Record Retention
ISO 13485 Clause 4.2.5 requires quality records to be kept at least for the lifetime of the medical device as defined by the organization, or as specified by applicable regulatory requirements, but no less than two years from the date the device was released. Most companies retain CAPA records well beyond two years because the defined device lifetime often runs longer. Keep both the completed template and its supporting evidence — raw data, charts, audit notes, test results — in your controlled document repository.
What to Do When the Check Fails
A failed effectiveness check means the corrective action did not eliminate the root cause. The CAPA stays open. Do not close it with a note calling the check “inconclusive” or “partially effective”; those characterizations are audit red flags.
- Reopen the investigation. Either the original root cause was wrong, or the corrective action did not adequately address it. Both possibilities require fresh analysis.
- Reassess risk and containment. Evaluate whether the ongoing nonconformance requires interim measures such as quarantining affected product, issuing field alerts, or increasing inspection frequency.
- Document the failure. Record what the check found, why the criteria were not met, and what happens next. Auditors specifically look for evidence that your system self-corrected when a CAPA failed rather than ignoring the data.4Food and Drug Administration. Warning Letters
Your workflow should have a hard stop that prevents closure when acceptance criteria are not met. Collecting effectiveness data and ignoring it is worse than not collecting it, because it proves you knew the fix was not working.
Mistakes That Draw Findings
FDA inspectors and ISO auditors see the same problems repeatedly on effectiveness check forms.
- Vague acceptance criteria. Every criterion should be a number, a threshold, or a zero-tolerance condition tied to the original failure mode.
- Verifying implementation instead of effectiveness. Confirming that a revised SOP was published or that operators were trained proves the action was taken; it does not prove the action worked. Measure the outcome.
- Unjustified sample sizes. Round numbers without a documented rationale invite questions. Tie the size to a recognized standard or a statistical calculation, and put that justification on the form.
- Verification periods that are too short. A week is not credible for failure modes that occur intermittently. Give the failure mode a fair chance to reappear.
- Closing despite failing data. When complaint rates climb or defects persist and the file closes anyway, a warning letter is the predictable outcome.
A well-built template forces the discipline the regulation expects: define what success looks like before you measure, pick a method proportionate to the risk, and produce a documented pass/fail result that either closes the loop or sends you back to fix what you missed.