Six repeat complaints, a nine-day closure, and a CAPA log that says "effective"

Separate the two concepts before judging the records.

• Correction = action to eliminate a detected nonconformity (ISO 9001:2015 clause 10.2.1 a, "react to the nonconformity and, as applicable, take action to correct it"). Quarantine, rework, scrap, and replacing the affected boards are corrections. So is training the named operator, if training only closes the immediate symptom.
• Corrective action = action to eliminate the cause so the nonconformity does not recur (10.2.1 b–f: evaluate the need for action to eliminate causes, determine and implement it, review effectiveness, update risks and opportunities determined during planning, make changes to the QMS if necessary).

Arvada performed corrections and labelled them corrective actions. Three pieces of evidence drive that conclusion:

1. The pattern contradicts the cause identified. Six recurrences across different lines and shifts make operator error an implausible single cause; the common factors are the process, the stencil/screen condition, and the material batch - none of which appears in the record.
2. No effectiveness review is demonstrable. "Closed, effective" with no post-action defect data at the same defect code is an assertion, not evidence.
3. 10.2.1 e–f were skipped. Nothing was updated in the FMEA or control plan, and no change to the QMS was considered, although the recurrence pattern is exactly the trigger for both.

Clause numbers follow ISO 9001:2015; verify the numbering of the edition in force for your exam, as the numbering of the 2026 revision is not identical.

Required documented information (10.2.2):

• the nature of the nonconformities and any subsequent actions taken;
• the results of any corrective action.

The standard does not require a specific form, an 8D, or a five-why template. It requires that these two things be retrievable, which is where Arvada's log fails: a record that says "retrained operator" does not describe the nature of the nonconformity at system level, and "closed, effective" is not a result.

Requests that test effectiveness:

1. Baseline and trend: the solder-mask registration defect rate for the previous 12 months, by line and shift, before and after each action. If the rate is unchanged, effectiveness was never demonstrated.
2. Cause analysis record: what alternatives were considered and rejected, and on what data. Look for whether machine parameters, stencil wear, and material batch were eliminated.
3. Coverage: evidence that the same process on the other two lines was examined, not just the line named in the complaint.
4. Update trail: the revision history of the FMEA and control plan showing a change dated after the third or fourth recurrence, plus the reasoning for the new inspection point.
5. Effectiveness review date and criterion: what number, measured when, would have shown the action failed.

A useful test question: if the same defect appears next month, what in these records would tell you the organization saw it coming?

Question 1 - scope of the cause. "Which other lines, shifts, machines and material batches run this same process, and what did you compare between them before deciding the cause?" A weak answer names only the line in the complaint, or says the process is "the same everywhere" without having compared parameters. A strong answer names the shared stencil supplier and the tension check that was identical on all three lines.

Question 2 - decision and residual risk. "What did your analysis conclude the cause was, and what residual risk remains if you are wrong? Where is that recorded?" This tests 6.1 and 10.2.1 e together. A weak answer restates the action ("we retrained") rather than the cause, and cannot say what risk was accepted. A strong answer names a mechanical cause, cites the parameter data, and shows a dated update to the risk register or FMEA with a residual-risk statement.

Question 3 - change to the system. "What changed in the QMS itself as a result - control plan, inspection interval, supplier requirement, competence criteria - and how will you know in 90 days whether it worked?" A weak answer points to a training record that already existed before the complaint. A strong answer points to a revised control plan revision number, a changed screen-tension interval with an owner, and a named metric reviewed at a set date.

Note that "we retrained the operator" can be a legitimate answer if competence evidence actually pointed that way - but with six recurrences across different shifts, the burden is on the organization to show why that cause survived scrutiny.