Why do corrective actions fail to stop the same complaint?

This answer covers corrective actions for repeat customer complaints about a supplied product, such as a chemical shipment, under a quality management system. It does not cover how to write the 8D report itself, and it does not replace sector rules such as medical device or pharmaceutical CAPA regulations, which add their own verification requirements.

What are the most likely reasons a corrective action does not stop a repeat complaint?

The most likely reasons a corrective action does not stop a repeat complaint are: the action names no cause link it blocks, containment was recorded as the fix, the success criterion was written after the results, and the action stayed on one product or site. Each of the four leaves its own trace in the corrective action record, so a reviewer can test for it without waiting for another complaint.

No named cause link

A corrective action that says "retrain operators" or "improve inspection" without naming the step in the cause chain it removes cannot be tested. If the real cause sits one link upstream, the complaint returns while the action is marked complete.

Containment taken as the fix

Sorting, re-filtering, relabeling or a tightened outgoing check stops bad material from leaving. The cause stays in place, so the defect rate is unchanged and the next lot that slips through becomes the next complaint.

Success criterion set after the results

A criterion chosen after the data is in can be fitted to whatever the data shows, so the action cannot fail the check. One CAPA practitioner notes that organizations "confuse verification of implementation and verification of effectiveness activities" (Durivage, 2020).

Action not extended to similar products and sites

ETQ lists failure to carry lessons "to other sites, products and processes" as one of five reasons corrective actions fall short (Ease.io, 2018).

How can you tell which link in the cause chain a corrective action actually blocks?

You can tell which link in the cause chain a corrective action blocks by writing the chain as a sequence of conditions from origin to complaint, then marking the one condition the action makes impossible or detectable. If you cannot mark a link, the action blocks none. Put the customer's complaint as the last link and work back to the earliest condition your team controls. For each link, name the evidence that would show the condition no longer occurs, such as a moisture reading, a log entry or a dated check record. The table below decides which flaw in the corrective action explains a repeat complaint.

The defect-rate row is easy to miss: when escapes drop and the defect rate at the source does not, the corrective action was containment. As an illustrative chain for haze in a drummed fluid: washed drums are not dried, water stays in the drum, the fill picks up water, the customer sees haze. An action that adds a drum dryness check blocks the second link; a tightened outgoing appearance check blocks none of them.

How do you set an effectiveness criterion before the corrective action runs?

You set an effectiveness criterion before the corrective action runs by writing four items into the corrective action record when the action is approved: the success criterion, the stop criterion, the person who judges, and the deadline for the verdict. A later criterion can be fitted to the result.

• Success criterion: a measurable result tied to the blocked link, such as no recurrence of the defect in the next three production lots (Durivage, 2020). Write the measure, the sample and the limit.

• Stop criterion: the result that ends the trial early and sends the case back to analysis, such as one confirmed recurrence from the same cause path (illustrative).

• Judge: the person who compares the result with the criterion and signs the verdict.

• Deadline: the date or lot count by which the verdict is due.

Durivage calls this a verification of effectiveness plan that documents "predetermined criteria," and writes that "if the predetermined acceptance criteria are not met, the CAPA fails" (Durivage, 2020). Keep verification of implementation separate: proof that the dryness check is in use shows the action happened, not that it worked. A criterion that only says "no further complaints" depends on the customer noticing and reporting, so add a measure your own team takes at the blocked link. Record both, in that order, so the record shows the criterion predates the data. ISO 13485:2016, a sector standard that ISO confirmed in 2025, also requires reviewing the effectiveness of corrective action taken (ISO 13485:2016, clause 8.5.2).

Who should judge whether a corrective action was effective, and by when?

The person who judges whether a corrective action was effective should be someone who did not design the action and who owns the measure in the success criterion, and the verdict is due on the date or lot count fixed before the action ran. In a supplier quality system, that can be the quality manager or the customer-facing quality engineer rather than the production lead who implemented the change.

The deadline depends on how fast the cause path can repeat. An action against a defect that appeared in one lot of ten needs enough lots for a recurrence to have a fair chance to show (illustrative). Practitioners on one quality forum raise the same question: how many opportunities should a corrective action get before it counts as effective (Elsmar Cove, 2012). Answer it in the record, not at closure.

Certification schemes set outer limits. Under the IATF Rules 6th edition, in force since 1 January 2025, the certification body assesses on site whether corrective actions for audit nonconformities worked: a minor nonconformity at the next regular audit or in a special audit, a major one in a special audit (TÜV NORD, undated). The same rules keep a 60-day deadline for nonconformity management, including the client's own verification of effectiveness (TÜV NORD, undated). Customer agreements may set shorter deadlines. Write the earliest of these into the corrective action record, and close the action as ineffective if the verdict date passes with the criterion unmet.

When should a repeat complaint reopen the root cause analysis?

A repeat complaint should reopen the root cause analysis when the corrective action blocked its named link, the defect rate at the source fell, and the same complaint still returned, because that pattern points to a second cause path. Reopen also when the stop criterion is met, or when the verdict deadline passes with the success criterion unmet.

Close the first corrective action as ineffective and open a new one that references it, so the history stays visible (Durivage, 2020). Do not rewrite the original success criterion to fit the new data. In the new analysis, start from the evidence the first action produced: which link it blocked, what the defect rate did, and which lots or sites still showed the complaint. IATF 16949:2016 clause 10.2.3 also asks that the corrective action consider "the impact on similar processes and products" and that documents such as the PFMEA and control plan be reviewed (SMMT Industry Forum, 2018).

Bring in a specialist from the product's maker or a test laboratory when the second cause path involves chemistry or a test method your team cannot check. Lubinpla's troubleshooting returns containment, a cause verdict, countermeasures and a success criterion fixed in advance, and leaves the effectiveness verdict to the people who run the action. Record the reopened case against the original complaint number.