Effective CAPA: Why Corrective Actions Often Miss the Mark — and How to Fix That

2026-10-04

Executive Summary

• A CAPA process that runs as a formality usually isn’t short of forms or people. The root cause analysis is being done superficially and the effectiveness check is being skipped.
• ISO 9001:2015 (clause 10.2) separates correction (dealing with what happened) from corrective action (eliminating the cause so it doesn’t recur). Confusing the two is the most common error.
• A CAPA loop that adds value goes through four steps: detect → analyze → correct → verify effectiveness. Closing records on time is not the finish line.
• An eQMS can enforce the full sequence, send reminders and surface repeat failures, but it won’t think about root causes for you.
• A checklist at the end helps you review your current CAPA process.

Closed on paper, still open in the plant

A scene many quality professionals will recognize: a corrective action record is completed, signed off, closed on time. Three months later, the same defect reappears. A new record is opened. The cause is written as “operator not attentive,” the action as “remind and retrain.” Closed. Three months later, the defect is back.

The system still passes audits, because the records are complete. But operationally, CAPA has failed at the one thing it exists to do: eliminate the cause so the problem doesn’t come back.

This article explains why that happens so often, and what a CAPA loop that works actually looks like.

What CAPA is and what it’s actually for

A word on terminology first. CAPA (corrective and preventive action) is the familiar name in many regulated industries. ISO 9001:2015 uses “nonconformity and corrective action” (clause 10.2) and no longer has a separate clause for preventive action as the 2008 edition did. In the 2015 edition the preventive role sits within risk-based thinking. Companies in practice still say “CAPA” for both ideas, and this article uses it that way.

The statements about the standard below are based on ISO 9001:2015, clause 10.2, summarized in practical terms. When a nonconformity occurs, the organization must:

• react to it: control it, correct it and deal with the consequences;
• evaluate the need for action to eliminate the cause, so it doesn’t recur or occur elsewhere, by reviewing and analyzing the nonconformity, determining its causes, and determining whether similar nonconformities exist or could occur;
• implement the action needed;
• review the effectiveness of any corrective action taken;
• update risks and opportunities if necessary, and make changes to the quality management system if necessary.

Corrective actions must be appropriate to the effects of the nonconformity. The organization must retain documented information as evidence of the nature of the nonconformity, the actions taken, and the results of any corrective action.

Read that list carefully and you’ll see the standard centers on cause and effectiveness, not on forms.

Why CAPA becomes box-ticking

From that list, it’s easy to see where CAPA tends to break. Across many manufacturing operations, six failures show up most often. (A general observation, not measured figures.)

1. Confusing correction with corrective action. Correction deals with what happened: scrap the faulty batch, rework, fix the record. Corrective action eliminates the cause so it doesn’t recur. A record that only describes reworking the batch has treated the symptom and left the cause alone.

2. “Human error” as root cause. “Operator not attentive” or “procedure not followed” is the easiest answer to write and the least useful. It stops exactly where the questions should begin: why did an ordinary person get this wrong under these conditions? Was the instruction clear? Does the equipment invite mistakes? Were they trained on the right revision?

3. “Retrain” as the default action. Retraining is easy to record and easy to prove it was done. But if the real cause lies in the document, the equipment or the process, retraining changes nothing.

4. No effectiveness review. This step often gets dropped because it comes late, after the record has already “finished” in form. Nobody remembers to go back and check.

5. No check elsewhere. The standard asks whether similar nonconformities exist or could occur elsewhere. This is rarely done: the record solves one case, not one type.

6. The metric is on-time closure. When the quality team’s indicator is “percentage of records closed on time,” people optimize for closing, not for eliminating causes.

Illustrative example (a composite scenario, not a real customer case): a packaging area has a recurring bag-seal defect. The first two records give the cause as “operator technique” and the action as “retrain.” The defect keeps appearing. On the third record the team asks further and finds that the changeover settings sheet at the line is an older revision than the one released. This time the actions are quite different: replace the sheet at the machine, change how obsolete copies are withdrawn, and check other lines of the same type. This is also where CAPA touches document control.

Root cause analysis: the most skipped step

ISO 9001 doesn’t prescribe a method for analyzing causes. Organizations choose: 5 Whys, a fishbone diagram, or another structured way of questioning. The method matters less than the depth of the questions.

A few questions that push analysis past “human error”:

• Why did these conditions allow the mistake?
• Did a document, the equipment or the process contribute?
• What was different from the times it was done right?
• If this cause is fixed, will the event definitely not recur?
• Does anywhere else share the same conditions?

A practical warning sign: if the root cause on the record could apply to almost any nonconformity (“not careful,” “lack of control”), it isn’t specific enough to lead to action that works.

Proportion matters too. Not every nonconformity needs deep analysis by a multi-person meeting. A single, minor lapse may need only correction and a note; a recurring or high-impact failure deserves the investment. The standard asks for action appropriate to the effects.

The effective loop: detect → analyze → correct → verify

From clause 10.2, a CAPA loop that adds value passes through four steps, each with a verification question.

The last step deserves emphasis. “Action taken” and “action effective” are different things. Only the second shows the CAPA loop is complete.

CAPA in eQMS

To be clear first: an eQMS doesn’t analyze root causes for you. A record saying “operator not careful” is still a weak record, even inside software.

What an eQMS changes is how much the loop depends on human memory:

It enforces the sequence. The process defines steps, owners and deadlines up front. A record can’t close without a cause analysis or an effectiveness check plan.

It reminds. The effectiveness review, often forgotten because it comes late, is scheduled and prompted by the system.

It adds context. Records link to related documents, batches, equipment, suppliers and training, making causes easier to trace.

It shows repeats. With records in one system, spotting “this is the third time the same type of defect has appeared in the same area” becomes feasible — nearly impossible when each record is a separate file.

It keeps evidence. Who analyzed, who approved, and what the effectiveness check found are recorded, supporting the requirement to retain evidence.

A deeper point: the root cause of many nonconformities lies in knowledge — who understands this equipment, why a setting was chosen, how it was handled last time. When that knowledge lives only in a few heads, every CAPA starts from scratch. This is where the quality system meets the problem of organizational knowledge.

A practical checklist

Use these questions to review your current CAPA process.

1. Does your process distinguish correction from corrective action, and do users understand the difference?
2. Are root causes on recent records specific enough to lead to action, or mostly “human error”?
3. What share of records have “retrain” as the action, and do any check whether the cause lies in a document, equipment or process?
4. Do you check elsewhere for similar nonconformities?
5. Does every action have a specific owner and deadline?
6. Is the effectiveness check scheduled when the record is opened, with real evidence of the result?
7. Can you spot repeat failures (same type, same area) from the pooled records?
8. Is action proportionate to the effects, avoiding heavy analysis for small things and light handling for big ones?
9. Does the quality team’s metric reward on-time closure more than problems not recurring?
10. Do CAPA findings lead to changes in documents, processes, training or risks where needed?

If three or more answers are “no” or “not sure,” the issue likely lies in how CAPA is run, not in missing forms or tools.

A note on ISO 9001 editions

This article refers to ISO 9001:2015. According to certification bodies, a revised ISO 9001:2026 was scheduled for publication in September 2026, with a transition period expected to run about three years and 2015 certificates remaining valid during it. Clause structure and wording may change; check against the edition your organization is certified to.

Is your CAPA process reaching causes, or just closing records?

→ Take the Knowledge Management Maturity Assessment to identify your maturity level and the first improvement to make.

Further reading:

• Document Control in eQMS: What It Really Means to Control a Document (previous)
• Nonconformance in ISO 9001: More Than an Error Report (next)
• Evidence-Based AI: When Answers Need to Be Verifiable (related)
• eQMS for Manufacturing SMEs: A Practical Guide (pillar)

Sources:

• ISO 9001:2015, clause 10.2 (nonconformity and corrective action): iso.org/standard/62085.html
• LRQA, ISO 9001 revision update: Publication date confirmed (certification body source; verify with ISO): lrqa.com

ISO 9001:2015 clauses are summarized in practical language and do not replace the text of the standard; compliance decisions should be checked against the official edition. The example is a composite illustrative scenario based on common patterns in manufacturing SMEs, not a specific customer case. The failure patterns described are general observations, not measured figures.