Corrective & Preventive Action (CAPA)

Almost every organization has a CAPA system. A great many have one that does not prevent anything.

The pattern is recognizable. CAPAs are opened, worked and closed within the target period. The metrics look healthy. And the same nonconformity reappears next quarter under a new number, because of one or more of these:

• Containment is recorded as correction. The affected product was sorted, the customer was protected, the CAPA was closed. Nothing about the process changed, so the process produced the same defect again.
• The root cause is a person. Investigation concluded operator error; the action was retraining; the CAPA closed. Retraining a competent person who made a predictable mistake in a poorly designed process changes nothing.
• Effectiveness was never verified. The action was implemented and the CAPA closed on implementation rather than on evidence it worked.
• The CAPA system is overloaded. Everything raises a CAPA, so investigators have no time to investigate anything properly, and the whole system degrades to administration.

That last one is worth dwelling on, because it is the most common and the least suspected. A CAPA system used for everything becomes a system that solves nothing.

• Clear triage — what warrants a full CAPA, what is handled as a simple correction, and what escalates immediately. Protecting investigator capacity is a design requirement, not an afterthought.
• Separation of containment from corrective action, recorded distinctly, so containment can never close a CAPA on its own
• Real root cause analysis — 5 Why, 8D, Ishikawa and Pareto, applied properly, with a discipline that pushes past operator error to the process condition that permitted it
• Action selection against the control hierarchy — design or error-proofing first, procedure and training last. Training as the default action is the signal of a weak CAPA system.
• Effectiveness verification before closure, with defined evidence and a defined interval — a CAPA closes when data shows it worked, not when the action was completed
• Preventive thinking — extending a confirmed fix to similar processes, products and lines before the same failure occurs there
• Trend analysis across CAPAs, so repeat patterns are visible rather than each event being treated as novel
• Metrics that measure the right thing — repeat-nonconformity rate and effectiveness-verification pass rate, not closure timeliness

Changing the metric changes the behaviour more than any procedural fix. When closure speed is measured, CAPAs close fast. When recurrence is measured, causes get found.

We design the system, train your team, and then work through your live CAPA backlog with you. That last part is deliberate. A redesigned CAPA procedure delivered as a document changes very little; sitting with your team through real investigations — where the pressure to close is genuine and the evidence is ambiguous — is what actually transfers the discipline.

Working the backlog also usually reveals the pattern. When you review a year of closed CAPAs together, the recurring causes become visible in a way they never are one at a time, and a small number of underlying process problems typically account for a large share of the volume.

CAPA sits at the centre of every standard we work with — ISO 9001, ISO 13485 where regulators scrutinize it particularly closely, IATF 16949 with its 8D expectations, and the food-safety schemes. If your CAPA system is weak, it is visible in every audit you take.

We bring 30+ years, 900+ organizations and a 100% first-time pass rate to this work.

• Containment closed as correction — the single most common failure
• Operator error accepted as a root cause, with retraining as the action
• Closing on implementation rather than on verified effectiveness
• Raising a CAPA for everything, overloading the system until nothing is investigated properly
• Measuring closure timeliness instead of recurrence
• No trend analysis, so a repeating cause looks like a series of unrelated events
• Never extending a confirmed fix to similar processes and products
• Root cause analysis performed by one person at a desk rather than with the people who were there
• Actions that stop at procedure and training when error-proofing was available
• A CAPA form so long that investigators spend their time completing it rather than investigating