Non-conformance to CAPA: a defect's full journey

Interviewer: So walk me through what happens when something goes wrong on the floor. A part comes out bad — then what?

QualityLead: Yeah, so first thing — "bad" can show up in a bunch of places, and where it shows up kind of colors everything that follows. It might be incoming inspection, a part from a supplier that fails at the dock. It might be in-process, an operator catches it mid-build. Could be final QC right before it ships. Could be a customer complaint, which is the one that makes everybody's stomach drop. Or it's an auditor pointing at it, which is worse. Same defect, very different temperature depending on the door it came through.

Interviewer: Okay. And the moment it's spotted — what's the very first move?

QualityLead: Log it. We open a nonconformance record — an NCR — in the eQMS. And I cannot stress this enough: nothing happens until that record exists. In a regulated shop, if it isn't written down, it didn't happen, and worse, it looks like you hid it. So what, where, how many, which lot, which spec did it violate. That record is the thread everything else hangs off of.

Interviewer: And then you start investigating?

QualityLead: No — and this is where people new to quality get it backwards. Before you investigate, you contain. You stop the bleeding. Quarantine the affected product, physically tag it, move it to a hold cage, flag it on-hold in the system so nobody can pull it or ship it. Because while you're off doing your clever root-cause analysis, you do not want that material walking out the door. Contain first, think second.

Interviewer: Makes sense. So it's locked down. Now what?

QualityLead: Now scope. How bad is this, how wide is it. How many units, how many lots, is it just this batch or has the problem been quietly happening for three months. And the big scary question — has any of it shipped? Because if it's all sitting in my hold cage, fine, that's a contained internal problem. If some of it's already at a customer, now I'm potentially in field-action, recall, customer-notification territory, and that's a whole different animal with regulators involved. So I really want the answer to be "no, nothing shipped."

Interviewer: And if something did ship?

QualityLead: Then it forks off into the recall world, which honestly is its own multi-week ordeal — but let's stay on the main line, assume we caught it in-house.

Interviewer: Sure. So it's contained, you know the scope. Next?

QualityLead: Severity and risk. We rate it — is this a safety thing, a regulatory thing, or is it cosmetic and the part still works fine. That rating does double duty: it drives what we do with the actual material, and it drives the bigger question of whether this becomes a full investigation or not. So it's a pivot point.

Interviewer: What do you mean "what we do with the material"?

QualityLead: So that's the disposition decision, and we don't make it solo — it goes to the MRB, the Material Review Board. QA, engineering, sometimes production and regulatory in the room. And we decide the fate of the held product. Four options, basically. Rework it — fix it, bring it back into spec, re-inspect. Scrap it — destroy it, document the destruction. Use-as-is — and that one's touchy, it means "yeah it deviates but it's still fit for use," and you'd better have a written, signed justification because an auditor will absolutely zero in on a use-as-is. Or return-to-supplier, if it's their defect.

Interviewer: And return-to-supplier — does that just end it for you?

QualityLead: For the material, sort of, but it kicks off its own thing — a SCAR, a Supplier Corrective Action Request. Now the supplier has to run basically this same process on their end and report back. So it branches into their world while we track it from ours.

Interviewer: Okay. So you've dispositioned the product. Are you done?

QualityLead: Sometimes! And this is the most important fork in the whole thing. The question is: do we need a CAPA? Is this a one-off, or is it systemic? Because here's the thing people get wrong in both directions. If it's genuinely a one-time fluke — a freak event, contained, low risk — you disposition it, you document it, you close the NCR, and you move on. You do not open a CAPA for every little thing.

Interviewer: Why not? Isn't more rigor better?

QualityLead: [laughs] You'd think. But if you open a CAPA for every minor blip, you drown. You end up with a backlog of two hundred overdue CAPAs, and then when the auditor comes, your overdue-CAPA list is the first thing they ask for and it tells them your quality system is out of control. So part of the discipline is knowing what does not need a CAPA. But — the flip side — if it's recurring, or systemic, or high-risk, and you close it as a one-off just to keep your numbers clean? That's how you get a repeat finding, and repeat findings are the cardinal sin. So it's a real judgment call.

Interviewer: Say it warrants one. You open the CAPA.

QualityLead: Right. Open it, link it to the NCR, and — this is the part that actually matters for it ever getting done — assign a real owner and a real due date. A CAPA with no owner is a CAPA that's gonna be overdue. Named human, date on the calendar.

Interviewer: And the owner's first job is?

QualityLead: Root cause. And this — if you take one thing from this whole conversation, take this — this is where everything lives or dies. The job is to get to the true root cause, not the symptom. So we run 5-Whys, fishbone, Ishikawa diagrams, whatever fits. And the discipline is to keep asking why. "The part was out of spec." Why? "The machine drifted." Why? "Calibration was overdue." Why? "There was no reminder in the system." Why? Oh — because our preventive maintenance schedule has a gap. Now you're at something real. If you stop at "operator error, retrained the operator," you have learned nothing and it'll be back.

Interviewer: That's the part you said people get wrong.

QualityLead: It's the part everyone gets wrong, including people who should know better. Shallow root cause. "Human error, recounseled the employee, closed." And then six months later the exact same defect shows up, and now the auditor sees you "fixed" it once already and it came back, and your credibility is gone. Repeat finding. So — true root cause, every time, even when it's slower and more annoying and points back at a process you own.

Interviewer: Okay. You've got the real root cause. Then?

QualityLead: Then the action plan, and it's two halves. Corrective action — fix what already happened, the immediate thing. And preventive action — stop it from recurring, including elsewhere. That second half is the one teams skip. Preventive means: okay, this happened on Line 2, but does the same failure mode exist on Line 4, on the other product, at the other site? Go fix it there too, before it bites you. Corrective is "patch the hole." Preventive is "are there other holes shaped like this one."

Interviewer: And actually changing things — how does that happen?

QualityLead: Through change control. You don't just walk over and edit the work instruction. Any change to an SOP, a work instruction, a drawing — it routes through formal change control, gets reviewed, approved, versioned. And critically, if the procedure changed, the operators get retrained and that retraining gets recorded. The training record is part of the package. No quiet edits.

Interviewer: Then you implement.

QualityLead: Then you implement — make the process change, install the fixture, run the retraining, requalify the supplier, whatever the actions were. And then you wait. Which feels weird, but you have to let it run.

Interviewer: Wait for what?

QualityLead: The effectiveness check. This is verification, and it's the step that separates a real quality system from theater. After some defined period — some number of units, some number of lots, a few months — you go back and ask: did it actually work? Did the defect stop happening? Not "did we do the actions," but "did the actions have the effect we wanted."

Interviewer: And if it didn't work?

QualityLead: Then you failed, and you own it. You reopen, you escalate, and you loop all the way back to root cause — because if the fix didn't take, odds are your root cause was wrong in the first place. You treated a symptom. So back to the 5-Whys, deeper this time. That loop is painful but it's the honest part of the system. A CAPA that closes without a real effectiveness check is just paperwork.

Interviewer: And if it did work?

QualityLead: Then we tie it off. Finalize the documentation, make sure the revised SOPs are in, make sure every training record is logged, attach all the evidence. Then it goes up for management review and QA approval — quality management signs off, and the bigger ones feed into the periodic management review meeting. And then we close it, with the full audit trail intact: detection, NCR, containment, disposition, root cause, actions, verification, approval, all chained together and retrievable.

Interviewer: Why so much emphasis on the trail at the end?

QualityLead: Because that closed record is the actual product. When an auditor shows up, they pull a closed CAPA and they pull the thread — show me the NCR, show me how you contained it, show me your root cause, show me the effectiveness data. If any link is missing, it doesn't matter how good the fix was, you've got a finding. The fix is for the product. The audit trail is for staying in business.

Interviewer: If you had to name the one thing people get wrong — what is it?

QualityLead: Two things, and they're cousins. The obvious one is shallow root cause — stopping at "human error" because it's faster and it doesn't implicate a process you're responsible for. That's how you get repeat findings. But the subtler one, the one that quietly kills quality systems, is treating the whole thing as a documentation chore instead of a learning loop. People grind out the paperwork to close the CAPA and hit their metrics, and they never actually ask "did this work and did we learn something." And you can always tell — the shop that treats it as paperwork has a fat backlog of overdue CAPAs and the same defects coming back around every year. The shop that treats it as learning has fewer CAPAs, deeper ones, and they close. The documentation's the same either way. The mindset is the entire difference.

Interviewer: That's a great place to stop. Thank you.

QualityLead: Anytime. Just don't make me pull a use-as-is justification on camera. [laughs]