Control of Nonconforming Product (ISO 9001 8.7)

Control of nonconforming product is the process of identifying material that does not meet requirements, holding it so it cannot be used by mistake, and formally deciding what happens to it. ISO 9001:2015 clause 8.7 requires you to identify, control, and disposition nonconforming outputs, and to keep records of the decision.

The idea is small and the failures are expensive. A part that fails a check is not the problem; a part that fails a check and then quietly moves to the next station is the problem. Everything in this article is about the gap between those two sentences: how you catch nonconforming material, how you stop it from moving, who gets to decide its fate, and what four decisions are even on the table.

What counts as nonconforming product?

Nonconforming product is any output that does not meet its specified requirements. That includes an out-of-tolerance dimension, a failed functional test, wrong material, missing paperwork, damage in handling, an expired shelf life, or a characteristic that drifted outside the limits on a control chart. The requirement can come from a drawing, a customer spec, a regulation, or your own internal standard. If the output misses any requirement it was supposed to hit, it is nonconforming, full stop.

A useful distinction: a nonconformity is the failed characteristic, and nonconforming product is the physical thing carrying it. You control the product. You investigate the nonconformity. The two live on different documents and get closed on different clocks, which is why a non-conformance report and a corrective action are not the same event.

What does ISO 9001 clause 8.7 actually require?

Clause 8.7 requires that nonconforming outputs be identified and controlled to prevent their unintended use or delivery, that you take action appropriate to the nature of the nonconformity, and that you retain records of what you found, what you did, who authorized it, and any concession obtained. The standard lists specific ways to deal with the material, and it insists on the paper trail behind the decision.

Clause 8.7 pairs with clause 10.2 (nonconformity and corrective action). Think of it as a division of labor: 8.7 governs the product in front of you right now, and 10.2 governs the problem so it stops coming back. You can satisfy 8.7 by scrapping one bad lot and never satisfy 10.2, which is why auditors read them together and why CAPA exists as a separate loop. Containment is not correction; sorting the current lot is containment, not corrective action.

Why segregate nonconforming material physically?

You segregate because human systems leak. A red tag on a skid that stays in the flow lane will eventually get grabbed by a hurried operator, a new hire, or a night shift that never heard the story. Physical separation, a marked-off quarantine cage, a "hold" area, a locked cart, removes the possibility rather than relying on everyone reading the tag correctly every time. The strongest control is the one that makes the wrong action physically harder than the right one.

Identification and segregation are two different controls doing two different jobs. Identification (a tag, a hold status in the system, a red bin) tells people the material is suspect. Segregation moves it somewhere it cannot re-enter production by accident. On a good line you do both, because a tag answers "is this okay?" and a quarantine answers "can I grab it anyway?" This is the same logic behind holding suspect incoming lots during incoming material inspection before they ever touch the floor.

What are the four disposition options?

Once material is identified and held, someone with authority decides its fate. In practice the choices reduce to four dispositions, and every quality system uses some version of them. The table below lays them out with the record each one demands.

Two of these deserve a warning. Use as-is is the disposition that quietly erodes a quality system: it is fast, it saves the lot, and if it becomes routine, your specification is now whatever the review board tolerates on a bad day. A concession should be rare, documented, time- or quantity-bounded, and reviewed for a pattern. If the same characteristic gets a concession every month, the honest move is to change the spec or fix the process, not to keep signing waivers. Rework hides a similar trap: reworked product must be re-inspected against the original requirement, and the rework step itself can introduce a new nonconformity, so it is not a free pass either.

Who decides the disposition?

The disposition is made by a defined authority, not by whoever is standing closest to the parts. Most organizations route anything beyond a routine scrap to a review board, historically a material review board, where quality, engineering, and sometimes the customer weigh in. The reason is separation of interest: the person under pressure to hit the shipment number should not be the sole person deciding that a marginal lot is "close enough."

The authority level should scale with the risk. A cosmetic minor defect might be dispositioned by a line quality tech within a written standard. A safety-related characteristic, or anything the customer flagged as special, needs a higher signature and often customer approval before any "use as-is" is even discussed. Writing down who can approve what, before the pressure hits, is what keeps disposition honest. Your QMS software should carry those approval rules so the record captures the right signatures automatically.

How do you build a control-of-nonconforming-product process?

You do not need a thick manual. You need a repeatable path from "this failed" to "the file is closed," with the authority and the records fixed in advance. Here is a working sequence.

1. Define what triggers a hold. List the checks whose failure creates nonconforming product: dimensional out-of-tolerance, failed functional test, missing traceability, out-of-control point on a chart. If a failure is not on the list, operators will guess, and guessing favors the shipment.
2. Identify immediately at the point of detection. Tag or flag the material the moment a check fails, before it moves. Capture what failed, how much, and the lot or serial, ideally in the same system that runs the check so nothing depends on someone walking a paper tag to a desk.
3. Segregate to a controlled location. Move held material to a marked quarantine area or lock its status so it cannot be pulled into the next operation. Make the wrong action physically or systemically harder than the right one.
4. Raise the record. Open a non-conformance report describing the nonconformity, quantity, and origin. This is the document clause 8.7 wants you to retain.
5. Route to the right authority. Send it to the disposition level that matches the severity. Minor stays low; critical or customer-special characteristics escalate. No single person under shipment pressure disposions a safety characteristic alone.
6. Disposition and execute. Choose use-as-is, rework, regrade, or scrap. Re-inspect anything reworked against the original spec. Record who authorized it and any concession obtained.
7. Feed the pattern into corrective action. One disposition closes one lot. Recurring dispositions of the same characteristic are a signal, not a nuisance; route them to CAPA so the problem gets fixed at the source.

Where control-of-nonconforming-product breaks down

The failures are boringly consistent across plants. Material gets a tag but never gets moved, so it drifts back into the flow. "Use as-is" becomes the house default because it protects the ship date, and the spec quietly rots. Rework happens without re-inspection, so a second nonconformity rides out the door on top of the first. And the records live in three places, a paper NCR binder, a spreadsheet, an email thread, so no one can see that the same defect has been dispositioned eleven times this quarter.

Every one of these is a visibility problem before it is a discipline problem. When holds, dispositions, and re-inspections are captured at the station in one system, a supervisor can see held material aging in quarantine, a quality manager can see concession frequency by characteristic, and the recurring-defect pattern that should trigger corrective action becomes obvious instead of buried. That is the quality intelligence Harmony's connected-worker platform is built to surface, on top of your existing process, with no rip-and-replace. Held product that never moves is also pure cost of quality: floor space, handling, and eventually a hurried decision made under deadline. The same nonconforming-product discipline applies whether you run continuous flow or discrete batch production where a single bad lot can strand an entire downstream schedule. Getting the disposition right is the difference between quality control and quality assurance and you can see the whole loop working in our CLS case study.