A failed check can create several urgent tasks at once. Product may need to be held, a line may need to be stopped, a setting may need correction and a cause may need investigation. Those tasks are related, but they solve different problems. Fixing the setting does not tell the site what happened to product made before the fix. Holding product does not explain why the setting drifted. A corrective-action case can record all the activity and still leave the underlying condition unchanged.
The first technical question is what the deviation means now: for product, process, evidence and current control. Immediate correction and containment then stabilise the situation while the site investigates. Corrective action changes the condition that allowed the failure, and effectiveness evidence tests whether that change works. The sequence need not be a rigid form or identical for every issue. It is a way to keep a current product decision from disappearing into a root-cause exercise and to keep a temporary protection from quietly becoming the permanent answer.
Establish the immediate exposure before analysing cause
When a non-conformity is identified, the site needs to describe it precisely. What requirement was not met, what evidence revealed it, when did the condition begin or become possible, and what product or process depended on that control? The scope may be uncertain at first. A last known good check, production record, equipment log or movement record can help bound it. Where evidence is missing, the site may need a wider protective boundary until it can narrow the concern. The answer should not wait for a polished root-cause diagram.
Consequence assessment asks more than “did the check fail?” It considers whether product safety, authenticity, legality or quality could be affected, whether material has moved, whether the process is currently under control and whether the records needed to decide are reliable. A failed inspection may be a detection problem as well as a product problem. A control could have performed correctly while its evidence is incomplete, or it could have failed in a way the record conceals. The site should distinguish those cases before making a status decision.
The immediate assessment needs an authorised route to action. Stop or restrict the activity where necessary, identify affected lots, protect them from unintended use and escalate material exposure. Record what is known and what remains uncertain. A controlled hold is not a final verdict on product; it preserves the decision while evidence is gathered. In some cases the investigation will show that exposure was limited. In others the initial boundary will widen. Either outcome is stronger when the site can explain how it moved from the first signal to its product-scope decision.
The boundary can extend beyond stock physically beside the failed control. A material used in several batches, a shared piece of equipment or a result used to release product may connect the same deviation to earlier work. The team should trace the dependency rather than draw a boundary only around what is easiest to see. This does not mean every nearby lot is automatically affected. It means the site can explain why each included or excluded period is supported by evidence, and what it did when that evidence was incomplete.
Correct the present problem without losing product status
Correction deals with the present problem: replace an incorrect label, adjust a setting, clean a surface or restore a functioning device. It can be essential to stop further exposure. It does not by itself address why the non-conformity arose or what happened before the correction. If a machine is reset after an out-of-limit reading, the product made during the uncertain period still needs a separate assessment. The correction record should show what was changed, when and how the restored condition was verified.
Some immediate action is directed at product rather than equipment. Affected stock might be isolated, inspected, reworked or given another authorised disposition based on the risk and evidence. Those actions resolve the status of particular material; they need not eliminate the condition that could create the same problem tomorrow. The site should keep product disposition and process restoration visible as distinct decisions. A batch can be disposed of while the causal investigation remains open, and a line can be ready to run while product from the previous period remains under hold.
An immediate correction can also alter evidence. Cleaning away residue, changing a setting or replacing a part may make it harder to reconstruct the failure. Protection of product takes priority, but the team should preserve relevant observations, settings, samples or records when doing so is safe and useful. A clear account of the as-found state can prevent the later investigation from having to guess what was wrong. That is why correction belongs within a controlled response, not as an unrecorded fix at the machine.
Give temporary containment a limit and an owner
Containment limits current exposure while the cause is being investigated or a durable change is prepared. It might mean segregating stock, restricting a line, increasing a relevant check or operating under a temporary engineering control. The measure must be capable of protecting the risk it is assigned. If a temporary manual check is meant to catch a failure, the site should know how often it occurs, what defect it can detect, who performs it, how affected product is identified between checks and what happens when a check fails.
The temporary state needs an owner, a review point and an exit condition. Without those, an interim instruction can become normal practice by inertia. Monitoring should show whether exposure remains controlled while permanent work is delayed. Monitoring the interim control helps decide whether a delayed action can remain open under controlled conditions and when escalation is needed. An overdue repair is not automatically an uncontrolled crisis, but it cannot be declared safe solely because a temporary measure was written on a form.
Containment is not corrective action by another name. Additional checks may catch defects but leave a drifting machine, inaccessible cleaning point or confusing instruction in place. They can also become less reliable as workload changes. The team should ask what happens if the interim measure itself fails, and whether the proposed duration and staffing are realistic. If the arrangement cannot reliably control the exposure, production or product movement may need restriction until a better option exists.
Investigate the condition that allowed the failure
Once immediate risk is controlled, the site can investigate why the non-conformity occurred. That begins with evidence: the as-found state, process conditions, maintenance history, instructions, training, prior cases and trend information. The depth and method should fit significance and recurrence. A minor isolated error may not need the same analysis as a failure that could affect product safety or has returned despite earlier action. The important discipline is to test explanations against the evidence, not to select a familiar category such as “operator error” before understanding the work.
Consider a constructed example: a seal-integrity check fails on one packing line. A technician adjusts a setting and subsequent test packs pass. Product since the previous credible good check is placed under controlled status and assessed. A temporary increased check may protect continued operation while the team investigates. The cause could be an equipment condition, material variation, setting control, changeover practice or a combination. It shows why the immediate correction, interim check and causal investigation are separate pieces of work.
An investigation should examine the conditions that made the failure possible or allowed it to persist. Did the setting change without being noticed? Were checks capable of detecting the failure mode and performed at a useful point? Was maintenance able to see early wear? Did the process change while instructions stayed the same? The answer may cross functions. A cause statement should be specific enough to guide an action that changes the pathway; otherwise the case risks ending with reminders or retraining that do not address the condition actually found.
Make corrective action address the condition that caused the failure
Corrective action addresses the reason for recurrence risk. If evidence shows a part repeatedly shifts under ordinary load, an engineering change may be needed. If the defect arises when a new material is introduced without adjusting a control setting, the material-change and setup process may need revision. If the check cannot detect the relevant failure, its method or frequency may need rethinking. The action should be linked to the demonstrated cause and assigned to people with the authority and resources to carry it through.
Planning should state what will change, who owns it, when it will be implemented, what evidence will show implementation and how exposure is protected meanwhile. A work order closed or a new instruction issued proves that a task happened; it does not yet prove the product-control outcome. Implementation may require communicating the changed control to shifts, suppliers or other lines and confirming that staff can use it. If the planned action is delayed, the decision to keep operating under interim control should be revisited with current evidence rather than carried forward automatically.
Ownership should follow the action's real dependency. A QA case owner may coordinate evidence and status, while Engineering must deliver a physical change and Production must use the revised operating method. A due date without access to the equipment or an agreed production window may be unrealistic. If resource or authority prevents completion, the site should escalate that constraint and reassess the interim exposure. Assigning every item to the person who opened the case can make the record look orderly while the necessary work remains outside that person's control.
The site should avoid confusing a wider action list with a better solution. Several actions can be appropriate, but each should have a reason. A generic “retrain everyone” step may add effort without changing a machine or a poorly designed hand-off. Conversely, an equipment repair may be insufficient if staff still have no way to recognise a new failure. A coherent corrective action matches the causal pathway and the normal conditions in which control must hold.
Test the change before calling the case closed
Verification has two stages. First, check that the action was implemented as intended. Then test whether it is effective against the original failure under representative operation. A newly fitted part may be present but not stable under load; a revised procedure may be approved but not used at a busy changeover. The effectiveness check should ask what evidence would disconfirm the belief that the cause was removed. That might involve observed runs, targeted records, relevant monitoring or absence of a previously repeated signal over a justified period. No universal fixed number of checks can establish effectiveness for every mechanism.
The constructed seal example might need evidence from ordinary production after the repair, not only a test pack immediately after adjustment. It may also need to show that affected stock was dispositioned and the temporary increased check was withdrawn only after the new control proved adequate. If the fault recurs or the test reveals a different failure mode, the case should be reopened or escalated. Effectiveness evidence informs the decision to close, reopen or escalate the case; completing the action list alone does not establish an effective control.
The final record should keep separate the product consequence, immediate correction, containment period, cause-linked action and effectiveness result. Recurrence and trend review can then challenge whether the site learned from the event or merely repeated it. A deviation is not stably controlled when activity has been documented. It is stably controlled when current exposure is resolved, interim protection has a deliberate end, and the changed system works against the cause that evidence identified.
