Nonconformance management: from the moment a fault is found to the moment it is closed

Raising one has to be easy or people will not bother

If reporting a fault takes an operator away from a running machine, requires a login they do not have, or reliably produces a difficult conversation, faults stop being reported and get quietly set aside instead. The system then shows an improving trend while the problem grows. Making the initial report short, available at the point of work, and free of blame is a design decision that determines whether anything else in the process functions. Detail can be added later by somebody with time; what matters at the moment of discovery is that the material stops moving and the event exists in writing.

Containment comes before understanding

The sequence that saves money is fixed. Secure the material first, in the plant, at the finished store, in transit and at the customer if it has already gone. Establish the boundary of what is suspect. Only then start asking why. Teams instinctively invert this because the technical question is more interesting than the logistical one, and by the time the analysis is finished half the suspect stock has been consumed. A useful rule is that no investigation begins until somebody has stated, in writing, which lots are affected and where they physically are.

Disposition is a decision with an owner and a deadline

Every report needs a route to one of a small set of outcomes: use as it stands, rework to specification, repair to an agreed condition, downgrade to another use, return to the supplier, or scrap. Each carries different authority requirements and different cost. Naming who may choose each option, in advance, removes the delay where material sits on hold while people work out whose decision it is. Attaching a deadline to the decision matters as much as attaching a name, because held material occupies space, ties up value and blocks a delivery while nothing happens.

Not every fault deserves an investigation

Treating every occurrence as a project buries the ones that matter. A workable filter uses severity and repetition: an isolated low-consequence event gets recorded and dispositioned, while anything severe, anything that reached a customer, and anything that has now happened more than once gets a full analysis. Publishing that filter prevents both failure modes, which are investigating everything until the queue collapses, and investigating nothing because the queue already collapsed. Recording the trivial events still matters, since the second occurrence is only visible if the first was written down. Keep the filter itself under review, because a threshold set when the plant was quiet will bury real signals once volumes rise.

The backlog is the measure worth watching

Open reports accumulate because raising one is quick and closing one is not. The count of open items, their age, and the value of material sitting on hold say more about how a plant is running than any defect rate. An ageing list of unresolved reports means material is deteriorating in a corner, capacity is tied up, and the same faults are being raised again while the earlier ones remain unanswered. Reviewing that list weekly with the authority to force a disposition is dull work and one of the higher-return meetings a quality manager can hold.

Frequently asked questions

Who should be allowed to raise a nonconformance?

Anybody who finds something wrong, including operators, maintenance staff and goods-in personnel. Restricting it to quality inspectors filters the information through the smallest group and delays discovery. What should be restricted is the disposition, since deciding what happens to suspect material calls for authority the person who found it may not hold. Separating those two permissions gives you broad detection with controlled decisions, which is the combination that works.

How quickly does a nonconformance need to be closed?

Containment should be complete within hours, because that is what stops the problem growing. The disposition decision deserves a stated target measured in days, driven by the cost of holding the material and the delivery it affects. Any deeper investigation runs on its own longer timescale and should not hold up the material decision. Plants that bundle all three into a single deadline tend to miss the urgent part while chasing the analytical one.

Should minor faults be recorded formally?

Record them, but keep the treatment proportionate. A one-line entry with the part, the fault, the quantity and the disposition is enough for low-consequence events, and it costs almost nothing. The value appears later, when a pattern emerges and the earlier occurrences are the evidence that it is a pattern rather than an isolated incident. What is not worth doing is opening a full investigation for each, which is how the process gets abandoned.