Preventive action: acting on a problem that has not happened yet

Where the warnings actually come from

Useful signals are more available than most plants realise. A characteristic drifting towards a limit without breaching it. A near miss where a defective part was caught by chance rather than by design. A fault on a similar component made by the same process. A supplier's own difficulties reported before they affect you. An audit finding on a process that has not yet produced a problem. Equipment behaving differently. None of these appear in the complaint log, which is why a plant that manages only by complaints has no preventive input at all and is permanently reacting.

Prioritising when nothing has actually failed

The list of things that could go wrong is always longer than the budget, so a ranking is unavoidable. The workable dimensions are how bad the consequence would be, how likely the failure is given what has been observed, and whether anything would catch it before it reached a customer. That last question changes priorities significantly, because a moderately likely failure with no detection route is more dangerous than a probable one that would be caught immediately. Ranking by likelihood alone puts effort into frequent nuisances and leaves the rare severe events untouched.

Learning from other people's failures without waiting for your own

A fault found on one part is evidence about every similar part, and a problem at one plant is evidence for the others in the group. The same reasoning extends outward: published failure information for a component type, a supplier's notification of a change, and warnings circulated through an industry body are all cheaper sources of learning than experience. What is needed is somebody whose job includes reading them and asking whether the same weakness exists here. Without that named responsibility the information arrives, gets acknowledged and changes nothing. Give that person a route to raise an item without owning the fix, or the reading quietly stops once the reading and the work become the same job.

Why this work is always first to be cut

Preventive effort competes for the same engineers as live problems, and a live problem always wins because somebody is on the phone about it. The consequence is predictable: preventive items sit on a list, get deferred repeatedly, and are eventually deleted during a review. The only reliable protections are structural. Ring-fence a proportion of engineering capacity, give the work its own review rather than sharing one with reactive items, and require a decision to defer to be recorded with a name against it, so deferral becomes visible rather than automatic.

Proving the value of something that did not happen

The perennial difficulty is that a success looks like nothing occurring, which is indistinguishable from having wasted the money. Two arguments carry weight with people holding a budget. The first is comparison: failures of a similar type elsewhere in the business, with their actual costs attached, showing what was avoided. The second is a shift in the pattern of work, where the proportion of engineering time spent on live problems falls over a sustained period. Neither is proof, and expecting proof for prevention is a standard that reactive work is never held to.

Frequently asked questions

How does preventive action differ from corrective action?

Corrective action responds to a fault that has occurred and is aimed at stopping it recurring. Preventive action addresses a potential fault before anyone has experienced it, based on a signal such as a trend, a near miss, an audit observation or a failure on a similar part. The practical difference is the evidence available: corrective work starts from a known mechanism, while preventive work starts from a projection and therefore needs a way of ranking what deserves attention.

What signals should trigger preventive work?

Anything showing a process moving towards trouble without yet producing a defect. Measurements drifting steadily within tolerance, a defective part caught by luck rather than by a designed check, an increase in equipment adjustment, a supplier reporting difficulties, a comparable fault on a related component, and audit findings on processes that have not yet failed. The common feature is that none of them appear in the complaint or scrap records, which is where most plants look exclusively.

How do we justify preventive spending to a finance director?

Use costed comparisons rather than abstractions. Take failures of the same class that the business has already suffered, attach the real cost including freight, sorting, engineering time and lost orders, and present the proposed work against that figure. Add whatever the detection gap analysis showed about how the problem would be found today. It is not proof, and it does not need to be, provided the comparison is drawn from your own history rather than from general claims.