There is a simplistic version of this argument: reactive is bad, preventive is good, and a competent team runs 80% preventive. It is wrong, and it costs money in the plants that believe it.
Both approaches have a cost. The job is deciding which one you pay, machine by machine.
What each one means
Reactive work happens after the failure. The machine stops, somebody reports it, a technician fixes it.
Preventive work happens before. A plan triggers on a date or on a meter reading — 500 running hours, 10,000 cycles — and the work order is written in advance.
A third category muddies the discussion: condition-based work, where the job depends on a measurement taken from the machine. That is still preventive; the trigger is a value rather than a date.
The hidden cost of preventive work
A preventive plan consumes technician hours, parts, and usually a planned stop. On a machine that is not critical, easy to repair and quick to replace, that cost exceeds the cost of simply letting it fail.
Worse, over-maintaining introduces its own faults. Every strip-down is a chance to reassemble something incorrectly. A bearing that was fine, replaced early by a technician in a hurry on a Friday, can fail sooner than the one it replaced.
The real test is criticality
The question is not "can this machine fail?" but "what happens when it does?" Three things to weigh:
- Production stoppage. Does this machine block a whole line, or is there a way around it?
- Safety. Does a failure put somebody at risk?
- Time to recover. Is the part on the shelf, or is it a six-week import?
A machine that scores on all three earns a preventive plan, even an expensive one. A machine that scores on none can reasonably run to failure — and there is no shame in saying so, as long as it is a decision rather than an oversight.
That is why criticality is a field on the equipment record in MainteNeat, not a note in the margin: it is there to settle this question.
Start with reactive work
The advice that surprises people most: do not build your preventive plans first.
A team starting out writes plans from what it thinks it knows — the machines people complain about, the ones with a bad reputation. That instinct is poorly calibrated. The memorable failures are not necessarily the ones costing the most in aggregate.
Run reactive work for two or three months, logging every job against its machine with the time spent and the parts used. You then have a list sorted by actual cost. Those are your preventive candidates, in order.
The inversion feels backwards but holds: well-recorded reactive work is the raw material for well-aimed preventive work.
The preventive work that never happens
The most common failure is not a bad plan. It is a good plan, written, scheduled, and pushed back every week because something urgent landed.
Two things help. First, the plan has to produce a real work order assigned to a named person a few days before it is due — not a line on a calendar nobody owns. Second, rescheduling must count from the date the job actually closed rather than the date it was created, or a monthly plan that slipped three weeks will ask for the same job again three weeks later.
The one number to watch
A single measure tells you whether the split is healthy: preventive compliance. Of the plans due this month, how many actually closed?
If that number is low, adding more plans will not help. You do not have a preventive maintenance problem — you have a capacity problem, and preventive work is simply the first thing to get dropped.