Most preventive maintenance programs are built the same way and fail the same way. Somebody lists the assets, applies a sensible-looking interval to all of them, loads the dates into a calendar, and within two quarters the schedule is being deferred quietly every time a breakdown eats a week. The failure is not discipline. It is that the program was never costed in hours against the technicians who had to carry it, and a schedule that cannot be staffed will always lose to a machine that has stopped. This page sets out the four decisions that make a program hold.
Decide the strategy per asset, not per site
Some assets earn a calendar interval, some a running-hours one, and some genuinely should be run to failure because they are cheap, spare and quick to swap. A site running one strategy across everything is over-maintaining most of its plant and under-maintaining the few things that would hurt. Write the decision on the asset so the next person can see it was a decision.
Set intervals from the manual, then move them on evidence
The manufacturer's figure is where you start. If nothing is ever found at the six-month check, the interval is too short; if something is found every time, it is too long. Both corrections need the found field on past work orders, which is the argument for filling it in.
Cost the program in hours before you load the dates
180 assets at four tasks a year at 1.1 hours is 60 jobs and 66 technician hours a month, $3,168 at a $48 loaded rate. Across three technicians that is 22 hours each and carryable; across one who also takes every call-out it is not, and the calendar will not tell you that.
Track deferral, not completion
A program at 95% completion with the same five jobs deferred every month is not at 95%. Deferred-and-rescheduled is the number that predicts the next breakdown, and it is the number a spreadsheet is worst at surfacing because nothing goes red.
Questions people ask about preventive maintenance program
How many PM tasks should an asset have?
As few as keep it running, which is usually fewer than a first pass produces. Every task is a recurring cost in hours, and tasks that never find anything are the first place to look when the program cannot be staffed.
Should the program cover every asset?
No. A register is a complete list; a program is a decision about which of those things are worth scheduled attention. Deciding some assets are run-to-failure is part of building a program, not a gap in it.
How do we get technicians to buy in?
Show them the deferral number and the found field, not the completion percentage. Technicians know which jobs never find anything; a program that removes those earns more credibility than one that adds paperwork.