Equipment preventive maintenance is easy to describe and awkward to start, because the obvious first move, setting intervals, is the third thing rather than the first. A site starting from nothing has a register that is probably incomplete, no explicit view of which assets matter most, and no history to set intervals from. Doing it in the wrong order produces a schedule that cannot be staffed and cannot be justified. This page is the first ninety days in the order that works, ending with the loop that turns borrowed intervals into your own.
Weeks one to four: finish the register
Walk the site. Every motor, pump, compressor, door, lift, fan and vehicle, with make, model, location and what has to be isolated to work on it. The assets you find that nobody owns are the ones this exercise exists to catch, and they are disproportionately the ones that fail badly.
Week five: rank the top ten by consequence
What halts if it fails, how long a replacement takes, whether there is a spare, whether anybody gets hurt. You do not need to rank all 180; you need to know the ten that matter, because they will absorb most of the attention and justify most of the budget.
Weeks six to eight: borrow intervals and cost the load
Manufacturer figures, adjusted where your duty is obviously harder. Then the arithmetic: 180 assets at four tasks a year at 1.1 hours is 66 technician hours a month, $3,168 at a $48 loaded rate on the worked example this site publishes. If that does not fit the people you have, cut scope now rather than discovering it in month four.
From week nine: record findings and start the loop
Every job, every time, what was found. In a year that record is what lets you lengthen the intervals that never find anything and shorten the ones that keep missing faults. Until then you are running somebody else's intervals, which is fine as a start and not fine as a destination.
Questions people ask about equipment preventive maintenance
How many assets before this is worth doing?
The ranking is worth an afternoon at any size. The formal schedule starts to repay itself somewhere around a couple of dozen assets and two people.
What if we cannot staff the schedule we designed?
Cut by criticality, not evenly. Protect the top ten and lengthen intervals on the rest, and write down that this is what you did and why.
How long before it shows?
Reactive share moves within two quarters if the schedule is being worked. MTBF moves more slowly, because it needs enough running time to be meaningful.