Predictive maintenance is service triggered by the measured condition of equipment, such as vibration, temperature, pressure, run hours or fault codes, rather than by a fixed calendar date. The aim is to service equipment shortly before it would fail and not much earlier.
It sits between two other approaches. Preventive maintenance services on a schedule whether it is needed or not. Reactive maintenance waits for a failure.
How to measure predictive maintenance
Predictive maintenance depends on a threshold. When a reading crosses it, a visit is created. The questions to measure are whether the threshold catches failures and whether it causes unneeded visits.
Catch rate = Failures preceded by an alert ÷ Total failures
False alert rate = Alerts with no fault found ÷ Total alerts
Say a customer's equipment produced 40 alerts last year, 10 of which found nothing wrong, and there were 6 failures, 4 of which had an alert beforehand. False alert rate is 10 ÷ 40, or 25%. Catch rate is 4 ÷ 6, about 67%.
Compare the cost of visits under each approach over the same period: preventive visits made, reactive callouts avoided, and alerts that turned out to be nothing.
Why it matters
Predictive maintenance changes the shape of your schedule. Preventive work is known weeks ahead. Predictive work arrives with days of notice: earlier than a breakdown, later than a calendar visit. That makes it semi-urgent work that still needs a slot near where your trucks already are.
If you take on predictive work, plan capacity for it the way you would for emergency calls, but with a longer window. The preventive maintenance scheduling guide covers mixing planned and condition-based visits on one board.
It also depends on data you may not own. Readings usually come from the equipment, a building system or the customer's monitoring. Without reliable readings and a clear asset management record, there is nothing to predict from.
A common mistake
Treating predictive maintenance as a replacement for all scheduled work. Many tasks, such as filter changes, cleaning and safety inspections, are cheaper to do on a calendar. Predictive approaches suit expensive equipment where both failure and unneeded visits are costly.
Related terms
- Preventive maintenance is the calendar-based alternative.
- Mean time between failures helps set and test thresholds.
- Asset management holds the equipment records predictions depend on.
Common questions
What is the difference between predictive and preventive maintenance?
Preventive maintenance follows a schedule. Predictive maintenance follows equipment condition. Preventive is simpler to run; predictive can avoid unnecessary visits when the data is good.
Does predictive maintenance need special sensors?
Not always. Some equipment reports fault codes or run hours already. Other approaches use readings technicians take on routine visits, trended over time.
Is predictive maintenance only for large companies?
It is more common with commercial and industrial customers that have monitored equipment. Smaller companies can apply the same idea by trending readings they already collect.