Key takeaways
- Preventive maintenance is triggered by time or usage: service the asset every month, every 500 hours or every 10,000 km, whether it looks worn or not.
- Predictive maintenance is triggered by the asset's actual condition: readings, trends and history indicate that a failure is likely, so work is scheduled before it happens.
- Preventive is simpler and works for most assets. Predictive pays off on critical, expensive assets where unplanned downtime hurts most.
- Most mature teams run a mix: preventive as the baseline, condition-based triggers and predictive analytics on the assets that matter most.
Four maintenance strategies at a glance
Preventive and predictive maintenance sit on a scale that runs from doing nothing until something breaks to anticipating failure from data:
| Strategy | What triggers work | Best for | Main drawback |
|---|---|---|---|
| Reactive (run-to-failure) | The asset breaks | Cheap, non-critical items with spares on hand | Unplanned downtime and rushed repairs |
| Preventive (time or usage based) | A calendar date or a usage count | Most assets with known service intervals | Some work is done before it is needed |
| Condition-based | A reading crosses a threshold | Assets with measurable wear indicators | Needs meters, inspections or sensors |
| Predictive | Data and trends forecast a likely failure | Critical, high-cost assets | Needs good historical data to be accurate |
Preventive maintenance: planned by the calendar or the meter
Preventive maintenance (PM) schedules work at fixed intervals. A chiller gets serviced every quarter, a generator every 250 running hours, an elevator every month. Intervals usually come from manufacturer recommendations, regulations and your own experience.
Why it works
- It is easy to set up and easy to explain to technicians, clients and auditors.
- It works without sensors, using only a calendar and a usage meter.
- It turns most maintenance into planned work that can be scheduled around operations, which is especially valuable in malls, hospitals and hotels where access windows are limited.
Where it falls short
Fixed intervals do not reflect how an asset is actually used. A pump running 24/7 in summer heat wears faster than the same pump idling in winter. Some PM work is done too early, and some failures still happen between visits.
In a CMMS, preventive jobs are generated automatically from recurring schedules (daily, weekly, monthly or yearly), so nobody has to remember to raise them. See Job Scheduler.
Predictive maintenance: planned by the asset's condition
Predictive maintenance (PdM) uses data to estimate when an asset is likely to fail, and schedules work just before that point. The data can come from sensor readings such as vibration, temperature or pressure, from meter readings, and from the asset's own maintenance history: how often it has failed, after how many hours and with which symptoms.
A simpler step on the way is condition-based maintenance: instead of forecasting, you set a threshold and let the system raise a work order when a reading crosses it. For example, a work order is created when filter pressure drop exceeds a set value.
Why it works
- Work happens when it is needed, which cuts unnecessary PM visits on healthy assets.
- It catches developing faults that a fixed schedule would miss between visits.
- It protects the assets where an unplanned stop is most expensive, such as production lines, critical HVAC, medical equipment and telecom sites.
What it needs
- Reliable data: readings, inspections or sensors on the assets you want to monitor.
- History: predictions improve as more failure and repair data is recorded against each asset.
- A system that turns a prediction into an actual work order, rather than a chart someone has to notice.
Which approach for which asset?
Decide asset by asset, not for the whole site. A simple way to sort them:
- 1
Rank assets by criticality
What happens if it stops? Consider safety, compliance, revenue, guest or patient impact, and repair cost and lead time.
- 2
Low criticality: reactive or light PM
Lighting, non-critical fixtures and cheap items with spares on hand often do not justify a schedule.
- 3
Medium criticality: preventive
Most building systems and equipment belong here, on time- or usage-based schedules.
- 4
High criticality with measurable wear: condition-based or predictive
Add meter or sensor triggers and predictive analytics on top of the PM baseline.
- 5
Review with data
Use MTBF, MTTR and PM compliance from your reports to tighten or relax intervals over time.
Running both in e-maintain
- Preventive: recurring work orders generated automatically on daily, weekly, monthly or yearly schedules through the Job Scheduler.
- Condition-based: job triggers in the Admin Portal create work orders automatically when meter readings, usage levels or performance thresholds are reached, and threshold alerts notify the right people.
- Predictive: predictive analytics uses historical asset data and performance trends to flag likely failures, and the AI assistant can monitor selected assets and estimate when each is likely to need maintenance, working alongside your scheduled PM.
- Measuring it: MTBF and MTTR tracking in Asset Management shows whether your strategy is reducing failures.
Frequently asked questions
Is predictive maintenance better than preventive maintenance?
Not across the board. Predictive maintenance is worth the extra data and effort on critical, expensive assets. For most other assets, a well-run preventive schedule gives most of the benefit at lower cost. The best programs combine both.
Do I need IoT sensors for predictive maintenance?
Sensors help, but they are not the only source. Meter readings, inspection results and the asset's own failure and repair history in your CMMS can also feed predictions, and condition-based triggers work with manual readings too.
What is the difference between condition-based and predictive maintenance?
Condition-based maintenance reacts when a reading crosses a set threshold. Predictive maintenance uses trends and history to forecast when that is likely to happen, so work can be planned earlier.
How often should preventive maintenance be done?
Start with manufacturer recommendations and any regulatory requirements, then adjust using your own data. If an asset keeps failing between PM visits, shorten the interval or add a condition trigger; if PM visits rarely find anything, the interval may be too short.