Preventive vs predictive maintenance: which lowers cost across a portfolio?

Preventive maintenance and predictive maintenance both aim to stop equipment failing, and both involve the same physical work: inspecting, servicing, replacing parts. The difference is when the work is triggered. Preventive runs on a calendar or a run-hours interval, every quarter, every so many cycles, whether the asset needs it or not. Predictive runs on the measured condition of the asset, doing the work when the data shows it is heading toward failure and not before. Across a portfolio of leased locations, that timing decision is where the cost difference lives.

The short answer to which one is cheaper: it depends on the asset. For a rooftop HVAC unit or a bank of commercial refrigeration, where a failure means a closed or degraded site and an emergency repair at premium rates, predictive timing usually pays. For a low-cost fixture that is quick and cheap to replace, or a fire and life safety system you are required by code to inspect on a schedule, preventive is the right and often the only choice. The cheapest program across a real portfolio is a deliberate mix, matched asset by asset, and it starts from the same place as what predictive maintenance saves across a portfolio.

What is the difference between preventive and predictive maintenance?#

Preventive maintenance is scheduled work. You set an interval based on the manufacturer’s guidance, the asset’s history, and a degree of caution, and you service to that interval regardless of the asset’s actual state. Predictive maintenance, also called condition-based maintenance, is triggered work. Sensors, building system data, and inspections track the asset’s condition, and the work happens when a measurable signal says failure is approaching.

Both beat reactive maintenance, where you wait for the breakdown and pay the emergency premium that comes with it. The reason both beat it is the same: a planned repair in a chosen window is cheaper than an unplanned one at 2 a.m. The reason predictive can beat preventive is subtler. A calendar cannot tell whether an asset is healthy or failing, so a preventive program is always guessing at the interval. Set it too long and failures slip through. Set it too short, which is the more common error, and you spend labor and parts servicing equipment that was fine.

Why calendar-based preventive quietly overspends#

A preventive program overspends in two directions at the same time, and most facilities teams only see one of them. The visible cost is over-servicing: technicians sent to inspect and replace parts on assets that showed no sign of trouble, billed in labor and parts every cycle. The hidden cost is the failures the schedule misses, because a fixed interval cannot see a fault that develops in week six of a quarterly cycle.

Reliability research has understood this for decades. The work behind reliability-centered maintenance, originally from the airline industry and later adopted by the US Navy and others, found that most components do not fail on a predictable, age-related schedule. For those assets, servicing on a calendar neither prevents the failures that matter nor avoids the spend on the ones that were never going to happen. The practical lesson for a portfolio is not that preventive is wrong, it is that a calendar is a blunt instrument, and using it on assets whose condition you could actually measure leaves money on the table in both directions.

Where predictive actually lowers cost#

Predictive maintenance lowers cost where two things are true: the asset’s condition can be measured before it fails, and a failure is expensive. According to the US Department of Energy’s Operations and Maintenance Best Practices Guide, a working predictive program runs about 8 to 12 percent below a preventive one. That figure is real, but it is an average over the right assets. On a low-cost fixture you can monitor all you like and never recover the cost of monitoring it.

For a multi-site occupier, the assets that clear both bars are familiar: HVAC and rooftop units, commercial refrigeration, elevators, and electrical infrastructure. These degrade in ways you can detect, and they are exactly the assets whose failure closes a store, spoils stock, or triggers a safety or compliance problem. Concentrating condition monitoring there, and leaving the long tail of cheap assets on a sensible preventive or reactive footing, is where the portfolio-level saving comes from. A share of avoided emergency cost also stops flowing through CAM and pass-through costs, which lowers recoverable occupancy cost across the estate.

Which method for which asset#

The decision is per asset class, not per portfolio. This is the working version most occupiers land on.

This is guidance, not a rule. Local codes, lease and landlord obligations, and your own failure history move individual assets between rows.

How to shift an existing program without re-tooling everything#

Most occupiers are not starting from nothing. You already run a preventive program, so the move is to redirect it rather than rebuild it.

1. 01Put your PM schedule and your work order history side by side. The history shows where failures and emergency spend actually happen, which is rarely where the schedule assumes.
2. 02Sort assets by failure cost and by whether failure is detectable in advance. That sort produces the method-per-asset view above.
3. 03Leave mandated preventive in place. Fire, life safety, and elevator inspections run on their required schedule no matter what the data says.
4. 04Move the high-cost, condition-detectable assets to condition-based monitoring first, and prove the saving before expanding.
5. 05Widen or drop intervals on healthy, low-criticality assets you have been over-servicing. This is often where the fastest saving sits.
6. 06Tie the plan to lease term and capital plans, so attention and capital follow the locations you are keeping rather than the ones you are about to exit.
7. 07Re-measure against your old PM baseline, then move to the next tier.

The takeaway: preventive versus predictive is not a contest with one winner. It is an allocation problem. Put predictive where failure is costly and measurable, keep preventive where a schedule is required or cheaper, and stop paying to over-service the assets that never needed it. Run on a platform where maintenance condition sits next to lease and location data, and the allocation gets easier, because the cost of a failure and the value of the location are visible in the same place.