The 4 Types of Maintenance Strategy Compared

The four types of maintenance strategy are reactive (run the asset until it fails, then fix it), preventive (service it on a fixed time or usage interval), predictive (service it when a measured condition says failure is coming), and reliability-centered maintenance (RCM, a framework that assigns one of the first three to each failure mode based on consequence). They are tools, not a scoreboard.

The most expensive mistake in maintenance is treating these as a ladder where every asset should climb toward predictive. A single plant runs all four at once: exhaust fans to failure, gearbox oil changes on a meter, main-drive bearings on vibration, and the whole mix chosen deliberately. This guide puts the four side by side, what each costs, what data each demands, and how to match one to an asset instead of picking a house religion.

What are the four maintenance strategies?

Every maintenance action a plant takes is one of four strategies, distinguished by what triggers the work.

Reactive maintenance also called run-to-failure or breakdown maintenance, does nothing until the asset breaks, then repairs or replaces it. As a default it is the most expensive way to run a plant: failures happen at the worst time, do collateral damage, and pull the crew into firefighting. As a deliberate choice for cheap, redundant, low-consequence assets, it is correct. The distinction is whether you chose it or drifted into it.

Preventive maintenance services the asset on a fixed schedule, a calendar interval or a usage meter, to catch wear before it becomes failure. It is simple to administer and right for components whose degradation tracks time or cycles. Its weakness is that it services on a guess: a machine that ran one shift and one that ran three get identical treatment, so you either over-maintain (wasting hours and risking maintenance-induced failure) or under-maintain. Building this well is its own discipline, covered in our preventive maintenance schedule guide.

Predictive maintenance services the asset when a measured condition crosses a threshold, vibration, temperature, oil chemistry, current draw. It is the most precise trigger because it responds to the actual asset, not a calendar, and it catches failures preventive schedules miss between intervals. It is also the most demanding: sensors, baselines, and someone who owns the readings. The lighter-weight version, checking condition on a route rather than continuously, is condition-based maintenance; the analytics-heavy version is predictive maintenance proper.

Reliability-centered maintenance is different in kind from the other three. It is not a trigger; it is the decision process that decides which trigger each failure mode deserves. Formalized in the SAE JA1011 standard, RCM works failure-mode by failure-mode and asks whether each one is worth preventing at all, and if so, by which method. The output of an RCM analysis is a maintenance plan that is part run-to-failure, part preventive, part predictive, each choice justified by the consequence of the failure it addresses.

Which strategy is cheapest?

None of them, universally, and that is the whole point. Total maintenance cost is the sum of two curves that move in opposite directions. Spend nothing on prevention and failure costs dominate. Spend everything on prevention and maintenance costs dominate. The cheapest posture for any given asset sits between the extremes, and where it sits depends on how much a failure of that asset actually costs.

This is why a plant-wide answer is always wrong. The exhaust fan with a spare on the shelf has a nearly flat failure-cost curve, its optimum is run-to-failure. The bottleneck filler that stops the whole line has a steep failure-cost curve, its optimum is predictive, and the sensors pay for themselves in one avoided stoppage. Matching strategy to the shape of each asset's curve is the work; it maps directly onto the maturity ladder in our equipment reliability guide.

How do you choose a strategy per asset? A 5-step method

Choose by consequence first, then by whether a practical trigger exists. Here is the sequence, which is a plant-floor compression of the RCM logic.

1. Rank the asset by criticality. What happens when it fails, safety, environmental, a line stoppage, a quality escape, or nothing much? A simple A/B/C ranking by consequence is enough to start. Everything downstream flows from this number.
2. For low-consequence assets, choose run-to-failure on purpose. If failure is cheap and safe and the asset is redundant, preventing failure is waste. Document the decision so nobody quietly adds a monthly PM back onto it. Run-to-failure is a strategy when chosen, a symptom when defaulted.
3. For the rest, ask whether a measurable warning exists. Is there a condition, vibration, heat, pressure, particle count, that rises before failure, with enough lead time to act? If yes, predictive or condition-based is on the table. This is the single most valuable question in the whole method.
4. Where no practical condition trigger exists, use preventive. If the failure is age- or usage-related and you cannot measure its approach economically, service it on a time or usage interval set from history and the manufacturer's guidance. Most PM tasks land here honestly.
5. Re-check with real failure data. A strategy set once and never revisited rots. PMs that never find anything want longer intervals; failures that keep surprising you want a condition trigger or a redesign. Feed the decision from your maintenance KPIs and close-out history every quarter.

Where do the strategies actually sit in a real plant?

Here is the same asset mix a food or CPG plant typically carries, with the strategy each earns and why. The pattern to copy is that the strategy column is not sorted by asset cost, it is sorted by failure consequence and warning availability.

Notice the ammonia compressor and the exhaust fan can cost wildly different amounts and still land where they do, because strategy follows consequence, not price tag. The whole mix is what a mature program looks like, and it is exactly the picture the equipment reliability maturity ladder describes when it says the goal is not maximum prevention but the right prevention.

What do the strategies pay off, and what does the data say?

The best public numbers on the economics come from the U.S. Department of Energy's Federal Energy Management Program guidance, maintained by Pacific Northwest National Laboratory (PNNL), plus the standards body that defines RCM.

• Moving from a heavily reactive posture toward planned, proactive maintenance offers savings that can exceed 30–40% of maintenance costs, depending on how reactive the starting point is (PNNL, O&M Best Practices: Maintenance Approaches).
• A functioning predictive program saves a further 8–12% over a program using preventive maintenance alone (same source), the reason mature plants graduate critical assets from time-based PM to condition triggers.
• RCM is a defined standard, not a vendor concept: SAE JA1011 sets the seven questions any process must answer to be called RCM, and PNNL documents RCM as the framework that blends all three strategies by consequence (PNNL, Reliability-Centered Maintenance).
• The people who execute any of these strategies are getting scarcer: the U.S. Bureau of Labor Statistics projects 13% employment growth (2024–2034) for industrial machinery mechanics, maintenance workers, and millwrights, with about 54,200 openings a year (BLS Occupational Outlook Handbook).

The honest counterweight: predictive is not free, and RCM done as a paperwork exercise burns months for nothing. The savings above assume the strategy is executed, not just documented. A predictive program with sensors nobody reads is more expensive than the run-to-failure it replaced.

How do these strategies connect to the rest of maintenance?

Strategy is the top of the stack; everything below it is execution. A strategy decision becomes a PM schedule entry or a condition route, which becomes a work order, and getting the work order type right is how you later measure which strategy each asset is actually running. Those work orders get prepared by planners and dispatched by schedulers through planning and scheduling and the whole loop lives in a CMMS. Operator-led first-line care through total productive maintenance extends the reach of every strategy by putting more eyes on more assets.

One practical blocker sits under all of it: choosing predictive or condition-based strategies requires data flowing off the equipment, and in most plants that data is trapped in PLCs, sensors, and paper logs that do not talk to each other. Pulling those sources into one searchable layer without ripping out the controls is the problem described on our platform overview and the shift from paper logging to live capture is what the CLS case study walks through. Get the data flowing and the higher-value strategies stop being aspirational. This coordination discipline is the same one that makes production scheduling stick: shared, current information beats heroics.