Preventive Maintenance: The Complete Guide

Ask ten maintenance managers what preventive maintenance is and you will get ten answers that all sound the same and mean different things. For some it is the statutory inspection calendar. For others it is whatever the CMMS generates on the first of the month. For a few it is a genuine engineering discipline built on failure modes and consequence. Over twenty-two years of CMMS, CAFM and EAM implementations across utilities, oil and gas, manufacturing, government and facility operations, I have seen every version of it, from programmes that quietly prevent millions of dirhams of failure to programmes that exist mainly to generate paperwork. This guide is the whole picture: the definitions, the economics, the build sequence, the actual schedules and checklists, and an honest account of where preventive maintenance stops being worth the money.

The message up front: preventive maintenance is not a volume exercise. A programme of 400 well-chosen, well-written tasks tied to real failure modes will outperform a programme of 4,000 inherited tasks every time, and it will cost a fraction as much to execute. The skill is not in creating PMs. Anyone can create PMs. The skill is in deciding which assets deserve one, what the task should actually ask the technician to do, and how often.

1. What preventive maintenance actually means

Preventive maintenance is work carried out at a predetermined interval, or against a predetermined usage threshold, with the aim of reducing the probability of failure or the consequence of failure. The defining characteristic is that the trigger is decided in advance rather than by the condition of the asset on the day. You service the pump every three months because you decided three months is the right interval, not because the pump told you it needed servicing.

That is worth stating plainly because the terminology in this field is a mess. The terms you will encounter, and what they actually mean:

• Preventive maintenance (PM), also written preventative maintenance: the umbrella term for scheduled, proactive maintenance. Both spellings are correct. British usage tends toward "preventative", American toward "preventive", and both appear in the same CMMS screens in the Gulf without anyone minding.
• PPM, planned preventative maintenance: the term that dominates UK and Middle East facilities management contracts. In practice PPM and PM mean the same thing, but PPM carries a contractual flavour: it usually refers to the agreed schedule of planned work annexed to an FM contract, against which the service provider is measured. When a client asks for "the PPM schedule", they want the annual calendar of planned tasks by asset and frequency.
• Planned maintenance: any maintenance that is planned in advance, which includes preventive work but also planned corrective work arising from an inspection. "Planned and preventative maintenance" as a phrase is usually just emphasis rather than two distinct things.
• Scheduled maintenance: work that has a date in the diary. All PM should be scheduled; not all scheduled work is preventive.
• Corrective maintenance: work to restore a failed or degraded asset. This is the work PM is supposed to reduce.
• Reactive maintenance / run to failure: no proactive intervention at all. As a deliberate choice on low-consequence assets this is entirely rational. As an accident it is how organisations end up firefighting.

The relationship between maintenance and preventive maintenance is simply that preventive is one of several strategies inside the wider maintenance function. It sits alongside corrective, predictive and condition-based work, and a mature operation runs all of them concurrently on different assets. The mistake is treating preventive maintenance as the goal rather than as one tool among several. For the wider system-of-record context, the differences between CAFM, CMMS, EAM and IWMS platforms are covered in the CAFM vs CMMS vs EAM vs IWMS comparison.

2. Why preventive maintenance pays, and where the argument is weaker than people claim

The business case for preventive maintenance rests on a straightforward asymmetry: a planned intervention on a healthy asset is cheaper, safer and less disruptive than an unplanned intervention on a broken one. That asymmetry shows up in several distinct places, and it is worth separating them because different sites gain more from different ones.

• Avoided downtime cost. A planned pump service happens in a window you chose, with parts staged and a technician scheduled. The same pump failing mid-shift means production or service interruption, emergency callout rates, expedited parts and whatever the consequential loss is. On a critical asset this single factor usually dominates the whole business case.
• Avoided secondary damage. Lubrication, filter changes and belt tensioning are trivially cheap. Running a bearing dry until it seizes can take out a shaft, a coupling and a seal. A large share of PM value is not preventing the primary failure but preventing the cascade.
• Labour efficiency. Planned work is batched, routed and resourced. Reactive work interrupts, travels, waits for parts and burns overtime. Wrench time on planned work is routinely a multiple of wrench time on reactive work, purely because of the logistics.
• Asset life extension. Assets that are cleaned, lubricated, aligned and kept within design conditions reach their design life. Assets that are not, do not. Deferred capital replacement is real money, it is just slow money that rarely gets attributed back to the maintenance budget.
• Statutory and insurance compliance. A great deal of PM exists because legislation, insurers or the fire authority require it. Lifting equipment examinations, pressure vessel inspections, fire system testing, electrical fixed-wiring inspection. The return here is not economic optimisation, it is licence to operate.
• Warranty protection. Manufacturer warranties are routinely conditional on documented servicing at specified intervals. A missing PM record is a rejected warranty claim, which is covered in more detail in the warranty management guide.

Now the part the vendor decks leave out. Preventive maintenance is not free of downside, and pretending otherwise damages your credibility with the finance director who has read more than one maintenance business case.

The honest position, and the one I take into every programme review, is that preventive maintenance is strongly positive on assets with age-related or wear-related failure modes and meaningful failure consequence, roughly neutral on low-consequence assets, and actively negative when applied intrusively to components that fail randomly. A programme that has never made those distinctions is almost certainly carrying tasks that cost more than they save.

3. The types of preventive maintenance task

Not all PM is the same kind of work, and the distinction matters when you write the task and when you decide the interval. In the schedules I build, every PM task falls into one of five categories.

The category that is most consistently under-used is the functional test. Protective devices, standby plant and alarms have what reliability engineers call hidden failures: they fail silently and you only discover it at the moment you needed them. A standby generator that will not start is not a problem until the grid drops. Functional testing is the only way to find hidden failures, it is almost always non-intrusive, and it is cheap. If a programme review turns up a thin PM budget, functional tests on protective systems are where I would spend the first money.

The category most consistently over-used is scheduled replacement. It is the most expensive, the most intrusive and the one most often applied out of habit rather than evidence. Replacing a component at a fixed age only makes sense if the component actually has an age-related failure pattern, and a surprising number do not.

4. How to build a preventive maintenance programme from nothing

This is the question I am asked most often, usually by someone who has just inherited a site with a partial asset list in a spreadsheet and a contractual obligation to produce a PPM schedule. The sequence below is the one I would follow, and the order matters more than the tooling.

• Step 1: build the asset register. You cannot schedule maintenance on assets you have not recorded. Walk the site, capture every maintainable asset with a unique identifier, location, make, model, serial and commissioning date. This is tedious, it is the single largest effort in the whole programme, and there is no way around it. Structure it properly from day one because retrofitting a hierarchy later is painful; the asset hierarchy design guide covers how to get the parent-child structure right.
• Step 2: classify by criticality. Rank assets by the consequence of their failure across safety, operational, financial, environmental and reputational dimensions. This single step determines where your PM effort concentrates, and skipping it is why so many programmes spread themselves thin. The method is in the asset criticality classification guide.
• Step 3: identify dominant failure modes for the critical assets. For each critical asset class, what actually goes wrong, and how? Bearing wear, seal leakage, contact erosion, filter blockage, corrosion, control failure. Use your own history if you have it, manufacturer documentation if you do not, and the experience of your technicians, which is usually the most accurate source and the least consulted.
• Step 4: choose a task that addresses each failure mode. For every failure mode worth managing, select the task type from the table above that will actually detect or prevent it. This is the step that separates an engineering programme from a copied one. If you cannot state which failure mode a task addresses, the task should not exist.
• Step 5: set the interval. Base it on the failure pattern, the usage rate, the manufacturer recommendation and any statutory minimum, whichever is most binding. Frequency selection is the single biggest lever on programme cost, and it is covered in depth in the preventive maintenance strategies guide, which compares time-based, meter-based and condition-based triggers properly.
• Step 6: write the task instruction. Turn each task into a numbered, unambiguous checklist with readings to record and pass/fail criteria. A task that says "service pump" is not a task, it is a hope. More on this below.
• Step 7: load it into the CMMS and level the schedule. Configure the PM records, generate the annual calendar, then flatten the peaks. A first-pass schedule always clusters absurdly, typically in January and at every quarter end. Spread the load across the year to match available labour.
• Step 8: measure, review, prune. After the first full cycle, review every task against what it actually found. Tasks that never find anything on non-critical assets are candidates for interval extension or deletion. The discipline of pruning is what keeps a programme healthy, and the PM programme design guide goes into why programmes bloat and how to reverse it.

5. A worked preventive maintenance schedule

Below is a representative PPM schedule for a mid-sized commercial or institutional facility. It is not a standard and it is not a substitute for manufacturer documentation or your own failure analysis, but it is a realistic starting frame that a planner can adapt. Frequencies assume normal duty in a Gulf climate, which means more frequent attention to filters, coils and anything exposed to dust and heat than a temperate-climate schedule would need.

Two notes on using a table like this. First, the frequencies are a starting point to be adjusted against your own failure history after the first annual cycle, not a fixed answer. Second, for statutory items the frequency is not yours to optimise. Where legislation, the insurer or the authority having jurisdiction sets an interval, that interval is the floor. For general FM task content and frequency benchmarking, the SFG20 maintenance standard is the most widely referenced library in the UK and Gulf FM market, and for fire-system testing regimes the NFPA codes are the usual reference where local codes adopt them.

6. Preventive maintenance forms, sheets and checklists that technicians actually use

A PM task is only as good as the instruction the technician receives. The most common weakness I find in programme audits is not the schedule, it is the task content: hundreds of PM records whose entire instruction is "carry out routine maintenance as per manufacturer recommendation". That is not an instruction, it is a liability. It cannot be executed consistently, it cannot be audited, and it produces no data.

A usable preventive maintenance form, whether it is a paper sheet, a PDF or a mobile CMMS checklist, has the same anatomy:

• Header block: asset ID, description, location, work order number, PM code, date, technician name, start and finish time.
• Safety preconditions: isolation required, permit reference, PPE, lock-out tag-out confirmation. This goes at the top, not the bottom.
• Numbered task steps: one action per line, in the order they should be performed, each with a tick box and a pass/fail or done/not-done outcome.
• Readings table: named measurements with units and an acceptable range printed next to the entry field. This is the part that turns a checklist into a data source, and it is the part most often missing.
• Defects found section: free text plus a flag to raise corrective work. Every inspection PM must have an exit route for the defects it discovers, or the inspection is pointless.
• Parts and consumables used: for cost capture and stock reconciliation.
• Sign-off: technician and, for critical or statutory work, a verifier.

Here is a reusable preventive maintenance sheet you can lift directly. This is a quarterly centrifugal pump service, one of the most common PM tasks in any facility, written the way I would want to see it in a CMMS.

Notice what that sheet does that a one-line instruction cannot. It captures five numeric readings per quarter, which after a year gives you a genuine condition trend on every pump in the estate without buying a single sensor. It forces a decision on the seal rather than leaving it to judgement. It gives defects a route into corrective work. And it records duration, which is what lets you cost the programme honestly later.

The same anatomy works for a forklift preventive maintenance checklist, which is worth including because mobile plant is where operator-led PM has the most leverage. The pre-use check below is designed for the operator, not the technician, and takes about four minutes.

7. How preventive maintenance works inside a CMMS

A PM programme of any size needs a system. Spreadsheet-driven PPM schedules work up to perhaps a few hundred tasks and then collapse, because the version control, the completion tracking and the history all degrade at once.

The mechanics are broadly the same across IBM Maximo, SAP PM, Hexagon EAM, Infor EAM, Planon and the mid-market tools such as MaintainX, Limble, Fiix, UpKeep and eMaint. The vocabulary differs, the model does not:

• Job plan / task list: the reusable content, the numbered steps, readings, estimated labour hours, trades and parts. Written once, referenced by many assets.
• PM record / maintenance plan: the link between a job plan and a specific asset or location, carrying the frequency, the next due date and the generation rules.
• Generation: the scheduler that turns due PM records into actual work orders, typically run nightly with a lead time so planners see work before it is due.
• Work order: the executable instance, assigned, scheduled, executed and closed. Understanding how PM work orders differ from corrective, emergency and project work orders in your type coding matters more than people expect, and is covered in the work order types guide.
• Meter / usage trigger: for hour or cycle based PMs, the reading source that drives generation instead of the calendar.
• History and failure coding: the closure data that makes the next programme review possible. Consistent structure here is what allows you to prove or disprove a task's value; see the failure codes guide.

Three configuration decisions cause most of the operational pain I see, and all three are easy to get right at the start and awkward to change later.

• Fixed versus floating next-due dates. If a quarterly PM is completed three weeks late, should the next one fall on the original anniversary or three months from actual completion? For statutory and seasonal work you want fixed. For wear-driven servicing you usually want floating. Getting this backwards produces either impossible schedules or slowly drifting compliance.
• Nested and hierarchical PMs. A monthly, quarterly and annual PM on the same asset should not all land in the same week with overlapping content. Nesting means the annual absorbs the quarterly and monthly scope so the technician does one visit, not three. Most platforms support this; most implementations do not use it.
• Route-based PMs. For large populations of small identical assets, fire extinguishers, emergency lights, FCUs, a route PM that covers fifty assets in one work order is far more efficient than fifty individual work orders. The trade-off is coarser per-asset history, so use routes for low-criticality populations and individual PMs for critical plant.

On platform choice, the honest answer is that PM functionality is not where the products differ much. Every serious CMMS does calendar and meter-based PM, job plans and route PMs competently. The differences that matter are integration, mobility, reporting and total cost, which is the angle taken in the CMMS buyer shortlist.

8. Preventive maintenance for IT assets and computer equipment

Preventive maintenance on computers and IT infrastructure is a category that deserves its own treatment because it is usually handled by a different team, on a different system, with very different assumptions. The failure modes are mostly not mechanical, which changes what is worth scheduling.

For a computer, the genuinely worthwhile preventive tasks are narrow:

• Thermal management: dust removal from fans, heatsinks and filters. This is the one classic mechanical failure path in a computer, and in a dusty environment it is the dominant one. Semi-annual in an office, quarterly or monthly in a workshop or plant floor.
• Storage health monitoring: SMART attribute review and SSD wear-level checks. This is condition monitoring rather than preventive work, and it is more useful than any fixed-interval task because drive failure does give warning.
• Backup verification: not taking the backup, which is automated, but testing that a restore actually works. This is a functional test on a protective system, and it is the single most under-performed PM task in IT.
• Patch and firmware currency: scheduled, tested and staged rather than automatic on production systems.
• UPS and battery testing: the most mechanical and most age-degrading component in the IT estate, and the one with a genuinely predictable end of life.
• Environmental checks: server room temperature, humidity, cooling redundancy changeover, leak detection. These belong to facilities but protect IT, and the handover between the two teams is where they get forgotten.

9. What to measure: the PM metrics that tell you the truth

Most PM reporting measures activity. Work orders raised, work orders closed, percentage complete. Those are useful operationally and useless strategically, because a programme can achieve one hundred percent PM compliance while reliability gets worse. The measures below are the ones I would put on a monthly pack, in roughly the order I would build them.

The pairing that matters most is PM compliance alongside PM yield. Compliance without yield means you are diligently performing tasks that find nothing. Yield without compliance means the tasks are valuable and you are not getting to them. Reporting both forces an honest conversation that neither number alone permits. For how these fit into a wider performance framework, including the SLA and contractual layer, see the FM KPI framework and the SLA matrix design guide.

10. When to stop at preventive and when to go further

Preventive maintenance is not the top of a ladder you should climb for every asset. It is the right answer for a large share of the estate, and the wrong answer for two groups at either end.

• Below preventive: run to failure. For low-cost, easily replaced, low-consequence assets with a fast corrective response available, scheduling maintenance costs more than the failures it prevents. A cheap extract fan in a back-of-house corridor does not need a PM. Declare run-to-failure deliberately, record the decision, and hold spares. What you must not do is let assets drift into run-to-failure by omission, because then you get the cost of failure without the saving of a considered decision.
• Above preventive: condition-based and predictive. For high-consequence assets whose failure modes develop detectably over time, fixed-interval work is a blunt instrument. Monitoring actual condition and intervening on evidence gives better reliability for less intrusion. The trigger-selection logic is in the preventive maintenance strategies guide, and the deeper treatment of failure prediction and remaining useful life is in the predictive maintenance guide.
• Sideways from preventive: design out the failure. Sometimes the right answer is not any maintenance strategy. If an asset fails suddenly with no detectable warning and high consequence, no schedule and no sensor will help. Redundancy, a design change or a different component is the answer. Maintenance strategy cannot fix a design problem, and recognising that boundary saves a lot of wasted effort.

A realistic distribution on a mature site looks something like this: a modest share of assets on condition-based or predictive regimes, concentrated on critical rotating and electrical plant; the majority on a lean, well-written preventive schedule; and a deliberate tail on run-to-failure with spares held. If your estate is one hundred percent preventive, you have almost certainly over-invested in the tail and under-invested at the top.

11. The mistakes I see in preventive maintenance programmes

These recur with remarkable consistency across sectors and platforms:

• Inheriting a programme without auditing it. The schedule arrives with the building or the contract and nobody ever asks whether the tasks make sense. Inherited programmes are almost always too large and too generic.
• Copying manufacturer recommendations wholesale. Manufacturer schedules are written conservatively to protect the manufacturer, not to optimise your total cost, and they assume duty and environment that may not be yours. Use them as an input, particularly where warranty depends on them, not as the answer.
• Task instructions that are not instructions. "Service as required" cannot be executed consistently or audited. If the technician has to guess, the outcome is a guess.
• No readings captured. A PM that records only a tick generates no trend and no evidence. Numeric readings are nearly free to capture and are the cheapest condition data you will ever own.
• Inspections with no defect route. Finding a fault and having nowhere to send it teaches technicians that reporting is pointless.
• Chasing compliance percentage as the goal. Compliance measures obedience to the schedule, not the value of the schedule. It is gameable and it is routinely gamed.
• An unlevelled schedule. Every annual PM due in January, every quarterly in the same week. Then everything slips, and the slippage is blamed on the technicians rather than the planner.
• Ignoring the technicians' knowledge. The people who repair the assets know which failure modes are real and which tasks are pointless. They are the cheapest and best source of programme improvement and they are consulted last.
• Never pruning. Programmes only ever grow, because adding a task is safe and removing one feels risky. Without a review cycle that deletes tasks, the programme bloats until compliance becomes impossible and everything is late.

The idea to walk away with

Preventive maintenance works when each task exists because a specific failure mode, on a specific asset whose failure actually matters, is being prevented or detected at an interval that makes economic sense. That is the whole discipline. Everything else, the software, the mobile app, the compliance dashboard, is machinery for executing and proving that judgement.

Which means the leverage is not in doing more preventive maintenance. It is in doing less of it, better targeted, with instructions specific enough that the work produces data, and with a review cycle honest enough to delete the tasks that never find anything. A lean programme on the right assets, executed properly and measured on yield as well as compliance, beats a comprehensive programme nobody can finish. Every time.

Final thoughts

If you are standing up a programme from scratch, resist the temptation to start with the software. Start with the asset register and the criticality ranking, because those two artefacts determine everything downstream and no platform will produce them for you. If you are inheriting a programme, do not start by improving compliance. Start by auditing the task list against the three-question test, because improving compliance on a bloated programme just makes you efficiently wasteful.

And whatever the state of your programme, the cheapest improvement available is almost always the same one: add numeric readings to your existing inspection tasks. It costs a technician thirty extra seconds per asset, it needs no capital and no sensors, and after one annual cycle you own a condition trend on your whole estate. That is the sort of unglamorous compounding work that separates maintenance organisations that improve from those that just keep busy.

Related reading: Preventive maintenance strategies (time vs meter vs condition), PM programme design: quality over quantity, Predictive maintenance and failure prediction, Asset criticality classification, Asset hierarchy design, Work order types in a CMMS, FM KPI framework, CMMS buyer shortlist.