Preventive Maintenance Plan
The program document, not a task list. Which assets get preventive attention and which are left to run to failure, why, what it costs, who delivers it, and how you will know within a year whether it worked.
- A strategy per asset, chosen against ten explicit tests
- Run to failure treated as a legitimate deliberate choice
- Hours unresourced and funding shortfall, calculated
- Ten measures with a baseline, because a plan is judged on them
Preventive Maintenance Plan
The program, not the task list
| # | How the strategy was chosen | Done | Evidence | By |
|---|---|---|---|---|
| 1 | Criticality assessed against safety, compliance, output and cost | |||
| 2 | Failure history reviewed for the last two to three years | |||
| 3 | Cost of prevention compared against cost of failure |
The document you will get. Download for the full, editable file.
Who this preventive maintenance plan is for
This is the one document in the set written to be approved rather than used, which changes what belongs in it.
The plan owner
Usually the maintenance planner or FM lead. Sections 4 and 5 are your work: a strategy per asset and the evidence behind each choice, which is what turns a list of intentions into a defensible position.
The budget holder
You read two rows first, hours unresourced and funding shortfall, and you are right to. A plan showing zero on both with no working behind it has not been costed, it has been hoped for.
The technicians who deliver it
The enablers section is about you: whether spares are held, procedures written, training current and access agreed. A plan that fails any of those twelve lines is not deliverable however good the strategy is.
Whoever reviews it a year later
The measures are your instrument. Ten of them with baselines and targets, reported monthly, so the review is a comparison rather than an argument about whether things feel better.
Which section matters the most in your sector
The structure holds across sectors. What changes is which consequence dominates the criticality assessment, and that is what decides where the money goes. If you run one of these, the sector page goes further than the template does.
- Data centres
Downtime cost per hour. When a failure is measured in minutes to over-temperature, the cost of prevention comparison stops being close and the strategy writes itself.
Data centre maintenance software - Healthcare
Statutory and insurance intervals as non-negotiable. Several assets have intervals you cannot trade against cost, and the plan has to show which ones were never up for discussion.
Healthcare maintenance software - Commercial real estate
Scope and what is explicitly excluded. Tenant and third-party assets you do not maintain need to be written down, because the assumption otherwise runs both ways at once.
Portfolio maintenance software - FM service providers
Roles and responsibilities. Under contract the plan is the document that says which party owns which measure, and an unnamed owner is a measure nobody reports.
FM service provider software - Retail and malls
What we will stop doing. Multi-site teams are the ones most often asked to add a program to an unchanged reactive load, and that line is where it gets resolved or quietly lost.
Retail maintenance software - Education
Access risk. Term dates compress the windows in which anything can actually be done, which belongs in the risks section rather than being discovered by the schedule.
Campus maintenance software
What a preventive maintenance plan should contain
A preventive maintenance plan is the program document that decides which assets receive preventive attention, under which strategy, at what cost. It records the objectives and their baselines, the criticality and strategy per asset with the evidence behind each choice, the resourcing and budget, who is accountable, and the measures the program will be judged on.
A. Fields specific to a preventive maintenance plan
| Field | What goes in it | Why it earns its place |
|---|---|---|
| Objectives with a baseline | What you are trying to change, measured, from what, to what, by when | The baseline is the field that gets skipped and the one that makes the rest work. Without it the review becomes an argument about whether things feel better than last year. |
| Scope, and what is out of it | Assets in scope, assets explicitly excluded and why, and tenant or third-party equipment | Naming the exclusions matters more than naming the inclusions, because an unstated exclusion is assumed to be covered by whoever is not covering it. |
| Criticality per asset | A rating, the consequence of failure, and the downtime cost per hour | Downtime cost is what converts an opinion into a comparison. An asset whose failure costs a known amount per hour can be argued about on arithmetic. |
| Strategy per asset | Run to failure, time-based, condition-based, predictive or redesign | Five options rather than a default. Run to failure is legitimate and deliberate for a cheap asset with a spare on the shelf; what is not legitimate is arriving there because nobody decided. |
| Ten selection tests | The evidence behind each strategy choice, with a source and a date | This is what makes the plan defensible. The eighth test, cost of prevention against cost of failure, is the one that changes decisions rather than confirming them. |
| Resourcing and budget | Hours required, available, contracted and unresourced, then the costs | Hours required comes from the schedule rather than from memory. Unresourced hours and funding shortfall calculate themselves, which is why an approver reads them first. |
| Roles and responsibilities | Seven roles, each with a person, a responsibility and a time commitment | Time commitment is the unusual column and the honest one. A plan that assigns 0.4 FTE of planning to somebody with no time freed is a plan that will not run. |
| Twelve enablers | The things that must be true for the plan to be deliverable at all | Asset register, task library, schedule, system of record, spares, procedures, training, access, KPIs, escalation and backlog rules. Any one missing undermines the whole. |
| Backlog rules | What carries forward and what is canceled, agreed in advance | The last enabler and the one that prevents the commonest failure. Without a rule, missed tasks accumulate until the backlog is larger than the program itself. |
| Ten measures | Each with a definition, a baseline, a target, a frequency and an owner | Compliance, schedule compliance, preventive to reactive ratio, MTBF, downtime, deficiency find rate, cost per asset, backlog, repeat failures and emergency work orders. |
| Deficiency find rate | How much each PM actually finds | The measure people forget and the one that tells you whether intervals are right. A PM that never finds anything is either too frequent or is not being done properly. |
| What we will stop doing | The trade being made to create room for this | The honest line. A preventive program added on top of an unchanged reactive workload does not happen, and naming the trade is what makes the plan credible to whoever signs it. |
Section 5 is what separates a plan from a wish, and its eighth line is where most plans would change if anybody filled it in. Comparing the cost of prevention against the cost of failure is uncomfortable because it occasionally produces the answer that a task is not worth doing, and maintenance culture is not built to welcome that. But a program that maintains everything equally is a program with no strategy, and it spends its scarce hours on the assets that are easiest to reach rather than the ones that matter. The five strategy options exist for the same reason. Run to failure sounds like negligence and is frequently the correct engineering answer for a cheap, non-critical, redundant asset with a spare on the shelf. Writing it down as a decision, with the evidence beside it, is what makes it defensible later. The failure mode is not choosing run to failure, it is arriving at it by accident and then having to explain that nobody had considered the asset at all.
B. What it looks like filled in
One row from the criticality table, and the arithmetic behind it. This is what section 5 is asking you to show for every asset that gets a strategy.
| Field | Entry | What it settles |
|---|---|---|
| Asset | Chiller CH-01 | Named, not a category |
| Criticality | A | Top band, so it gets the scrutiny |
| Consequence of failure | Server room over temperature within 20 minutes | Minutes, not inconvenience |
| Downtime cost per hour | 1,400.00 | Turns the argument into arithmetic |
| Strategy | Condition-based | Chosen, not defaulted |
| Why this strategy | Vibration and approach temperature trend well | Feasibility tested before it was assumed |
The last row is doing the work that most plans skip. Condition-based maintenance is the fashionable answer and it is only available where the asset actually gives you a signal you can trend. The sixth selection test asks you to establish that feasibility before assuming it, and here somebody has: vibration and approach temperature both trend usefully on this chiller, so monitoring will give warning rather than noise. Had those signals been poor, the honest answer would have been time-based, and a plan that had written condition-based anyway would have quietly delivered neither. The downtime figure is the other half. At 1,400 an hour with over-temperature twenty minutes away, the cost of prevention comparison is not close, and nobody has to relitigate this asset's budget line every year.
Word for the version that goes to an approver with the strategy and evidence written out, Excel for the resourcing arithmetic, the criticality table and the Plan Review tab that tracks the measures period by period, PDF for circulation. Free, and yours to rebrand.
How do you write a preventive maintenance plan?
This is written to be approved, so work it in the order an approver reads it: what it is for, which assets and why, what it costs, and how you will know. Six steps.
Set objectives with a baseline against each
What the program is trying to change, how it will be measured, where it stands now, the target and the date. Capture a baseline even if you have to estimate it, and say that it is estimated. An objective with no baseline cannot be reviewed.
Draw the scope, including what is out of it
Assets and systems in scope, what is explicitly excluded and why, the sites covered, tenant or third-party assets you do not maintain, and the interfaces with statutory compliance, capital replacement and shutdown planning.
Assign criticality and a strategy per asset
A criticality rating, the consequence of failure in concrete terms, the downtime cost per hour, then one of five strategies: run to failure, time-based, condition-based, predictive or redesign. Record why that strategy and not another.
Show the evidence behind the choices
Ten tests, each with evidence and a date: criticality assessed, failure history reviewed, failure modes identified, manufacturer intervals obtained, statutory intervals identified, condition monitoring feasibility established, spares checked, prevention cost compared against failure cost, redundancy considered, operator input taken.
Cost it, resource it, and name who does what
Scheduled hours from the schedule rather than from memory, against in-house capacity, contracted hours and the unresourced remainder. Then labor, contract, materials and tooling costs against the approved budget. Then the seven roles with a real time commitment each.
Work the enablers, the measures and the trade
Twelve enablers, each in place or with a gap, an owner and a date, including the backlog rule. Ten measures with baselines and owners. Then the risks, ending with the honest line: what you will stop doing to make room for this.
Time-based versus condition-based maintenance
Section 4 asks for a strategy per asset, and for most assets the real choice is between these two. Getting it wrong in either direction is expensive in a different way.
| Aspect | Time-based | Condition-based |
|---|---|---|
| What triggers the work | The calendar, whatever the asset is doing | A measured signal crossing a threshold |
| What it needs to work | An interval you can defend | A signal that actually trends on that asset |
| Where it wastes money | Servicing things that did not need it yet | Monitoring that never tells you anything new |
| How it fails | A failure between intervals | A threshold set wrong, or a signal nobody reads |
| Best suited to | Wear-driven items and statutory tasks | Expensive assets that degrade with warning |
| The honest default | Correct far more often than people admit | Only once feasibility has been established |
Condition-based is the answer everyone wants and time-based is the answer most assets need. Monitoring only pays where the asset degrades in a way you can actually measure and where somebody will read the measurement, and the sixth selection test exists precisely because that feasibility gets assumed rather than established. A plan that specifies condition-based on an asset with no usable signal delivers neither strategy: the time-based tasks were dropped because monitoring was supposed to replace them, and the monitoring produces a trend nobody can interpret. Meanwhile statutory tasks settle the question on their own. Where an interval is set by law or insurance, it is time-based whatever the condition data says, and the plan should record it as non-negotiable rather than re-deriving it. The useful discipline is to treat condition-based as something an asset has to earn on evidence, and to be relaxed about time-based being the right answer most of the time.
When the template starts to feel limiting
A plan is a document about a year, and a year is exactly the timescale a document handles worst. Four places it strains.
Review dates that pass quietly
The plan sets a review frequency and an out-of-cycle trigger, both written in a file that gets opened when somebody remembers. Plans are most often reviewed after the thing they were meant to prevent.
The measures need assembling by hand
Ten measures with baselines and targets, reported monthly. Every one has to be compiled from somewhere else, so the reporting becomes a task in itself and is the first thing dropped when the month is busy.
Strategy and the work it generates live apart
Section 4 says condition-based for a chiller. Whether anything is actually monitoring it, and whether anyone acted on a trend, is in a different system and connected only by a person's memory.
The plan cannot see its own delivery
Hours required came from the schedule at a point in time. As assets are added and contracts change, the resourcing arithmetic in the plan silently stops describing the program being run.
What running this in Facilio looks like
The template is the paper version of this program. The strategy per asset is the same; the difference is that it generates the work, and the measures come back from that work rather than being compiled by hand.
Work Completion Validator
The measures reflect work actually done
PM compliance is only meaningful if completed means completed. Work Completion Validator holds each closure against its evidence, so the figure the plan is judged on is not inflated by jobs ticked rather than done.
Ops Performance Intelligence
Reviews and backlog rules apply themselves
Review dates, out-of-cycle triggers and the agreed backlog rule run against live data, so a compliance figure below target for two periods surfaces as the trigger it was written to be.
Audit Report Intelligence
Ten measures assemble themselves
Compliance, preventive to reactive ratio, deficiency find rate, backlog and repeat failures come from the work records, so the monthly report is a read rather than a compilation exercise.
Contractor Work Tracker
The contracted half of the plan is visible
Hours contracted, what was delivered against them and which measures the contractor owns sit against the contract, so the split between in-house and contracted stops being a gap in the reporting.
Hallucination-free by design. Atom AI answers from the records in your tenant rather than generating plausible text, so an empty field reads as empty rather than filled in for you.
Frequently asked questions
What is a preventive maintenance plan?
A preventive maintenance plan is the program document for a site or portfolio. It sets out what the program is trying to achieve with a baseline and target for each objective, which assets are in scope and which are deliberately excluded, the criticality and maintenance strategy chosen for each asset with the evidence behind that choice, what it costs and who delivers it, and the measures it will be judged on.
It is not a task list and not a calendar. The tasks live in a preventive maintenance checklist and the timing in a schedule. The plan is the document that decides which assets earn those tasks in the first place.
What should a preventive maintenance plan include?
Plan details and version control, objectives each with a baseline and target, and a scope that names what is excluded as well as what is covered. Then the criticality table with a consequence of failure and a downtime cost per hour, a strategy per asset, and the ten tests showing how each strategy was chosen.
Then delivery: resourcing and budget with unresourced hours and any funding shortfall calculated, seven roles with real time commitments, twelve enablers that decide whether the plan can be delivered at all, ten measures with baselines, the risks including what you will stop doing, and the governance and review arrangements.
What is the difference between a preventive maintenance plan and a schedule?
The plan decides which assets get preventive attention, under which strategy, at what cost, and how success will be measured. It is written to be approved and reviewed annually. The schedule places the resulting tasks into months and is worked all year.
Sequence matters. The plan sets strategy, the checklist supplies the task library, and the schedule sets the timing. Writing a schedule without a plan produces a calendar nobody can defend when asked why a particular asset is being maintained at that interval.
Is run to failure ever the right strategy?
Yes, frequently. For a cheap, non-critical asset with redundancy and a spare on the shelf, running it to failure and replacing it is often the correct engineering and financial answer, and the template lists it as one of five legitimate strategies.
What is not legitimate is arriving at run to failure by accident, because nobody assessed the asset. The difference between a deliberate choice and an oversight is entirely in whether it is written down with the reasoning beside it, which is what section 5 is for.
How do you measure whether a preventive maintenance plan is working?
With ten measures, each given a definition, a baseline, a target, a reporting frequency and an owner: PM compliance, schedule compliance, the preventive to reactive ratio, mean time between failures on critical assets, unplanned downtime hours, deficiency find rate per PM, PM cost per asset, backlog in hours, repeat failures within ninety days, and emergency work orders.
The one most often left out is deficiency find rate, and it is the most diagnostic. A PM that consistently finds nothing is either running too frequently or is not actually being carried out, and neither is visible in a compliance figure.
How much detail does the budget section need?
Enough that the two calculated rows mean something. Scheduled hours required, in-house hours available, hours to be contracted and hours unresourced, then in-house labor, contracted cost, materials and spares, tools and software, the total, the approved budget and the shortfall.
An approver reads hours unresourced and funding shortfall first. A plan showing zero on both with no working behind it has not been costed, and the usual consequence is a program that quietly under-delivers from about the second quarter.
What is the most common reason these plans fail?
Being added on top of an unchanged reactive workload. The plan is approved, nothing is taken away, and the preventive work loses every scheduling contest with a breakdown because breakdowns are louder.
That is why the risks section ends by asking what you will stop doing to make room. It is the least comfortable line in the document and the one that most determines whether the program survives its first busy quarter. The backlog rule in the enablers is the second: without an agreed rule on what carries forward and what is canceled, missed tasks accumulate until the backlog outgrows the program.
Can I edit and rebrand this template?
Yes. It is free to use, edit, rename and put your own logo on, internally or for clients. No attribution required.
The Word version is the one to edit if you want to match your own criticality bands, approval levels or governance language. The Excel version is the one to use if you want the resourcing arithmetic and shortfalls to calculate themselves, plus the Plan Review tab, which tracks all ten measures period by period so the annual review is a comparison rather than a recollection.
In one paragraph
A preventive maintenance plan decides which assets earn preventive attention and which do not, and it is written to be approved rather than used day to day. Give every objective a baseline, even an estimated one, because a plan with no baseline gets judged on how busy everybody looked. Set a criticality and a strategy per asset from the five available, and treat run to failure as the legitimate deliberate choice it often is for a cheap redundant asset, while never arriving there by accident. Show the evidence behind each choice through the ten tests, and take the eighth seriously: prevention that costs more than the failure it prevents is a habit rather than a strategy. Cost it properly, since an approver reads unresourced hours and funding shortfall before anything else. Work the twelve enablers, agree the backlog rule in advance, and answer the hardest line honestly by naming what you will stop doing to make room.
The template is the floor, not the ceiling
Take the plan; it will get a program approved and give the annual review something real to measure against. When compiling ten measures by hand every month becomes the reason the review slips, a connected CMMS closes it: strategy attached to assets, the work it generates tracked to completion, compliance and find rate and backlog read from the records rather than assembled, and the resourcing picture updating as the estate changes.
