Facilities Maintenance Management: A Guide

Most people do not choose facilities maintenance management. It arrives. A building opens, a portfolio grows, a colleague leaves, and one morning you are the person who owns the lifts, the chillers, the fire alarm panel, the leaking roof, the contractor who never turns up, and a mailbox full of complaints. Nobody hands you a manual, because the discipline is taught almost entirely by apprenticeship and the literature that does exist is mostly software marketing. This guide is the manual I would want if I were starting again: what the function actually consists of, in what order to build it, and which parts you genuinely cannot skip.

The message up front: facilities maintenance management is not a collection of trades, it is an information discipline. You are managing a register of assets, a calendar of obligations, a queue of requests, a pool of labour, a stock of parts and a budget, and every failure of the function traces back to one of those six records being wrong. Get the asset register and the statutory calendar right and everything else becomes tractable. Get them wrong and no amount of software, contracting or effort rescues you.

1. What facilities maintenance management actually covers

Facilities maintenance management is the practice of keeping the physical fabric and the engineering systems of a building or portfolio safe, compliant, functional and financially predictable across their whole life. That is a broad definition, and deliberately so, because the job in a single office block looks very different from the job across forty retail units, a hospital or a university campus. What stays constant is the structure of the work.

The conventional split in the industry, and the one you will meet in every tender document and every service contract, is hard services and soft services.

• Hard services are the fixed engineering and fabric elements: heating, ventilation and air conditioning, electrical distribution, lighting, plumbing and drainage, lifts and escalators, fire detection and suppression, generators and uninterruptible power supplies, building management and controls, water systems, roofing, facade and structure. These are the systems that carry statutory obligations, that cost serious money when they fail, and that require competent engineering judgement rather than just a schedule.
• Soft services are the recurring people-delivered services that keep the place usable: cleaning, waste, pest control, landscaping, security and manned guarding, front of house, catering support, porterage and internal moves. They are high frequency, high visibility and heavily people dependent. They generate most of the day-to-day complaints and comparatively little of the technical risk.

The split matters for practical reasons rather than semantic ones. Hard services are managed by asset and by obligation: you maintain a specific chiller on a specific regime because of what it is and what governs it. Soft services are managed by area and by frequency: you clean a floor to a standard so many times a week. Those are two different management models, two different kinds of specification, and frequently two different supplier markets. Confusing them is why so many maintenance regimes end up either over-specified on cleaning or dangerously thin on statutory engineering work.

A third category deserves separating out, even though contracts usually bury it inside hard services: statutory and compliance maintenance. This is work you do because a law, a regulation, an insurer or a licence requires it, not because you judged it economically sensible. It behaves differently from everything else, and section five deals with it on its own.

Finally, understand what facilities maintenance management is not. It is not project work or fit-out, although it frequently inherits the consequences of both. It is not space and occupancy management, although in many organisations the same person owns it. And it is not a procurement exercise: how you contract for delivery, whether you bundle everything into a single integrated provider or hold a dozen specialist contracts, is a separate decision that sits on top of a well-defined maintenance requirement. That contracting question is covered in the integrated facility management model explainer, and I would not touch it until the requirement itself is defined.

2. The asset register: the foundation nobody wants to build

If you take one thing from this guide, take this. The asset register is the single record that determines whether every other part of the function works. A maintenance schedule is a list of tasks against assets. A budget is a forecast of spend against assets. A compliance file is evidence of work done on assets. A lifecycle plan is a projection of replacement for assets. A work order is an instruction to attend an asset. If the register is wrong, incomplete or invented, all of those outputs are wrong in ways that are very hard to detect, because they will still look plausible.

Here is the honest part. Building an asset register is tedious, unglamorous, physically tiring survey work, and everybody wants to skip it. Every organisation I have worked with has at some point proposed a shortcut: take the consultant's handover list, take the previous contractor's schedule, take the drawings, take an export from whatever system the last provider used. All of those are useful starting inputs and none of them is a register. Handover lists describe what was designed, not what was installed or what has since been replaced. Contractor schedules describe what that contractor was being paid to visit, which is a commercial artefact, not an inventory. Drawings age badly. The only way to get a register you can rely on is for competent people to walk the building and record what is physically there.

What a usable asset record contains, at minimum:

Two design decisions will haunt you if you get them wrong: how deep to go, and how to structure the hierarchy. On depth, the practical test is whether you would ever raise a work order, hold a spare, or record a statutory certificate against the item. If yes, it is an asset. If no, it is a component of one. Registering every valve and every light fitting individually produces a register nobody maintains; registering an entire plant room as one asset produces history so coarse it tells you nothing. On structure, the parent and child relationships between site, building, system and equipment are a genuine design exercise with long-term consequences, and I have written that up separately in the asset hierarchy design guide. Read it before you start surveying rather than after.

3. The maintenance types in plain terms

The vocabulary here is badly abused, so it is worth setting out plainly. There are four types you need to distinguish, and the difference between them is what triggers the work.

• Reactive or corrective maintenance is triggered by something being broken or reported. Somebody notices a fault, or an alarm fires, and you respond. It is unavoidable and it is not automatically bad: for cheap, non-critical, quickly replaced items, waiting for failure is the rational choice.
• Preventive or planned maintenance (PPM) is triggered by elapsed time or accumulated usage. Quarterly filter changes, annual service, every five hundred running hours. It is predictable, budgetable and auditable, and its weakness is that it is blind to actual condition, so you will service healthy plant and occasionally still be surprised.
• Condition-based maintenance is triggered by a measured parameter crossing a threshold: differential pressure across a filter, vibration amplitude, temperature rise, water quality reading. It removes the waste of fixed intervals on assets whose condition genuinely varies, but it requires the measurement to exist and someone to act on it.
• Statutory or compliance maintenance is triggered by regulation, licence or insurance condition. Frequency and scope are set externally, not by your judgement of risk or economics, and the deliverable is usually a certificate or record rather than an improvement in condition.

Deciding the mix is the core judgement of the job, and it is made asset class by asset class rather than portfolio-wide. The decision inputs are consequence of failure, cost of the maintenance versus cost of the failure, whether the failure gives any detectable warning, and whether an external obligation removes your discretion entirely.

A rough shape that holds up in most commercial portfolios: statutory work is fixed and comes first, a substantial planned programme covers the significant plant, condition-based methods are applied selectively where the measurement is cheap and the variation is real, and reactive work is what remains rather than what dominates. If you want the strategic comparison in more depth, including where predictive methods sit above condition-based, that is covered in the preventive maintenance scheduling guide.

4. The statutory and compliance layer

This section carries a caveat that genuinely matters. Statutory maintenance requirements vary substantially by jurisdiction, by building type, by occupancy and by use, and they change. Nothing in this guide is legal or compliance advice, and the specific obligations that apply to your buildings must be verified locally, with competent professional advice and against the current regulations, codes, licences and insurance conditions in force where you operate. What follows are the generic categories that exist in most regulatory regimes, offered so that you know what to go and check, not as a list to implement.

The categories that commonly carry statutory or mandatory inspection and maintenance obligations include:

• Fire detection, alarm, suppression and means of escape, including detection systems, sprinklers and other suppression, extinguishers, emergency lighting, fire doors, dampers and escape routes.
• Electrical installations and equipment, including fixed installation inspection and testing, portable appliance regimes where they apply, and protective devices.
• Lifting equipment, including lifts, escalators, hoists, cradles and lifting accessories, typically with periodic thorough examination by a competent person.
• Pressure systems and gas installations, including boilers, pressure vessels, compressed air receivers and gas appliances.
• Water systems and hygiene, including cold and hot water services, cooling towers, and the monitoring and control regimes that manage waterborne risk.
• Ventilation and indoor air, including local exhaust systems, kitchen extract and ductwork hygiene.
• Standby and emergency power, including generator and uninterruptible power supply testing regimes.
• Hazardous materials and building fabric registers, where the building age and jurisdiction require a managed register and periodic reinspection.

There are three operating rules I would apply to this layer regardless of jurisdiction. First, obligations attach to assets, so every applicable regime should be recorded against the asset in the register, not held in a manager's head or in a contractor's schedule. Second, the deliverable is evidence, which means a dated record, the name and competence of who performed the work, the result, and any defects raised and their closure. Third, competence is part of the requirement: many of these tasks can only be discharged by a suitably qualified or accredited person, and a signature from someone who does not hold that competence is not compliance.

5. Building a PPM schedule from scratch

Once you have a register and a view of your obligations, the planned maintenance schedule is what turns them into a year of work. The mistake almost everyone makes on a first attempt is to invent it. Somebody sits with a spreadsheet and assigns intervals from memory, experience and guesswork, and the result is a schedule with no defensible basis, inconsistent depth between asset classes, and no way to explain to a finance director or an auditor why the chiller is quarterly and the pump is six-monthly.

The better approach is to start from a published maintenance task library and adapt it. A structured library gives you, for each asset type, a defined set of tasks, a recommended frequency, an indication of the skill level required, and in the better ones a flag for which tasks are statutory or safety critical. That means your schedule starts from an industry baseline you can cite rather than from personal preference. SFG20 is the best known example of this kind of library in the building services world; manufacturer operation and maintenance manuals are the other primary source, and for specific plant they should override a generic library. Professional bodies such as IWFM and IFMA publish good-practice material worth reading alongside.

A workable sequence for building the schedule:

• Group the register by asset type. You are not writing tasks for four hundred assets, you are writing them for perhaps thirty asset types and applying them.
• Pull the baseline tasks and frequencies for each type from your chosen library, and overlay the manufacturer requirements for any plant still in warranty or with unusual specification.
• Mark the statutory tasks separately so that they are visibly non-negotiable and cannot be deferred in a budget squeeze without an explicit decision.
• Adjust frequency by criticality and environment, not by preference. A critical asset in a harsh, dusty, coastal or continuously loaded environment earns a shorter interval; a duplicated standby unit in a clean plant room may justify a longer one. Record the reason for every deviation from the baseline.
• Level the workload across the calendar. A schedule that puts everything in January and March is a schedule that will not be delivered. Spread annual tasks across the year, respect seasonal logic (cooling plant before the hot season, heating and drainage before the wet one), and check the resulting labour hours per month against the resource you actually have.
• Write down what completion means. A task with no defined checklist and no recorded readings produces a tick, not evidence. This is where a lot of otherwise decent schedules quietly fail.

The detailed mechanics of turning this into a live, resource-levelled schedule with routes, checklists and meter triggers are covered in the guide to building a preventive maintenance schedule. For the purposes of this guide, the point is simply that the schedule is derived, documented and defensible rather than invented.

6. The helpdesk and the service request flow

Planned work is what you control. Reactive work is what arrives, and the way it arrives determines how the function is perceived. Most of an organisation's opinion of facilities management is formed by what happens between a person reporting a problem and that problem being resolved, which means the request flow deserves real design attention even in a small operation.

The flow has six steps, and each one is somewhere I have seen operations lose control:

• Intake. One route in, or a small number of clearly published routes. The failure mode is an operation where requests arrive by email, phone, corridor conversation and message to a technician's personal number, so nothing is counted and nothing can be improved.
• Triage and classification. Someone competent decides what it is, what asset or location it relates to, and what priority it carries. Classification at intake is what makes the data useful later.
• Prioritisation. A published priority scheme with a small number of levels, each with a defined response and rectification target, and, critically, written rules for which situations fall into which level. Without the rules, priority becomes a negotiation and the loudest requester always wins.
• Assignment. To an in-house technician, a specialist contractor, or a queue. Assignment should be a routing rule, not a daily act of improvisation.
• Execution and recording. What was found, what was done, what parts were used, how long it took, and whether it is fixed or a temporary measure pending follow-up.
• Closure and feedback. The requester is told it is done. This step costs almost nothing and does more for the reputation of a maintenance function than any other single change.

On priorities, keep the scheme short. Four levels is usually enough: an emergency tier for safety, security and major service loss; an urgent tier for significant impact on operations; a routine tier for everything that matters but can be planned into a normal working week; and a scheduled tier for work batched to a next planned visit. Targets should be times you can actually hit, and separated into response (someone attends and makes safe) and rectification (the problem is resolved), because the two are genuinely different promises. If you are formalising these into a contractual or internal service agreement, the structure and the traps are set out in the SLA matrix design guide.

7. The labour model and why it drives everything

Who does the work is the structural decision that shapes cost, response, flexibility and the amount of management effort the function needs. There are three models and the right answer is almost always the third.

• In-house. Directly employed technicians. You get familiarity with the buildings, immediate response, discretionary capacity for small improvements, and much better retention of institutional knowledge. You pay for fixed headcount whether the work is there or not, you carry training and competence management, and you cannot cover every specialism.
• Fully outsourced. Contractors deliver everything under specification. You convert fixed cost to variable, buy specialist competence on demand, and transfer some risk. You also lose day-to-day visibility, spend real effort on specification and contract management, and discover that anything not written into the specification is a variation with a price attached.
• Mixed. A small in-house core covering general trades, first response, minor works and supervision, with specialist contracts for lifts, fire systems, controls, water hygiene, high-voltage work and anything requiring accreditation. This is what most well-run operations converge on, because it matches the reality that some work is continuous and generic while some is periodic and specialist.

The important insight is how much the labour model determines. Your PPM schedule has to be achievable by the resource you have, so labour capacity constrains frequency. Your response targets are only credible if someone is available to meet them, so the model constrains the service agreement. Your spares holding depends on whether your own technicians or a contractor carries the stock. Your data quality depends on whether the people doing the work are inside your system or reporting through a portal. Deciding the schedule first and the labour model afterwards is the wrong order, and it produces schedules that are quietly never delivered.

One caution on contractor competence: for statutory and safety-critical work, accreditation and qualification are not administrative preferences, they are part of the requirement. Verify them at appointment, record them against the contract, and diarise their expiry, because an expired certification discovered during an audit is indistinguishable from never having had one.

8. Spares, stores and the small-parts problem

Spares management is where a lot of otherwise competent operations leak time and money quietly. The symptom is familiar: a technician attends, diagnoses correctly, and then cannot complete the job because the part is not held, so the visit is wasted, the job goes into backlog, and somebody pays a premium to expedite a component that costs very little.

The basic discipline is modest and worth doing even without software. Identify the consumables your planned schedule consumes predictably, filters, belts, lamps, seals, gaskets, chemicals, and hold them because the demand is known in advance. Identify the critical spares whose absence causes real service loss on high-criticality assets, and hold those on a risk basis rather than a usage basis. Everything else is bought on demand. Then insist that parts consumption is recorded against the work order, because that record is the only thing that tells you what you actually use and it feeds directly into both budgeting and failure analysis.

The two failure modes are opposite and both common: a storeroom full of obsolete stock bought for plant that was replaced years ago, and a storeroom with nothing in it because stock was cut as an easy saving. The deeper mechanics of stock control, reorder logic and MRO catalogue structure are covered in the spare parts and MRO inventory guide.

9. Budgets, the planned to reactive ratio, and lifecycle planning

A facilities maintenance budget has three distinct parts, and keeping them separate is what makes the numbers defensible.

• Planned and statutory maintenance. Derived directly from the schedule: tasks, frequencies, labour hours, materials, and specialist contract values. This part should be forecastable to a close tolerance, because you know in advance exactly what is due.
• Reactive and corrective. Based on your own history of reactive volume and cost, held as a provision rather than a line-by-line plan. If you have no history, this is the first reason to start recording work orders properly.
• Lifecycle and capital replacement. Derived from the register: install dates, expected lives, condition grades and replacement cost estimates, projected forward to show what falls due in each of the next several years.

The planned to reactive ratio is the headline diagnostic of maintenance maturity. Expressed as the proportion of maintenance labour hours or cost spent on planned work versus unplanned response, it tells you whether you are managing the estate or being managed by it. A reactive-heavy operation is expensive in ways that do not all appear on the maintenance line: premium call-out rates, wasted travel, disruption to the occupier's business, accelerated asset deterioration, and staff who spend their whole week firefighting and therefore never get to the work that would reduce the firefighting.

Be realistic about how fast that ratio moves. Shifting a genuinely reactive operation towards planned dominance takes more than one budget cycle, usually two or three, and the reason is arithmetic rather than attitude. To do more planned work you need capacity, but your capacity is currently consumed by reactive work, and the reactive work only falls once the planned work has been running long enough to improve asset condition. There is a period in the middle where you are funding both, and if that overlap is not explicitly budgeted and explained to finance up front, the improvement programme gets cut at exactly the point it starts costing money and has not yet started saving it. Say this at the beginning, not in month eight.

Lifecycle planning is the part most likely to be missing entirely in a young operation, and the part finance values most once it exists. A ten-year projection of what needs replacing, when, and at roughly what cost turns facilities from an unpredictable annual argument into a planned capital programme. It is also the single most persuasive argument for having built the asset register properly, because a lifecycle plan is nothing more than the register sorted by remaining life.

10. Data and reporting: making the function defensible

Facilities maintenance has a structural political problem: when it is working, nothing happens, and nothing is hard to take credit for. The defence against that is evidence, which means the function has to produce a small set of numbers consistently, from records created as a by-product of doing the work rather than assembled by hand at month end.

The minimum reporting set I would establish first:

• PPM completion against plan, for the period, with statutory tasks reported separately and never netted into the overall percentage.
• Compliance status by regime: what is due, what is complete, what is overdue, and where the evidence is held.
• Reactive volume and mix, by priority, by building, by asset type, and by trade.
• Response and rectification performance against the published targets, by priority level.
• Backlog, counted and aged, and separated into safety-relevant, condition-relevant and cosmetic. An uncounted backlog is a liability nobody has quantified.
• Planned versus reactive ratio, trended over time rather than reported as a single figure.
• Cost per building and per square metre, which is what lets you compare buildings, spot outliers and answer benchmarking questions credibly.

Two warnings. Reports built by hand every month will be abandoned within a year, so the measurement has to be a by-product of the work record. And a measure that nobody makes a decision from should be dropped, because every metric you publish is a promise to act on it. A wider treatment of metric selection, targets and the politics of reporting is in the FM KPI framework.

11. Where software fits, and when a spreadsheet stops being enough

Software is not the foundation of this discipline, it is the mechanism that makes the foundation operable at scale. A spreadsheet is a perfectly honest starting point, and for a single small building with a handful of assets and one technician it may remain adequate for years. What a spreadsheet cannot do is enforce a process, hold an audit trail, work on a phone in a plant room, or answer a question about the state of the estate without someone rebuilding the answer by hand.

The signals that a spreadsheet has stopped being adequate are fairly consistent:

• More than one person needs to update the same records, and you have started emailing versions around.
• You cannot produce compliance evidence for a given asset and period in a few minutes.
• Reactive requests are arriving faster than anyone can log them, so the log is no longer complete.
• Work history is not reliably attached to assets, so you cannot see what has happened to a given piece of plant.
• Technicians are recording work on paper that someone retypes later, which means the data is always days old and partly wrong.
• Monthly reporting has become a multi-day manual exercise.
• You have added a second site, or contractors who need controlled access to your records.

On the categories, the short version is that the same underlying need is sold under several labels with different emphasis, and the sensible sequence is to define your requirement first and pick a category second, not the other way around. I have written the selection process up separately in the guide to choosing facility management software, and the operational view of what a maintenance system does inside a facilities team in CMMS for facilities management. Both assume what this guide argues: that software implemented on an unverified register and an undocumented schedule digitises the existing mess rather than resolving it.

12. A realistic first twelve months

If you have inherited an operation that is reactive, under-documented and possibly non-compliant, the temptation is to fix everything at once. The pattern that actually works is sequential, because each phase depends on the one before it.

A word on system implementation and where it lands in this sequence. It is tempting to buy software in month one because it feels like decisive action. In practice the most useful placement is around the middle: late enough that the register survey and the schedule design are underway so you have something real to load, early enough that the schedule and the request log go live in the system rather than being migrated later. What you should not do is put the purchase in month one and treat the implementation as the improvement programme. The register, the compliance calendar and the schedule are the improvement programme. The system is where they live.

The idea to walk away with

Facilities maintenance management is a records discipline wearing overalls. Six records, well kept, produce a function that is safe, predictable and defensible: the asset register, the statutory obligations calendar, the planned maintenance schedule, the request and work history, the spares position, and the budget with its lifecycle projection. Every serious problem in the field traces back to one of those six being incomplete, out of date or invented.

That is a genuinely encouraging conclusion, because it means the improvements that matter most are within your control and do not require a large budget. Surveying the estate properly, writing down what governs each asset, deriving a schedule from a published baseline, and logging every request through one route are all things a competent person can start this month. They are also the things everyone wants to defer in favour of something that feels more strategic, which is precisely why so many operations remain reactive for years.

Final thoughts

If you are new to this and want an order of operations, it is this. Find out what your statutory obligations are and whether you can evidence them, because that is the only category where the gap cannot be recovered later. Then build the asset register, properly, once. Then derive the schedule from a task library rather than from memory. Then decide the labour model, the priority scheme and the spares holding, in that order, because each depends on the previous. Then measure, and only then worry about which software category you belong in.

Be patient with the reactive ratio. It moves over years, not quarters, and the middle period where you are funding both the firefighting and the prevention is the part to explain honestly at the start rather than defend later. And be sceptical of anyone who tells you the discipline can be fixed by a purchase. The unglamorous records work is the job. Everything else is a way of doing it more efficiently once it is done.

Related reading: How to choose facility management software, Integrated facility management explained, Asset hierarchy design, Building a preventive maintenance schedule, SLA matrix design for FM operations, Spare parts and MRO inventory, FM KPI framework, CMMS for facilities management.