Ask a maintenance team to split last month's work orders into planned and unplanned and you will get an answer. Ask them what rule they applied and the answer falls apart. Some will count every preventive task as planned and every corrective repair as unplanned. Some will count anything with a date in the system as planned. Some will count anything the supervisor knew about on Monday morning as planned. Those are three different measurements wearing the same label, which is why the resulting percentage tells you almost nothing. The fix is not a better definition of one term. It is recognising that two separate axes are in play, and learning to read a job's position on both.
The message up front: planned versus unplanned describes whether the work was prepared before it started. Preventive versus corrective describes whether the item had already failed. These are independent axes, so all four combinations are real, and the combination most organisations barely recognise is planned corrective work: the asset has failed or is failing, but the job is scoped, parted, permitted and windowed before anyone picks up a spanner. That quadrant is where the biggest available improvement usually sits.
1. Two axes, not one scale
The single most useful correction I make in maintenance workshops is to stop people drawing one line with "planned" at the good end and "unplanned" at the bad end, with preventive and corrective somewhere along it. That picture is wrong in a way that quietly damages decisions for years.
Here are the two axes, stated as plainly as I can:
- Axis one: prepared or not. Was this job worked out in advance? Was the scope defined, were the parts and tools identified and secured, was the labour and duration estimated, was access and permitting arranged, and was a window agreed with whoever owns the asset? If yes, it is planned. If the crew arrived and started figuring it out in the field, it is unplanned, regardless of how urgent or important it was.
- Axis two: failed or not. Had the item already lost its required function, or was it detectably on its way there? If yes, the work is corrective. If the item was still performing and the work was done to stop it from failing, the work is preventive.
Nothing about the first axis determines the second. A failure can be handled with careful preparation. A preventive task can be thrown at a crew with no preparation at all. Once you see that, the whole vocabulary settles down, and the arguments about whether a particular job "counts as planned" mostly evaporate.
This mapping also sits reasonably close to the formal terminology. EN 13306:2017, the European maintenance terminology standard published by CEN (there is no ISO equivalent), structures maintenance as preventive, splitting into predetermined and condition-based with predictive treated as a form of condition-based, and corrective, splitting into immediate and deferred. That deferred corrective category is close to what practitioners mean when they say planned corrective work: the failure has occurred, but the repair is held back and prepared rather than attacked on the spot. If you want the dependability vocabulary underneath all of this, IEC 60050-192:2015 is the formal home of the failure, fault and repair terms, and ISO 14224:2016 is the reference for how maintenance and reliability data on equipment should be collected and exchanged. I would read all three as scaffolding rather than scripture, and I would not attempt to quote clause wording from documents most readers have not bought.
The test that settles most arguments
To decide whether a job is planned, ask one question: before the technician started, did somebody already know what was going to be done, with what, for how long, and in what window? If the answer needs a "well, roughly", the job was not planned. Urgency does not enter into it. Neither does whether the asset had failed.
2. The four quadrants, with worked examples
Cross the two axes and you get four quadrants. All four exist in real operations, though they exist in very different volumes, and three of them are usually invisible in the reporting because the reporting only has two buckets. The examples below are hypothetical and illustrative, chosen because they are the shapes I see most often rather than because they describe any particular site.
| Quadrant | What it looks like | Worked example (illustrative) | What it costs | How to get more or less of it |
|---|---|---|---|---|
| Planned preventive prepared · not failed |
A routine task issued from a recurring schedule, with a checklist, a known parts kit and a known duration. | Quarterly service on an air handling unit. Filters, belts and a bearing grease kit are drawn from the store the day before; the unit is taken off line during a pre-agreed low-occupancy window; the checklist and readings come back with the closed work order. | Labour and materials, plus the productive availability given up during the window. Modest and forecastable. | More, but only where justified. Volume is not the aim; a bloated preventive programme is its own waste. Get the task content right first. |
| Planned corrective prepared · failed or failing |
A known defect that is deliberately held, prepared properly, and executed in an agreed window. Close to what EN 13306:2017 calls deferred corrective maintenance. | A chiller condenser pump is found with a rising vibration signature and a weeping seal. It is still running and there is standby capacity, so the repair is not done that afternoon. Instead the seal kit and coupling are ordered, the isolation and permit are prepared, a two-technician four-hour job is written, and the work is done in an agreed window eleven days later with the standby unit carrying the load. | Very little beyond the repair itself, because the waiting, improvising and expediting have all been removed. The residual cost is the risk carried during the deferral period, which is why the deferral has to be a judgement, not a habit. | Much more. This is the quadrant with the largest unrealised gain in most organisations. It requires a defect-reporting route, a disciplined backlog, and the nerve to hold a repair rather than swarm it. |
| Unplanned preventive not prepared · not failed |
The rarest quadrant. An opportunistic preventive task taken because the asset happened to be available, with no preparation behind it. | A production line is down for an unrelated electrical fault. While the crew waits, someone decides to grease and inspect a conveyor drive that was not due for six weeks, using whatever consumables are on the trolley. Genuinely useful, entirely unprepared. | Usually small, and sometimes negative, because opportunistic access is free downtime. The real costs are an unrecorded or badly recorded task and a schedule that now contains a task that has quietly already been done. | Keep it, but capture it. Opportunistic work is sensible. The fix is not to suppress it but to record it against the asset and reset the schedule, otherwise it corrupts the history. |
| Unplanned corrective not prepared · failed |
The breakdown. What everyone actually means when they say "reactive". Work begins before anyone knows the scope. | A domestic water booster set trips at 06:40 with no standby. A technician is pulled off a scheduled task, diagnoses a failed pressure transducer, finds no spare, sources one by courier at premium cost, and completes the repair at 15:20. Two planned jobs slip a week, and one of them was the inspection that might have caught this. | The largest and most under-measured of the four. The repair is the smallest part of it; see section 6. | Less, but never zero. Reduce it by converting failures into prepared work and by catching defects earlier, not by reclassifying it. |
Read that table twice and you will notice something about the second row. Planned corrective work is not a compromise or an accounting trick. It is a distinct operating mode in which failure has already happened and the organisation still keeps control of the response. Most sites have far less of it than they could, because their instinct on hearing about a defect is to send someone immediately. For the full treatment of the corrective side of the axis, see the corrective maintenance guide, and for the unprepared end, the reactive maintenance guide and the breakdown maintenance guide.
3. What "planned" actually requires
The word planned is used so loosely that it has almost stopped carrying information. In most systems it means "has a date". That is not planning, and the gap between the two is a very common cause of jobs that start and then stall.
Work is planned when all five of the following are true before the technician starts:
- A defined scope. Somebody has written what is to be done, to what standard, and what "finished" looks like. Not "investigate pump 3" but the specific intervention, with the acceptance condition.
- Identified parts, materials and tools. Not just listed, identified: part numbers confirmed against the actual asset, availability checked, and reserved or staged. A job whose parts list says "gaskets as required" is not parted.
- An estimated duration and labour requirement. How many people, of which trades, for how long. Without this the job cannot be scheduled honestly, and the schedule becomes a wish list.
- Access and permits arranged. Isolation identified, permit to work raised or prepared, keys and escorts organised, confined space or height provisions in place, the area cleared. In regulated environments this is the step most likely to stop a job dead on arrival.
- An agreed window. Agreed with the person who owns the asset or the space, not merely entered in a system by the maintenance department and hoped for.
Work that has a date in a system but fails any of those five is scheduled, not planned. Say it plainly to your own team, because the distinction is what separates an organisation whose schedule holds from one whose schedule is fiction by Wednesday. A scheduled and unplanned job typically reaches the asset, discovers a missing part or a missing permit, and returns. It then appears in the reporting as planned work that was completed late, which is a flattering and misleading description of what happened.
Where this definition is inconvenient
Apply the five tests honestly to an existing backlog and the planned proportion will drop, possibly sharply. That is uncomfortable to present, and it is the reason many organisations never apply the tests. If you do apply them, say clearly that the figure fell because the definition tightened, not because performance fell. A measurement that becomes stricter and a performance that becomes worse look identical in a trend line, and confusing the two has cost more than one planner their credibility.
4. Planning and scheduling are different functions
The second conflation, and the one that explains a surprising number of stalled improvement programmes, is treating planning and scheduling as one job. They answer different questions, they need different skills, and combining them in one overloaded person usually means the scheduling gets done and the planning does not.
Planning is what and how. Scheduling is when and who. Planning happens once per job and is technical. Scheduling happens continuously and is logistical.
| Responsibility | Planning (what and how) | Scheduling (when and who) |
|---|---|---|
| Core question | What work is required, and how should it be done? | When will it be done, and by whom? |
| Scope definition | Owns it. Writes the task steps and acceptance condition. | Takes it as given. Does not redefine scope. |
| Parts and materials | Identifies, verifies against the asset, requests and reserves. | Confirms availability on the target date before committing the job. |
| Labour | Estimates trades, headcount and duration. | Allocates named people against available capacity. |
| Permits and access | Determines what is needed and prepares the paperwork. | Times the isolation and the permit to the agreed window. |
| Coordination with operations | Establishes what shutdown or access the job requires. | Negotiates and books the actual window. |
| Time horizon | Works ahead of the schedule, typically on the backlog. | Works the near horizon: next week, next day, today. |
| Failure mode if absent | Jobs arrive at the asset and stall. Dates hold, work does not. | Well prepared jobs sit in the backlog and never get executed. |
Read the last row carefully, because it is the diagnostic. An organisation with a scheduler and no planner produces a full, tidy schedule of jobs that keep failing on arrival, and concludes that its technicians are unproductive. An organisation with a planner and no scheduler produces beautifully prepared job packs that nobody ever books. The first pattern is far more common, and it is usually misdiagnosed as a discipline or attitude problem rather than a missing function. If you have someone titled planner who spends the morning reallocating today's crews around an emergency, you do not have a planner. You have a second scheduler.
Whichever system you run this in, the mechanism is the same: the work order types and statuses have to be able to distinguish a job that is prepared from a job that is merely dated. If they cannot, no amount of reporting will recover the distinction. The work order types guide covers how to structure that, and any competent maintenance system will support it; this is a configuration and discipline question rather than a product question.
5. The real cost of unplanned work
When organisations cost a breakdown they typically cost the repair: the hours booked and the parts consumed. That is the smallest component. The costs that matter are the ones that never reach the work order.
- Waiting time. The technician is on site and unable to proceed: waiting for a part, for an isolation, for a permit signature, for the area to be cleared, for someone to confirm whether the asset can be taken off line. This is paid time producing nothing, and in unplanned work it frequently exceeds the wrench time.
- Expedited parts. Premium freight, out-of-contract supply, and sometimes the wrong item because it was ordered from a description rather than from a verified part number. The unit cost difference between a stocked part and an expedited one is rarely small, and it is almost never attributed back to the absence of planning.
- Rework from improvisation. A repair done with the tools and materials that happened to be available is often a repair that has to be revisited. Unplanned work generates a disproportionate share of repeat visits, and because the second visit is logged as a new job, the link back to the improvised first attempt disappears from the data.
- Safety exposure. This is the one I would put first. Unprepared work means hazards identified in the field under time pressure, isolations arranged in a hurry, permits chased rather than prepared, and people improvising access. Preparation is a safety control, not just an efficiency measure, and an organisation running mostly unplanned work is carrying a risk profile it has not consciously accepted.
- The knock-on to the schedule. This is what turns one breakdown into a lost week. Pulling two technicians onto an emergency does not just cost the emergency. It displaces the scheduled work they were on, which slips, which pushes the following week's schedule, which means preventive tasks are deferred, which raises the probability of the next failure. That feedback loop is the actual mechanism by which organisations become trapped in reactive mode, and it is invisible if you only ever cost the breakdown in isolation.
There is a widely repeated industry claim that reactive work costs some multiple of planned work, and you will see specific multipliers quoted with great confidence in vendor material. I decline to put a number on it. The commonly cited figures are an unsourced industry assertion, they vary wildly between the sources that repeat them, and the true ratio on your site depends on your labour model, your stocking policy, your contract structure and your production consequence. The direction is not in doubt and the mechanism above explains why. The magnitude is something you should derive from your own records, not import from a slide. For the downtime and backlog measurement that supports that derivation, see the backlog and downtime tracking guide.
6. The goal is not zero unplanned work
This needs saying firmly, because the opposite is implied by many improvement programmes, and the implication does real damage.
Some failures are genuinely unforeseeable. A random electronic failure, a manufacturing defect surfacing years in, an external event, a human error during operation: these have no detectable precursor and no amount of planning capacity converts them into prepared work. Some assets, meanwhile, do not justify the analysis required to foresee their failures. A low-consequence, cheap, quickly replaced item is correctly left to fail, and the resulting job is correctly unplanned. Requiring it to be planned would cost more than the failure.
The serious risk is behavioural. An organisation that sets an aggressive planned-work target and manages people against it does not usually get more planned work. It gets more work classified as planned. The mechanisms are well known to anyone who has audited maintenance data: the emergency gets raised as a planned job after the fact; a standing job number absorbs reactive hours; a job is backdated into last week's schedule; the definition of planned quietly loosens until "we knew about it that morning" qualifies. At that point the figure improves, the reality does not, and the organisation has lost the instrument it was using to steer.
What I would not do
I would not publish a target planned-work percentage, and I would be sceptical of any benchmark figure offered for one. There is no credible universal target: the right proportion depends on your asset mix, your failure modes, your redundancy, your criticality profile and your contractual obligations. A figure that is right for a redundant district cooling plant is wrong for a single-line production facility. Set direction against your own measured baseline and explain the movement, rather than chasing someone else's number.
7. How to move work from unplanned to planned
The transition is not a policy announcement. It is five capabilities, and the last one is the one that gets skipped.
- Backlog discipline. A backlog is not a list of things that are late. It is the inventory of identified work from which planning draws. It needs to be sized, aged, prioritised and actively groomed, and it needs entries that are specific enough to plan against. A backlog of vague one-line requests cannot be planned no matter how much planner time you throw at it.
- A functioning defect-reporting route. This is the supply line for planned corrective work, and it is where most organisations are weakest. Technicians and operators see developing problems constantly. If reporting one is slow, or invites an argument, or results in the reporter being told to just fix it now, they stop reporting. The route has to be fast, low friction, and it has to visibly lead to prepared work rather than to an immediate callout.
- Condition monitoring where it buys warning time. The point of condition monitoring in this context is not sophistication, it is lead time: converting a future breakdown into a defect you know about weeks early, which is exactly the raw material for planned corrective work. Deploy it where the failure modes are detectable and the warning interval is long enough to be useful, and do not deploy it where failures are sudden. The preventive, predictive and reactive comparison covers that strategy choice properly, and it owns that territory; my concern here is only that whichever strategy you choose feeds prepared work rather than alarms.
- Planning capacity. Planning is a distinct role with distinct output, and it needs enough of the week to produce job packs ahead of the schedule. An organisation that wants more planned work and has allocated nobody to prepare it has made a wish, not a plan.
- Protection of planner time. Here is the honest observation: planner time is the first thing consumed by today's emergency, and it is consumed invisibly. The planner is the most knowledgeable person available, so when something breaks they get pulled in, and the work they were preparing for three weeks out does not get prepared, which guarantees more emergencies in three weeks. Breaking that loop requires someone senior to protect planner time explicitly and accept a slightly worse response today for a materially better position next month. In practice this single decision determines whether the improvement holds or quietly reverses within two quarters.
The wider reliability work sits alongside this rather than inside it. Reducing failures at source is a different project from preparing the failures you do get, and both are needed. See equipment reliability and how to improve it and what reliability engineering actually is for that side, and the preventive maintenance guide for building the routine programme that supplies the first quadrant.
8. How it is measured, and why the figures mislead
Two measures carry this topic, and both are conceptually simple and practically fragile. I am deliberately describing them without target values.
The planned-work proportion expresses prepared work as a share of total work, usually by labour hours rather than job count, because counting jobs treats a ten-minute lamp change and a two-day pump overhaul as equals. It answers: how much of our effort was prepared before it started? Its weakness is that it depends entirely on the classification rule, so it is trivially gameable by loosening the rule, and it is not comparable between two organisations or even between two years in the same organisation unless the rule is written down and stable.
Schedule compliance expresses how much of the committed schedule was actually executed in the committed period. It answers: does our schedule mean anything? Its weakness is the mirror image: it improves if you commit to less, or if you add work to the schedule after doing it. A site can show excellent compliance on a schedule that contains a fraction of the work that needed doing. The detailed treatment of that measure belongs to the PM KPIs and schedule compliance guide; the conceptual point here is only that the two measures must be read together, because each one covers for the other's failure mode.
How to read a suspiciously good figure
A planned-work proportion that jumps sharply in one period rarely means the organisation learned to plan in one period. Planning capability moves slowly because it is built from backlog quality, defect reporting and planner capacity, none of which change in a month. A sudden jump usually means reclassification. Before celebrating it, ask what changed in the rule, who raises the jobs, and whether emergencies are being logged after the event.
The idea to walk away with
Planned versus unplanned is not the same question as preventive versus corrective, and treating them as one scale is what makes the whole topic muddy. Prepared or not is one axis. Failed or not is the other. They are independent, all four combinations occur, and the one that deserves far more attention than it gets is planned corrective work: failure has happened, and the organisation still controls the response because the job was scoped, parted, permitted and windowed before anyone started.
Two follow-on points are worth holding alongside it. Planned means prepared, not dated, and work that is merely scheduled will stall on arrival. And planning is not scheduling, so an organisation with a scheduler and no protected planner will keep producing tidy schedules full of jobs that cannot be executed.
Final thoughts
If you want one practical exercise from this, take last month's work and sort it into the four quadrants rather than two buckets, applying the five tests of preparation honestly. The planned figure will probably fall. The useful discovery will be how small the planned corrective quadrant is, and how much of what sits in unplanned corrective was actually known about days before anybody touched it. That gap between "known about" and "prepared for" is not a technology gap and it is not a budget gap. It is a gap in who was given the time to prepare, and it is the most reliably available improvement in maintenance operations.
And resist the pull toward a target. The organisations that make real progress on this do not chase a percentage. They fixed the defect-reporting route so developing problems surfaced early, protected a planner's week from today's emergency, and let the proportion move where it moved. The number followed the capability, which is the only order in which it means anything.
Disclosure
Alongside advisory work I also build a CMMS and CAFM platform, so I have a commercial interest in this category. Nothing above is a recommendation for it, and no vendor named here has paid for inclusion or had any editorial input. Weigh the analysis accordingly.
Stuck in reactive mode?
Independent advisory on maintenance planning and scheduling structure, backlog and defect-reporting design, work order taxonomy, and measurement that does not mislead. 22+ years across utilities, oil and gas, manufacturing, government and facility operations.
Book a conversationRelated reading: Corrective maintenance: complete guide, Reactive maintenance: complete guide, Breakdown maintenance: complete guide, Preventive vs predictive vs reactive, Maintenance backlog and downtime tracking, PM KPIs and schedule compliance, Work order types in a CMMS.
Primary sources: CEN-CENELEC (EN 13306:2017 maintenance terminology), IEC (IEC 60050-192:2015 dependability vocabulary), ISO (ISO 14224:2016 reliability and maintenance data; ISO 55000:2024 and ISO 55001:2024 asset management). All are paywalled in whole or part; check the current published edition before relying on any terminology in a contract or specification.
Muhammad Abbas
CMMS / CAFM Manager & Independent Advisor · 22+ years across enterprise CMMS, EAM, CAFM and ERP implementations in utilities, oil and gas, manufacturing, government and facility operations.
Work with me