property management operations research
Measuring Preventive Maintenance Completion Without Counting Emergency Repairs
A measurement design for preventive maintenance completion, separating scheduled tasks from emergency repairs, with reproducible denominators, deferral states, sampling, and honest limits.
| 6 sources
Table of contents
Research question and scope
This brief asks how a property management team can report preventive maintenance completion without inflating the number with emergency repairs. The question sounds like an accounting detail, but it changes what the metric means. A report that counts every completed work order as preventive maintenance will look strong in a month full of emergency calls, even if no scheduled task was finished. The goal here is a rate that describes planned work and can be reproduced by another reviewer.
The unit of analysis is one scheduled preventive maintenance task for one asset at one property during a defined period. A task may be an inspection, a filter change, a belt replacement, a test, a lubrication, or any other recurring activity the manager has written into the maintenance calendar. The scope excludes emergency repairs, resident-requested repairs, capital replacements, and work that was never scheduled in the first place. Each task should carry a stable identifier so the same item can be followed across periods.
The method is documentary. It can show how many scheduled tasks existed, how many were completed on time, how many were deferred, and where the record disagrees with the vendor invoice. It cannot prove that a completed task was performed correctly or that the interval was technically appropriate. Those are inspection and engineering questions that require a different kind of evidence.
Why completion rate hides the definition problem
Preventive maintenance is usually reported as a percentage: completed tasks divided by scheduled tasks. The denominator is where most of the trouble lives. If the schedule changes during the month, if tasks are added retroactively, or if emergency work is mixed into the same queue, the percentage can move without any change in actual performance. A rising rate may simply mean the definition of the population changed.
Teams also tend to count a task as complete when a technician visits, even if the visit produced a note rather than a completed service. The gap between attended and completed is small in a well-run program and large in a struggling one. A useful measure has to distinguish the two. Otherwise the report rewards activity and hides unfinished work behind a status label.
Finally, emergency repairs are not failures of preventive maintenance in every case. Some emergencies are unavoidable, and some are caused by resident behavior or weather. Blending them into the preventive metric makes the operation look better during a bad month and worse during a quiet one. Keeping the populations separate is more honest and more useful for decisions.
Define the scheduled task population
The schedule should be the source of the denominator. Before the period begins, list every recurring task with its asset, property, interval, and next due date. Record the version of the schedule that applied during the period, because a schedule edited after the fact can erase the tasks that were missed. A dated export is usually enough to make the population reproducible.
Distinguish a standing task from a one-time assignment. A quarterly backflow test is a standing task with a recurring due date. A one-time inspection requested by an owner is project work and should not enter the preventive completion rate. If the team wants to track both, it should report them separately rather than combining them into a single number that answers neither question.
Define what counts as due. A task is due when its scheduled date falls inside the reporting period, even if it was completed early. A task completed before the period for a due date inside the period should still be counted in that period's numerator and denominator. This convention prevents a team from improving the rate simply by doing next month's work early and removing it from next month's population.
Separate planned work from emergency work
The cleanest separation is a work order type that is set at creation and not changed later without an audit note. Preventive work orders should be generated from the schedule, not opened by a technician after an emergency visit. If a scheduled task is discovered during an emergency call, the scheduled task should still be tracked as its own record with its own due date. The emergency record and the planned task should reference each other without merging.
Emergency work should have its own queue, its own response measures, and its own reporting. A common mistake is to close an emergency work order as preventive maintenance because the technician also changed a filter. That closes the emergency accurately but misstates the preventive program. The filter change belongs to the scheduled task, and the emergency visit belongs to the emergency record.
A conversion between the two types should require a reason and an approver. A scheduled inspection that becomes an emergency repair is a real event, and the record should show the change rather than hiding it. The conversion should also preserve the original due date so the reviewer can see that the planned task did not happen as planned. A converted task that later receives a new due date should keep both dates in the history.
Record deferral and cancellation states
Completion is not a binary. A task can be completed on time, completed late, deferred with a new date, cancelled because the asset was removed, or still open past its due date. Each state should map to a different follow-up and a different treatment in the rate. Collapsing them into complete or incomplete hides the difference between a managed deferral and a missed task.
A deferral is legitimate when it has an owner, a reason, and a new date that the manager has accepted. The record should show all three. A task that is repeatedly deferred without a new date is drifting, and the report should show it as open rather than as scheduled work in progress. That visibility is often more valuable than the headline completion rate.
Cancellation needs its own evidence. If an asset was replaced, sold, or taken out of service, the schedule should be updated and the cancellation linked to the change. A cancellation without a reason looks like a missed task in a later audit, even when the underlying decision was correct. The note should name the person who approved the change and the source record that supports it.
Build denominators that survive a schedule change
Schedules change, and the measurement should survive those changes without becoming a moving target. Use the schedule version that was in force at the start of the period for the denominator, then disclose any tasks added or removed during the period in a separate reconciliation. That approach keeps the original commitment visible while acknowledging real change. The reconciliation should be attached to the report so a reader can see exactly which tasks moved.
If a new property is added mid-period, the cleanest option is to report it separately for its first partial period. Folding a new portfolio into an existing rate can make the rate drop for reasons that have nothing to do with performance. A separate line for new properties lets the reader compare like with like, and it gives the team a fair starting point. The same treatment applies when a property changes management mid-period.
The same logic applies to a property that leaves the portfolio. Remove it from the forward schedule, but keep its completed tasks in the period where they occurred. A reviewer should be able to reconstruct the population from the schedule export, the additions and removals log, and the final count. If those three do not reconcile, the rate is not ready to publish.
Sampling and reconciliation
A full count is best when the volume is manageable. When it is not, sample by property and by task type rather than selecting the easiest files. Record the selection rule and the population size before the sample is drawn. A sample that always includes the same well-run property will produce a rate that says more about the sample than about the program.
Reconcile the maintenance calendar against vendor invoices and completed work orders. Count tasks that the calendar marks complete but that have no supporting invoice, photo, or checklist. Count the reverse as well, where a vendor invoice exists for a task the calendar never scheduled. Both directions reveal process defects that a simple percentage misses.
Repeat the reconciliation after the first repair cycle. The purpose is to confirm that the fix changed the record, not just the report. If the same task type keeps failing, the cause may be an unrealistic interval, a vendor capacity problem, or a schedule that was never loaded correctly. The next repair should target that cause rather than adding another reminder.
What the measure cannot prove
The completion rate cannot prove that a task was performed to a professional standard. It can show that a work order was closed with the required evidence, but a closed order is not the same as a safe asset. Inspections, spot checks, and vendor quality reviews are separate controls that answer a different question. A manager should treat a high completion rate as a prompt to sample the work, not as proof that the work was good.
The measure also cannot prove that the scheduled interval was correct. A team can complete every task on time and still miss a failure that the interval never addressed. Engineering guidance and manufacturer instructions should inform the schedule, while the completion rate only describes adherence to that schedule. Confusing the two can create false confidence.
Finally, the measure cannot prove that deferrals were justified. It can show that a deferral had an owner and a new date, but the decision to defer may still have been wrong. A manager should review deferrals with safety or compliance implications individually, especially where the asset affects habitability or a regulated system. The data supports that review; it does not replace it.
Practical takeaways
- Set the denominator from a dated schedule export before the reporting period begins.
- Keep preventive work orders separate from emergency and resident-requested repairs.
- Track on-time completion, late completion, deferral, cancellation, and open past due as distinct states.
- Reconcile the schedule against invoices and completed work orders in both directions.
- State the limits of the measure and review safety-related deferrals individually.
Sources
- U.S. Department of Housing and Urban Development, "Real Estate Assessment Center." https://www.hud.gov/program_offices/public_indian_housing/reac
- U.S. Department of Housing and Urban Development, "NSPIRE." https://www.hud.gov/program_offices/public_indian_housing/programs/ph/inspections/nspire
- U.S. Bureau of Labor Statistics, "Property, Real Estate, and Community Association Managers." https://www.bls.gov/ooh/management/property-real-estate-and-community-association-managers.htm
- U.S. Census Bureau, "Housing Vacancies and Homeownership." https://www.census.gov/housing/hvs/index.html
- U.S. Environmental Protection Agency, "ENERGY STAR for Buildings and Plants." https://www.energystar.gov/buildings
- Internal Revenue Service, "How long should I keep records?" https://www.irs.gov/businesses/small-businesses-self-employed/how-long-should-i-keep-records
- PropertyManagementBiz, "How to Audit a Preventive Maintenance Schedule with a Property Management VA." https://propertymanagementbiz.com/blog/property-management-preventive-maintenance-schedule-audit-va