Facility Maintenance KPIs Every Building Owner Should Track

The monthly maintenance report arrives as a PDF. PM compliance is near perfect. Average response time is down again. Every cell is green.

Then a rooftop unit fails in July, and you ask for its history. The quarterly PMs were all marked complete. Each one was closed in a few minutes, with no filter size, no readings and no photo. The report was accurate. It measured the wrong things, and nobody looked at the records underneath it.

This guide covers facility maintenance KPIs from the owner’s side of the table: the person who receives the report, not the person who writes it. For each KPI you get the formula, the data it needs, how it gets gamed and how to catch that. Then you get a one-page report template and contract terms that keep the numbers honest. If you are building or re-bidding a preventive maintenance program, these are the measures to write into it.

Which facility maintenance KPIs tell an owner the building is being maintained?

The 14 below, grouped around five questions. Is scheduled work done? How fast is broken work fixed? Is equipment more reliable? What does upkeep cost? Is risk building up?

Federal O&M guidance from Pacific Northwest National Laboratory (PNNL), written for building owners and their O&M teams, sets the principle. Use as few KPIs as possible while still detecting problems. Balance leading indicators, which warn before things degrade, against lagging ones, which record what already happened.

KPI	Question it answers	Type	Better when
PM compliance	Is scheduled work done on time?	Leading	Higher
Planned maintenance %	Do we choose the work, or does the building?	Leading	Higher
Emergency work order share	How much is crisis work?	Lagging	Lower
Response time	How fast does someone arrive?	Lagging	Lower
Resolution time	How fast is it fixed?	Lagging	Lower
SLA attainment	Did the provider meet the contract?	Lagging	Higher
Backlog and aging	Is work piling up?	Leading	Smaller, younger
Reopen / first-time fix	Was it fixed right?	Leading	Fewer reopens
Repeat failures, MTBF	Is equipment more reliable?	Lagging	Fewer, longer
MTTR	How long are key systems down?	Lagging	Lower
Cost per square foot	What does upkeep cost?	Lagging	Stable, explained
Cost by asset	Which equipment eats the budget?	Lagging	Flags replacements
Deferred maintenance, FCI	Is risk accumulating?	Leading	Lower
Tenant satisfaction	Do occupants notice?	Lagging	Fewer repeats

A word on targets. Maintenance-software blogs quote PM compliance targets anywhere from 80 to over 95 percent, usually without a source. The Society for Maintenance & Reliability Professionals (SMRP) publishes standard definitions and best-in-class targets for more than 70 metrics in its Best Practices Metrics, 7th edition as of October 2026 and free to members. This guide quotes a target only where a public body publishes one, and names it. Elsewhere, direction and trend matter more than any absolute number.

Preventive, predictive or reactive: why does the mix matter?

Reactive maintenance fixes equipment after it fails. Preventive maintenance services it on a calendar or runtime schedule. Predictive maintenance services it when a measured condition says it needs it. The mix shows how much work you choose and how much the building chooses for you.

The Department of Energy’s Federal Energy Management Program (FEMP) cites studies that put preventive savings at as much as 12 to 18 percent over a purely reactive program, and predictive savings at 8 to 12 percent over preventive alone. PNNL’s maintenance approaches guidance, adapted from FEMP, gives this benchmark mix for federal O&M programs:

Approach	Typical range	Target goal	Best in class
Reactive	40–60%	20–30%	Under 10%
Preventive	30–50%	25–35%	25–35%
Predictive	10–15%	30–40%	45–55%

PNNL notes that what a site can achieve depends on its mission and systems. Most commercial buildings will not instrument everything. A practical program is preventive at its core, with a condition-based layer where it pays: filters changed on pressure drop instead of the calendar, or thermographic scans of electrical gear.

One definition trips up owners. “Planned” is not the same as “preventive.” PNNL defines planned maintenance as work whose procedure, labor, materials and tools were defined and assured before it started. A well-planned pump replacement after a failure is planned, but not preventive. Ask which definition your report uses.

Planned maintenance percentage = hours on preventive, predictive and planned corrective work ÷ total maintenance hours × 100. Reactive share is the unplanned remainder. Both need every work order typed at creation and labor hours logged. Where trades do not report hours, a count-based version is a weaker fallback.

How do you calculate PM compliance, and how does it get gamed?

PM compliance = PM work orders completed by their due date (plus any grace window the contract defines) ÷ PM work orders due in the period × 100. It is usually the headline number, and the easiest to inflate.

The data behind it: a PM schedule tied to each asset, a due date fixed at generation, a system-stamped completion time, and checklist results with readings and photos. Our guide to building a preventive maintenance program covers how the schedule gets built.

Illustrative example: 80 PM work orders fall due in March and 74 close on time. Compliance is 92.5 percent. The useful question is what those 74 records contain.

How it gets gamed	How to detect it
Tasks closed without the work	Minutes-long completions; no readings or photos
Batch closing	Dozens closed in one hour, or on the last day
Due dates moved out of the month	Due-date change log; PMs due vs generated
PM schedules deactivated	PM count vs asset register, month over month
Checklists thinned	Task steps today vs at contract start
Inspections that never find anything	PM findings rate near zero

Add one companion measure, the PM findings rate: the share of PMs that generate a corrective follow-up, plus on-time completion of those follow-ups. A program that never finds a loose belt or a weeping valve is either over-scheduled or not being performed.

Split code-required inspections and tests from discretionary PMs. Fire alarm tests and elevator inspections are not a percentage. Each one was done, with a report on file, or it was not.

How should response time, resolution time and SLA attainment be measured?

Response time runs from request received to a qualified person on site, or another first action the contract defines. Resolution time runs from the request to verified completion. SLA attainment is the share of work orders, by priority, that met the contract’s targets. Report each by priority, with the median and the slowest cases rather than the average, because averages hide the tail that tenants remember.

The clock definitions do most of the work:

• Start: when the request entered the system through the portal, the phone line or a logged email. Not when a dispatcher first opened it.
• Response stop: technician on site, shown by a site photo or check-in, not a status changed from the office.
• Resolution stop: work complete with evidence, not “vendor scheduled.”
• Pauses: parts on order or no tenant access only, each visible on the record with a reason.

Emergency work order share = emergency-priority work orders ÷ all work orders × 100. A rising share means the preventive program is not catching failures. Penalize it directly, though, and emergencies quietly become “urgent.” Read the priority-change log and spot-check descriptions; a ceiling leak coded as routine is a finding. Clocks started at “acknowledged” and stretched pauses show up the same way: in the raw records, never in a summary.

Tenant satisfaction is the lagging check on all of these. PNNL lists occupant complaints per period as a lagging KPI. Count complaints from every channel, repeat complaints from the same space, and a short survey after closure. Complaints that arrive by email and never become work orders will flatter the result.

What do work order backlog and aging tell you?

Backlog is work not completed by its required-by date. Aging is how long each open item has waited. Together they show whether the team keeps up and whether old problems are quietly abandoned.

PNNL uses that definition and suggests a target for federal programs of under 5 percent of work orders in backlog status, trending downward. For an owner, the more useful view is a count of past-due work orders by priority and age band, measured from the original request date. An in-house team can also express backlog in weeks of work: estimated open hours ÷ weekly available labor hours.

How it gets gamed	How to detect it
Mass cancellation of old tickets	Cancellations by month, with reasons
Parked in “awaiting owner”	Every such item on your decision list
Closed and reopened to reset age	Age from the first request on that issue
Moved to “deferred” with no estimate	Each deferral in the deferred register

The last row matters most. A deferral is a legitimate decision, but it belongs in the deferred maintenance register with a cost estimate and a date, not out of sight.

Do MTTR, MTBF and first-time fix rate work for buildings?

Yes, for critical equipment with enough history. For a single building, a count of repeat failures per asset is often more honest than a mean.

MTTR (mean time to repair) = total downtime ÷ number of failures, for a defined asset or class. MTBF (mean time between failures) = total operating time ÷ number of failures. Both definitions come from the PNNL KPI guidance. The catch is sample size. Two boilers may see a handful of failures a year, and an average of two events is noise. Use trailing 12 months, pool identical equipment across a portfolio, or report repeat failures: corrective work orders per asset over the trailing year.

First-time fix rate = corrective work orders resolved on the first visit, with no return or reopen for the same issue inside a window the contract defines ÷ corrective work orders completed × 100. Its mirror is the reopen (rework) rate. PNNL suggests under 3 percent of work orders in rework status for federal programs, trending downward.

The classic game is ticket splitting. The first visit closes as “diagnosed.” The repair opens as a new ticket, and the return visit as a third. Response time, MTTR and first-time fix all improve while the tenant waits longer. To catch it, link every work order to an asset and a location. QR asset tags make that link reliable at the equipment. Then count work orders on the same asset or space within the window, and require follow-ups to reference their parent.

Which maintenance cost KPIs should an owner track?

Track three: cost per square foot for budgets, cost by asset for repair-or-replace decisions, and the split between planned and reactive spend.

Maintenance cost per square foot = total maintenance spend within a defined scope ÷ rentable or gross square feet, stating which. Scope matters more than arithmetic. Labor, parts, contracted services, and whether janitorial or utilities are in or out all change the number. Our facility management cost per square foot guide covers how to define it so comparisons mean something.

Cost by asset = trailing 12-month repair spend per asset, set beside its estimated replacement cost and age. A short list of the top assets by repair spend is one of the most useful pages an owner can get. It is the evidence behind decisions like repairing or replacing commercial HVAC.

Cost KPIs get gamed in three ways:

• Repairs reclassified as capital projects to flatter the operating number.
• Vendor invoices paid outside the maintenance system, so cost by asset understates.
• Deferral dressed as savings. Cutting PM lowers spend this year and raises it later.

Catch all three with a quarterly reconciliation of maintenance-system spend against accounts payable, and by reading cost beside the deferred register.

How do you track deferred maintenance and the facility condition index?

Deferred maintenance is maintenance and repair that was not performed when it should have been and was put off to a future period. That is the federal accounting definition in FASAB SFFAS 42. The facility condition index (FCI) = cost of deficiencies ÷ current replacement value. Lower is better.

The National Park Service rates an FCI under 0.10 as good and above 0.50 as serious for its own assets. That is one agency’s scale, not a rule for private buildings, but it shows how the index is read.

The data comes from condition assessment: periodic physical inspections that record deficiencies and estimate the cost to correct them. FASAB names these, along with life-cycle cost forecasts, as ways to measure deferred maintenance. For an owner, FCI is an annual number. The deferred register behind it is a living document: each item has an estimate, the date it was deferred and who decided.

The index is easy to flatter. Deficiencies never written down do not count. Low repair estimates shrink the top of the fraction, and a generous replacement value grows the bottom. The check is an assessment by someone other than the party being measured, repeated periodically and fed into the capital plan and reserve study.

Why do these KPIs require a CMMS with timestamps?

Every KPI above is a calculation over event records: when the request arrived, when someone got there, which asset, what type of work, how many hours and dollars. A spreadsheet or inbox cannot produce system-set timestamps or a history of who changed a due date. A summary PDF cannot be audited at all.

Record field	Feeds	Protects against
Request received time	Response, resolution, SLA	Typed times
Priority, with change history	SLA, emergency share	Downgrades
Work type	Planned %, emergency share	Reclassification
Asset ID and location	MTBF, repeats, cost by asset	Ticket splitting
Due date, with change history	PM compliance, backlog	Moved due dates
On-site and completion times	Response, resolution, MTTR	Remote status changes
Labor hours and cost	Planned %, cost KPIs	Off-system spend
Photos and readings	PM and repair verification	Pencil-whipping
Parent work order link	First-time fix, PM findings	Splitting

The owner also needs the raw records, not only the report. Our article on what a CMMS is covers the basics. Our piece on what “software included” really means explains why a monthly PDF from the provider’s own system is “unverifiable, because you are reading a summary authored by the party being summarized.”

Levaru’s in-house CMMS platform is included with service and built around the owner’s view: photo-documented work orders, QR tags that tie major equipment to its service record, PM schedules that generate work orders, and client portal access. Owners with portal access can export the asset register and work-order history for their buildings themselves, so an owner can compute these KPIs from the records instead of taking a summary on trust.

What should a one-page monthly maintenance KPI report look like?

One table with the same rows every month. Show the numerator and denominator beside each percentage, the trend against the trailing 12 months, and a target only where the contract defines one. Add an exceptions list and attach the raw export. A template to adapt:

KPI	This month	Last 12 mo.	Contract target	Count (n / N)
PM compliance, all	__%	trend	__	__ / __
PM compliance, code-required	__%	trend	All	__ / __
PM findings rate	__%	trend	—	__ / __
Planned maintenance %	__%	trend	__	hours
Emergency WO share	__%	trend	—	__ / __
Response, by priority	median / slowest	trend	__	__
Resolution, by priority	median / slowest	trend	__	__
SLA attainment, by priority	__%	trend	__	__ / __
Past-due WOs, by age band	__	trend	—	oldest: __ days
Reopen rate	__%	trend	__	__ / __
Repeat-failure assets	__	trend	—	list
Maintenance cost / sq ft, YTD	$__	trend	budget	scope note
Deferred items added / closed	__ / __	trend	—	$ estimate
Tenant complaints, repeats	__ / __	trend	—	__

Below the table, list the exceptions:

1. Every missed code-required inspection or test, with the reason and the new date.
2. Every emergency work order, with a one-line cause.
3. Past-due work orders by priority, oldest first.
4. The top assets by trailing-12-month repair spend.
5. New deferred items, with estimates.
6. PM schedules added or removed, and due dates changed.
7. Cancelled work orders, with reasons.

Then read the pairs. High PM compliance with near-zero findings calls for a records sample. Faster response with rising reopens suggests splitting. Falling cost with a growing deferred list is deferral, not savings.

How do you put KPIs into a service contract without perverse incentives?

Contract on a few well-defined measures, computed from records you can access. Pair each one with a counter-measure, and tie money to verified trends rather than a single month.

Federal practice is a useful model. Under FAR 37.601, performance-based service contracts include measurable performance standards, the method of assessing performance against them, and incentives where appropriate, matched to those standards. Agencies pair this with a quality assurance surveillance plan. An owner can borrow the idea as a one-page verification exhibit.

1. Define every KPI in an exhibit: formula, clock start and stop, priority definitions, grace windows, pause rules and exclusions.
2. Name the source of truth: the system records, with owner access and raw export. The provider’s summary reports the measurement. It is not the measurement.
3. Reserve audit rights: a monthly sample of closed work orders, and site walks that check tagged equipment against its records.
4. Pair the metrics: PM compliance with findings and evidence, response with reopens, backlog with cancellations, cost with the deferred register.
5. Never pay for volume. Paying for work orders closed or PMs completed rewards splitting and over-scheduling.
6. Never punish honest findings. A provider that records deficiencies gives you an accurate deferred register. Penalize what it hides, not what it reports.
7. Use trends and cure periods. A missed target triggers a corrective plan before any fee consequence. Code-required inspections are the exception; a miss there needs action now.
8. Review the definitions every year. PNNL recommends reviewing KPI definitions and calculations at least annually.

Our guide to writing a facility management RFP shows where these exhibits fit in a solicitation.

What changes for buildings in DC, Virginia and Maryland?

The KPIs are the same across the region. What changes is who outside the building sees its performance, and which duties the law places on the owner. This is general information, not legal advice. Confirm current rules with each agency.

• District of Columbia. Private buildings of 10,000 square feet or more submit annual energy and water benchmarking to DOEE by May 1, for public disclosure. Neglected HVAC maintenance shows up as energy use, in public, and feeds the District’s building energy performance standards.
• Maryland. Under the state’s Building Energy Performance Standards, commercial and multifamily buildings of 35,000 square feet or more submit an annual benchmarking report, with performance standards arriving in the 2030s. Montgomery County also covers most buildings of 25,000 gross square feet and up, with reports due to its Department of Environmental Protection by June 1 each year.
• Virginia. The maintenance part of the Uniform Statewide Building Code says the owner shall provide and maintain buildings, systems and equipment in compliance unless the tenant is responsible, and localities may choose to enforce it. Map the lease’s responsibility split onto KPI scope, and list tenant-maintained equipment separately so it does not silently drop out.

Two report lines follow. Add benchmarking submission status for covered buildings; our BEPS deadlines guide covers the energy side. And track code-required inspections as their own all-or-nothing line, using a building compliance calendar, because inspection regimes differ by jurisdiction.

Frequently asked questions

What are the most important facility maintenance KPIs?

For an owner, the core set is PM compliance with a findings check, planned maintenance percentage, emergency work order share, response and resolution time by priority, backlog aging, reopen rate, repeat failures per asset, cost per square foot, cost by asset, and the deferred maintenance register. Together they show whether scheduled work happens, whether repairs hold, what upkeep costs and whether risk is growing.

What is a good PM compliance rate?

No single number fits every building. SMRP publishes best-in-class targets for members, and vendor blogs quote figures from 80 to over 95 percent, often without a source. What matters more is that the number is real. Every code-required inspection should be complete, and each closed PM should carry readings, photos and a believable duration. A trend that improves on verified records is worth more than a high figure nobody has checked.

What is the difference between response time and resolution time?

Response time measures how long it takes for a qualified person to arrive, or take the first contracted action, after a request is received. Resolution time measures how long until the problem is fixed and verified. A provider can respond quickly and still resolve slowly, for example when parts are not stocked. Report both by priority, define exactly when each clock starts and stops, and show the slowest cases as well as the median.

What is a facility condition index?

The facility condition index is the estimated cost of a building’s deficiencies, its deferred maintenance, divided by its current replacement value. Lower means better condition. The National Park Service, for example, rates its own assets under 0.10 as good. The index depends entirely on a thorough condition assessment and honest repair estimates, so it is best calculated annually from an assessment by someone other than the party whose work it measures.

How often should maintenance KPIs be reviewed?

Review operational KPIs monthly: PM compliance, response, resolution, backlog and reopens. Review reliability and cost KPIs quarterly, because single months are noisy. Calculate the facility condition index annually from a condition assessment. Review the KPI definitions themselves at least once a year, as PNNL’s federal O&M guidance recommends, because a stale definition quietly measures the wrong thing.

Should KPIs be tied to payment in a facility management contract?

They can be, carefully. Tie money to a small number of clearly defined measures computed from records the owner can access and export. Pair each one with a counter-measure, and use trends and cure periods rather than single-month penalties. Avoid paying for volume, such as tickets closed. Never penalize a provider for reporting deficiencies, because that teaches it to stop reporting them.

Can I track maintenance KPIs without a CMMS?

You can approximate a few with spreadsheets, but not reliably. Most maintenance KPIs depend on system-set timestamps, asset links, work-type classification and change histories, which hand-kept records cannot provide. If a provider runs the system, the minimum is owner access to the underlying records and a raw export you can recompute from. Without it, you are trusting the report rather than measuring the building.