Buildings that stay in good condition for decades have one thing in common. Someone writes down what got fixed, when it was fixed, and why it broke. Record keeping is the least interesting part of commercial property maintenance, and it decides more than the budget does.
In August the Government Accountability Office published an audit of how the Department of Defense keeps up its buildings. The Pentagon owns more than 736,000 buildings worth an estimated $2.6 trillion. Its repair backlog hit roughly $285 billion in fiscal year 2025. Some of that is a money problem, since the Pentagon aims to fund 90 percent of its maintenance needs and funds about 80 percent.
GAO found a second problem that has nothing to do with funding. The work order data across those bases is, in GAO's words, generally unreliable for determining the overall effectiveness of maintenance performed. So the largest property owner on earth can't say whether its own repairs are working. Well run buildings keep a maintenance record accurate enough to plan from. Without one, every budget and staffing decision is a guess.
Each service in the report sets repair deadlines by category. Emergency work gets 24 hours, urgent work gets five to seven days depending on the branch, and routine work gets 30 days. Its officials told GAO that strict deadline rules can lead to false reporting. Maintenance offices may close work orders early to look like they hit the target.
The trap has nothing to do with the military. A close time target puts pressure on closing, and closing is the one step a tech fully controls. The repair itself waits on parts, on access to a suite, and on a second trade showing up. So the number turns green while the drywall stays open, and the next person to pull up that asset learns nothing from it. Speed is the one target a team can hit without touching the building.
Most maintenance management software rollouts walk straight into that trap. Reports get better before buildings do, because a report measures whatever moves easiest. Average close time drops and the monthly deck looks sharper. Nobody in the room can say whether the chiller is healthier than it was a year ago.
GAO talked to 37 customer groups across eight bases, and 22 of them gave the same answer. Emergency work usually gets done on time, and everything else struggles. That split shows up in office and retail portfolios for the same reason it shows up on a base. Emergencies get escalated by whoever is affected. Routine work has nobody pushing it, so it slips.
Deferred routine work gets expensive. An official at Minot Air Force Base told GAO about a roof leak that still hadn't been repaired. It was reported roughly three years before the visit. At Naval Station Norfolk, GAO saw dead fire suppression and alarm systems in barracks and aircraft hangars in January 2025. The barracks fix was scheduled for February 2026. People stood fire watches for more than a year. Neither of those started as an emergency, and both of them ended as one.
A roof leak is a 30 day routine item right up until the deck rots, and then it becomes a capital project. Building operations teams lose ground exactly this way, one deferred routine item at a time, over years rather than weeks. The work order log is the only place that loss shows up while it's still cheap to reverse.
At intake, every request needs an asset tied to it, so history piles up against a chiller or a rooftop unit instead of a floor number. A complaint logged as "it's hot on four" and pointed at RTU-3 becomes the third entry in a pattern. Capital requests get approved on patterns, and a floor number never produces one.
At close, four fields carry nearly all the value. What was wrong, what was done, which parts went in, and how long it took. A line reading "replaced condenser fan motor, 2.5 hours, second failure this year" is a sentence of next year's capital plan. The technician is the only person who can write it. A one word close erases the visit.
The Marine Corps audits its own work orders. GAO found its guidance tells maintenance offices to inspect a set share of completed work orders each week and judge the quality of the work performed. Sampling closed tickets is dull work, and nothing else stops a closing standard from decaying. Five tickets a week is enough to change what people write.
Once the record is clean, repeat visits outrank every other number in property maintenance. Count how many times a single asset generates a work order in twelve months. Response time describes how your team is doing. Repeat visits describe how your building is doing, and that is the one that turns into a budget line. A rooftop unit on its fourth call of the year is near the end of its life, and no inspection report will say so this clearly.
Grading on repeat visits also drains the reward for closing early. A job closed before it's finished comes back as a callback, and that callback lands on the same asset within weeks. So the metric and the outcome point the same way, and honest reporting stops costing the team anything.
The staffing outlook makes an honest record more valuable every year. The Bureau of Labor Statistics counted about 1.6 million general maintenance and repair workers in 2025, with roughly 148,700 openings projected each year through 2035. Most of those openings come from replacing people who retire or leave the trade. BLS puts real growth at just 4 percent across the whole decade. You're refilling the same seats, over and over.
Every exit carries building knowledge out the door. One engineer knows which valve sticks in February. Another knows which tenant calls about heat every October. Both are holding a maintenance record in their head. Written into the system, that knowledge outlives the person and trains whoever comes next. Left unwritten, it leaves with them, and your next hire relearns it at your expense.
Good building maintenance software makes the honest close the easy one. If tagging an asset takes three taps and a photo, techs tag the asset. If it takes a laptop back at the office two hours later, the details stay in a pocket and the record thins out. Most of the gap between a good maintenance program and a poor one is how hard the system makes it to file the record.
That's why so much facility maintenance software lets buyers down. The reporting looks good and the intake is slow, so the field notes never get entered. Building operations software earns its keep when a tenant request, a preventive task, and a vendor visit all land on the same asset, and each one requires the same fields.
Cove was built around that idea. When Cove rolled its building operating system out across the EQ Office portfolio, it replaced three separate legacy systems. Work orders, preventive maintenance, and inspections moved onto a single record per asset. EQ is a Blackstone company whose holdings have included Willis Tower, and a forty building portfolio and a single building need the same thing from a record.
Any team can start on this in a week, without buying anything. Tie every request to an asset, and require four fields before a job can close. Read five closed tickets every Friday, and grade the team on repeat visits rather than speed. Do that for ten years and you'll have what the best run buildings have, which is a record that tells you what to fix before it fails.
Repeat work orders per asset over twelve months. Response time measures your team, while repeat visits measure your building, and the second one is what turns into a capital request. It also removes the reward for closing a job early, because an early close comes back as a callback on the same equipment. Track it monthly by asset and the pattern shows up long before the failure does.
Because closing a ticket is the one step a technician controls completely. Air Force officials told the Government Accountability Office that strict deadline rules can produce false reporting. Maintenance offices may close work orders early to look like they met the target. A close time goal measures paperwork speed. Sampling a share of closed work orders each week, the way Marine Corps guidance requires, is the cheapest correction available.
Four things: what was wrong, what was done, which parts were used, and how long it took. All of it ties to a specific asset rather than to a floor or a suite. That combination lets you count repeat visits, catch warranty claims, and price a replacement with real history behind it. Anything shorter costs the same labor hour and leaves you nothing.
The Bureau of Labor Statistics projects about 148,700 openings a year for general maintenance and repair workers through 2035. Retirements and people leaving the trade drive most of that. Every exit takes unwritten building knowledge with it. A full asset history means your next hire starts with the equipment's whole record. The alternative is rediscovering it during an outage at two in the morning.