How I run preventive maintenance instead of emergencies

Most maintenance money gets spent reacting. A supply line lets go on a Saturday, a furnace quits in the first cold week of the year, a water heater fails into the unit below it, and the price of the repair gets set by how urgent it is rather than by what the work is worth. A preventive calendar is the only thing I have found that shifts that ratio.

Why the unplanned version costs so much more

The repair itself is rarely the expensive part. The expensive part is everything attached to it: an after-hours call-out rate, whichever vendor happens to answer rather than the one on the bench, and the secondary damage already done by the time anyone arrives. Water is the clearest example, because it does not stay in one unit.

The Insurance Information Institute puts water damage and freezing at 27.6 percent of homeowners insurance claims as of 2022, second only to wind and hail by frequency, with an average claim severity of $13,954 across the 2018 to 2022 period. About one in 60 insured homes files a water damage or freezing claim. Those are homeowner figures rather than multifamily ones, but the braided supply line under a rental sink is the same part as the one under an owner-occupied sink, and in a rental the claim usually arrives with a displaced resident attached to it.

Planned work lets you pick the vendor, the rate, and the day. Unplanned work picks all three for you, and it never picks well.

What actually goes on the calendar

I keep the recurring list short enough that it gets done. A calendar with 40 line items per unit becomes decoration by the second quarter. These are the items I have seen return the most for the least amount of technician time:

• HVAC filters: on a fixed interval, with the interval set by system type and filter depth rather than by habit.
• HVAC service: once before cooling season and once before heating season, including condensate lines.
• Water heaters: age on the asset register, relief valve checked, and a replacement plan for anything past its expected life.
• Dryer vents: lint trap housing and the full duct run to the exterior termination, not just the screen.
• Smoke and carbon monoxide alarms: tested, batteries handled, and manufacture dates logged.
• Plumbing quick look: under every sink, behind the washer, and at the angle stops, for the slow drip that becomes a claim.
• Gutters and drainage: cleared, and the grading around downspouts checked so water leaves the building.
• Exterior envelope: caulking, flashing, and any penetration that has opened since the last look.
• Shutoffs: main and unit-level water shutoffs located, labeled, and confirmed to actually turn.

That last item looks trivial and is not. The middle of a burst-line emergency is the wrong moment to discover that nobody knows where the shutoff is or that it seized years ago.

Seasonal work has to be timed to weather, not to the month

Seasonal items get skipped because they have no obvious trigger until the day they are already late. The fix is to schedule backwards from the weather a portfolio actually gets. A northern property does its heating check in September, not in November when the first cold call has already come in. A Gulf coast property clears condensate lines in April and worries very little about hose bibs.

Winterization is the highest-stakes version of this in cold climates: hose bibs drained and covered, unheated crawl spaces insulated, minimum heat maintained in vacant units, and someone physically confirming the vacant thermostat is set and the power is still on. Vacant units are where freeze losses concentrate, because nobody is there to notice the temperature dropping.

The rest is four passes a year, each anchored to a date that lands before the season rather than during it.

Filters and alarms are the cheapest items with the best ratio

Two tasks on the list cost almost nothing and carry more weight than their price suggests.

The first is the air filter. The Department of Energy's Energy Saver guidance is that routinely replacing or cleaning air filters can lower an air conditioner's energy consumption by 5 to 15 percent. Where the resident pays the utility bill, that saving is theirs, which some owners treat as a reason to deprioritize it. The part that matters to the owner is different: a clogged filter restricts airflow and makes the system work harder for longer, and the compressor being protected belongs to the owner.

The second is the dryer vent. The U.S. Fire Administration reports that failure to clean was the leading factor contributing to the ignition of clothes dryer fires in homes from 2018 to 2020, at 31 percent, drawing on National Fire Incident Reporting System data. A duct run that has not been cleared in five years is a known, documented, and entirely preventable fire cause sitting inside the unit.

On alarms, the item most programs miss is age. The U.S. Fire Administration is direct about it: smoke alarms need replacing 10 years from the manufacture date, and they belong in every bedroom, outside each separate sleeping area, and on every level of a home including the basement. A test tells you the unit responds today and nothing about a sensor that is eleven years old. So the manufacture date goes on the asset record alongside the test date, which turns alarm replacement into a budgeted line rather than an inspection finding.

Access is the real constraint

The work is not hard. Getting into 40 occupied units is hard, and it is where preventive programs quietly die.

Entry notice requirements are set by state and sometimes local law, commonly written notice a day or two ahead, and they need checking against the actual statute rather than assuming. Past that legal minimum, what makes access work is treating it as a scheduling problem rather than a permission problem. I batch by building and by day, publish the window well ahead of the notice deadline, offer a real second date, and say plainly what the technician will do and how long it takes.

Then I make the visit count. A technician already standing in the kitchen can change the filter, test the alarms, look under both sinks, check behind the washer, and eyeball the water heater in one trip. That combined visit is the cheapest inspection in the whole operation. More on how I run maintenance intake and triage →

Spreading the load so no single week is impossible

The mistake I made early on was scheduling by task: every filter in the portfolio in one week, every dryer vent the next. That produces a few weeks a year nobody can staff and ten months where nothing recurring happens. Now I split the portfolio into groups by building, and each group carries its own rotation. The weather-driven items keep fixed anchor dates because they have to, and everything else distributes across the year. Vendor capacity holds, and a missed week becomes a small catch-up rather than a lost quarter.

Every task gets a named owner, a date window, and a definition of done that includes evidence. A work order closed with no photo and no note is an unverified claim rather than a completed task. Photos go against the unit and the asset, so next year's technician can see what last year's found. More on how I qualify and score the vendor bench →

The numbers that keep the program funded

Preventive maintenance is the first budget line questioned, because its whole purpose is producing an absence of events. So I track it in a way that shows the absence:

• Completion rate per cycle, because a program running at 60 percent completion is not a program.
• After-hours and emergency work orders per unit per year, tracked as a trend rather than a single number. This is the metric that should fall.
• Repeat calls on the same asset, which usually says replace rather than repair.
• Asset age register for water heaters, HVAC systems, roofs, and major appliances, so capital replacement gets planned into a budget year instead of arriving as a surprise.
• Planned cost per unit against unplanned cost per unit, which is the comparison an owner actually responds to.

Those five numbers go into the monthly owner report with a sentence on what moved and why. An owner who can see emergency calls trending down against a stable preventive spend stops treating that line as discretionary. More on how I build the monthly owner report →

Where AI helps and where it does not

AI runs the scheduling, the memory, and the chasing. It builds the annual calendar from the asset register and the unit list, generates work orders on their windows, drafts entry notices with the right dates, chases anything open past its window, and flags assets nearing the end of their expected life so replacement enters the budget early.

What it does not do is decide. Whether the stain under a sink is old and dry or currently active, whether a twelve-year-old water heater on a top floor should be replaced now rather than next year, whether a resident's hesitation about entry is worth a phone call. Those are judgment calls made by someone who has seen the unit. The calendar can be automated. The reading of what the technician found cannot. More on the AI stack behind this →