Preventive Maintenance Planning 101

Every building already has a maintenance program

The only question is whether yours was designed or whether it accumulated.

Programs that accumulate look the same everywhere. Intervals were inherited from a spreadsheet nobody can source. Half the tasks exist because a vendor suggested them. The equipment that would shut the building down if it failed gets the same attention as a hallway exhaust fan. And when something breaks at 4:45 p.m. on a Friday, the true state of the program becomes obvious.

A designed program is not more paperwork. It is a short chain of decisions: which assets matter, what could go wrong with them, how often somebody should look, what that person should measure, and what you will do with what they find. This guide walks that chain in eight steps, using publicly available codes, standards, and federal data rather than vendor marketing.

Four ways to maintain a building

Preventive maintenance is one strategy, not the only one. Mature programs assign a strategy per asset instead of applying one philosophy to everything.

The goal is not maximum preventive maintenance. It is the least maintenance that keeps consequence-bearing assets from failing unexpectedly.

Step 1 — Build an asset register you can trust

Nothing downstream works without this. Every maintainable asset needs a unique identifier, a location in a consistent hierarchy (site, building, floor, room, system, asset), nameplate data, an install date, an expected service life, and a pointer to its warranty, service contract, drawings, and manuals.

Do not try to inventory everything in month one. Start with the systems whose failure stops the building: heating and cooling plant, electrical distribution and switchgear, elevators, water-based fire protection, domestic water and booster pumps, roofing and drainage, and the building automation system itself.

Step 2 — Rank criticality before you write a single task

Criticality is what converts an equipment list into a plan. Score each asset on consequence and on likelihood of failure, then multiply. Consequence should account for more than repair cost:

• Life safety and code exposure. Does failure create a hazard or an inspection finding?
• Occupant and tenant impact. How many people notice, and how fast?
• Revenue or service interruption. Does failure stop rent-generating or mission activity?
• Replacement lead time. A part on a twenty-week lead time is a different risk than one on a shelf.
• Redundancy. N+1 changes the answer completely.

Sort the results into three tiers. Tier A gets preventive plus condition monitoring plus spares. Tier B gets a preventive schedule. Tier C can often be run to failure on purpose, which is a decision rather than an oversight.

Step 3 — Set intervals from three sources, in order

Frequency is where most programs go wrong, because it is where guessing is easiest to hide. Work through three sources in priority order.

First, obligations. Inspection, testing, and maintenance intervals written into codes, standards, and regulations are not negotiable and should be loaded into your schedule before anything else. These are also the tasks most likely to be tracked in a separate silo and then missed during an audit.

Second, the manufacturer. Warranty conditions and published service intervals set your floor for new and in-warranty equipment. Record the source of each interval in the task record so a future planner can tell obligation from opinion.

Third, your own history. After twelve to eighteen months of clean records you can defend changing an interval. If quarterly inspections on a Tier C unit have found nothing three years running, semiannual is a legitimate decision. If a Tier A pump keeps failing between services, the interval was wrong or the task was.

Step 4 — Write procedures a new technician can execute

A preventive maintenance task is a work instruction, not a title. Each one should carry the ordered steps, tools and parts, required personal protective equipment and energy-isolation points, an estimated duration, and — most importantly — acceptance criteria expressed as measurements.

• Replace "check belt" with a tension value and a wear description.
• Replace "inspect filter" with a differential pressure reading and the change threshold.
• Replace "test operation" with amp draw, supply and return temperatures, or discharge pressure.
• Require photos on anything you might have to prove later, especially life-safety work.
• Give technicians a short, fixed list of failure codes so the data can be grouped.

Step 5 — Do the labor math honestly

Add up the estimated duration of every task at its planned frequency and you have annual preventive maintenance hours. Compare that to real capacity, not gross headcount. A full-time technician has roughly 2,080 gross hours, and after paid leave, training, travel between sites, meetings, and interruptions, planning at 55 to 70 percent productive time is realistic in most portfolios.

Then price it. Federal wage data gives you a defensible starting point: general maintenance and repair workers had median pay of $48,620 per year, or $23.38 per hour, in 2024, across about 1.63 million jobs, with roughly four percent employment growth projected through 2034. Facilities and administrative services managers had median pay of $106,880. Your fully burdened cost per hour is meaningfully higher than any of those figures once benefits, tools, vehicles, uniforms, training, and supervision are loaded in.

Escalate multi-year plans rather than freezing today’s rate. Employment cost index data showed compensation costs for civilian workers up 3.4 percent over the twelve months ending March 2026, with benefits rising slightly faster than wages.

If planned preventive hours exceed available capacity, the program will fail quietly: technicians will triage, the schedule will slip, and by month four nobody will trust the completion report. Fix scope, frequency, sourcing, or headcount before you publish the schedule, not after.

Step 6 — Schedule for the building, not the calendar

• Route the work. Group tasks by location so a technician clears a floor or a mechanical room in one visit instead of crossing the site six times.
• Respect occupancy. Noisy, disruptive, or shutdown-dependent work belongs in low-occupancy windows, and lease language often dictates notice periods.
• Pre-condition for seasons. Cooling readiness before the first heat wave, heating and freeze protection before the first hard freeze, roof and drainage clearing before the wet season.
• Hold reserve capacity. Leave twenty to thirty percent of weekly hours unscheduled for corrective and emergency work, or that work will consume preventive time by force.
• Level the load. Spread annual tasks across the year rather than stacking them in the month the program launched.

Step 7 — Measure seven things, not seventy

Trend all seven against your own history first. Comparing a single month against an industry average tells you almost nothing; comparing this quarter against the same quarter last year tells you a great deal. If you do want external context, use consistently defined public data and state your normalization basis — the approach we walk through in the 2026 maintenance benchmarks report.

Step 8 — Review annually, with life-cycle economics

Once a year, revisit criticality tiers, intervals, task content, and sourcing. Repair-or-replace arguments should be settled with life-cycle cost analysis rather than gut feel: the National Institute of Standards and Technology publishes life-cycle costing methodology and free analysis software for building systems and energy projects, which is a defensible framework for comparing cumulative repair spend, energy penalties, and downtime exposure against the annualized cost of replacement.

Five ways preventive maintenance programs fail

1. Template intervals with no criticality ranking. The program over-maintains trivial assets and under-maintains the ones that will actually hurt.
2. No asset hierarchy. History cannot be attributed to equipment, so nothing can be analyzed or defended at budget time.
3. Tasks closed without readings. Three years later there is no evidence base for changing anything.
4. No capacity check. The schedule was arithmetic fiction from day one.
5. Compliance work tracked separately. Code-driven inspection and testing lives in a binder or a vendor portal, so gaps surface during an inspection instead of on a Tuesday.

A ninety-day start

Days 1–30: establish ground truth

• Inventory Tier A candidates with nameplate data and photos.
• Pull every code-required inspection and testing obligation into one calendar.
• Baseline your reactive labor share and current backlog, however ugly the numbers are.

Days 31–60: design

• Score criticality and assign a strategy per asset.
• Write measurable procedures for your top twenty assets.
• Set intervals, record their source, run the capacity check, and publish the schedule.

Days 61–90: run and tune

• Execute weekly, review completion and backlog weekly, not monthly.
• Audit ten closed tasks a week for real readings.
• Adjust intervals where evidence supports it and report a one-page scorecard to leadership.