Maintenance KPIs: The Metrics That Actually Matter

Maintenance KPIs: The Metrics That Actually Matter

Walk into most maintenance departments and you'll find one of two things. Either nobody measures anything and every decision is a gut call, or somebody built a dashboard with forty metrics on it that nobody looks at. Both are the same problem wearing different clothes: no clear line between the number and the decision it's supposed to drive.

Maintenance KPIs are only worth tracking if they change what you do. A metric that goes up or down and prompts no action is decoration. This guide covers the maintenance KPIs that actually earn their place — what they tell you, how to read them, and which ones to start with if you're building measurement from scratch. It's written from the shop floor, not the vendor brochure.

We treat the three big reliability metrics — MTBF, MTTR, and OEE — in dedicated guides, and link to them below. This is the hub: how the whole set fits together and where to begin.

Leading vs Lagging: The Distinction That Organises Everything

Before the list, one idea that makes the rest make sense. Maintenance KPIs split into two kinds.

Lagging indicators tell you what already happened. Downtime hours, mean time between failures, cost per asset — these are the scoreboard. They're honest and they matter, but by the time they move, the events that caused them are weeks gone. You can't manage a failure that's already occurred.

Leading indicators tell you what's coming. Schedule compliance, planned maintenance percentage, backlog size — these describe the health of your process right now. They move before the failures do. If schedule compliance drops this month, breakdowns rise next quarter. Leading indicators are where you actually steer.

The mistake most teams make is drowning in lagging metrics while ignoring leading ones. You end up with a beautifully documented history of problems you did nothing to prevent. Track both, but if you're time-poor, put your attention on the leading indicators — they're the ones you can still do something about.

The Maintenance KPIs Worth Tracking

Here's the shortlist. Not fifteen metrics — the ones that consistently drive decisions in real operations.

Two things about that table. First, the targets are directional, not gospel — a haulage fleet and a food line have different economics, and "world-class" numbers come from operations that spent years earning them. Chase the trend, not the benchmark. Second, you do not need all eight on day one. More on where to start below.

Schedule Compliance (PM Compliance)

The single most useful leading indicator, and the one to start with. Schedule compliance is the percentage of planned maintenance completed on time:

Schedule compliance = (PMs completed on schedule ÷ PMs due) × 100

If you scheduled 100 preventive tasks this month and finished 78 in their window, you're at 78%. Aim for 90% or better. Below that, your PM programme is theatre — the schedule exists on paper but the work isn't happening, and the failures it was meant to prevent are still coming.

What makes this metric powerful is that it's honest early. When compliance slips, it's usually because the crew is being pulled onto breakdowns — which means reactive work is crowding out planned work, which guarantees more breakdowns next month. It's the first domino. Watch it weekly.

One caveat: a PM completed late, or ticked off without actually being done, corrupts the number. Compliance is only as trustworthy as the discipline behind closing a work order. If your team can close a task from the car park, the metric lies to you.

Planned Maintenance Percentage

Closely related, and the clearest single measure of whether you're in control or firefighting. It's the share of total maintenance hours spent on planned work versus reactive:

Planned maintenance % = (planned maintenance hours ÷ total maintenance hours) × 100

Under roughly 80% planned, you're in reactive mode — expensive, stressful, and self-perpetuating. Reactive work costs three to five times more per job than the same work planned: overtime, expedited parts, collateral damage, and the production you lose while the line sits. We put real numbers on that in our breakdown of the hidden costs of manufacturing equipment downtime. Moving this number up is the highest-leverage thing most maintenance teams can do.

Maintenance Backlog

Backlog is all identified work that's approved but not yet done, usually expressed in crew-weeks — how many weeks it would take your current team to clear it working normally. Counterintuitively, zero backlog is a bad sign: it means you're doing everything the instant it's raised, which means no planning and probably no PM discipline. A healthy backlog sits around two to four weeks. It gives the planner a buffer to sequence work sensibly and buy parts in advance.

The danger is a backlog that grows without limit. That's a leading indicator of a team underwater — and the point at which good work order management practices stop being optional and start being the thing keeping you afloat.

MTBF — Mean Time Between Failures

MTBF is the average running time between failures on a repairable asset. Higher is better; a rising MTBF is the clearest evidence your preventive programme is actually improving reliability rather than just consuming hours. It's a lagging indicator — you're measuring failures that already happened — but tracked over time it tells you whether the trend is going your way. The calculation, the traps, and how to use it per-asset are covered in our dedicated guide to MTBF calculation.

MTTR — Mean Time to Repair

MTTR is the average time to restore an asset after it fails — from the moment it goes down to the moment it's producing again. Lower is better. Where MTBF measures reliability, MTTR measures your response: parts availability, technician access, documentation, spares on the shelf. A team can't always stop a failure, but it can nearly always recover faster. We cover what to include in the clock and how to cut it in the guide to MTTR and reducing mean time to repair.

OEE — Overall Equipment Effectiveness

OEE rolls three things into one score — availability, performance, and quality — and is the standard efficiency metric on a production line. World-class is 85%+, but the absolute number matters less than which of the three components is dragging you down, because that's what tells you where to act. The full method, worked example, and the common ways OEE gets gamed are in our guide to overall equipment effectiveness.

The Metrics That Look Good and Do Nothing

Half of getting KPIs right is refusing to track the wrong ones. A few to be wary of.

• Total work orders closed. Rewards volume, not value. A team can close a hundred trivial jobs and ignore the one failure that shuts the plant. Closing more work orders is not the goal; keeping equipment running is.
• Raw downtime hours with no denominator. Ten hours of downtime means nothing without knowing the planned running time or which asset. Use availability or per-asset figures instead.
• Cost reduction as a standalone target. The cheapest maintenance department is one that does nothing — right up until the day it does. Cost only means something alongside a reliability metric. Cutting spend while MTBF falls isn't a saving, it's a deferred bill.
• Anything the crew can trivially game. If closing a task requires no evidence, compliance and completion numbers drift from reality. The metric has to be anchored to something real — a photo, a reading, a signature, a scan at the asset.

The pattern is the same each time: a number that can improve while the actual situation gets worse. If a KPI can move the wrong way in disguise, it's not measuring what you think.

Where to Start: A Sane First Three

If you're building measurement from nothing, don't stand up an eight-metric dashboard. Nobody will trust it and nobody will maintain it. Start with three:

1. Schedule compliance — is planned work actually getting done? (leading)
2. Planned maintenance percentage — are you in control or firefighting? (leading)
3. MTBF on your critical assets — is reliability improving where it counts? (lagging)

Those three cover the essential question from both ends: are we running a disciplined process (the two leading indicators), and is it producing results (the lagging one)? Get them trustworthy — meaning the underlying work orders are closed honestly — before you add anything. A dashboard people believe with three metrics beats a dashboard people ignore with fifteen.

And they only work if the data underneath is clean. KPIs are downstream of work orders, PM records, and an accurate asset register. If technicians close jobs from memory a week late, or half the assets aren't in the system, no metric you build on top will be worth reading. Measurement is the reward for good record-keeping, not a substitute for it.

Turning KPIs Into Decisions

The point of all this isn't the dashboard. It's the weekly conversation where someone looks at schedule compliance slipping to 82%, asks why, and finds the crew's been firefighting a chronic bad actor for three weeks. The metric didn't fix anything — it pointed at the problem early enough to act. That's the whole job of a maintenance KPI: shorten the distance between something going wrong and someone noticing.

A CMMS earns its keep here by generating the numbers as a by-product of work already being logged, rather than someone rebuilding a spreadsheet every month. When work orders, PM schedules, and the asset register live in one system, schedule compliance and MTBF are a report, not a research project.

If you're trying to get maintenance metrics off spreadsheets and into something that updates itself, book a call and we'll walk through how teams set this up in AssetOS — usually with the first three KPIs live in a week.