Maintenance KPIs

Any maintenance metric can be improved by doing the wrong thing, and a function measured without that in view will eventually be managed into looking good rather than into being good. This is not cynicism about maintenance teams; it is a structural property of measurement. PM compliance rises if the schedule is made less demanding, or if jobs are closed without the asset being visited. Backlog falls if work is canceled rather than completed. Mean time to repair shortens if only the straightforward failures get logged and the awkward ones are handled informally. Planned work ratio improves if reactive jobs are retrospectively reclassified. Each of these is a rational local response to being measured, and none of them involves anybody behaving dishonestly; they are what happens when a number becomes a target and nobody has written down what a good version of that number would also be consistent with. That is the whole function of section three. It pairs every metric with the uncomfortable alternative reading and, more usefully, with a specific check: audit a sample of completed preventive tasks against the asset, look at what proportion of the backlog is being closed as canceled, compare logged repair times against the jobs people actually remember. The second structural decision worth protecting is reporting backlog in crew weeks. Backlog expressed as a job count is genuinely unreadable, because two hundred jobs might be two weeks of work or a year of it, and the figure moves when job sizes change without the workload changing at all. Expressed as crew weeks it becomes immediately legible and, set beside the rate at which new work arrives, it answers the only question anybody actually has: is this clearing, holding, or quietly growing.

This is a competent maintenance function having an ordinary period, which is exactly the condition under which a scorecard is most likely to be misread. Ninety one percent PM compliance against a target of ninety five is amber and worsening, and the comment against it on the scorecard is two technicians on long-term absence, which is the right way to use that column: the number explains itself and nobody needs to invent a theory. Availability at ninety seven point two and emergency work at nine percent are both healthy. The two lines worth a question are wrench time and backlog. Thirty one percent wrench time means roughly two thirds of available hours go somewhere other than tools-on-task, and while that figure is typical it is typical for reasons that are fixable: travel between sites, waiting for parts, waiting for access, and the paperwork around a job rather than the job. It is the single largest capacity opportunity on this sheet and it does not involve anybody working harder. Backlog at four point one crew weeks is only meaningful alongside the rate at which new work arrives, which is why the headline section asks for crew hours available rather than backlog alone; four weeks of work with three weeks of new work arriving every month is a backlog that will never clear, and four weeks with two is one that will. Note what is not here: nothing about the building itself, its running cost per unit of floor, or how the people in it feel about it. That is a facility management KPI scorecard, which is a different document for a different reader. The stoppage data behind time between failures and time to repair is captured on the downtime log, and the completion record behind PM compliance is the preventive maintenance log. For the question of which of these metrics are worth tracking in the first place, there is a guide on how to choose which maintenance metrics to track.

Nothing in the right-hand column requires anybody to act dishonestly, which is precisely why it is so common. Each of those readings is the predictable consequence of measuring something and attaching consequences to it, and the people closest to the number are usually the last to notice the drift because every individual decision was locally reasonable. A planner who removes an unachievable task from the schedule has improved the schedule. A supervisor who closes a job that was attended but not fully completed has recorded what substantially happened. A technician who fixes something quickly and informally, without raising a work order, has solved the problem faster than the process would have. Repeat those decisions for a year and the scorecard is green and the asset base is deteriorating. The defense is not tighter policing, which produces its own distortions; it is pairing each metric with a specific check that looks at the thing rather than at the record. Audit a sample of completed preventive tasks against the asset itself. Look at what share of the closed backlog was closed as canceled. Compare logged repair durations against the jobs people actually remember from the period. Three checks of that kind are worth more than another month of compliance reporting, and they take a morning. The other reason this page exists separately from a general facilities scorecard is vocabulary. Schedule attainment, PM yield, wrench time and cost as a percentage of replacement asset value are reliability-engineering terms with precise meanings, and softening them into general facilities language loses the precision that makes them worth measuring. The broader scorecard that measures the building rather than the maintenance function is the facility management KPI scorecard, and it is deliberately a different document for a different reader.