The Measures Worth Tracking

Maintenance produces a great deal of measurable activity and very little of it indicates whether anything is improving.

The measures that matter

Emergency versus scheduled proportion. The share of work orders that are unplanned. A falling emergency share is the clearest evidence a preventive programme is working, and a rising one is the earliest warning it is being eroded.

Total cost per site per year, including consequential loss where estimated. Comparable across sites, adjusted for age and volume.

Cost per asset, which is what identifies the units that should be replaced.

Repeat visit rate. Work orders on the same asset for the same fault within a defined window. High rates indicate diagnostic quality problems or unaddressed causes.

Preventive completion rate, by site and by vendor. Scheduled tasks completed on time. This is the measure that reveals a programme existing only on paper.

Open work order age. Anything past its priority commitment. The most useful operational view.

Response time achieved against agreed, by vendor.

Downtime hours by asset, where the equipment matters to service.

Eight measures. That is enough.

The measures that mislead

Total work order count. Falls when the programme improves and falls when reporting collapses. Indistinguishable without other measures.

Average cost per work order. Falls when you do more cheap routine work, which is good, and falls when you defer expensive necessary work, which is not.

Technician utilisation. Relevant to a service contractor, irrelevant to an operator who contracts the work.

Mean time between failures, across a mixed estate of different assets and ages. Statistically meaningless at restaurant volumes.

First-time fix rate as a headline, which is worth watching per vendor and is easily gamed by closing and reopening work orders.

Anything averaged across sites without adjusting for age and volume, which compares a new kitchen with a twenty-year-old one and concludes the second is badly managed.

Reading them together

Single measures are ambiguous; combinations are not.

Emergency share falling and total cost falling — the programme is working.

Emergency share falling and total cost rising — you are doing more preventive work than it is saving. Check whether the schedule is over-specified.

Emergency share rising and cost flat — preventive work is being skipped or done badly.

Total work orders falling and emergency share rising — reporting has collapsed and only failures are being reported. This is the pattern to watch for after a rollout loses momentum.

Repeat rate rising at one vendor — a diagnostic problem.

One site consistently above the group on cost — old equipment, a weak vendor, or a management problem. The work order history distinguishes them.

What to report and to whom

To the maintenance owner, weekly: open work orders by age, anything past commitment, emergencies this week.

To operations leadership, monthly: emergency share, cost per site, completion rate, outlier sites.

To finance, quarterly: total cost against budget, replacement backlog, consequential loss.

To the board or ownership, annually: trend on emergency share and total cost, capital replacement plan, compliance position.

Keep each report to one page. A maintenance report nobody reads is a report that does not protect the budget.

The baseline problem

The value of a preventive programme is failures that did not happen, which cannot be measured without knowing the rate before.

Measure the year before you change anything. Emergency call-outs, product loss, downtime, total cost.

Without a baseline, the programme's value is unprovable and it will be cut in the second year when the failures have stopped and the spend is visible.

This is the single most common reason good maintenance programmes are abandoned, and it is prevented by a quarter of measurement before the programme starts.

Data quality first

Every measure above depends on work orders that identify the asset and describe the fault.

Before building a dashboard, check that the last fifty work orders can answer what was wrong and what was done. If they cannot, the metrics will be precise and meaningless, and the effort belongs in close-out discipline rather than in reporting.

The one-page monthly report

What the maintenance owner sends to operations, every month, on one page.

Top line: emergency share this month and the twelve-month trend. One number and one line chart.

Open work orders past commitment, by site. A short list, and empty is the goal.

Cost by site, this month and year to date, with the three outliers named.

Preventive completion rate, by site, showing anything below target.

Compliance exceptions. Anything expired or expiring within sixty days.

Significant events. Two or three lines on anything that stopped service or cost more than a threshold.

One action requested, if any.

Nothing else. A monthly maintenance report that runs to six pages is a report that is filed unread, and the budget is defended by the report being read.