Maintenance KPIs: Definitions, Formulas, and Data Checks for Plant Teams
Define maintenance KPIs before setting targets. Learn how to choose a numerator, denominator, period, exclusions, and data owner for common plant maintenance measures, with a free KPI definition worksheet.
A KPI starts with a decision, not a dashboard
A maintenance KPI should help a team decide what to investigate or change. Before choosing a chart, name the decision: for example, whether overdue PM work needs a scope review, whether the weekly schedule is realistic, or whether a recurring failure warrants engineering attention. A metric that has no owner or follow-up action is usually just a number on a screen.
The same metric name can hide different rules. Define the event, asset population, time period, data source, calculation, exclusions, and accountable owner. Preserve the definition alongside the result so a later report or software change does not quietly alter what is being measured.
Preventive maintenance compliance
One practical definition is eligible PM work orders completed within the agreed completion window divided by eligible PM work orders due in the reporting period, multiplied by 100. For example, if 39 of 46 eligible PM work orders were completed inside the defined window, the result is 84.8% for that example period. That arithmetic is not a recommended target.
Document how the measure treats cancelled tasks, approved deferrals, work completed late, partially completed work, and tasks generated by meter readings. Keep a separate overdue list so exclusions do not hide work that still needs a decision. The owner should check whether the source dates and closure records represent actual completion rather than data-entry timing.
Schedule compliance
Schedule compliance may be calculated by work-order count or by estimated craft hours. The count method gives a short inspection and a multi-day repair equal weight; an hours method better reflects planned labor but depends on reasonable estimates and accurate time records. Choose the method for the decision being made and keep it consistent across reporting periods.
Define when the weekly schedule becomes fixed, how approved break-ins are recorded, what happens when work moves within the week, and whether partial completion counts. Read the result with break-ins, deferrals, unavailable equipment, material shortages, and actual crew capacity. SMRP publishes separate schedule-compliance measures for hours and work orders, so label which definition your plant uses.
Planned work percentage
A common operational measure compares labor hours on planned work with total maintenance labor hours for the same population and period. State how your site classifies work as planned or unplanned, and decide how emergency work, contractor hours, routine operator tasks, and indirect labor are handled. Do not compare percentages built from different work categories as though they were equivalent.
A rising planned-work share does not by itself prove that asset reliability improved. Pair it with measures of repeat failures, emergency work, PM effectiveness, and operational impact. Investigate classification changes, missing time entries, or changes in the work mix before attributing movement to a process intervention.
Mean time between failures and repair duration
For a defined asset group, a practical MTBF calculation is operating time divided by the number of qualifying failures in the same period. Define the failure event, whether recurrent trips count separately, how standby assets are treated, and whether operating hours are measured or estimated. Do not compare unlike equipment duties or small samples without showing those limits.
“MTTR” is used for different repair-time concepts. If the useful question is hands-on repair duration, calculate active repair hours per completed repair and keep waiting for parts, access, or operations release separate. If the business question is time until service is restored, measure elapsed restoration time as a distinct interval. Name the start and stop events so teams do not compare unlike values.
Maintenance backlog and data-quality checks
Backlog can mean open work orders, estimated labor hours, planned work, or work ready to schedule. State the included statuses and show hours separately from work-order count. If expressed in weeks, use the same craft population for planned backlog hours and net weekly capacity; do not combine raw notifications with executable work and call the result capacity coverage.
Before publishing a KPI, sample records back to their source work orders. Check missing dates, duplicate assets, cancelled or reopened jobs, changing priority codes, zero-hour estimates, and late closure entry. Display the reporting window, numerator, denominator, exclusions, source, and definition version with each recurring report.
Set targets only after the measure is stable
Use a baseline that covers a representative operating period and note seasonality, production duty, asset changes, outages, staffing, and data-quality limits. Agree a locally meaningful target with the people who own the process, and record the business reason and review date. A competitor’s “best in class” percentage is not automatically suitable for your plant.
The downloadable KPI definition worksheet helps document the calculation, data source, inclusion rules, exclusions, owner, and action when a result changes. Complete it with maintenance, operations, and finance or reliability stakeholders as appropriate, then validate the definition against a sample before publishing the dashboard.
Further reading
These references offer definitions and examples; confirm the method and data rules that apply to your organization.