Asset Maintenance KPIs: Downtime, Reliability, and Cost

Asset maintenance KPIs measure equipment downtime, failure frequency, repair cost, and unfinished maintenance work. Read them together: low repair spending can reflect reliable equipment, deferred work, or missing cost records. A useful comparison identifies which physical assets need investigation and shows the records behind that conclusion.

Choose measures with clear time boundaries

Start with availability, mean time between failures, repair time, corrective cost, and backlog. Each needs a definition, period, and owner, as the KPI glossary explains.

Pacific Northwest National Laboratory's maintenance guidance distinguishes reactive, preventive, and predictive maintenance approaches. Preserve those categories in your export so a preventive inspection does not become an equipment failure in the report.

Observed availability = Operating hours / Scheduled production hours × 100

MTBF = Operating hours / Recorded failures

Average active repair time = Active elapsed repair hours / Repairs completed

Corrective cost per repair = Corrective repair cost / Repairs completed

Backlog weeks = Remaining labor hours / Available maintenance labor hours per week

Here, observed availability includes planned and unplanned downtime within scheduled production time. Unscheduled weekends sit outside that window. Retain this definition each period.

Active repair time is a clock duration; technician labor hours measure effort. The entire outage can also include waiting for parts or production access.

Assemble the files without counting a failure twice

Export work order ID, asset ID, failure-event ID, work type, priority, status, opened date, required date, completed date, labor cost, parts cost, contractor cost, and remaining labor estimate. Bring scheduled and actual operating hours from the production or equipment log.

Count distinct failure-event IDs, because several work orders can describe one breakdown. Sum labor and material transactions without repeating a work order total beside each parts row.

Downtime counts overlapping out-of-service intervals only once. Separate active repairs, waiting, and planned work where records permit it.

Use a fixed period for operating results and a dated snapshot for backlog.

A worked manufacturing example

A US packaging manufacturer reviews three comparable machines for September 2026. Each has 160 scheduled production hours. Every failure in this example was repaired within the month, with one corrective work order per failure; there are no overlapping repair intervals or unfinished failures.

The following results use those rows. Percentages and hours are rounded to two decimal places; calculations retain unrounded values.

Fleet MTBF divides total exposure by total failures for comparable machines; it does not predict the next shift. NASA's reliability-centered maintenance guide discusses failures per operating hour and distinguishes repair time from broader downtime.

The time check is 452 operating + 28 downtime = 480 scheduled hours. The 28 − 18 = 10 downtime hours outside active repair must match waiting and planned-work records.

Add repair costs and backlog

These are costs assigned to the same completed corrective repairs. They exclude preventive work and equipment purchases.

The cost proof is 1,800 + 1,200 + 500 = $3,500, tied to repair transactions. Dividing by the six repairs gives 3,500 / 6 = $583.33 per repair, rounded to cents.

At September 30, the approved open-work backlog reconciles as follows. Added and completed hours are estimates moving into and out of the backlog, rather than actual technician hours consumed.

With no cancellations or estimate revisions, 96 + 64 − 40 = 120 hours. Backlog is 120 / 60 = 2.00 crew weeks. NASA's guide expresses backlog in estimated labor hours divided by available weekly hours; separate ready work from jobs still awaiting parts or approval before making staffing decisions.

Decide who investigates what

The maintenance manager owns failure definitions and repair records. Production verifies scheduled hours and stoppages; finance checks repair costs; the planner owns remaining estimates and the backlog snapshot.

Machine A has more failures and lower availability, while active repair time is the same across machines. Investigate its failure causes and non-repair downtime before assuming technicians need to work faster. Machine C's highest cost per repair comes from only one repair, so review the job details before calling it a persistent cost problem.

Connect operational findings to manufacturing cost per unit when output changes, and use manufacturing operational ROI to assess proposed spending. Repair costs alone do not establish the financial value of a reliability improvement.

Where it goes wrong

• Calendar hours as exposure. A machine scheduled only for production shifts should not receive operating credit for idle weekends.
• Open repairs omitted. Completed-job averages can look better while long outages remain unresolved. Show open outages separately.
• Labor hours called downtime. Crew effort and elapsed asset outage are different quantities.
• Backlog measured only by count. A small inspection and a major overhaul are not equivalent workloads.
• Zero failures entered as zero MTBF. Report the observed operating hours and absence of failures instead of dividing by zero.

Check your maintenance files in Covirage

Upload work orders, operating-hour records, and repair-cost exports; Covirage maps the columns, and its tools compute and check the reliability, cost, and backlog comparisons. The external AI model explains the results with every figure traced to its rows. Create a free Covirage account, which takes about a minute, then set up your business and bring a sample of your own data.

See Covirage for industrial manufacturers for the industry context.