Maintenance KPIs Guide

Introduction

A maintenance metric is useful when it changes a decision. A dashboard that shows green arrows but cannot explain the denominator, exclusions or underlying work is a reporting product, not a management system.

Maintenance leaders need a balanced view of demand, planning, execution, reliability, resource and data quality. No single KPI proves maintenance effectiveness. High PM compliance can coexist with repeat failures; low MTTR can coexist with rushed repairs; a smaller backlog can reflect deleted work instead of improved delivery.

This guide explains how to define a trustworthy KPI set and use it in operating routines. Product-specific Titan MMS reporting should be added only after verification.

Table of Contents

1. KPI design principles
2. Demand and backlog
3. Planning and schedule
4. Preventive maintenance
5. Reliability and downtime
6. Resource and cost
7. Data quality and adoption
8. Dashboard and governance design
9. FAQs

KPI Design Principles

For every metric, document:

• Decision it supports.
• Business-facing unit and cohort.
• Numerator and denominator.
• Included/excluded work, assets, sites and time.
• Source systems and refresh.
• Owner and review frequency.
• Target/baseline basis.
• Drill-down fields.
• Likely gaming or misinterpretation.

Prefer a small set with actionable drill-down. Segment by criticality, site, work type, asset class and planned/unplanned status when aggregation hides risk.

Use a metric dictionary. If one site calculates PM compliance by due date and another gives a grace period, the enterprise average is not comparable until policy is harmonized.

Demand and Backlog Metrics

New work demand

Count requests/work identified by source, priority, asset and type. A rise can mean deterioration, better reporting or a new inspection program. Interpret with context.

Backlog size and age

Backlog can be measured in work orders or estimated labor hours. Hours usually better represent workload, but only if estimates are reasonable. Show age bands and priority/criticality.

Decision: Does the team have capacity and is high-risk work waiting too long?

Trap: reducing backlog by mass-closing unvalidated requests.

Ready backlog

Separate work ready to schedule from work waiting for scope, parts, engineering, access or approval. This makes planning constraints visible.

Emergency/break-in work

Define emergency narrowly—work requiring immediate response to an unacceptable consequence. Track share of labor/work and causes. If every request is urgent, priority has no meaning.

Planning and Schedule Metrics

Planned work percentage

Possible definition:

planned maintenance labor hours / total completed maintenance labor hours

Define what qualifies as planned: scope, labor estimate, parts, instructions, safety/access readiness. Merely scheduling a work order does not make it planned.

Schedule compliance

Possible definition:

scheduled work completed in the agreed period / scheduled work committed

Track labor hours as well as work-order count where job sizes vary. Capture reasons: equipment unavailable, emergency break-in, parts, labor, scope change or production priority.

Decision: Is the weekly plan realistic and protected?

Trap: scheduling only easy work to raise compliance.

Wrench time

Direct hands-on time can reveal waiting and coordination, but measurement can be intrusive and misused. Prefer process sampling and removal of systemic delays rather than individual productivity surveillance.

Preventive-Maintenance Metrics

PM compliance

Possible definition:

PM work completed within policy window / PM work due

Define due, grace window, deferral, cancellation and critical PM. Separate statutory/safety-critical tasks.

Trap: closing incomplete PM to meet target.

PM effectiveness

Compliance shows execution, not whether tasks control failure. Evaluate:

• Findings/defects discovered.
• Corrective work generated and resolved.
• Failures occurring between PMs.
• Tasks consistently returning no useful finding.
• Repeat failures linked to task quality or interval.

Preventive-to-corrective balance

Work mix can show maturity, but a universal target is unsafe. New defect-identification programs can temporarily increase corrective work while improving risk control.

Reliability and Downtime Metrics

Availability

A common form is:

uptime / (uptime + downtime)

But define planned time, standby, changeover and external loss. Use bottleneck/business context.

MTBF

operating time / number of defined failures

Use for comparable repairable assets and stable definitions. An average can hide early-life failures, wear-out and different operating conditions.

MTTR

Define whether it measures active repair, restoration or the entire downtime event. Break down response, diagnosis, waiting, access and repair for action.

Repeat failure

Define same asset/component/failure mode within a selected window. This can expose ineffective repairs or unresolved causes, but needs usable failure coding.

Bad actors

Rank assets by consequence, frequency, duration and maintenance cost—not one measure alone. A Pareto is a starting point; validate attribution.

Resource, Parts and Cost Metrics

• Labor hours by planned/emergency, asset, work type and site.
• Overtime and contractor usage with demand context.
• Parts stockouts affecting scheduled or emergency work.
• Emergency purchases/expedite events.
• Inventory accuracy and obsolete/slow-moving stock.
• Maintenance cost by asset or output, with finance-aligned definitions.
• Estimate accuracy for planned work.

Cost reduction is not automatically positive. Deferred work can reduce current spend and increase risk. Pair cost with reliability, backlog and compliance.

Data Quality and Adoption Metrics

A dashboard is only as trustworthy as its records. Monitor:

• Work linked to valid assets/locations.
• Completed work with required symptom/cause/action.
• Labor and parts completeness.
• PM with current job plans and owners.
• Assets with criticality/class/status.
• Duplicate or inactive master data.
• Mobile records synchronized successfully.
• Supervisory rejection/correction patterns.

Use adoption measures by role and workflow, not raw logins. For example, percent of assigned work completed through the intended process or percent of critical inspections with required readings.

Build the Dashboard Around Decisions

Daily/frontline

Safety/critical exceptions, emergency work, overdue critical tasks, today’s schedule, unavailable parts/access and unassigned demand.

Weekly planning

Ready backlog, schedule compliance/reasons, next-week capacity, PM due, critical parts and major defect status.

Monthly reliability/management

Downtime/failure Pareto, repeat failures, PM effectiveness, work mix, backlog risk, resource/cost, data quality and improvement action status.

Each visual needs an adjacent interpretation and owner. Provide drill-down from metric to events/work orders. Do not use color alone to indicate status.

Targets

Set targets from risk, policy, historical capability and improvement plan. Avoid copying an industry “world-class” value with different definitions. Use control limits/trends where helpful and distinguish target from forecast.

Governance Routine

1. Data owner validates exceptions and completeness.
2. Maintenance/operations review the decision metrics.
3. Owners investigate material movements.
4. Actions are recorded with due dates and expected mechanism.
5. The next review checks action and outcome.
6. Metric definitions change only through governance and are versioned.

When behavior changes around a target, look for gaming. If PM compliance rises suddenly, sample job evidence. If MTTR falls, check repeat failures. Metrics should invite learning, not punishment.

FAQs

What are the most important maintenance KPIs?

A balanced minimum often covers risk/backlog, planning/schedule, PM, reliability/downtime, resource/parts and data quality. Select based on decisions.

What is a good PM compliance rate?

No universal number is safe. Define policy windows and criticality, baseline performance and set a target consistent with risk and capacity.

Is lower MTTR always better?

No. It can indicate faster restoration, but rushed repair may create repeat failures. Break down the timeline and pair with quality/repeat measures.

How many KPIs should a dashboard have?

Enough to manage the operating routine without hiding the story—often a small top-level set with drill-down. Avoid decorative metrics.

Can a CMMS calculate these metrics?

Many can calculate maintenance metrics, but verify formulas, source data, filters and drill-down. Cross-system measures may require MES/ERP/data integration.

How often should KPIs be reviewed?

Exceptions daily, planning weekly and reliability/management monthly is a common pattern, adapted to risk and operation.