Maintenance systems & execution
Fix the System
The operational field manual: prepare the work, protect the schedule, and close the feedback loop.
KPI Cascade / From result to responsibility
Pick a result.
Follow the connections.
See how a plant result connects to the work. Trace a path, find its owner, and take one practical change into your next review.
Your selected path
Click a card to trace another connection.
01 Plant resultWhat needs to improve?
02 Maintenance driverWhat could influence it?
03 Role measureWhere can someone act?
Explore the lit connections. Faded drivers belong to a different result.
Your pathAvailability→Schedule Compliance→Parts Kitting Rate
Why these connect & how to measure them
Completing committed maintenance can protect availability. Check whether missed jobs are contributing to the asset's downtime.
A scheduled job can stall if its materials are not ready. Check kit readiness against parts-related schedule misses.
Asset Availability
Example target ≥ 95%
Percentage of scheduled production time that equipment is available to run. A plant result influenced by maintenance, operations, and the agreed downtime definition.
- How to measure
- (Scheduled Time − Downtime) / Scheduled Time × 100. Track by asset, line, and plant using CMMS downtime tracking.
- Why review it
- Availability is the bridge between what maintenance does and what the plant needs. It's where maintenance earns — or loses — credibility with operations.
Watch for
- Not distinguishing between planned and unplanned downtime
- Inconsistent downtime recording across shifts
- Excluding changeovers or minor stops from the calculation
Schedule Compliance
Example target ≥ 90%
Percentage of scheduled work completed in the week it was scheduled. Where planning meets reality.
- How to measure
- Work orders completed on schedule / work orders scheduled × 100. Measure weekly. Don't count mid-week additions.
- Why review it
- If you can't execute the schedule, you don't have a planning system — you have a wish list.
Watch for
- Allowing the schedule to be broken without escalation
- Padding the schedule so compliance looks good but throughput suffers
- No weekly schedule-break analysis to understand why jobs slip
Parts Kitting Rate
Example target ≥ 90%
Percentage of scheduled jobs with all required parts staged before the start date.
- How to measure
- Jobs with parts kitted on time / total scheduled jobs requiring parts × 100.
- Why review it
- Missing materials can interrupt ready work. Physical kitting makes shortages visible before the job starts and can reduce avoidable trips and waiting.
Watch for
- Parts listed on the work order but not physically staged
- No lead-time buffer for ordering before scheduled date
- Storeroom not aligned with the weekly schedule
Targets are examples. Agree definitions and targets for your assets, risk, and operating conditions.
04 / Put it to work
Availability → Schedule Compliance → Parts Kitting Rate
Confirm materials before schedule release.
- Owner
- Planner / storeroom
- Review cadence
- Before the weekly schedule is released; review parts-related misses at the weekly schedule review.
- What to check together
- Check each job requiring parts against the physical kit and work order. Record shortages, assign a resolution owner, and hold unready work out of the committed schedule. Compare those misses with schedule compliance.
This is a pathway to investigate, not proof of cause. Review the role measure, driver, and plant result together before judging whether the change helped.
Keep the work connected
Take the chain into your next review.
Fix the System connects preparation, scheduling, execution, and feedback. Use its companion worksheets to record the evidence and agree the next change.
Put Fix the System to work
One improvement. Three useful views.
- 01Explore
Choose the right tool
Start with your role and the foundations your plant has in place.
Maintenance Tool Roadmap - 02Explore
Trace what moves the KPI
Find the supporting systems and the first change to investigate.
KPI System Map - 03You are here
Give the change an owner
Connect the result to a role measure, a practical action, and a review.
KPI Cascade Framework
Use the book’s cycle and playbook picker to choose a response, then bring the relevant worksheet to the team. These connections guide an investigation; they do not prove a cause.
The complete reference
Browse all 22 metrics.
The catalog keeps every definition and book reference. The guided path above shows a connected subset, with one action at a time. All targets below are examples.
Plant results
Overall Equipment EffectivenessExample target ≥ 85%
Availability × Performance × Quality. A combined view of downtime, speed, and quality losses relative to the defined production opportunity.
- How to measure
- Pull availability from CMMS, performance and quality from the production system. Measure per line, roll up to plant.
- Why review it
- OEE tells you where the losses are hiding — in downtime, slow running, or defects. Review the components to decide which loss to investigate first.
Watch for
- Measuring OEE without breaking it into its three components
- Using it as a stick rather than a diagnostic tool
- Comparing OEE across unlike assets without normalising
Asset AvailabilityExample target ≥ 95%
Percentage of scheduled production time that equipment is available to run. A plant result influenced by maintenance, operations, and the agreed downtime definition.
- How to measure
- (Scheduled Time − Downtime) / Scheduled Time × 100. Track by asset, line, and plant using CMMS downtime tracking.
- Why review it
- Availability is the bridge between what maintenance does and what the plant needs. It's where maintenance earns — or loses — credibility with operations.
Watch for
- Not distinguishing between planned and unplanned downtime
- Inconsistent downtime recording across shifts
- Excluding changeovers or minor stops from the calculation
Maintenance Cost per UnitExample target Trend ↓
Total maintenance spend divided by production output. Are you getting more output per dollar spent?
- How to measure
- Total maintenance cost (labour + materials + contractors) / production units. Track monthly, trend quarterly.
- Why review it
- Cost per unit connects maintenance to the P&L. It shifts the conversation from 'spend less' to 'spend smarter.'
Watch for
- Cutting cost without tracking the impact on availability
- Not separating capital vs. maintenance spend
- Using cost as the only lens — ignoring the reliability it buys
Maintenance Safety RateExample target Zero harm
Total Recordable Incident Rate for maintenance activities. Safety is the constraint every other metric operates within.
- How to measure
- (Recordable incidents × 200,000) / total hours worked. Track weekly, report monthly.
- Why review it
- Review incident patterns alongside work preparation and hazard controls. A rate alone does not identify the cause or establish that work is safe.
Watch for
- Tracking lagging indicators only — no near-miss reporting
- Pressure to not report to keep the number clean
- Disconnecting safety from maintenance process quality
Maintenance drivers
Planned vs. Unplanned WorkExample target ≥ 80% planned
The most telling metric in maintenance. Shows whether you're running the work or the work is running you.
- How to measure
- Planned work orders / total work orders × 100. Track weekly. Don't count PMs as 'planned' unless they were truly scheduled.
- Why review it
- The work mix can reveal dependence on reactive work. Compare it with actual labor, material costs, and equipment losses rather than assuming a fixed cost multiplier.
Watch for
- Counting all PMs as 'planned' — even calendar-triggered firefighting
- Not tracking emergency vs. urgent vs. planned separately
- Setting the target without building the planning process to achieve it
Schedule ComplianceExample target ≥ 90%
Percentage of scheduled work completed in the week it was scheduled. Where planning meets reality.
- How to measure
- Work orders completed on schedule / work orders scheduled × 100. Measure weekly. Don't count mid-week additions.
- Why review it
- If you can't execute the schedule, you don't have a planning system — you have a wish list.
Watch for
- Allowing the schedule to be broken without escalation
- Padding the schedule so compliance looks good but throughput suffers
- No weekly schedule-break analysis to understand why jobs slip
PM ComplianceExample target ≥ 95%
Percentage of preventive maintenance completed on time and to standard. Not just 'done' — done right, on time.
- How to measure
- PMs completed within tolerance window / PMs due × 100. Track weekly by area. Include quality checks.
- Why review it
- On-time, effective PM can support reliability. A missed task warrants a risk-based review; completion alone does not prove that failures are being prevented.
Watch for
- Tracking completion without checking quality
- PM lists that haven't been reviewed in years
- Closing PMs as 'done' without actually performing the work
Backlog WeeksExample target 3–5 wks
Ready-to-schedule backlog ÷ weekly crew capacity. Shows whether workload and resources are in balance.
- How to measure
- Sum estimated hours in ready-to-schedule backlog / available crew hours per week. Measure weekly.
- Why review it
- Compare ready demand with actual crew capacity and asset priorities. The example range is a discussion starting point, not proof that a backlog is balanced.
Watch for
- Not estimating job hours — making the calculation meaningless
- Including jobs that aren't truly ready (waiting for parts, permits)
- Letting backlog grow without ever prioritising or purging
Mean Time Between FailuresExample target Trend ↑
Average operating time between equipment failures. Shows whether your equipment is getting more or less reliable over time.
- How to measure
- Total operating hours / number of failures. Track per asset class. Trend monthly or quarterly.
- Why review it
- An MTBF trend can indicate changing reliability. Check operating exposure, failure definitions, and asset mix before attributing a change to the program.
Watch for
- Not defining 'failure' consistently across the plant
- Averaging MTBF across unlike assets to get a meaningless number
- Tracking MTBF without linking it to root cause elimination
Mean Time to RepairExample target Trend ↓
Average time from failure to return to service. Measures response and repair capability.
- How to measure
- Total downtime for repairs / number of repair events. Track by asset class. Trend monthly.
- Why review it
- Long repair events can include diagnosis, access, material, repair, or restart delays. Split the elapsed time to find the constraint before choosing an action.
Watch for
- Treating MTTR as a speed metric rather than a process quality metric
- Not tracking the components (diagnosis, parts, wrench time)
- Pressuring for faster repairs without addressing root causes
Role measures
Bad Actor Elimination RateReliability EngineerExample target ≥ 3/qtr
Chronic failure modes addressed per quarter through root cause analysis and implemented fixes, with effectiveness verified over a defined follow-up period.
- How to measure
- Count bad actors where root cause was identified, fix implemented, and no recurrence for 90+ days.
- Why review it
- Closing repeat-failure actions can reduce recurring losses. Verify effectiveness through recurrence and operating exposure, not just a completed action count.
Watch for
- Analysing failures but not driving fixes to completion
- Working on interesting problems instead of the highest-cost failures
- No tracking system for RCA completion and effectiveness
PM Tasks ReviewedReliability EngineerExample target ≥ 20/mo
PM tasks audited for value, frequency, and content — resulting in changes, additions, or deletions.
- How to measure
- Track PM changes in CMMS. Count additions, deletions, frequency changes, and content updates separately.
- Why review it
- An unmanaged PM program grows until it's bloated, ineffective, and ignored. Continuous optimisation keeps it sharp.
Watch for
- Reviewing PMs but not making changes (analysis paralysis)
- Optimising without failure data to guide decisions
- No formal PM review cycle — it happens ad hoc or not at all
Failure Coding AccuracyReliability EngineerExample target ≥ 90%
Percentage of closed work orders with accurate failure mode, cause, and remedy codes.
- How to measure
- Audit a sample of closed work orders monthly. Score for completeness and accuracy of coding.
- Why review it
- Missing or inconsistent failure data weakens reliability analysis. Check coding quality before drawing conclusions about repeat failures.
Watch for
- Relying on technicians to code without training or feedback
- Too many code options — technicians pick the first close match
- No feedback loop from reliability analysis back to coding quality
Job Plan CoveragePlanner / SchedulerExample target ≥ 80%
Percentage of scheduled work with a complete job plan (steps, parts, tools, permits, estimated hours).
- How to measure
- Scheduled work orders with complete plans / total scheduled × 100. Audit weekly.
- Why review it
- A work order without a job plan isn't planned — it's a request. Job plans turn planners from order-takers into force multipliers.
Watch for
- Job plans that are just a title and priority code
- Planners spending time on urgent work instead of planning next week
- No library of standard job plans for recurring work
Parts Kitting RatePlanner / SchedulerExample target ≥ 90%
Percentage of scheduled jobs with all required parts staged before the start date.
- How to measure
- Jobs with parts kitted on time / total scheduled jobs requiring parts × 100.
- Why review it
- Missing materials can interrupt ready work. Physical kitting makes shortages visible before the job starts and can reduce avoidable trips and waiting.
Watch for
- Parts listed on the work order but not physically staged
- No lead-time buffer for ordering before scheduled date
- Storeroom not aligned with the weekly schedule
Backlog Age (% > 90 days)Planner / SchedulerExample target ≤ 10%
Percentage of backlog waiting more than 90 days. Aging work can reveal gaps in priority, readiness, or ownership; some deferrals may be deliberate.
- How to measure
- Work orders aged > 90 days / total backlog × 100. Review monthly.
- Why review it
- Stale backlog erodes trust. If work orders sit for months, people stop writing them — and defects go unreported.
Watch for
- No regular backlog review and purge cycle
- Keeping old work orders 'just in case'
- Planner too busy with reactive work to groom backlog
Wrench TimeSupervisorExample target ≥ 55%
Percentage of the shift spent with tools on equipment. The direct measure of how well the system supports the people doing the work.
- How to measure
- Work sampling: observe technicians at random intervals. Categorise as wrench time, travel, waiting, admin, breaks.
- Why review it
- Use representative work sampling to locate waiting, travel, and support gaps. Improve the conditions for work rather than assuming a universal baseline or pushing people to work faster.
Watch for
- Using wrench time as a productivity whip, not a system diagnostic
- Not addressing root causes of low wrench time (travel, waiting, unclear scope)
- Measuring once and never again
Schedule Break RateSupervisorExample target ≤ 10%
Percentage of scheduled work displaced by unplanned work during the week.
- How to measure
- Scheduled jobs displaced / total scheduled × 100. Track with break codes explaining why.
- Why review it
- A break displaces a commitment. Record the reason and actual consequences so recurring disruption can be addressed without assuming a fixed cost penalty.
Watch for
- No break code system to track why the schedule broke
- Operations breaking the schedule without escalation
- Supervisor absorbing the break without pushing back
Daily Job Completion RateSupervisorExample target ≥ 90%
Percentage of assigned daily work completed by end of shift.
- How to measure
- Jobs completed / jobs assigned per day × 100. Track by crew. Review at next-day morning meeting.
- Why review it
- Low daily completion calls for a review of estimates, job readiness, interruptions, and other constraints before judging individual performance.
Watch for
- Assigning work without checking job plan readiness
- Not adjusting when emergency work hits
- No end-of-shift debrief to understand why jobs didn't finish
First-Time Fix RateTechnicianExample target ≥ 85%
Percentage of jobs completed correctly on the first visit — no rework, no return trip.
- How to measure
- Work orders closed without re-open within 30 days / total completed × 100.
- Why review it
- Rework is waste. A low first-time fix rate signals gaps in diagnostics, job plans, parts, or skills.
Watch for
- Not tracking rework as a separate metric
- Blaming technicians for rework caused by incomplete job plans
- No debrief process for repeat failures
Work Order Feedback QualityTechnicianExample target ≥ 90%
Percentage of completed work orders with useful feedback — what was found, what was done, conditions observed.
- How to measure
- Audit closed work orders. Score for: failure code, cause code, action taken, conditions noted.
- Why review it
- Technician feedback is the raw material for reliability analysis. Without it, the reliability engineer is flying blind.
Watch for
- 'Fixed it' as the only work order note
- No mobile device or easy way to capture feedback in the field
- No one reads the feedback, so technicians stop providing it
Safety Observations SubmittedTechnicianExample target ≥ 1/tech/wk
Number of safety observations, near-misses, or hazard reports submitted per technician per week.
- How to measure
- Count submissions per technician per week. Track the trend, not the absolute number.
- Why review it
- Reports can reveal hazards and barriers to speaking up. Low counts alone do not establish disengagement or safety; check access to reporting, response quality, and crew feedback.
Watch for
- Making the submission process too cumbersome
- Not closing the loop — hazard reported but nothing happens
- Using the metric as a quota rather than a culture indicator
The books behind the tools
Build the system behind the result.
Leadership & accountability
Leadership Isn't Personal
The leadership context: clear standards, supported ownership, and early responses to system drift.