Maritime Maintenance KPI: The 15 That Matter

Every shipowner measures something. Very few measure things that change a decision. Between the chief engineer's spreadsheet, the superintendent's monthly report and the annual management review, the same fleet can produce three different versions of the truth about its technical condition — and none of them says clearly whether to bring a drydocking forward, reinforce a critical spares holding, or revise a maintenance interval.

A maritime maintenance KPI is only worth having if it meets three conditions: it can be calculated with no extra effort from data already captured on board, it is comparable from vessel to vessel, and it is tied to a specific decision that someone is empowered to take. An indicator that triggers no decision is not an indicator — it is meeting-room decoration.

This article covers 15 indicators that genuinely earn their place, grouped into four families: reliability, execution, cost and compliance. For each one you will find the definition, the formula, the unit, the calculation frequency, a target range presented honestly as an observed field benchmark rather than a standard, the decision it triggers, and the calculation trap that quietly corrupts it. A full summary table, a three-tier review rhythm and an FAQ close the subject.

Why measure — and why most dashboards are useless

The decorative dashboard trap

The pattern is familiar. An operator rolls out a planned maintenance system, switches on the reporting module, and six months later has eighteen charts nobody looks at. Why? Because the indicators on display answer the question “what can we measure?” instead of “what do we need to decide?”.

Three symptoms give away a decorative dashboard. First: nobody can say what would happen if the indicator turned red. Second: the number gets commented on in the meeting but no action is raised behind it. Third: the figure is disputed at every review, which means the definition is not shared between ship and shore.

The rule: one KPI, one owner, one decision

Before adding an indicator to a dashboard, force yourself to complete this sentence: “If X exceeds Y, then Z does W within T days.” If you cannot complete it, the indicator can stay in the database, but it has not earned screen space.

This discipline has a practical consequence: an operational shipboard dashboard should carry no more than five or six indicators. The fifteen KPIs below are not meant to share a single screen — they are meant to be distributed across three review tiers (weekly on board, monthly at fleet level, quarterly at management level) described later in this article.

Family 1 — Reliability: is the equipment holding up?

KPI 1 — MTBF, mean time between failures

Definition: average running time of an item of equipment between two failures that cause a loss of function.
Formula: cumulative running hours ÷ number of failures over the period.
Unit: hours.
Frequency: calculated monthly, read as a trend over 6 to 12 months.
Order of magnitude: there is no universal figure. MTBF should only be compared against itself over time, or between sister vessels on similar trades. On a well-maintained emergency generator, several thousand hours is common; on a hard-worked separator or ballast pump the figure can be ten times lower without anything being wrong.
Decision triggered: an MTBF that declines over three consecutive periods forces you to reopen the equipment file — wrong interval, spare part quality, work procedure, operating conditions. This is the classic signal that justifies moving from corrective to condition-based maintenance on that specific machine.
Calculation trap: using calendar hours instead of actual running hours. An auxiliary engine at rest does not age like an engine under load. Without reliable running hour counters, MTBF is a ratio with the appearance of rigour and none of the substance. Second trap: aggregating unrelated equipment into a vessel-level MTBF that means nothing.

KPI 2 — MTTR, mean time to repair

Definition: average elapsed time between the occurrence of a failure and the equipment's actual return to service.
Formula: total corrective downtime hours ÷ number of corrective interventions.
Unit: hours.
Frequency: monthly.
Order of magnitude: the absolute value matters far less than its breakdown. On fleets that measure carefully, waiting time — for a spare, a technician, or an operational window — very often accounts for the majority of MTTR, with the repair itself taking only a fraction of the elapsed time.
Decision triggered: always break MTTR into four segments — diagnosis, waiting for parts, labour, testing and return to service. If waiting for parts dominates, the problem belongs to inventory and criticality, not to maintenance. If diagnosis dominates, the problem is documentation or competence.
Calculation trap: starting the clock when the work order is raised rather than when the failure was observed. The gap between the two — sometimes several days at sea — is precisely the interesting information.

KPI 3 — Failure rate

Definition: how often failures occur, normalised against the equipment's actual exposure.
Formula: (number of failures ÷ running hours) × 1,000.
Unit: failures per 1,000 running hours.
Frequency: monthly, consolidated quarterly by equipment family.
Order of magnitude: to be built fleet by fleet. The useful output is not the overall figure but the Pareto ranking: on most vessels, five to ten items of equipment account for the large majority of failures.
Decision triggered: every quarter, publish the “top 5” contributing equipment and give it a proper root cause analysis, instead of spreading effort thinly across the whole asset register held in the Equipment module.
Calculation trap: counting every corrective work order as a failure. A gasket replaced during a routine round is not a breakdown. Define failure as a loss of function and hold that definition, otherwise the figure inflates mechanically as soon as the crew starts recording properly.

KPI 4 — Technical availability

Definition: share of time during which the vessel or equipment is fit to perform its mission, measured against the time it was required.
Formula: (required hours − downtime hours) ÷ required hours × 100.
Unit: percentage.
Frequency: monthly.
Order of magnitude: on a commercially trading vessel, excluding planned off-hire, anything below 95 % generally deserves examination, while a ferry in peak season or a unit on a charter with performance penalties should be aiming considerably higher. These are field observations, not a standard.
Decision triggered: dropping below the contractual threshold forces a trade-off — bring a drydocking forward, install redundancy, or renegotiate the operating profile.
Calculation trap: including or excluding planned stoppages shifts the result by several points. Decide once, write the rule into the procedure, and do not change it mid-year.

Family 2 — Execution: is the plan actually being delivered?

KPI 5 — PMS compliance (planned maintenance completed on time)

Definition: share of planned maintenance tasks due in the period that were completed inside their tolerance window.
Formula: (planned work orders closed within the window ÷ planned work orders due) × 100.
Unit: percentage.
Frequency: weekly on board, monthly at fleet level.
Order of magnitude: mature fleets typically sit between 90 and 95 %, with a hard 100 % requirement on critical equipment tasks and on anything tied to the safety management system. A score persistently below 85 % rarely means a careless crew — it almost always means an overloaded plan.
Decision triggered: below 85 %, the correct response is to re-scope the plan, not to lecture the vessel. Compare the theoretical planned maintenance workload against the hours actually available, then revise the intervals using a structured method for building the planned maintenance programme.
Calculation trap: bulk closing at month end, which produces an excellent score and zero information. Second trap: not having defined the tolerance window at all. Around ±10 % of the interval is common practice; what matters is that it is written down and configured in the system.

KPI 6 — Schedule compliance

Definition: the organisation's ability to execute work in the week it was scheduled — a measure of planning quality, not to be confused with KPI 5.
Formula: (scheduled hours actually completed in the scheduled week ÷ hours scheduled for that week) × 100.
Unit: percentage.
Frequency: weekly.
Order of magnitude: a realistic benchmark in ship operation is 75 to 85 %. Counter-intuitively, a score above 95 % week after week is suspicious: it usually means the schedule is deliberately under-loaded.
Decision triggered: below 70 %, the weekly schedule is fiction. Cut the scheduled volume to what is genuinely executable given port calls, watchkeeping and rest hour requirements.
Calculation trap: measuring in number of tasks rather than hours. Ten ten-minute visual inspections do not offset one 40-hour overhaul that slipped.

KPI 7 — Maintenance backlog in hours, and backlog ageing

Definition: identified, approved but not yet executed workload, expressed in hours and converted into weeks of capacity, complemented by the share of backlog older than 90 days.
Formula: sum of estimated hours on open work orders ÷ weekly capacity of the technical team. Ageing: (backlog hours open more than 90 days ÷ total backlog hours) × 100.
Unit: hours, weeks of workload, percentage.
Frequency: weekly.
Order of magnitude: two to four weeks of workload is a healthy backlog — it gives planning flexibility. Below one week, you are planning day to day and the organisation is purely reactive. Beyond six weeks, control is lost and the backlog becomes a graveyard. For the ageing component, staying under 10 to 15 % beyond 90 days is a reasonable benchmark.
Decision triggered: past six weeks, only three honest levers exist — bring in a riding squad, defer work into the next drydocking, or purge work orders that have become irrelevant. Pretending is not one of them.
Calculation trap: counting backlog in number of work orders. “87 open work orders” means nothing: it could be three days or six months of work. The corollary is that a work order without an hours estimate is unusable — make the field mandatory.

KPI 8 — Planned versus corrective ratio

Definition: how maintenance effort splits between planned work and unplanned work.
Formula: (planned and condition-based maintenance hours ÷ total maintenance hours) × 100.
Unit: percentage.
Frequency: monthly.
Order of magnitude: in steady state, something in the range of 70/30 to 80/20 in favour of planned work is what well-run fleets show. Chasing 95 % planned makes no sense: it would mean over-maintaining.
Decision triggered: when corrective work persistently exceeds 40 % of hours, identify the three items of equipment driving it and treat those specifically, rather than adding planned tasks across the whole fleet.
Calculation trap: measuring in number of work orders instead of hours, which flattens reality. Another common trap: reclassifying a corrective job as planned after the fact because “it was due anyway”. Lock the work order type at creation.

KPI 9 — Spare parts waiting time

Definition: average delay between a work order raising a requirement for a part and that part being physically available on board.
Formula: sum of (date received on board − date requirement raised) ÷ number of order lines.
Unit: days.
Frequency: monthly, segmented by criticality and trading area.
Order of magnitude: no global figure is meaningful — short-sea coastal trading and a South Atlantic campaign are not comparable. Segment instead: for a critical spare, more than seven to ten days is an operational problem; for a standard consumable delivered at the next port call, three to six weeks is common and acceptable.
Decision triggered: any critical item whose real lead time systematically exceeds the target should move to on-board stock with a defined reorder point. That is the direct link between this KPI and MRO inventory management.
Calculation trap: starting from the purchase order date instead of the date the requirement was raised. The internal approval delay — often the longest segment — then disappears from the measurement, even though it is exactly the part the office controls through its Purchasing module.

Family 3 — Cost: where does the money actually go?

KPI 10 — Maintenance cost per vessel and per running hour

Definition: fully loaded maintenance cost normalised against a unit of use, so that vessels become comparable.
Formula: (external labour + parts consumed + service contracts + tool hire) ÷ running hours of the main plant, or ÷ operating days.
Unit: currency per running hour, or per operating day.
Frequency: monthly, consolidated quarterly.
Order of magnitude: no external benchmark transfers from one fleet to another. The only valid comparison is internal: sister vessels, same trade, same operating profile. A gap of more than 20-25 % between two identical vessels is a strong signal.
Decision triggered: investigate the sister-vessel gap before touching the overall budget. It is almost always explained by a local practice — heavier use of external technicians, intervals modified on board, or a difference in recording discipline. Practical levers are covered in our article on reducing vessel maintenance costs.
Calculation trap: mixing committed and invoiced amounts from one month to the next, which produces an unreadable sawtooth. Another classic omission: technician mobilisation, travel and standby costs, which sometimes exceed the cost of the job itself.

KPI 11 — Share of emergency purchases

Definition: proportion of purchasing carried out on an emergency basis, i.e. outside the normal quotation and lead time route.
Formula: (value of orders flagged urgent ÷ total maintenance purchasing value) × 100.
Unit: percentage, preferably by value.
Frequency: monthly.
Order of magnitude: staying under 10 to 15 % by value is achievable for an organised fleet. Above 20 %, the operator is structurally paying for air freight and for the absence of competitive quotation.
Decision triggered: a high rate is not a purchasing problem, it is a symptom. Trace it back to the cause: minimum stock breach, planned maintenance not delivered, or an unanticipated failure. The fix sits upstream.
Calculation trap: not defining “urgent”. Without an objective criterion — requested lead time below a written threshold, or use of express freight — everyone ticks the box according to mood and the indicator stops measuring anything.

KPI 12 — Inventory value and turnover

Definition: capital tied up in spare parts and the speed at which that capital turns, complemented by the share of dormant items.
Formula: turnover = annual consumption at cost ÷ average inventory value. Dormant = (value of items with no movement in 24 months ÷ total inventory value) × 100.
Unit: currency, turns per year, percentage.
Frequency: quarterly.
Order of magnitude: in marine MRO, roughly 1 to 2 turns per year is normal — structurally slower than shore-based industry, and that is expected. A dormant share above 25 %, on the other hand, calls for a clean-up.
Decision triggered: sort dormant items into three categories — insurance spares to keep despite no movement, items transferable to another vessel in the fleet, and items to write off. That logic is developed in our guide to spare parts management at sea.
Calculation trap: lumping consumables and critical spares into a single ratio. Consumables should turn fast; a propulsion insurance spare may never turn at all, and that is precisely its purpose.

KPI 13 — Cost of unplanned downtime

Definition: full cost of unplanned stoppages, including lost earnings and not merely the repair invoice.
Formula: (unplanned downtime hours × hourly operating cost) + direct extra costs (express freight, technician mobilisation, towage, contractual penalties, passenger or cargo re-routing).
Unit: currency per period, and percentage of the maintenance budget.
Frequency: monthly, reviewed quarterly at management level.
Order of magnitude: the trend and the comparison with the planned maintenance budget matter more than the absolute value. This is the one KPI on the list that speaks immediately to a finance director.
Decision triggered: this is the indicator that unlocks investment. Costing a single unplanned stoppage in full is usually enough to justify an insurance spare, a vibration monitoring programme, or an earlier drydocking.
Calculation trap: counting only the repairer's invoice. This is the error that kills most maintenance investment cases: it makes the failure look far cheaper than it actually was.

Family 4 — Compliance: is the vessel staying in class and in certificate?

KPI 14 — Certificates and surveys due within 30 days

Definition: number of certificates, class surveys and crew documents falling due within 30, 60 and 90 days, complemented by the number of expiries that actually occurred.
Formula: count by rolling window. Control indicator: number of documents expired as at today, which must remain strictly zero.
Unit: number of documents.
Frequency: weekly.
Order of magnitude: the target is not negotiable — zero expiries, and 100 % of items entering the 90-day window carrying an assigned, dated action.
Decision triggered: entering the 90-day window triggers the survey booking, the service order or the certificate of competency renewal; an item at 30 days with no open action triggers immediate escalation to the superintendent and the Designated Person Ashore.
Calculation trap: an incomplete register. Many operators track the vessel's statutory certificates but forget equipment certification — liferafts, extinguishers, EPIRB, lifting appliances, breathing apparatus cylinders — as well as the certificates of competency and medical certificates held in the Crew module. A properly structured certificate register must cover all three families.

KPI 15 — Open audit findings and corrective action closure rate

Definition: the stock of findings raised by internal audits, ISM audits, port State control inspections and class surveys, and the speed at which the associated corrective actions are cleared.
Formula: count of open findings by severity (observation, non-conformity, major non-conformity) and average age in days. Closure rate = (corrective actions closed within the agreed deadline ÷ corrective actions falling due in the period) × 100.
Unit: number, days, percentage.
Frequency: monthly.
Order of magnitude: zero open major non-conformities, an on-time closure rate above 90 %, and an average age of open findings below 60 days are demanding but achievable benchmarks.
Decision triggered: any corrective action past its deadline escalates automatically to the Designated Person Ashore. A finding that recurs from one audit to the next is no longer a corrective action issue: it is a procedure that needs rewriting, as covered in our analysis of how the ISM Code and a maintenance management system fit together.
Calculation trap: administrative closure with no evidence of effectiveness. An action closed without objective evidence — photograph, reading, updated procedure, training record — will reopen at the next audit. Second trap: treating internal audit findings with less rigour than external ones, which empties internal auditing of its purpose.

The 15 KPIs at a glance

KPI	Family	Formula	Unit	Frequency	Field benchmark	Decision triggered
1. MTBF	Reliability	Running hours ÷ number of failures	h	Monthly	No universal value; read the trend	Revise interval or spare quality if declining over 3 periods
2. MTTR	Reliability	Downtime hours ÷ number of interventions	h	Monthly	Break it down rather than benchmark it	Act on inventory if waiting for parts dominates
3. Failure rate	Reliability	(Failures ÷ running hours) × 1,000	failures / 1,000 h	Monthly	Internal baseline to be built	Root cause analysis on the top 5 contributors
4. Technical availability	Reliability	(Required hours − downtime) ÷ required hours × 100	%	Monthly	Examine below 95 % excluding planned off-hire	Bring drydocking forward or add redundancy
5. PMS compliance	Execution	Planned WOs closed in window ÷ WOs due × 100	%	Weekly / monthly	90-95 %; 100 % on critical equipment	Re-scope the plan below 85 %
6. Schedule compliance	Execution	Hours done in scheduled week ÷ hours scheduled × 100	%	Weekly	75-85 %; above 95 % means under-loaded	Rebuild a realistic schedule below 70 %
7. Backlog and ageing	Execution	Open WO hours ÷ weekly capacity; % beyond 90 days	weeks, %	Weekly	2 to 4 weeks; under 10-15 % beyond 90 days	Riding squad, defer to drydocking, or purge
8. Planned / corrective ratio	Execution	Planned hours ÷ total hours × 100	%	Monthly	70/30 to 80/20 measured in hours	Move top contributors to condition-based
9. Spare parts waiting time	Execution	Received on board − requirement date, averaged	days	Monthly	Segment by criticality and trading area	Hold critical slow-movers on board
10. Cost per running hour	Cost	Full maintenance cost ÷ running hours	currency/h	Monthly	Compare sister vessels only	Investigate a gap above 20-25 %
11. Emergency purchase share	Cost	Urgent order value ÷ total value × 100	%	Monthly	Under 10-15 % by value	Trace the upstream cause above 20 %
12. Inventory value and turnover	Cost	Annual consumption ÷ average stock; % dormant	turns/year, %	Quarterly	1 to 2 turns per year; dormant under 25 %	Write off, transfer, or keep as insurance spare
13. Cost of unplanned downtime	Cost	Downtime hours × hourly cost + direct extras	currency, % of budget	Monthly / quarterly	Read the trend, not the absolute value	Build the reliability investment case
14. Certificates due in 30 days	Compliance	Count by 30 / 60 / 90-day window	number	Weekly	Zero expiries, no exceptions	Book the survey or escalate to the DPA
15. Audit findings and closure	Compliance	Open findings by severity; on-time closures ÷ due × 100	number, %	Monthly	Zero major NC; closure above 90 %	DPA escalation and procedure revision if recurring

Without reliable source data, no KPI is worth anything

A wrong indicator is more dangerous than no indicator, because it manufactures false confidence. Before building a dashboard, check these six foundations.

Real running hours. Without regular counter readings, every usage-normalised indicator — MTBF, failure rate, cost per hour — is invalid. Counter reading must be a scheduled recurring task, not a goodwill entry.
Honest timestamps. Failure observed, work order raised, work started, work finished, equipment returned to service: five distinct timestamps, captured at the real moment. That is the prerequisite for an interpretable MTTR.
A written definition of failure. Total loss of function, observed degradation without loss of function, anomaly with no consequence: three categories are enough, but they must be identical across the fleet.
A stable equipment hierarchy. Equipment that changes code or parent breaks its own history. Freeze the numbering — SFI or an equivalent structure — before you start measuring.
A mandatory hours estimate. Without it there is no usable backlog, no schedule compliance and no credible planned/corrective ratio. A rough but systematic estimate is worth infinitely more than a precise but optional one.
Offline data capture. At sea, data that cannot be entered at the moment of the job gets entered from memory three weeks later, or not at all. That is why the Smart Sailors mobile app works without connectivity and synchronises at the next port call.

Key takeaway — A KPI only exists if it has an owner, a threshold and an associated decision. Five or six indicators tracked seriously deliver more than fifteen charts reviewed once a quarter. Start with backlog in hours, PMS compliance and certificates due within 30 days: those three alone reveal most of what matters about a vessel's technical health.

The right review rhythm: ship, fleet, board

Weekly on board — 20 minutes

Chaired by the chief engineer with the chief officer, at the end of the week. Four indicators only: backlog in hours and weeks of workload, schedule compliance for the week just ended, planned maintenance due in the next two weeks, and certificates or documents due within 30 days. Expected output: next week's schedule and the list of blockers to escalate ashore.

Monthly at fleet level — 60 minutes

Chaired by the superintendent with the chief engineers. Indicators: MTBF and MTTR on critical equipment, failure rate and top 5 contributors, PMS compliance, planned/corrective ratio, spare parts waiting time, cost per running hour, emergency purchase share, open audit findings. Expected output: interval revision decisions, minimum stock adjustments, corrective actions assigned with dates.

Quarterly at management level — 45 minutes

Chaired by the technical director with the owner and the Designated Person Ashore. Indicators: technical availability by vessel, cost of unplanned downtime, maintenance cost compared across sister vessels, inventory value and turnover, consolidated compliance position. Expected output: budget trade-offs, drydocking planning, reliability investment decisions.

This split avoids the most common failure: presenting an owner with weekly execution indicators that swing too much to carry a strategic decision, or conversely asking a chief engineer to run his week off a quarterly cost per running hour.

Building the dashboard in practice

One screen, eight tiles, three colours

The Smart Sailors Dashboard module aggregates in real time the data produced by the other modules — maintenance, equipment, inventory, purchasing, certificates, counters — with no re-keying. The design rule is simple: one page, eight tiles maximum, a three-state colour code (compliant, under watch, action required), and a vessel / vessel group / fleet filter.

Every tile must drill down to the underlying record. A superintendent seeing a seven-week backlog must be able to reach the list of work orders concerned, sorted by age, in two clicks. An indicator you cannot drill into will be disputed and, in time, ignored.

Compare, do not judge

Comparing vessels is only meaningful when the scope is identical: same operating profile, same equipment hierarchy, same recording rules. Otherwise it produces unfair league tables that penalise exactly the crews who record best — a vessel that reports many small defects will always look less reliable than one that reports none. This is the main risk of the exercise, and it is prevented by comparing time series rather than instant rankings.

From measurement to forecasting

Once the fifteen KPIs have been stable for two or three quarters, they stop being a report card and become a basis for projection: forecast workload for the next drydocking, next year's spares budget, required technical manning. The Forecast module is built on exactly this history. That is the point where measurement starts paying for itself — but it presupposes several months of disciplined recording, which is the real entry cost.

FAQ

How many KPIs should a vessel really track?

Five or six in the weekly on-board review, no more. The fifteen indicators presented here are distributed across three review tiers: the vessel tracks execution and deadlines, the office tracks reliability and cost, management tracks availability and consolidated compliance. A shipboard screen showing fifteen figures stops being read within a month.

What is the difference between PMS compliance and schedule compliance?

The first measures whether planned tasks that fell due were completed inside their tolerance window: it is a measure of conformity with the maintenance plan. The second measures whether the work scheduled for a given week was actually done that week: it is a measure of planning quality. You can report 95 % PMS compliance while running a completely fictional weekly schedule, because everything gets caught up at the end of the period.

How do you calculate MTBF without running hour readings?

You do not calculate it properly. The fallback is to use operating days as the denominator, which remains acceptable for continuously running equipment but becomes misleading for intermittent machinery. The right answer is to establish a recurring counter reading task: a few minutes a week that determine the validity of three of the fifteen indicators.

Is a large backlog always a bad sign?

No. A backlog of two to four weeks of workload is healthy: it gives planning margin and allows work to be grouped sensibly. A near-zero backlog often signals under-reporting of defects, which is more worrying. What really matters is the ageing profile: a stable three-week backlog with nothing older than 90 days is far better than a one-week backlog containing jobs open for two years.

Which indicators does an ISM auditor look at first?

PMS compliance on critical equipment, the traceability and justification of any deferrals, the tracking of certificate and certificate of competency expiry dates, and the closure rate of corrective actions raised at previous audits. An auditor is not looking for a perfect number: he is looking for consistency between what the safety management system claims and what the records show.

How long before KPIs become reliable after a new system goes live?

Allow three to six months for usable execution indicators — backlog, PMS compliance, schedule compliance — and twelve months for credible reliability and cost indicators, which need enough history to smooth out one-off events. The first months of figures will mostly serve to correct the underlying master data, which is already a useful result. A structured implementation shortens that period noticeably.

Conclusion

Measuring fleet maintenance is not a reporting exercise: it is a way of making decisions faster and with less guesswork. The fifteen indicators above are only worth what the discipline around them is worth — a written definition, an owner, a threshold, a decision. Pick three, hold them for six months, then expand. That is vastly more effective than deploying fifteen charts nobody will read.

Smart Sailors is a maritime maintenance management system built by seafarers, now in service on more than 400 vessels. Its 12 modules feed the indicators described in this article automatically, with no double entry, through a mobile app that works offline. Review our pricing or book a 30-minute demo to see your own dashboard take shape — the trial is free for 30 days.