The maintenance backlog is one of the clearest signals an asset-intensive operator has about the health of its operating model. Yet in most organisations it is treated as an accounting artefact: a large number, quoted at the monthly review, argued about at budget time, and rarely used as a management tool between those moments. Treated properly, backlog management is an asset management discipline. It exposes where work identification is strong or weak, whether planning is producing schedulable jobs, and whether asset criticality is actually influencing what crews touch this week. Ignored, the same backlog quietly redistributes risk onto assets that no one meant to defer.
This piece sets out what a maintenance backlog is once you look past the headline number, what a healthy backlog looks like, and the governance that keeps it that way. It is written for reliability leads, maintenance managers and asset directors who are tired of month-end conversations that end with “the backlog is up again” and no clarity on whether that is good, bad, or a symptom of something else entirely.
What the backlog actually is
Backlog is not a single number. Any competent view of it separates at least three things.
- Total backlog: every open, non-completed work order in the CMMS, regardless of state. This is the number executives usually see. On its own it is close to useless.
- Ready backlog: work that is fully planned, materials confirmed, permits achievable, and cleared for scheduling. This is the number planners and schedulers actually use.
- Deferred or waiting backlog: work that is legitimate but blocked, whether by shutdown windows, long-lead spares, capital dependency, or engineering change.
The SMRP body of knowledge treats these as distinct metrics for a reason. Work identification, planning, scheduling and backlog management are separate disciplines that show up in the same list of work orders, and mixing them together hides where the operation is actually struggling. A high total backlog with a healthy ready backlog is a very different problem from a low total backlog where almost nothing is ready to schedule.
Sitting alongside those states, three attributes matter for every work order in the backlog: age, criticality of the parent asset, and work type (corrective, preventive, condition-driven, statutory, project). Without those attributes, the backlog cannot be prioritised in any way that connects to reliability outcomes.
What “healthy” looks like
Industry benchmarks are consistent enough to be useful. SMRP references a target of roughly two to four crew-weeks of ready-to-work backlog per craft as the range where planners can build efficient schedules without running the crew day-to-day. Below that, the schedule is reactive by default: planners are placing whatever they can find rather than sequencing work by priority. Well above it, ready work sits so long that estimates go stale, permits expire, and the same job is effectively re-planned several times before it is done.
The exact number matters less than the shape. A healthy backlog usually shows:
- A ready backlog inside that two to four week band per crew, stable week on week.
- A clear ageing profile, with most ready work under 90 days and very little corrective work on critical assets over 180 days.
- A preventive maintenance completion rate that is not being propped up by closing PMs late, or worse, cancelling them to protect the metric.
- A visible, small pool of deferred work with a named owner and a scheduled review date on each item.
A backlog that ages quietly is a warning that criticality is not driving prioritisation. It is worth working from asset criticality into the backlog view itself, so that a two-year-old corrective on a Class A pump is treated as a governance event, not a data point.
The failure modes
Most backlog problems are not caused by too little labour. They are caused by governance gaps.
Work identification without triage. Defect notifications flow in from operators, inspections and condition monitoring, but nothing filters them. Everything becomes a work order. The backlog inflates, planners spend their time on jobs that will never be scheduled, and real work gets buried.
Planning that never finishes. Work orders sit in a “waiting for planning” state for weeks. Ready backlog is starved. The crew works reactively while the total backlog grows.
Silent PM deferral. Preventive maintenance is closed on paper to hit a completion KPI, or cancelled with weak justification. The backlog looks controlled. Failure rate on the affected asset class rises three to six months later, and the connection is not made.
No ageing view. Old work is invisible unless someone runs a report. Statutory items age past their due date without anyone noticing until an audit or an incident.
Backlog owned by no one. The CMMS is the system of record, but the accountability sits with a maintenance manager whose bonus is tied to output, not to backlog quality. Every conversation about backlog is a conversation about resource, never about work identification, planning quality or criticality.
Governance that holds
A backlog stays healthy when a small number of things run on a fixed cadence.
- Weekly backlog review at planner and supervisor level, focused on ready backlog by crew, ageing, and the top ten oldest items on Class A and B assets. Not a data-quality meeting, a work sequencing meeting.
- Monthly backlog governance at asset manager level, focused on trends by work type, deferred work with named owners, and any statutory or safety-critical item beyond its target date.
- A written prioritisation rule, tied to criticality and consequence. Not a matrix on a slide, an actual rule the CMMS can apply consistently. This is where a well-designed maintenance strategy either shows up or is exposed as narrative.
- A hard rule that PMs are not closed to manage the metric. Late or missed PMs are a legitimate data point. Silent closure destroys the reliability history that ISO 14224 style analysis depends on.
- A single backlog owner at asset director or head of maintenance level, with the mandate to reject work orders that should not exist and to escalate work orders that have aged past tolerance.
None of this requires new software. It requires the CMMS to be configured to expose the right views (ready backlog by crew, ageing bands, work type mix, backlog by criticality) and a leadership team that treats those views as operational, not reporting.
What good looks like a year in
Operators that run backlog as a discipline notice the same pattern within about twelve months. The total backlog number stops moving in dramatic swings. Ready backlog settles into a defensible band per crew. The ageing profile flattens: very little critical corrective work over 180 days, statutory work rarely past its due date, deferred items known and named. Reactive work, measured as a share of total labour hours, drops. And the monthly review conversation shifts from “how big is the backlog” to “what does the backlog tell us about work identification, criticality and PM design this quarter.”
That shift is what backlog management as an asset management discipline actually delivers. The number becomes secondary. The signal becomes the point.