What exactly is the backlog?
The maintenance backlog is all identified work not yet completed. It is neither good nor bad in itself: too small a backlog can mean problems are not being reported, while an endlessly growing one signals an imbalance between demand and capacity. The right question is not "how many orders are open?" but "how many weeks of work do they represent, and what do they contain?"
Measure in weeks, not in order count
The most useful measure is backlog in weeks: total estimated hours of open orders divided by the weekly capacity actually available (labour hours minus historical emergency work and absences). A backlog of 4 to 6 weeks is generally considered healthy: enough identified work to plan efficiently, without lead times becoming unacceptable.
Clean before you reduce
- Remove duplicates: the same problem reported by three people counts once.
- Close work already done but never closed in the system.
- Separate "ready to execute" work (planned, parts available) from blocked requests (waiting on parts, engineering or budget).
- Re-estimate old orders: hours estimated two years ago are often worthless.
In several plants where we work, the apparent backlog shrinks by 20 to 40% after a simple cleanup — without a single extra hour of work being performed. That is a typical example, not a guarantee: the result depends on the initial state of the data.
Prioritizing without closing overdue orders
Mass-closing overdue orders to "clean up" the backlog destroys operations' trust and makes real work disappear. The right approach combines three criteria: risk (equipment criticality and consequence of deferral), age of the request, and grouping effect (executing work on the same equipment or area together). Blocked work must stay visible, but in a separate category that does not inflate the executable backlog.
A clean backlog measured in weeks is a management tool; a backlog bloated with duplicates and ghost orders is just a frightening number.