Prioritising reactive work without a shouting match
Priority set by the person reporting produces a queue in which everything is urgent. Priority set by rule produces one that can be planned.
Reactive maintenance arrives faster than it can be done, so something has to determine the order. In most organisations that something is a combination of who asked, how forcefully, and how recently. For a product-side comparison related to the cost of changing priorities repeatedly, see task switching cost before deciding which data the operation actually needs.
This is not a moral failing of the people involved. It is what happens when there is no rule, because in the absence of a rule the only available inputs are volume and proximity.
Priority is a function of two things
Consequence and time sensitivity, and they are genuinely separate.
Consequence is what fails to happen because of this fault: safety, statutory compliance, business operation, comfort, appearance. Time sensitivity is how quickly the consequence escalates: a leak worsens by the hour, a failed lamp in a well-lit corridor does not. For a wider operational and compliance reference, consult OSHA hazard prevention guidance.
A matrix of these two, with three or four bands on each axis, produces a priority that can be assigned by whoever logs the job and defended afterwards. It also produces the occasionally uncomfortable result that a high-consequence, low-urgency job outranks a low-consequence, high-urgency one — which is correct and is exactly what an unstructured queue gets wrong.
Reporters should describe the impact accurately. They should not select the priority, because every priority field offered to a requester eventually reads 'urgent'.
Attach a target, and mean it
A priority band with no time attached is a label. Each band needs a response target and a rectification target, and they are different: attending within four hours and fixing within four hours are very different commitments, and conflating them is the main reason maintenance targets are perceived as broken.
Response is when someone attends and makes the situation safe or understood. Rectification is when the fault is resolved. Publishing both, and reporting against both, gives an honest picture — many jobs are attended promptly and then wait for a part, and a single combined figure hides that entirely.
Safety-related work sits outside the queue
Anything with an immediate safety implication — an exposed conductor, a leak on a stair, a failed fire door, a gas smell — does not enter the priority system. It is attended immediately and made safe, and the rest of the queue moves.
This needs stating explicitly, because a well-designed priority matrix creates a temptation to run everything through it. The matrix exists to order work that can wait; the point of a separate route is that some work cannot.
Watch the low-priority tail
A functioning priority system produces a predictable pathology: low-priority jobs are never reached. They accumulate, age, and eventually someone deletes the backlog because it has become meaningless.
Two mechanisms help. Ageing, where a job's priority rises after a defined period, so nothing waits indefinitely. And a reserved share of capacity — a defined slot each week for low-priority work, taken in age order. Both are small commitments that prevent the queue quietly becoming a two-tier system where anything not urgent is never done.
Group by location and by trade
Strict priority order is inefficient because travel and access dominate short jobs. Doing three low-priority jobs on the fourth floor while already on the fourth floor is better use of the hour than returning twice more.
The practical compromise is priority banding with geographic and trade grouping inside each band. Higher-priority work still goes first; work of equal priority is sequenced to minimise movement.
Report the queue, not just the completions
Most maintenance reporting counts jobs closed, which describes activity rather than position. The more informative figures are how many jobs are open, how old the oldest is, and whether the total is rising or falling.
A department closing more jobs than ever while the backlog grows is not improving; it is being overwhelmed slightly faster. That is a resourcing conversation, and it needs the queue figures to be had at all.