Availability is the headline number in asset-heavy operations. It appears in board packs, contracts and improvement programmes.
It is also, in most organisations, a number nobody can reproduce, because the definition changes depending on who is counting.
The same outage, three durations
Consider a unit that stops overnight:
- When did it start? When the fault occurred, when the alarm was acknowledged, or when someone raised a ticket? Those can be hours apart.
- When did it end? When the repair finished, when the unit was returned to service, or when it was back at full output? Also hours apart.
- Did it count at all? If it happened during a window where nothing was scheduled, some definitions exclude it and some do not.
Maintenance reports the repair duration. Operations reports the production loss. Both are right. They are answering different questions using the same word.
Why the ambiguity persists
Because each definition is locally useful. Maintenance genuinely needs repair time to manage its crews. Operations genuinely needs production loss to manage output. The problem only appears when the numbers meet in a report and are treated as one.
It is also, quietly, convenient. A definition that can be selected after the fact is a definition that can be selected to suit the answer — rarely dishonestly, usually just by picking the source that seems most reasonable given what everyone already believes.
Settle it explicitly
The work is unglamorous and mostly not technical:
One definition per metric, written down, including what starts the clock, what stops it, and what is excluded.
Planned and unplanned counted separately, always. Combining them hides the only distinction that matters for improvement.
The clock started by an event, not a person. If the start time is when someone raised a ticket, you are measuring reporting behaviour as much as availability.
Exclusions listed and stable. Every operation has legitimate exclusions — weather, grid constraint, third-party. Each is defensible, and the list needs to be fixed in advance rather than extended when a month looks bad.
Restatement allowed and visible. Numbers improve as investigation completes. A figure that quietly changed is worse than one that changed with a note.
What it unlocks
Once the definition holds still, ordinary questions become answerable: which assets account for most of the loss, whether the same fault recurs, whether an intervention worked, and how much of the total is waiting rather than working.
That last one is usually the surprise. In many operations more downtime is spent waiting — for a part, a permit, an engineer, a window — than on the repair itself. Nobody sees it while the metric is repair duration, and it is often the cheapest thing to fix.
Start here
Take last month’s availability figure and ask two teams to reproduce it independently.
If they cannot, the number is not measuring the operation. It is measuring which system was opened first.
If your availability numbers do not survive scrutiny, get in touch, or read about our energy and utilities work.