Guide6 min read
Setting response and completion SLAs for facilities: a priority matrix
A service level agreement for facilities is two promises per job: how soon someone will respond and how soon it will be finished. Most teams that struggle with SLAs have either too many priority levels, targets nobody believes, or no agreed rule for what counts as which priority.
By the FM360AI team
Two clocks, not one
- Response time: from the request being logged to someone taking ownership and acting on it.
- Completion time: from the request being logged to the work being finished.
They measure different things. A fast response with slow completion points to parts or access problems. A slow response with fast completion points to triage. Reporting only one hides which of the two you have.
Define priorities by consequence
Priority should describe what happens if the job is not done, not how insistent the person reporting it is. Four or five levels are enough.
| Priority | Meaning | Examples |
|---|---|---|
| Critical | Immediate risk to people, or loss of a whole building's service | Exposed live wiring, gas smell, flooding, fire system fault |
| High | A space cannot be used, or damage is getting worse | No cooling in an occupied room, active leak, blocked only washroom |
| Normal | A fault that inconveniences but has a workaround | One of several lights out, dripping tap, sticking door |
| Low | Cosmetic or improvement work | Paint touch-up, signage, minor adjustment |
An example matrix
These figures are illustrative. Yours should come from what your team has actually achieved over the last few months, tightened gradually.
| Priority | Response target | Completion target |
|---|---|---|
| Critical | 15 to 30 minutes | 4 hours, or made safe |
| High | 1 to 2 hours | 24 hours |
| Normal | 4 to 8 hours | 3 days |
| Low | 1 day | 7 to 10 days |
Refine by category and location
One matrix rarely fits everything. A lift fault and a broken blind may both be high priority to the person affected, but they need different targets. The usual refinement is to set a default per priority and then override it for particular categories or locations: tighter for life safety systems and for critical rooms, looser for work that depends on an outside contractor.
When work has to pause
Some delays are not the team's doing: waiting for a part, for access to a room or for an approval. Decide in advance whether the clock stops in those cases, and require a reason each time a job is put on hold. Without that rule, either the numbers are unfair to the team or hold becomes a place to hide late jobs.
Common mistakes
- Everything is urgent. If more than about one job in ten is critical, the definition is too loose.
- Targets copied from a contract template. Start from your own history.
- No owner for response. Response only means something if a named person takes the job.
- Measuring but not looking. A compliance figure nobody reviews changes nothing.
- Counting closure, not quality. Track reopened jobs and feedback alongside compliance.
In FM360AI
FM360AI has separate response and completion SLA policies. Each can be scoped by priority, category, sub-category and location, with a due period in minutes, hours or days. When a ticket is created, the most specific matching policy of each type is attached, the ticket shows a live countdown for both, and a policy can be set to pause while the ticket is on hold. The dashboard's SLA watch lists tickets that are overdue or have four or eight hours left.
FM360AI measures SLA targets in elapsed time from ticket creation. If your targets are defined in working hours, set the periods with that in mind.
