An SLA policy sets how fast your team should answer and resolve, and what happens when time is running out. Each ticket it covers gets countdown clocks you can see in the list and in the conversation header.
Create a policy
You see the SLA tab with permission to view SLA policies, and change it with permission to manage them.
- 1First response
- 2Resolution
- 3Clock starts at
Open Inbox, then Settings, then the SLA tab, and select New policy.
Name the policy ("Name this policy").
Choose a Queue and a Priority. Leave the priority on Any priority to cover all of them, and the queue on Every queue to cover all.
Set the Targets: First response and Resolution. Optionally turn on Next response.
Set the Clock behaviour, who to notify, and any escalation steps (below).
Select Save policy.
A policy's queue is fixed when it's made. To cover another queue, add another policy.
Targets and clocks
| Clock | Counts |
|---|---|
| First response | Until someone replies to the customer for the first time. |
| Next response | Starts when a customer writes again after the first reply, and clears on your next reply. |
| Resolution | Until the ticket is resolved. |
Enter each target in seconds, minutes, hours or days.
Clock behaviour
- Follow business hours counts time only while the queue is open. With Every queue, each ticket uses its own queue's hours.
- Pause while pending stops the clocks while you wait on the customer. Snoozing always pauses them.
- Warn before the deadline sets how early a clock is marked at risk. Left empty, it's a fifth of the target, and never under 10 minutes.
Where the clocks start
Under Clock starts at, choose when the first-response and resolution clocks begin:
- When the ticket is made — time the customer spent with an AI agent doesn't count.
- At the customer's first message — time with the AI agent counts, so a ticket can arrive already late.
Tickets already open keep the clock they started with.
Who to notify
Choose people from those who work the queue, and teams, separately for When a clock is at risk and When a clock is missed. The assignee is always told. Put a supervisor or an escalation team on missed clocks. To pick individual agents, the policy needs a queue; a policy for every queue has no members to pick from, though teams still work.
Escalation
Escalation is what the system does by itself when a clock reaches at risk or missed. It never messages the customer. Select Add a step, choose when it runs, then Add an action:
- Set priority
- Add label
- Move to queue
- Assign team (retired for new steps; existing ones keep running)
What agents see
A ticket's header shows each clock, for example "First response: due in 5m", "overdue", or "paused". The list shows the next one. Agents are notified when a clock is SLA running out or SLA missed, and the conversation records "SLA breached" in its events.
Reading the results
The SLA report shows your hit rate, which clocks you miss and by how much. See The SLA report.