The SLA report shows whether you're keeping your promises. It covers conversations under an SLA policy: how often every clock was met, which clock you miss, how late, and where.
Open Inbox → Reports → SLA. Set the range, grouping and filters as described in Report filters and export. This tab is the only one with a Policy filter. If your backend doesn't filter by policy yet, a note says "the figures cover every policy".
Targets, clocks and escalation are defined in SLA policies.
Are we keeping our promises?
Tiles compare with the previous period. A conversation counts as a miss when any of its clocks ran out.
- 1Hit rate
- 2Missed
- 3Breached right now
| Tile | Meaning |
|---|---|
| Hit rate | Conversations whose clocks were all met, such as "180 of 200 conversations". |
| Under SLA | Conversations covered by a policy in the period, and how many policies. |
| Missed | Conversations that missed any clock, with the count per clock: first response, next response, resolution. |
| Breached right now | Live: open conversations past a deadline, with how many are at risk. |
View misses and Open the desk jump to the list or the Inbox.
Is it getting better or worse?
Hit rate per period is drawn against a fixed 95% target line, not a target read from your SLA policies. The line is thin where a period has fewer than 5 conversations. Met and missed shows conversations per period by outcome, so a quiet week can't hide behind a high rate.
Which clock do we miss, and by how much?
Clocks missed counts misses per clock; a conversation can miss more than one. How late groups missed clocks by how long past the deadline they were met, or still are, so a five-minute slip and a two-day silence don't count the same. Lateness isn't recorded for the next-response clock.
Where do we miss?
Hit rate By policy, By queue and By agent, lowest first. The agent is the conversation's assigned agent, its last assignee, which is kept after it's resolved. It isn't necessarily who held it when the deadline passed. Rows with fewer than 5 conversations are shown but not ranked, marked n=3. When a policy filter is on, the by-policy card is hidden.
What is at risk right now?
Not tied to the range. A Clocks table lists, per clock, how many conversations are At risk (inside the warning lead) and Breached (past due and still open). A conversation can be late on two clocks; the last row counts each once.
Longest overdue lists open conversations past a deadline, furthest first, filtered by queue, team, agent, label and channel, each with "{duration} late". You need permission to read conversations to see it; otherwise you're told so. When nothing is overdue it says "Nothing overdue right now." Open the desk shows them in the Inbox.
All misses
A table of every conversation that missed a clock in the period: Policy, Conversation, Agent, Clock, Late by, Status (Still open or Resolved) and Opened. Filter by clock with All clocks and the clock chips. Select a conversation to open it.
You need the SLA breaches permission to see this list. Without it the table says "You need the SLA breaches permission to see the list."
When the clock starts
Each policy decides where its first-response and resolution clocks start: when the ticket is made, or at the customer's first message, which counts time with an AI agent. The tab notes which applies to yours, for example "Every SLA policy starts its clocks when the ticket is made." First-response times in the other reports always count from when the ticket was made, so they won't match this tab if a policy starts at the first message.
Export
Export downloads the missed-conversation list as CSV or Excel for the same range and filters, and respects the clock filter on the table.