Ticket lifecycle
The states a ticket moves through, who moves it, and what happens at each step.
States
| State | Meaning | Set by |
|---|---|---|
| Open | Needs someone to act | Created, or a customer replied |
| Pending | Waiting on the customer | Agent |
| Resolved | Done | Agent, or the AI when it fully answers |
There is no separate "closed". Resolved is terminal until a customer replies, which reopens the ticket automatically.
What happens when a ticket is created
- Spam and virus checks — for email, anything failing is dropped before it reaches your queue.
- Deduplication — the same inbound message arriving twice creates one ticket, not two.
- Contact matching — the sender is matched to a contact, or a new one is created.
- Intent classification — the AI labels the request and assigns a team.
- SLA clock starts — first-response and resolution targets are stamped on.
- Assignment — an agent is picked by round robin, unless routing is off.
- Autopilot — the AI answers, drafts, or escalates according to your settings.
All of this happens server-side within a second or two of the message arriving.
Priority
Priority drives the SLA targets. New tickets start at medium. Raise it manually, or let SLA breach escalation raise it for you — a ticket that blows its first-response target is escalated and reassigned, on the assumption that the current owner is not on it.
Reopening
A customer replying to a resolved ticket sets it back to open and marks it unread. The SLA clock does not restart from zero; the ticket's history is continuous, which is what you want when measuring whether you actually resolved something the first time.
Resolving well
- Resolve when the customer's problem is solved, not when you have replied
- Use Pending while waiting on them, so your open queue reflects real work
- Leave a note if the resolution is unusual — the next agent inherits your reasoning
Related