What counts as a crisis?
A request is a crisis when the student's message trips the crisis signal listPhrases that indicate danger to the student themselves or to others, used to flag a request as a crisis.: phrases that indicate danger to themselves (self-harm or suicide) or to others (threats toward people or the institution). It is not a judgement you have to make in the moment; the system flags it.
That last point is the one worth internalising. You are not the detector. The crisis level does not depend on you reading a message carefully enough to catch the danger in it, which means a heavy inbox or a tired afternoon cannot cause a crisis to be missed.
It's detected, not assigned
Detection runs twice: in the student app as the message is typed, and again on our server when it is sent. The server has the final say and can only escalate, never downgrade. So a crisis is caught even if the student's app missed it or the priority was tampered with.
That is why you should trust the crisis flag. If a ticket is marked crisis, it met a real threshold, and it pins to the top of your inbox for that reason.
The signal list is tuned to over-detect rather than miss something, so some crisis tickets will turn out to be false alarms. That is intended. A false positive costs you a careful look at a request that was fine; a false negative costs a student who needed help and did not get flagged. The system is built to make the cheaper mistake.
Crisis versus urgent versus normal
- Normal is the default: a help request that needs a timely, human reply.
- Urgent is an intermediate priority the system can carry, but there is no console action to set it today. For requests that need faster attention, triage with status and assignment instead.
- Crisis is the danger threshold, and the only level that pages people in real time: email, plus Slack or Teams where your institution has connected them.
The clearest way to hold these apart: normal and urgent are about how soon a reply is needed, while crisis is about danger. Urgent is a priority the system can carry, not a label you set from the console; crisis is a detected threshold the system applies, and only crisis reaches out and pages a human in real time.
What it means for you
When a crisis lands, the student is already being shown immediate-support resources in the app. Your job is the human follow-through: acknowledge the ticket so the team sees it is handled, then follow your institution's crisis procedure.
Because the list over-detects, you will sometimes open a crisis ticket that turns out to be fine. Respond to it as genuine until you have confirmed otherwise. Confirming a false alarm costs minutes; assuming one and being wrong does not have an undo.
The full path, from detection through every alert channel to what you see in the inbox, is in What fires when a crisis is detected.
Common questions
Can I mark a request as crisis myself?
Yes. If a ticket reads as crisis to you but the detector didn't flag it, open the ticket and click Mark as crisis (with an optional reason, saved to the audit log). The escalation pages your team the same as auto-detection would: email, plus Slack or Teams where your institution has connected them. Priority can only ever move up, whether automatic or staff-initiated, never down. See the limits of automated detection for when this action is the right one to reach for. For requests that need faster attention but are not crisis, there is no console action to set urgent today; triage with status and assignment instead.
What makes crisis different from urgent?
Urgency is about how soon a reply is needed; crisis is about danger. Urgent is an intermediate priority the system can carry, with no console action to set it today, so triage with status and assignment instead. Crisis is a detected threshold the system applies, and it is the only level that pages people in real time: email, plus Slack or Teams where your institution has connected them.
This crisis ticket looks like a false alarm. What do I do?
Respond to it as real until you have confirmed otherwise. The list is tuned to over-detect, so some flagged requests will be fine. The cost of looking at a false positive is far lower than the cost of dismissing a real one.
Related
The fastest answer is usually one question away.