Support Ticket Triage

Triage is three decisions. The category says what the ticket is about and feeds reporting. The team says who should work on it. The priority says how soon. They are related but not the same: a billing question can be urgent, and a security issue can arrive labelled as a login problem.

Ask them as three separate Choice questions. Each returns one label, a probability for every label and a confidence value, so you can auto-assign confidently categorized tickets while still reviewing uncertain priorities.

Labels are plain text. To add a category, add an option and describe it. The third example uses an online store’s labels and teams with the same priority definitions, which is how most teams adapt triage to a new product line or region.

Describe what belongs in each label, not just its name. “Account access: login, password, 2FA or suspicious sign-in activity” sorts edge cases better than “Account”. Keep labels mutually exclusive, or the probabilities will split between them.

Angry wording does not make a ticket urgent, and calm wording does not make it routine. Define priority levels by business impact and time: security risk or outage, money or a deadline within a day, a problem with a workaround, a question. The instructions tell Jev to judge impact rather than tone.

The second example shows why this matters. A customer who cannot log in after an unrequested password reset is reporting a possible account takeover. Jev assigns it to security with urgent priority, although the subject line only says “Can’t log in”.

Run a few hundred past tickets through Batch with the labels your team actually used, and compare. Look at which labels are confused and whether the confusion is in the descriptions or the data. TypeSafe’s intent-routing pattern sends low-confidence requests to a person rather than forcing a label.

Start by auto-filling fields for agents to confirm, then automate the categories that match your labels reliably. Re-check when you add a product, a team or a label.