Tag the topic, one per ticket
The most useful tag answers “what was this about?”: billing, login, shipping, bug, feature request. Pick one topic tag per ticket, so counts add up to the total. Start with ten or fewer.
Add a few flags for what is special
Separate from topics, keep a short list of flags: vip, refund, escalated, outage-2026-10. Flags may overlap on a ticket. Keep them to cases you will actually report on.
Write the rules down in one place
Each tag gets a name in lowercase with hyphens, and a one-line definition. “billing: questions about invoices, charges and plans. Refund requests get both billing and refund.” Put it where agents can see it.
Let automation do the first pass
A rule that tags by keyword, or an AI that reads a new ticket and adds tags your team already uses, saves the agent a click and makes tagging more consistent. An agent still fixes the wrong ones.
Review once a month
Look at the tag list. Merge near-duplicates (“login” and “log-in”), rename vague ones, and remove tags used fewer than a few times. Many help desks let you fold two tags into one across every ticket, so cleaning up costs minutes.
Use them
Count tickets per tag each week, watch the top five, and ask what would make each smaller. Tags nobody reports on are noise. Tags that drive a weekly decision pay for themselves.