Support ticket tags: turning a busy inbox into a report

Updated 2026-09-24 · 5 min read · By the Convo360 team

What tags are actually for

Tags are not a filing system. Nobody goes looking for a single ticket by tag; they search by name or number. The only reason to tag at all is to answer a question at the end of the month: what was this month actually about?

That reframing changes what you tag. If the purpose is a monthly pattern, then tags must describe the customer's problem in the customer's terms. "Delivery delay" is a good tag. "Escalated to ops" is not, because it describes what you did rather than what happened.

The test is whether a tag could appear in a sentence the customer would recognise. "A fifth of our tickets last month were about delivery delays" is a real finding. "A fifth were escalated to ops" tells you about your own process and nothing about the business.

This also explains why tagging so often dies within a month. Teams introduce it as an obligation, the tags describe internal process, and no report is ever produced from them. Agents are not lazy; they correctly detect that the work has no consumer and stop doing it.

Designing a short list

Start with six to ten. A D2C brand might use:

• delivery delay
• damaged item
• wrong item
• return
• refund
• size or fit
• payment
• product question

That is enough to describe almost every ticket and short enough that agents remember it without reading a list. Resist adding a tag for every new situation; a tag used twice is noise in a report.

Review quarterly. If one tag is on sixty per cent of tickets it is too broad and should be split. If a tag has been used four times in three months, delete it.

Avoid tags that encode severity or ownership. Those already exist as priority and owner, and duplicating them in tags means two places to keep in step, one of which will rot. Tags should carry the information the other fields cannot.

If two tags are regularly applied together, that is usually a sign one of them should not exist. 'refund' and 'return' on the same ticket every time means your team thinks of them as one thing, and the report should too.

Keeping tags clean automatically

The failure mode of every tagging system is drift. One agent types "Billing", another "billing", a third "billing " with a trailing space. Your monthly report then shows three separate rows for one thing, each too small to notice.

Convo360 normalises this for you. Tags are lower-cased, trimmed and de-duplicated when saved, so those three become one. The tag box also suggests what your workspace has already used, ordered by how often, which nudges agents towards existing tags rather than inventing new ones.

There is a cap of twelve tags per ticket, which sounds generous and is deliberately so. A ticket with twelve tags has effectively none, and the cap exists to catch that rather than to constrain normal use.

In Convo360, tags also drive the worklist filter, so the list above your tickets shows every tag in use with its count. That count is doing quiet work: a tag with three uses in two months is visible as a mistake without anybody having to audit anything.

Reading the monthly pattern

Open the worklist, filter by each tag, and write down the count. Five minutes of work, once a month. What you are looking for is not the biggest number but the biggest number you can do something about.

In each case the fix is outside support. That is the point of tagging: the support queue is often the earliest and cheapest sensor a business has for problems that started somewhere else.

Cross the tags with other numbers once you have three months of them. Satisfaction score by tag tells you which problems make customers angriest rather than which are most common, and those are rarely the same. Reopen rate by tag tells you which problems your team cannot actually fix, which is often a resourcing question rather than a support one.

Making tagging actually happen

Tags fail when they are optional and unrewarded. Three things make them stick.

First, keep the list short enough to remember. Second, tag at resolution rather than at creation, because that is when you actually know what the ticket was about. Third, use them visibly. If the team never sees a report built from their tagging, they will correctly conclude it does not matter and stop.

A five-minute item in the weekly review, showing the tag counts and one action taken because of them, is usually enough to keep the habit alive. It also surfaces disagreements about what a tag means early, while they are cheap to fix.

Do not backfill months of old tickets. The value is in the trend from here, and an afternoon spent tagging history is an afternoon not spent on the problem the tags just revealed.

Keep one person as the owner of the tag list. Shared ownership of a vocabulary means no ownership, and within a quarter you will have both 'delivery delay' and 'late delivery'. One person pruning quarterly is enough to keep it honest, and the job takes about twenty minutes.

Step by step

1. Write six to ten tagsDescribe the customer's problem, not your internal process. 'delivery delay', not 'escalated to ops'.
2. Tag at resolutionApply one or two tags when the ticket is resolved, which is when you actually know what it was about.
3. Use the suggestionsThe tag box in Convo360 suggests what your workspace already uses, ordered by frequency. Pick rather than type.
4. Filter and count monthlyFilter the worklist by each tag, note the counts, and look for the largest number you can act on.
5. Take one action and show itFix one upstream cause each month and report it back to the team, so tagging visibly matters.

Frequently asked questions

Related reading

Available across India — see WhatsApp & AI calling automation by state and city.