Conversation tagging· Support operations
Conversation Tagging — Labels That Make Chats Countable
Quick answer: A tag is a short label, such as refund or shipping_delay, attached to a conversation. One chat is just a story; a thousand tagged chats tell you what customers keep asking about. Tagging answers three questions: which chats belong together, which ones need a particular team or follow-up, and what is changing week to week. It only works when the tag list is short and everyone uses the same words.
How tags get applied
| Method | How it works | Good for | Watch for |
|---|---|---|---|
| Agent tags by hand | The agent picks tags during or after the chat | Nuance only a person sees | Skipped when the queue is busy; different agents pick different tags |
| Rules and workflows | A rule adds a tag when conditions match, such as channel or message text | Consistent, simple facts | Only as good as the conditions you write |
| Keyword auto-tagging | The platform adds tags from words in the message | Volume without effort | Common words create noise tags |
| AI suggestions or classification | A model suggests or applies tags from the conversation | Large volumes, varied wording | Needs checking against real chats before you trust it |
Most teams combine them: rules for facts the system already knows, agents for judgement, and automation to catch what agents miss.
What the big platforms document
- LiveChat. Admins and the owner create tags in Settings, with names up to 70 characters, and can limit a tag to one group so only that group's agents see it. Agents add tags from the bottom of a chat in Chats or Archives, can pick several at once and pin frequent ones. An optional reminder asks agents to tag a finished chat when they archive it, and once at least 100 chats are tagged, agents start getting tag suggestions on the Team, Business and Enterprise plans. Archives and reports can be filtered by tag, and a Tags usage report shows the most common topics. See LiveChat.
- Zendesk. Tags are words or word combinations added to tickets, users and organizations; a user's or organization's tags are copied onto their tickets. Only letters and numbers, dashes, underscores, colons and forward slashes are allowed, and multi-word tags need a connector, so
late_orderandlate-ordercount as two different tags. Automatic ticket tagging adds tags from keywords in the ticket description, and Zendesk warns that common words such as "and" or "the" make it much less useful. Tags drive views, reports, macros, triggers and automations. See Zendesk. - Intercom. Intercom separates three tools: topics (keyword-based, automatic, can be applied to past conversations, built for trend reports), conversation attributes (structured fields such as a predefined list, usable as required fields before closing) and tags (the most flexible, applied to individual replies by teammates or by Workflows). Its rules article shows a Workflow adding a tag when a message contains words like "bug" or "error". Intercom recommends using all three together. See our Intercom review.
Tags, fields and topics: the difference
A tag is free-form and quick to add, which is its strength and its weakness: nothing stops five spellings of the same idea. A structured field (Zendesk ticket fields, Intercom conversation attributes) offers a fixed list, so every chat gets exactly one reason from the same menu, which makes cleaner reports. A topic in Intercom's sense is applied by keyword rules across all conversations, including old ones. A practical split: use a required field for the single main contact reason, and tags for extra flags such as vip, outage_nov or bug_reported.
Why tagging goes wrong
- Too many tags. Agents scroll past a long list and pick whatever is near the top, or nothing.
- Duplicates and spellings.
refund,refundsandrefund_requestsplit one reason into three report lines. - Tagging in the rush. Tags added from memory after a busy shift are guesses.
- No owner. Nobody removes old tags, so the list grows forever.
- Tags nobody reads. A tag that never appears in a report or view is effort with no return.
The cure is a short list with written definitions, one owner, and a monthly look at what got used. No source we read gives an ideal number of tags, so judge by whether agents can choose quickly and reports stay readable.
How tags connect to chatbots
Tags matter more once a bot handles part of the traffic. A tag added when the bot hands over, such as bot_handoff, shows which questions it could not finish, and tags on bot-only chats show what it resolves alone. Tags can also feed chat routing: a rule that tags billing can send the chat to the billing team. For reading the conversations themselves, see chat transcripts and the conversational analytics guide.
FAQ
What is conversation tagging in live chat? Adding short labels to chats so they can be filtered, routed and counted later, for example every chat about refunds or every chat affected by an outage.
Who adds tags to chats? Agents by hand, rules that add a tag when conditions match, or keyword and AI auto-tagging. Most teams use a mix.
What is the difference between a tag and a ticket field? A tag is a free-form label you can add several of; a field offers a fixed list and usually holds one value. Fields give cleaner reports for the main contact reason; tags suit extra flags.
Can chats be tagged automatically? Yes, on many platforms. Zendesk tags from keywords in the ticket description, Intercom Workflows add tags by rules, and LiveChat suggests tags once enough chats are tagged. Check automatic tags against real conversations before relying on them.
How many tags should a support team use? No source we read sets a number. Keep the list short enough that an agent can choose in a few seconds, and remove tags nobody uses.
Are tags visible to customers? The vendor articles we read describe tags as internal labels for the team; none says customers see them. Do not put anything in a tag you would not want a customer to read, in case it is exported or shared.
Related terms
- Live chat — the channel most tagged conversations come from.
- Chat routing — sending chats to the right agent, often using tags.
- Chat transcript — the full record a tag summarises in one word.
- Intent recognition — how a bot classifies a single message.
- Chatbot analytics — the reports tags feed into.
- Sentiment analysis — automatic labels for tone rather than topic.
- Unified inbox — where tags from every channel meet.
Sources
- LiveChat, Tagging chats and tickets — livechat.com/help/tagging-chats-and-tickets/, read 14 November 2026. Source of who creates tags, the 70-character limit, group tags, the archive reminder, suggestions after 100 tagged chats and the Tags usage report.
- Zendesk, Using tags — support.zendesk.com/hc/en-us/articles/4408888664474, read 14 November 2026. Source of tags on tickets, users and organizations, automatic ticket tagging and tags in views, reports and workflows.
- Zendesk, Working with ticket tags — support.zendesk.com/hc/en-us/articles/4408835059482, read 14 November 2026. Source of allowed characters, multi-word tags and the common-words warning.
- Intercom, How and when to use conversation topics, attributes and tags — intercom.com/help/en/articles/7126365, read 14 November 2026. Source of the three-way split and the recommendation to combine them.
- Intercom, Tag conversations automatically with rules — intercom.com/help/en/articles/6512873, read 14 November 2026. Source of Workflow tagging by message content and channel.
- Chatbotscape evaluation methodology. /methodology (continuously updated).