Understanding Topics & Intents
Smart Tagging automatically labels every conversation and ticket with a Topic (broad category) and an Intent (specific request). This gives you a clear, consistent view of what customers ask and how your agent performs for each type of issue.
Key definitions
| Term | Definition | Example |
|---|---|---|
| Topic | A broad category of issues | Billing, Orders, Account |
| Intent | A specific customer request within a topic | Reset password, Cancel order |
Example: A customer asking “Can I change my delivery address?” might be tagged as Topic = Shipping and Intent = Change delivery address.
Why smart tagging matters
Clear tags make it easier to:
- See your top issues without manual sorting.
- Compare CSAT, audit scores, and AI resolution by issue type.
- Prioritize the highest impact Responses or Workflows to improve.
- Keep reporting consistent as volume grows.
How smart tagging works
Applied uses three steps to keep tags accurate and up to date:
-
Clustering (build the starting taxonomy)
Applied groups similar customer questions and suggests Topics and Intents. If you already have a taxonomy, add it first so tagging stays familiar. Descriptions and examples help the model understand how you define each topic and intent. -
Tagging (apply labels to new conversations)
For each new conversation or ticket, Applied selects the best fit Topic and Intent and attaches them to the record. Tags show up in the Inbox and are available in Analytics, Audits, and Test Coverage. -
Gap fill (capture new issues)
When a conversation does not fit any existing intent, Applied creates a new intent under the closest topic. This keeps new or emerging issues from being lumped into a generic bucket. You can review, rename, merge, or remove these intents anytime.
Where tags appear
- Inbox and Tickets: See tags on each record, filter by topic or intent, and spot spikes in a specific issue type.
- Analytics: Compare volume, CSAT, and AI resolution by topic or intent.
- Audit & QA: Focus audits on high volume or low quality intents.
- Test Coverage: Organize scenarios and coverage by intent so you know what is tested and what still needs attention.
Managing your taxonomy
Go to Settings -> Workspace -> Topics & Intents to:
- Create topics and add intents under each topic.
- Edit names, descriptions, and example queries to improve accuracy.
- Merge duplicates to keep reporting clean.
- Bulk add topics and intents with a CSV (Topic,Intent).
If you already use a taxonomy today, start by importing or creating those topics and intents so your reporting matches how your team already talks about customer issues.
Tips for higher quality tagging
- Keep topics broad and easy to understand; make intents specific.
- Avoid overlapping intent names such as “Refund request” and “Return refund”.
- Add a few example queries for each intent to teach the model your wording.
- Review new intents on a regular cadence and merge or rename when needed.