Plan requirement
| Subscription | Suite Growth or higher |
| Also required | Copilot |
| Access | Agent |
Agents see the labels and can correct them. Those corrections are the cheapest quality improvement available.
What appears on the ticket
- The topic, as a field.
- The language.
- The sentiment.
- Any entities that were extracted.
They sit alongside the ordinary fields, and they can be changed like any other.
Correcting a wrong label
Change the field. The ticket is then routed and reported correctly from that point, which matters for anything downstream that reads it.
Why the corrections matter beyond the ticket
A pattern of corrections on one topic tells you the topic description is wrong, or that two topics overlap. That is the signal you would otherwise have to find by sampling.
Ask agents to correct rather than to work around, and check the pattern monthly.
Tell them what it is for
Agents who do not know why a field is there tend to ignore it. Explaining that it routes work and shows what the team is being asked about turns it into something they will maintain.
Sentiment is context, not instruction
A ticket marked negative is a hint to read carefully, not an instruction to escalate. Sentiment on short factual messages is frequently wrong in both directions.
They do not have to use it
Triage adds information. An agent who handles the ticket well without looking at the labels has done nothing wrong, and the value is mostly in routing and reporting rather than in the agent's own reading.
Comments
0 comments
Article is closed for comments.