Plan requirement
| Subscription | Suite Team or higher, Support Team or higher |
| Access | Admin |
Where the AI agent numbers live, what to look at first, and how to avoid reading a rise in traffic as a success.
Find it
Reporting in the left sidebar of the AI agents workspace. It reports on what the agents handled without any building, and refreshes hourly.
Four tabs: Overview, Contact reasons, Custom resolutions and Conversation journey.
Look at these first
- Resolution share, against last month. On Overview, alongside understood conversations, assisted escalations and the bot satisfaction score.
- Where handovers happen, early or late in the conversation. That is the Conversation journey tab, which shows the paths and where they end.
- Which subjects the agent handles, and which it consistently passes on. The Contact reasons tab, with how each use case performs and how your knowledge sources are doing.
The third one is the useful one
Subjects consistently handed over are your work list. Either an article is missing, or one exists and is not usable, and both are fixable in an afternoon.
That list is the difference between an agent that gets better each month and one that plateaus.
Set every number against the period before
A month in isolation is not information. Resolution counts rise with traffic, so a busy month looks like progress even when the share fell.
Split by channel and language
Aggregate figures hide a channel that is performing badly, and a language with almost no content will drag averages without ever being visible.
Then read some conversations
The dashboard shows where to look; the transcripts show what happened. Working from numbers alone leads to changing settings when the cause is an article.
Share it monthly
Resolution share, handover rate, and the subjects to document next. Three lines, sent to whoever decides where effort goes, is what keeps the knowledge work funded.
Comments
0 comments
Article is closed for comments.