Analytics
Chats, visitors, feedback, topics and sentiment, compared with the period before.
On this page
Analytics shows what your agents are doing and how customers feel about it. Analytics in the sidebar covers every agent together; each agent's own Analytics tab covers just that agent.
Choose 7, 30 or 90 days. Every number is compared with the period before, so you see the trend, not just the total. Playground tests are never counted.
The numbers
| Metric | What it tells you |
|---|---|
| Chats | Conversations started |
| Messages from visitors | How much people are asking |
| Thumbs up / down | How visitors rated replies (turn the thumbs on in the widget's settings) |
| Visitors (daily, weekly, this month) | How many different people used the chat |
| Stickiness | Daily ÷ monthly visitors: how often people come back |
| Chats over time | The daily trend against the previous period |
| Channels | Where conversations happen: widget, help page, messaging apps… |
| Actions | Which actions ran most |
Today's numbers are counted live. Earlier days are summarised nightly, and those summaries are kept even after the conversations themselves are deleted by your retention setting.
Topics and sentiment
Once a conversation goes quiet, ikazebot reads it once and notes:
- Topic: what the visitor wanted, for example "Refunds" or "Delivery times".
- Sentiment and emotion: how they felt, for example Grateful, Confused or Frustrated.
This reading costs you no credits.
You can shape the topics to match how you think about your business:
- Rename a detected topic.
- Merge into… to combine duplicates ("Refund" and "Money back").
- Add a topic you want tracked.
- Delete one that is not useful.
- Freeze topics to stop new ones being created, so every conversation is filed under your list.
Helpdesk numbers
If your team answers conversations, the Helpdesk section shows, per member: conversations replied, replies sent and solved. It also shows median first response, median time to solve, conversations created and still open, and your CSAT (satisfaction) score from the rating survey.
Using analytics to improve
- 1Spot a rising topic or many 👎
- 2Read those conversationsin the Inbox
- 3Fix knowledge or instructions
- 4Watch the trend next period
- Many thumbs down on one topic: the knowledge for it is missing or unclear. Add a Q&A pair.
- One action dominates: maybe a page on your site should answer that directly.
- Frustration rising: check whether the agent hands over to a person soon enough.
Related
Something not working as described here? See Troubleshooting.