Website Chatbots
Reviewing chats and closing knowledge gaps
The dashboard figures, reading conversations, and the report that tells you what to write next.
Last updated September 2, 2026
A chatbot left alone gets worse. Half an hour a month reading what it was asked is what turns it from a novelty into a lead source.
The dashboard
| Figure | Tells you |
|---|---|
| Visitors | How many people talked to it. |
| Enquiries Submitted and Enquiry Rate | How many turned into a lead. |
| Avg Messages / Chat | How much of a conversation it holds. |
| Answer Rate | How often it could actually answer. |
| Positive Sentiment | How people felt about it. |
There are also panels for the most common requests and the top tribes over the last 30 days — a free summary of what your customers are asking about.
Reading conversations
The Chats tab lists every conversation with its tribe, message count, sentiment, whether it was answered and whether it produced an enquiry. Open one to read it in full. Test conversations are marked, and spam is hidden unless you ask to see it.
Knowledge gaps are the useful screen
This lists the questions your chatbot could not answer, with the topic and the tribe. It is a to-do list for your website, written by your customers.
What it costs
Chatbot conversations use AI, billed through your AI token allowance like everything else. The analysis that produces the sentiment and knowledge gap figures is billed too — a small cost, but it means a very busy chatbot contributes to your AI usage in two ways.