Chatbot Metrics

Understand how your AI chatbot is performing: how many conversations it handled, how often it escalated to a human, and how many leads it produced.

How to get there: Go to Analytics in the top menu → select a chatbot from the widget selector.

Core Metrics

Conversations Started

A conversation is counted each time a visitor sends the first message to the chatbot. Returning visitors who continue an old session do not count as a new conversation.

Messages Sent

The total number of messages exchanged in chatbot conversations (visitor + bot). A high messages-per-conversation number usually means the bot is doing useful back-and-forth.

AI Credits Used

Each AI response consumes credits based on the model and message length. Compare this against your plan's monthly credit allowance.

Escalation Rate

The percentage of conversations that were handed off to a live agent. A high escalation rate may signal:

  • Missing or thin training content
  • The bot lacks confidence on common questions
  • Visitors prefer talking to humans (which is fine for some businesses)

See Choosing Training Sources to improve coverage.

Lead Conversion Rate

The percentage of conversations that produced a captured lead (email, phone, or other contact field). This is the headline metric for most businesses.

Knowledge Quality Score

Boei automatically scores each conversation on whether the bot answered the visitor's question. Conversations are grouped into:

  • Answered well — bot resolved the question confidently
  • Partially answered — bot gave incomplete information
  • Could not answer — bot did not know the answer

Use this to spot training gaps. Drill into "Could not answer" conversations to see what topics need new training content.