AI Chatbot Guide v1
AI Chatbot
Version 1

Analytics

Copy link

Analytics is a powerful feature that provides a snapshot of your bot's performance, user interactions, and feedback during a 14-day period of your choice. It helps you understand how users are engaging with your bot, identify areas for improvement, and optimize your bot's performance.


Bot activity

Copy link

This section provides an overview of your bot's activity, including the number of active bot channels, total bot messages, and average bot messages per channel.

NameDescription

Active bot channels

Number of channels where your bot is actively engaging in. Active means the bot sent at least one message in that channel.

Total bot messages

Total number of bot message sent across all channels within an application.

Average bot message per channel

Average number of bot messages sent per channel.

User feedback

Copy link

This section provides insights into user feedback on your bot's responses, including total feedback count, feedback rate per message, and feedback breakdown.

NameDescription

Total feedback

Total count of user feedback on bot response. Thumbs up count as helpful, and thumbs down count as unhelpful.

Feedback rate per message

The ratio of bot messages to user feedback, indicating how often users give feedback to the bot's message.

Feedback breakdown

Shows whether user sentiment towards the bot's performance is mostly positive (helpful) or negative (unhelpful).

Performance highlights

Copy link

This section highlights the top function calls and knowledge base sources used in your bot's responses.

NameDescription

Top function calls

The five most frequently used function calls in bot responses.

Top knowledge base

The five most frequently retrieved knowledge base sources in bot responses, including file and URL.

Message credit usage

Copy link

This section provides the cumulative amount of message credits used on a daily basis.

The message credit consumption for each LLM model is as shown below:

ModelMessage credit consumption

GPT-3.5-turbo

1 message credit / bot message

GPT-4-turbo

10 message credits / bot message

Claude 3

3 message credits / bot message

Llama 2 70B

1 message credit / bot message

Solar

1 message credit / bot message

  • Bot message = All bot response messages, including the welcome message and suggested replies. Your message credit quota is for the organization. It cumulates all message credits used for all applications within your organization.

Top function calls

Copy link

This section provides insights into the most frequent function calls used in your bot's responses.

NameDescription

Function call success rate

The percentage of successful and unsuccessful function calls. For further information on possible error codes, see this page.

Functions to response ratio

The ratio of functions called to bot response, indicating how often function calls are triggered in your bot responses.

User feedback for function calls

Copy link

This section provides insights into user feedback on your bot responses using function calls, including total feedback count, feedback rate per message, and feedback breakdown.

NameDescription

Total feedback

Total count of user feedback on bot response. Thumbs up count as helpful, and thumbs down count as unhelpful.

Feedback rate per message

The ratio of bot messages to user feedback, indicating how often users give feedback to the bot's message.

Feedback breakdown

Shows whether user sentiment towards the bot's performance is mostly positive (helpful) or negative (unhelpful).

To learn more, see Sendbird's AI chatbot product page.