Analyze History and Improve Your Chatbot

Discover how to use past conversations to understand your users and perfect your assistant’s responses.

Introduction

Your DoxyChat chatbot is not static: it must evolve alongside your visitors’ needs. The History tab is a goldmine of information. It allows you to see exactly what your users are asking, how the bot responds, and most importantly, to identify “gaps” in your knowledge base.

1. View Conversations

To access the exchange log:

1. Go to your Dashboard.
2. Select the relevant chatbot.
3. Click on the History tab.

You will see a chronological list of all chat sessions appear. Each line presents a summary: the date, the first question asked, and the duration of the exchange.

Click on a session to open it. You can then reread the entire dialogue as it unfolded, see the exact sources that the AI used for each response, and understand the context of each question.

2. Understanding Feedback (Likes / Dislikes)

In the chat interface, your users have the ability to rate each AI response by clicking on a Thumbs up (👍) or a Thumbs down (👎).

These signals are crucial for your maintenance:

• 👍 Positive: The user is satisfied. The response was relevant and the source correct. This is an indicator of your bot’s good health.
• 👎 Negative: The user did not obtain the desired information. This may mean that the response was wrong, incomplete, or that the chatbot responded “I don’t know.”

Tip: In your history dashboard, you can filter conversations to display only those containing Negative Feedback. That’s where you should start to improve your bot quickly.

3. Analyze to Improve (Improvement Cycle)

The goal of analysis is not just to read, but to act. Here are three common scenarios and how to remedy them:

Case A: The Chatbot Responds “I Don’t Know”

• Diagnosis: The user asked a legitimate question, but the information is not in any of your imported documents.
• Action: Create a small text or Word file with the answer to this question and import it into the Sources tab.

Case B: The Chatbot Gives Outdated Information

• Diagnosis: The user asks for a price, the bot gives the old 2023 rate because it is based on an old PDF.
• Action: Go to the Sources tab, delete the outdated document and replace it with the updated version.

Case C: The Chatbot Misunderstood

• Diagnosis: The question was ambiguous or your documents are contradictory.
• Action: Use Custom Intents (Intelligence tab) to force a specific response to this type of question, or rephrase the relevant passage in your source documentation to make it clearer.

By dedicating a few minutes per week to this analysis, you will transform your chatbot into an infallible expert.