Skip to main content
All CollectionsUse cases
How to analyse my AI chatbot’s performance on Enterpret?
How to analyse my AI chatbot’s performance on Enterpret?
Team Enterpret avatar
Written by Team Enterpret
Updated over 2 weeks ago

Are you using an AI chatbot for customer support? While these bots are powerful, they're not perfect. Enterpret can help you understand exactly how your chatbot is performing and identify clear opportunities for improvement.

Before creating your analysis dashboard, two important setup considerations:

  1. Identify Your Chatbot: Most AI chatbots operate as a seat within your helpdesk. Look for identifiers like agent email or specific metadata fields that distinguish bot from human interactions.

  2. Track Performance Tags: Most probably you already have a set of tags that either the AI bot or your human agents are adding to the tickets. If you don’t we recommend following the structure given below

Recommended Performance Tags:

  • Rick - Knowledge Gap ❓: Shows when bot couldn't answer due to missing information in its knowledge base

  • Rick - Ignored ⛔: Indicates when bot's didn’t respond to the customer's query

  • Rick - Handed Off 🔄: Tracks when bot appropriately transferred conversation to a human agent

  • Rick - Resolved ✅ : Marks conversations where bot successfully addressed the customer's need

  • Rick - Out of Scope 🎯: Identifies queries that fall outside the bot's intended capabilities

  • Rick - Replied 💬: Shows when bot provided an initial response to customer query

  • Rick - Escalated ⚡: Flags when human intervention was needed after bot attempted to help

  • Rick - Incorrect Path 🚫: Indicates when bot took wrong conversational direction with customer

  • Rick - Pending Resolution ⏳: Tracks conversations requiring additional follow-up or resolution

  • Rick - Needs Improvement ⚙️: Marks responses that were relevant but could be enhanced

  • Rick - Training Required 📚: Highlights areas where bot needs additional pattern recognition or responses

  • Rick - Perfect Response ⭐: Identifies ideal bot interactions to use as best practice examples

Pro Tip: Replace "Rick" with your chatbot's name in these tags and add them to your helpdesk if you don’t have similar tags already.

Now, let's build a dashboard that helps us

track:

  • Overall bot effectiveness

  • Knowledge gaps

  • Resolution rates

  • Areas needing improvement

  • Training opportunities

Step by Step Instructions

Create a Dashboard

  1. Let’s create a dashboard called “Rick ( Replace this with your own AI bot’s name) Performance Analysis”

  2. Start adding the quantify analysis listed below to put together a report. Use text block to add the section titles

Adding Quantify Charts

  1. Click on Add Quantify on the Dashboard

  2. One by One do the analysis mentioned below and add it to the dashboard

Quantify Analysis to Add

1. Bot Activity Overview

Let’s start by putting together a chart to tracks overall bot usage. This will also help spot any system issues or unexpected downtime.

Total Conversation Volume

Plot: Source 

Filter: Agent Email = [Bot Email]

View: Trend + Anomalies

Rick Tags All Time

Plot: Tags ( from your helpdesk metadata) 

Filter: Agent Email = [Bot Email]

View: Bar chart

Rick Top Reasons

Plot: Reasons 

Filter: Agent Email = [Bot Email]

View: Bar chart

Rick vs Humans

Plot: Reasons 

Use Compare
Filter A: Agent Email IS ANY OF [Bot Email]
Filter B: Agent Email ISN'T ANY OF [Bot email]

View: Bar chart

2. Resolution Success

Resolved vs Handed-Off Patterns

Shows which topics your bot handles well vs needs human help with.

Plot: Reasons 

Use Compare
Filter A: Agent Email = [Bot Email] AND Tag = "Resolved"
Filter B: Agent Email = [Bot Email] AND Tag = "Handed Off"

View: Bar Chart

3. Knowledge Gap

Topics Needing Training

Identifies areas where bot needs knowledge base updates.

Plot: Reasons 

Filter: Agent Email = [Bot Email] AND (Tag = "Knowledge Gap" OR Tag = "Training Required")

View: Bar Chart

4. Escalation Analysis

Incorrect Handling

Shows where bot is misunderstanding user intent or needs better routing.

Plot: Reasons 

Use Compare
Filter A: Tag = "Incorrect Path"
Filter B: Tag = "Escalated"

View: Bar Chart

5. Quality Tracking

Helps track if your improvements are increasing quality responses.

Perfect vs Needs Improvement

Plot: Reasons 

Use Compare Filter A: Tag = "Perfect Response"
Filter B: Tag = "Needs Improvement"

Duration: Last 3 months

View: Trend Chart Switch to table view on Dashboard

6. Scope Analysis

Topic Coverage

Shows which new topics you might want to train your bot on.

Plot: Reasons 

Use Compare
Filter A: Tag = "Out of Scope"
Filter B: Tag = "Resolved"

View: Bar Chart

7. Response Issues

Ignored vs Replied

Identifies potential bot responsiveness issues.

Plot: Reasons 

Use Compare
Filter A: Tag = "Ignored"
Filter B: Tag = "Replied"

View: Trend

Pro Tip: Set up regular dashboard subscriptions to track these metrics weekly. Look for topics that move from "Knowledge Gap" to "Resolved" to validate your improvements.

Did this answer your question?