TAU Platform

Analytics

Automatic reports produced by Leo for each conversation — sentiment, quality, resolution, and more.

Analytics

The Analytics section contains the reports that Leo automatically produces for each of your assistant's conversations. Without you having to manually read a single message, Leo evaluates the entire interaction — customer sentiment, problem category, response quality, whether it was resolved or not.

Access it from the sidebar under Analytics.

💡 Leo plan feature

Automatic analytics are part of the Leo plan. If you see a locked screen, reach out at meajuda@taubot.ai to learn about plan options.


What Leo analyzes in each conversation

For each conversation, Leo generates a complete report with:

FieldWhat it shows
SummaryOverview of what was covered in the conversation
CategoryType of interaction (sale, support, question, etc.)
SubcategoryBreakdown of the category
SentimentWhether the customer was positive, neutral, or negative
ScoreService quality rating (0–10)
PriorityWhether that conversation deserves team attention
ResolvedWhether the customer's goal was met
TransferredWhether human intervention was needed
TagsKeywords identified automatically
Improvement suggestionWhat could have been done better

In addition to the standard fields, you can define custom criteria — specific questions that Leo answers for each conversation according to your business needs.


How to use it in practice

The Analytics dashboard presents all evaluated conversations in a table. You can:

  • Filter by period, sentiment, category, score, or status
  • Sort by any column
  • Customize which columns appear — useful for focusing on what matters to your business
  • Open any conversation to see the full history alongside Leo's analysis

Start with the extremes: filter by low score to identify where the assistant is falling short, and by negative sentiment to find customers who deserve your team's attention.


Custom criteria

This is the most powerful part. You define questions and Leo answers them for every conversation automatically. Examples:

  • "Did the customer show purchase intent?"
  • "Did the assistant mention the loyalty program?"
  • "Did the conversation result in a booking?"
  • "Did the customer complain about price?"

The answers appear as extra columns in the table — making it possible to track business-specific metrics at scale, without manual review.


Configuring analytics

Go to Settings → Evaluations to:

  • Enable or disable automatic analytics
  • Include conversations without AI — to evaluate purely human interactions as well
  • Continue after the plan limit — additional analytics generate extra charges
  • Add custom criteria

🚀 Quality at scale, without manual effort

With dozens or hundreds of conversations per week, reading everything is impossible. Analytics solves this:

  • You know the overall service quality without opening a single conversation
  • You identify patterns — recurring topics, critical hours, profiles of dissatisfied customers
  • You detect opportunities — conversations with purchase intent that weren't converted
  • You track quality evolution over time

It's like having a quality analyst reviewing every interaction, 24 hours a day.

On this page