Use Cases reads quality along the axis of what your bot is built to do — each use case you designed in the bot builder — rather than along the topic the user raised. It's where you see which use cases trigger, and how well each one resolves. AI-computed widgets carry an AI badge.

All three global filters apply, plus per-chart Daily / Weekly / Monthly granularity.

Overview

How are use cases distributed?

Four KPI tiles — Use cases volume, Deflection tags, Handover requests, Misunderstanding — and a stacked area chart, Use Case vs Louis distribution, splitting conversations into Use Case only / Louis only / Both. That split is the quick read on how much of your traffic the generative agent is now touching versus scripted use cases.

Metrics over time

How do use cases perform?

A bar chart with a metric toggle — Volume / CSAT / CX Score / Resolution rate — and a selector for up to five use cases at once. This is where you watch a specific use case move after you change it: pick the metric, pick the use case, compare before and after.

Detailed performance

How does each use case compare?

A sortable table in two column groups:

• Use case: Name, Conversations, CSAT
• Conversation-level: Resolution, CX Score, Handover %

The conversation-level columns are the AI metrics — Resolution and CX Score — measured on every conversation the use case appeared in, not only the voted ones.

Drill-down: click a row to open that use case's conversations, pre-filtered.

Needs attention

Which use cases need improvement?

A shortlist of use cases flagged automatically — tagged Low CSAT, Low CX Score, Low resolution or High handover, each with its mini stats. This section only appears when there's something to flag; a clean bot shows nothing here, which is the good state. When it does appear, it's your ready-made worklist — pair it with How to improve your resolution.