📘

Part of AI Quality Monitoring. This page describes the Overview tab, screen by screen. For what each number means, follow the links to the metric pages.

Overview is the headline of your dashboard. It answers, in one screen: how the bot is performing, how AI compares to scripted, how much activity you have, and what users are talking about. AI-computed widgets carry an AI badge.

All three global filters — Languages, Channels and the date range — apply here, and the over-time charts each have their own Daily / Weekly / Monthly granularity control.

Conversation Flow

How do conversations travel through the system?

A Sankey diagram of the end-to-end journey: entry channel → topic → treatment → outcome. The legend splits flows into Automated / Handovered and Resolved / Unresolved. It's the fastest way to see where volume concentrates and where it leaks.

Drill-down: click any node or link to open the conversations behind that path, pre-filtered. This is AI-powered, so on a date range before 1 January 2026 it shows the "AI analysis unavailable" placeholder.

Health

How is the bot performing?

The KPI row, in two tiers:

• Primary: Resolution (%), CX Score (/10), Handover rate (%) — each with a trend badge.
• Secondary: Knowledge Gap (%), Satisfaction score (/10), Automation rate (%), Conversations (count).

Resolution, CX Score and Handover rate carry the AI badge; Knowledge Gap is AI-computed too.

📘

Automation rate is the "no-handover" number

If you're used to reading resolution as "the bot answered without a human", that view lives here, as Automation rate — not as Resolution. AI Resolution counts a correct handover as resolved; the two are deliberately different.

AI Impact

How does AI compare to scripted?

A usage donut showing the split between conversations handled by an AI step and conversations handled by scripted flows, followed by four side-by-side comparison cards: Resolution (%), Knowledge Gap (%, lower is better), CX Score (/10) and CSAT (/5). Each pair is labelled AI vs Scripted, with the sample size (n =) behind each bar. When a comparison doesn't have enough conversations to be trustworthy, the card says Not enough CSAT data instead of showing a number.

This is the section to open before deciding where to put a Gen AI step next — see How to improve your resolution → AI vs scripted. Read it as a signal, not a controlled experiment: AI steps are enabled on specific use cases, so the gap mixes "what AI does" with "where AI is deployed".

Volume

How much conversation activity?

A grouped bar chart plotting Conversations and Customers over time, plus tiles for Conversations, Customers, Messages / conversation and Avg. duration. No AI here — this is raw activity, and it covers your full history (no 2026 cutoff).

Louis

How is the AI agent performing?

Five KPI cards focused on the AI agent: Involvement (% of conversations Louis took part in), Resolution (/%), CX Score (/10), CSAT (/10) and Handover rate (%). Involvement carries the AI badge.

Drill-down: Involvement opens the conversations Louis handled; CSAT opens the negatively-rated ones; Handover opens the handed-over ones.

Topics

What are users talking about?

Two stacked blocks:

• Topics by volume — a treemap sized by conversation volume (each cell's tooltip shows conversations + CX /10), with an Aggregate / Over time toggle. Over time switches to a stacked area chart where you can compare up to five topics.
• Topic details — a sortable table: Topic, Volume, CX Score, Resolution, Knowledge Gap. This is the table to sort when you want to rank topics by any AI metric.

Both carry the AI badge. Drill-down: click a treemap cell or a table row to open that topic's conversations. For what to do next, see Topics and the How-to guides.