Analytics

Open Analytics in your agent's menu and pick a period: last 7, 30 or 90 days, this month, last month, the last 12 months or a custom range. Each number is compared with the period of the same length just before it. Playground and test chats are never counted.

Chats

• Total conversations and total messages, with a chart over the period.
• Answer rate, Customer rating, Handed to team (chats the AI passed to a person), Leads (collected in chats) and Credits used.
• Thumbs up / down on AI replies.
• Active customers — daily and weekly active users, and Stickiness (daily ÷ monthly active customers: how many of the month's customers come back on a typical day). The footer splits customers into new and returning. A signed-in customer counts once across devices and chats; WhatsApp and other channel customers count by their number or account.
• Chats by channel, Chats by country (a world map from the visitor's connection) and Actions used.

Click View these conversations › to open the matching chats.

Topics

Every chat gets a short topic (Refund request) and a broad category (Billing) a couple of minutes after it goes quiet. The treemap shows categories sized by the number of chats, with their topics inside; ⚠ marks topics where many customers were unhappy. Click a category to zoom in. Regroup with AI sorts all your topics into a few clean categories and relabels past chats.

Sentiment

Sentiment over time (positive, neutral, negative) and Where customers are unhappy — the topics and chats with the most negative sentiment, so you know what to fix first.

Ask AI

Type a question — What do customers complain about most?, What are people asking about delivery?, Why did customers ask for a person? — and get an answer drawn from your chats in that period, with links to the conversations it used. It reads each chat's summary plus the customer messages that use your question's words. Each question uses credits like one reply; earlier questions are kept below.