How Analytics works

Analytics is built from the same conversations you work in every day, not from a separate dataset assembled somewhere else. That is why you can click any number and read the actual threads behind it, and why a figure you disagree with is something you can settle in about ten seconds.

Open Analytics from the main navigation.

The dashboards you get

• Overview. Your daily glance: how much came in, how much your AI handled, how long resolution takes.
• AI Performance. How well your agent covers and finishes conversations, and where it stops.
• Team Performance. Your people, with your AI shown alongside them as a benchmark rather than in a separate report.
• Workload. Throughput. New, replied, closed.

You cannot delete these, but you can reshape any of them and build your own. Anything in the sidebar marked coming soon is not live yet.

The three controls that drive everything

One bar sits at the top of every dashboard, and it changes every panel at once rather than one chart at a time.

• Date range. The period you are asking about.
• Aggregation. Whether to read it by day, week or month. Weekly is usually the honest resolution for support: daily is noise, monthly hides the week something broke.
• Filters. Channel, category and agent, among others.

Filtering is the step most people skip, and it is what turns a number into a decision. "Resolution time" across everything is not actionable. "Resolution time for shipping questions on WhatsApp" tells you where to look.

Every number opens

Click a chart, a figure, a slice of the funnel, or a cell in a table, and you get the conversations that produced it, still inside your current range and filters.

This is the most useful habit to build here. A metric that looks wrong is almost never wrong; it usually means the definition is not what you assumed. Opening the conversations tells you which of those two it is immediately.

Why your numbers will not match your inbox exactly

Three reasons, all deliberate, and knowing them saves you an afternoon of doubting the data.

• It refreshes about every 15 minutes. A conversation you closed a moment ago is not in there yet.
• Spam is excluded from everything except the spam count itself.
• Time metrics are medians, not averages, and they run on plain clock time rather than your opening hours. So a Friday-night message answered on Monday counts the weekend, which is worth remembering before you read a response time as bad.

Getting the data out

Export the current view to CSV and you get exactly what you are looking at, range and filters included. Stay on the page while it downloads.

Who can see it

Analytics is for the roles that manage the business rather than work the queue: Manager, Admin and Owner. If a teammate reports it missing, check their role before anything else.