Usage Analytics

Usage Analytics is your platform-wide consumption dashboard. Use it to keep an eye on how much your organization is using the AI platform: request volume, token usage, cost, latency, and success-versus-error rates. You can view totals across the whole deployment, then drill down to see which users and departments are driving the activity.

Before you start

To open Usage Analytics, you need the View Analytics permission. With it, you’ll see Usage Analytics in the left sidebar under Monitor.

A few things depend on the scope of your permission:

If you have the View Analytics permission granted organization-wide, you’ll see everything, including the Overview tab.
If your analytics permission is limited to specific departments, you can still open the page, but the Overview tab is hidden. You’ll land on the By User tab instead, and the By User and By Department tabs show data only for the departments you administer.

Finding your way around

The page has a header with the title and a filters toolbar for choosing a date range. Below that is a row of tabs plus a Refresh button on the right.

The tabs are:

Overview — high-level totals and trend charts (shown only if your permission is organization-wide).
By User — a searchable, sortable, paginated table of per-user activity.
By Department — the same kind of breakdown, grouped by department.
By Model — a placeholder for now; it shows a “coming soon” message and no data yet.

Switching tabs moves focus to the new panel so it’s easy to keep working from the keyboard.

Setting the date range

Every view shares the date range you pick in the header.

Choose a preset: Last 7, Last 30, Last 90, or This Month. The start and end dates fill in automatically.
To set your own window, choose Custom. Start and end date fields appear so you can enter exact dates.

When you change the date range, the dashboard reloads to match. On the Overview tab this happens right away; on the By User and By Department tabs there’s a brief pause before the table refreshes, so rapid changes don’t trigger a flurry of reloads.

Reading the Overview tab

The Overview tab gives you the big picture in three parts.

KPI cards

At the top is a grid of summary cards:

Total Users — with the number of active users beneath it.
Total Requests — with the average requests per user beneath it.
Total Tokens
Total Cost — shown in USD.

Several cards include a small growth badge comparing the current period with the previous one of the same length (for example, “vs previous 30 days”). A badge appears only when there’s an actual change to report.

Top Models

When model data is available, a Top Models table lists the most-used models with their Model name, Provider, Requests, Tokens, and Cost. If there’s no model data for the selected range, this table is hidden.

Charts

The Charts section shows four trend charts side by side:

A multi-metric chart of requests over time, with tokens and latency in the tooltip.
Usage Growth — requests as bars with tokens as a line.
API Reliability — success versus error counts over time.
Cost Trend — cost over time.

Use the granularity pills (Hour, Day, Week, Month) to change how the trend lines are grouped. The charts redraw to match.

Each chart has a View data table button. Click it to open the underlying numbers as a table in a pop-up, with columns suited to that chart (for example, Date, Requests, Tokens, and Latency for the multi-metric chart). Close it with the close button, by clicking View data table again, or by pressing Escape. Focus returns to the button you opened it from.

Reading the By User tab

This tab lists per-user activity in a table. For each user you’ll see:

User Name, Email, and Department (a dash if none is set)
Requests, Success, and Errors
Tokens and Cost (USD)
Avg Latency (in milliseconds)
Last Activity (or “Never” if the user hasn’t been active)

A stats row shows the total number of users and how many are currently displayed.

To work with the list:

Search by name or email. Results filter as you type, and a clear button resets the search.
Choose how many rows to show per page: 10, 20, 50, or 100.
Sort by clicking a column header (Name, Email, Requests, Tokens, Cost, Avg Latency, or Last Activity). Each click cycles through descending, ascending, and unsorted.
Move through pages with the Prev and Next buttons, which disable at the first and last page.

Reading the By Department tab

This tab works much like By User, but each row is a department. You’ll see the Department name, number of Users, Requests, Success, Errors, Tokens, Cost (USD), and Avg Latency (ms). A stats row shows the total number of departments and how many are displayed.

You can search by department name, choose a page size (10, 20, 50, 100), sort by department name, and page through results with Prev and Next.

Refreshing the data

The dashboard keeps itself current automatically, quietly refreshing the active tab’s data every couple of minutes in the background. You can also force an immediate update with the Refresh button on the right of the tab bar. Refresh updates just the active tab without taking over the whole page; it’s briefly disabled while the update is in progress.

Good to know

Loading and empty states. The Overview tab shows a full-page spinner on first load and an error card with a Retry button if something goes wrong; individual charts show their own spinner and a friendly empty graphic when there’s no data for the range. The By User and By Department tables show a spinner only on first load and an italic message in the table when there’s nothing to show — with different wording for “no results match your search” versus “no data yet.”
Overview is organization-wide; the breakdowns are scoped. The Overview totals and charts cover the entire deployment, while the By User and By Department tables cover only the departments you administer. That’s why the same page can mix an organization-wide view with department-scoped views.
The By Model tab isn’t live yet. It’s a placeholder and doesn’t load any data for now.
Growth badges and the comparison window are based on the length of the date range you’ve selected, so a 30-day range compares against the previous 30 days, and so on.

Part of the Control Hub Guide.