Analytics

Analytics helps workspace admins and managers understand how their workspace consumes credits. Use it to check whether consumption is on pace, find changes over time, and identify the agents, members, or other contributors driving usage. Go to Admin > Analytics to open the page.

​

Choose a period

Use the selector at the top right to analyze:

• This cycle: the current cycle. This is the default.
• Last 7 days: a rolling seven-day window.
• Last 30 days: a rolling 30-day window.
• Last 90 days: a rolling 90-day window.

The period applies to the summary, chart, attribution table, comparisons, and raw-data export. Below the page title, Analytics shows:

• The start and end dates of the selected period. When This cycle is selected, these are the boundaries of the current credit cycle.
• The number of active members out of all workspace members. Members count as active when they consume credits during the selected period.
• When Analytics data was last updated. For example, Updated 2h ago means the figures were refreshed two hours ago. More recent activity may not appear yet.

Daily chart boundaries use UTC.

​

Check whether consumption is on track

When This cycle is selected, the status banner compares the percentage of the cap used with the percentage of the cycle elapsed.

• On target means the remaining credits are likely to cover the rest of the cycle at the current pace.
• Off target means consumption is ahead of pace and deserves attention.
• Critical means the remaining credits are unlikely to cover the rest of the cycle.

Select Manage in Usage to review the workspace cap and other credit controls. See Credit management for details. The summary cards show:

• Used this period: total credits consumed during the selected period. For This cycle, the card also shows the percentage and total value of the cap.
• Top agent: the agent with the highest credit consumption, together with its share of all consumption during the period.

Cap status and cap values do not appear for the rolling seven-day, 30-day, and 90-day periods.

​

Investigate consumption over time

The Consumption chart provides two views of the selected period.

​

Find spikes with the Daily view

Daily displays credits consumed each day as stacked bars. The bars are broken down using the dimension selected in Attribution. For example, selecting Models in Attribution changes the chart from an agent breakdown to a model breakdown. The chart displays the five highest-consuming contributors individually. Others combines the remaining contributors. Hover over a bar to see the credits attributed to each contributor and the total for that day. The Today (partial) marker identifies the current, incomplete day. Its bar is faded because today’s usage is still being added, so the total may increase. Avoid comparing it directly with completed days.

​

Follow the trajectory with the Cumulative view

Cumulative shows how actual consumption builds over the period. For an unfiltered This cycle view, the chart also shows a target line that spreads the cap evenly across the cycle. Hover over the chart to compare actual and target consumption and see how far consumption is ahead of or behind the target. The target line is not displayed for rolling periods or while an Explore filter is active.

​

Filter the analysis

Select Filters in the Explore section to narrow the chart, Attribution, and raw-data export by:

• Agents
• Members
• Groups
• Models
• Tools
• Skills
• Sources
• API keys

You can select several values in one category. A record can match any selected value within that category. When you combine categories, records must match every selected category. Select Apply to update the analysis. Applied filters appear beneath the Explore heading, where you can remove individual values or clear them. Filters remain applied when you change the period or Attribution tab, but reset when you leave or reload the page.

​

Find what drives consumption

Attribution ranks contributors for the selected period and filters. Choose the dimension that answers your question: Member consumption is attributed to the member who triggered it. Group consumption is attributed to every group that member belongs to, so the same consumption can appear under more than one group. The selected Attribution tab also controls the breakdown used by the Daily chart.

​

Read the attribution table

Use the search field to find a value in the active tab. Results are initially ranked by Total credits, from highest to lowest, and are paginated when necessary. The table includes:

• Consumption share: the contributor’s percentage of all credits in the selected scope, not only the credits represented by the visible rows.
• Total credits: credits attributed to the contributor.
• Credits / message: average credits generated by each originating user message, including all model and tool work that message triggers. This metric is shown for agents, members, groups, models, sources, and API keys.
• Credits / invocation: average credits per invocation for tools and skills.
• Vs Prev: the change from the previous period. This cycle compares the current cycle with the previous cycle. Rolling periods compare with the immediately preceding window of the same length. -- appears when there is not enough previous data to calculate a percentage.

Select a sortable column heading to reverse its order.

​

Drill into a contributor

Select a row or its chevron to expand it. The breakdown shows up to three leading models, tools, and members where they provide an additional perspective. Select View all in an expanded breakdown to apply the original row as a filter and open the corresponding Attribution tab. For example, from an agent row, select View all under models to see every model used by that agent. Use the funnel icon at the end of a row to add that contributor directly to the Explore filters. Select it again to remove the filter.

​

Download raw data

Select Download raw data above Attribution to generate a CSV for the selected period and Explore filters. If no matching export is ready, opening the panel starts one automatically. You can also select New export to generate a fresh file. The export contains consumption records with metadata such as timestamps, conversation and message identifiers, agents, members, groups, models, sources, API keys, tools, skills, credit components, status, and execution time. The active Attribution tab, table search, and table sort order do not restrict the export. Generated exports are kept for a maximum of 15 days.

​

Metric definitions

​

Export analytics through the API

For programmatic reporting, use GET /api/v1/w/{wId}/analytics/export with a workspace admin API key. This API export is separate from the raw-data export generated in the Analytics page. See the Dust API documentation for authentication and endpoint details.