Usage Analytics
The usage analytics section in the Portal provides detailed visibility into your organization's AI consumption. Track token usage, cost breakdown, and request counts across models, users, and time periods.
Token Usage Charts
The main usage dashboard displays interactive charts showing:
- Daily token usage: Bar chart of input and output tokens per day
- Usage trend: Line chart showing consumption trends over the billing period
- Model distribution: Pie chart showing token usage split by model
Charts can be filtered by date range to zoom into specific time periods.
Cost Breakdown
The cost view translates token usage into estimated dollar amounts using the configured per-model rates:
By Model
See how much each model contributes to your total spend. This breakdown helps you:
- Identify which models account for the majority of spend
- Find opportunities to shift workloads to more cost-effective models
- Validate that tier-based routing is working as expected
By User
View spending attributed to individual team members. The per-user breakdown shows:
- Total tokens consumed (input + output)
- Estimated cost for the billing period
- Number of requests
- Most-used model
Privacy: Per-user analytics show aggregate usage metrics only. Individual conversation content is not visible in the usage dashboard — only token counts, model selections, and costs.
Request Counts
Track the total number of AI requests by:
- Request type: Chat completions, skill runs, assistant executions, engine runs
- Model: Requests per model
- Status: Successful, rate-limited, DLP-blocked, or errored
Billing Period
Usage analytics are organized by billing period (calendar month by default). The dashboard displays:
- Period start/end dates: Current billing period boundaries
- Days remaining: Days left in the current period
- Current spend: Total estimated cost so far
- Budget utilization: Percentage of monthly cap consumed
- Projected spend: Estimated total for the period based on current trajectory
Exporting Data
Usage data can be exported for external analysis or reporting:
- CSV: Tabular export of usage data with all fields
- JSON: Structured export for programmatic processing
Exports respect the current date range filter and include all available metrics.
Next Steps
- Configure cost controls and budget caps
- Optimize costs with model routing
- Review audit logs for detailed request-level data