Skip to main content

​
Configuring Usage Alerts

Navigate to organization settings

Access your organization settings from the dropdown in the top navigation bar.

Open Billing tab

Select the Billing tab in the organization settings.

Find Usage Alerts section

Scroll to the Usage Alerts section within the Billing page.

Add alert thresholds

Click Add Alert or Configure Alerts (depending on whether alerts exist).Enter the event threshold that should trigger an email notification. Examples:
  • 1,000,000 events (1M)
  • 5,000,000 events (5M)
  • 10,000,000 events (10M)
Multiple thresholds: Add several alerts for staged warnings (e.g., 5M, 7.5M, 9M for 10M monthly budget).

Save configuration

Click Save to activate the usage alerts.Counter start: Event counting begins from the moment you save the configuration. Historical events created before this point are not counted toward thresholds.

Verify recipients

Ensure the correct people have Owner or Admin roles in the organization—only these roles receive alert emails.Learn more about RBAC →
Usage alerts only count events created after the alert was configured or last updated. If you modify alert thresholds, the counter resets from the update time.

​
Alert Behavior and Timing

​
What Triggers Alerts

Events counted:
  • Traces: Each trace created counts as one event
  • Observations: Each observation (LLM call, tool use, retrieval) within traces counts as one event
  • Scores: Each score (evaluation result, user feedback, annotation) counts as one event
Total event count = traces + observations + scores Example: An LLM request creates 1 trace with 3 observations (system prompt, user query, LLM response) and 2 scores (quality eval, user feedback) = 6 events counted toward usage alert thresholds.

​
Event Counter Reset Conditions

  1. Billing cycle start: Automatic reset at the beginning of each billing cycle (monthly)
  2. Alert modification: Updating alert thresholds resets the counter immediately
  3. Alert deletion and re-creation: Deleting and adding alerts resets counters
Important: Modifying existing alerts resets counters, potentially delaying expected notifications if you’re close to a threshold.

​
Troubleshooting

Possible causes:
  1. Insufficient permissions: Only Owner and Admin roles receive alerts. Check your role.
  2. Email filtering: Alerts filtered to spam or promotions folder. Check junk/spam folders.
  3. Alert configured mid-cycle: Counter only includes events after configuration. Events before configuration don’t count.
  4. Alert updated recently: Updating alerts resets counters. Threshold may not have been reached again post-update.
  5. Billing cycle reset: If alert triggered in a previous cycle, it won’t trigger again until the next cycle.
Resolution:
  • Verify your role (Organization Settings > Members)
  • Check spam folders and add ABV emails to whitelist
  • Review alert configuration timestamp in Billing settings
  • Check if alert triggered earlier in the current billing cycle (won’t re-trigger until next cycle)
Possible causes:
  1. Counter reset: Alert was recently created or updated, resetting the counter. Threshold based on post-update usage, not full billing cycle.
  2. Usage spike: Sudden usage spike crossed threshold faster than expected based on typical usage patterns.
  3. Threshold miscalculation: Threshold doesn’t match actual budget or expected usage.
Resolution:
  • Check alert configuration/update timestamp to determine if counter reset recently
  • Review usage dashboard to identify usage spikes or anomalies
  • Recalculate threshold based on actual pricing and budget
  • Adjust thresholds if current settings don’t align with usage patterns
ABV usage alerts are organization-wide only, not per-project.Alternative approaches:
  1. Manual monitoring: Regularly review per-project usage in the ABV dashboard
  2. API-based custom alerting: Use the Public API to query per-project usage programmatically and implement custom alert logic (email, Slack, PagerDuty)
  3. Separate organizations: For completely isolated billing and alerts, use separate organizations for different teams or projects (incurs organizational overhead)
Learn more about the metrics API →

​
Related Topics

Data Retention

Control storage costs by automatically deleting old observability data

Metrics API

Query usage metrics programmatically for custom alerting and reporting

Role-Based Access Controls

Manage who receives usage alerts through organizational roles

Audit Logs

Track configuration changes to usage alerts and investigate anomalies