power automatemonitoringflow healthmicrosoftpower platformautomation

Power Automate Flow Monitoring: Tracking Health, Errors, and Run History

How to monitor Power Automate flows effectively — understanding run history, common error patterns, connection failures, and tools for ongoing health tracking.

6 min read

Get the extension

MakerOps Radar works directly inside Power Automate in your browser.

Why Flow Monitoring Matters

Power Automate flows run in the background, and failures are easy to miss. A flow that sends daily reports, processes form submissions, or synchronises data between systems may fail silently for days before someone notices the data is stale or records are missing. Proactive monitoring catches failures early and reduces the impact of broken automation.

Understanding Run History

Every flow maintains a run history accessible in the Power Automate portal. Each run shows:

  • Status: Succeeded, Failed, Cancelled, Running
  • Start and end time: exact timestamps for duration analysis
  • Triggered by: the specific trigger event that initiated the run
  • Action results: the output of each action in the flow

When a run fails, the run history expands to show which action failed and provides the error message from that action. Most errors are one of a small number of types: connection failures, data type mismatches, missing required fields, or throttling from a connector.

Common Error Patterns

Connection failures are the most common cause of flow failures. Connections use OAuth tokens that expire or become invalid when a user's password changes, permissions are revoked, or the service account used to create the connection is disabled. Re-authenticating the connection resolves these failures. For production flows, use service accounts or service principals rather than individual user accounts as connection owners.

Throttling occurs when a connector's API rate limits are exceeded. Microsoft 365 connectors (SharePoint, Outlook, Teams) have per-user throttling limits. Flows that process large batches in tight loops frequently hit these limits. Solutions include adding delays, processing in smaller batches, or upgrading the Power Platform licence.

Expression errors happen when expressions reference columns or fields that no longer exist, or when data arrives in an unexpected format. Dynamic content from adaptive cards, form submissions, or external APIs is a frequent source of type mismatches.

Setting Up Failure Notifications

Configure a flow to send an email or Teams message when another flow fails. Power Automate supports this through the "When a flow run fails" trigger available in the "Power Automate Management" connector. Setting up a notification flow for your most critical automation takes a few minutes and provides immediate awareness of failures.

Monitoring Across an Environment

For administrators managing dozens or hundreds of flows across an organisation, the Power Automate admin portal provides an environment-level view. The Center of Excellence (CoE) Starter Kit includes Power BI dashboards that aggregate flow health metrics, error trends, and connection status across all environments in a tenant.

Browser-Based Monitoring

For developers and power users who want a lighter-weight monitoring layer without setting up the full CoE toolkit, browser extensions can surface flow health directly in the browser while working in the Power Automate portal. MakerOps Radar provides a browser side panel showing run history, error trends, and connection status for flows in your environment — useful for quickly checking the health of flows without navigating through the portal's nested menus.

Long-Term Health Tracking

Beyond reacting to individual failures, monitor trends over time: which flows fail most frequently, whether error rates increase after a SharePoint change or a connector update, and which connections are most fragile. Patterns in the error history often reveal structural issues — a flow that fails 20% of the time may have a race condition, a missing null check, or a dependency on a fragile external service that needs a more resilient design.