Alerts

Use alerts to surface operational conditions that need attention without continuously watching Monitoring.

Alerts help bring important operational conditions to your attention without requiring someone to continuously watch Monitoring.

They are most useful when you already know what kind of event deserves attention.

Start from the condition, not the notification

A useful alert begins with an operational question such as:

• Did a production execution fail?

• Has work been waiting longer than expected?

• Is a particular process producing repeated errors?

• Has a critical automation stopped running?

Define that condition first. Then decide who should be notified and what they are expected to do.

Use alerts for actionable exceptions

An alert should normally cause a clear next action.

Good examples include:

• investigate a failed production execution;

• review unusually slow work;

• check repeated negative feedback;

• inspect activity matching a known risk condition.

Avoid alerting on information that nobody is expected to act on. Too many non-actionable alerts make real exceptions easier to miss.

Connect alerts to Monitoring

Alerts are based on activity conditions. Use Monitoring activity to define and verify the view of work you care about before relying on it as an alert condition.

When an alert fires, use the underlying activity and traces to understand what actually happened rather than treating the notification itself as the evidence.

Review alert quality over time

Operational conditions change.

Periodically check whether an alert is still useful:

• Does it identify a real exception?

• Is the right person receiving it?

• Does it fire too often?

• Are repeated alerts exposing a structural issue that should be fixed instead?

A recurring alert can be a signal that the underlying Agent, Workflow, data or process boundary needs improvement.