What is an Anomaly Impact?
An anomaly impact is the total sum of deviations between the expected values of the metric (represented by the baseline sleeve) and the actual metric values across the entire duration of the anomaly. Visually, it represents the contracted area between the baseline and the metric curve, as highlighted in orange in the screenshot below:
- If a monetary Impact was set for the alert measure, the minimum impact is calculated based on the monetary Impact.
Why Should I Use Minimum Impact in Anomaly Alerts?
The minimum Impact feature enables users to trigger alerts based on calculated business impact, without committing to any minimum duration, delta, or score thresholds. With minimum Impact, notifications are triggered only when the financial or business impact reaches a user-defined threshold. For example:
- "Alert me only if the impact reaches 1,000 lost impressions"
- "Alert me only if the impact exceeds $1,000 in revenue loss"
This approach lets you define the minimum impact level that warrants your immediate attention - essentially answering the question: "What's the smallest business impact that would make me get up from my desk?"
Enabling Impact
- The Impact condition is only applicable for Anomaly type alerts.
Best Practice and Limitations:
-
When setting a minimum impact, the best practice is to set the Minimum Duration, Score and Absoulte Minimum Delta to their lowest possible values (effectively ignoring them). It is advisable, however, to assign a Percent Minimum Delta value of at least 10%.
For example, this is the recommended setup for a 5m-scale Impact alert:
- If functions are used in the alert expression, both a measure and a relevant stream must be explicitly stated:
