Skip to main content

Detect process anomalies


Detect process anomalies

Anomalies are found in variants that are deviating from the standard business process by following a different pattern of activities and vary in processing time.

Anomaly detection allows you to identify data points, cases, or activities that do not conform to the expected pattern. These unexpected behaviors might not occur frequently but can signify potential threats to your business such as frauds or system errors.

Key concepts

Ensure that you understand the following basic concepts documented in Basic concepts of Zuora ProcessAI:

  • Anomaly
  • Variant
  • Case

Detect anomalies

Navigation: Click the Anomaly Detection in the top left pane to detect the anomalies that lie within variants. Different variants are displayed in the anomaly_icon tab in the Process Data pane.

The table in this tab contains the following columns:

  • Variant: The variant number.
  • Variant %: The percentage of the current variant in all variants.
  • # Of Activities: The total number of activities in each variant.

Click Anomaly_icon  to switch to the scatter plot to identify anomalistic data points. Red dots in the scatter plot are the variants that contain detected anomalies, while the blue dots represent the healthy cases.    

You can click each of the red dots to check the variant details. Upon clicking a red dot, the corresponding variant is displayed in the Process Map pane.

For example, when you click the red dot in the scatter plot in the following screenshot, the corresponding variant is displayed. You can then start to analyze the anomaly of this variant based on the data in the process map. You can switch between KPIs as a variant could have more than one issue.