Create or Configure a Metric Alert

Learn how to create new metric alerts or configure existing alerts.

Not available to Embedded Partners

Access requirements

Custom profiles with one of the following capability sets:

  • Data (Write, Simple), Model (Read, Simple), and Access to all data such as the Super Admin permission

  • Data (Write, Simple), Model (Read, Simple), and Data Validation - All Data

Reach out to your administrator for access.

Overview

Metric alerts allow you to set an expected range for changes in metric values over time.

Example: Let's say your organization sets Headcount to an expected range of -2% to 2%. Any changes to your organization's headcount value that exceeds 2% in any direction will trigger the metric alert when a new data version is generated and cause an issue to be created. If the Headcount alert is critical, any changes exceeding 2% in any direction will fail the data version and require an admin to review the data version.

Create a metric alert

  1. In a project, on the navigation bar, click Validation > Data Alerts.
  2. Click Create Alert.
  3. In the Alert Creation dialog, select Metric Alert, and then click Next.
  4. Select a metric from the list and click Next.

  5. Type a display name and description.

  6. Click Create.
  7. Configure the alert. For more information, see Configure a metric alert below.

Configure a metric alert

Specify the exact conditions of the alert.

  1. In a project, on the navigation bar, click Validation > Data Alerts.
  2. Click Metric Alerts and select the alert to configure.
  3. Enable or disable the alert. If enabled, the alert will check every data load job for metric values outside the expected range.
  4. Select a comparison type to use for the alert.
    • Month to Month: Indicates that the new value should be compared to last month's value.
    • DV to DV: Indicates that the new value should be compared to the previous production data version.
    • Value: Indicates that the new value should be compared to a specific value.
  5. In Lower bound, type a value to be the lower bound of the metric's expected range of change.
  6. In Upper bound, type a value to be the upper bound of the metric's expected range of change.
  7. Optional: In Exclude last X months, type a number of historical months to exclude from the alert. For example, if you don't want to validate the past three months for the new data load, type 3 in this field.
  8. Optional: To exclude certain time periods from the alert, in Excluded time periods, click Add. Set a date range during which the metric alert will not create issues. The end date is inclusive. For example, if you don't want the alert to create issues for March 1 to March 15, add an excluded time period from 2025-03-01 to 2025-03-15. You can add as many time periods as required.
  9. If the alert should fail the data version, enable the Critical toggle. If the alert is triggered in data load jobs, the data version will fail and require an admin to review the data version.
  10. When finished, publish your project to production.

Example: Month to Month versus DV to DV

Let's say your organization's Headcount is 1,000 at the end of last month.

This month, you publish two data versions:

  • DV 1: Headcount is 1,010.
  • DV 2: Headcount is 1,030.

With Month to Month, the alert compares the new value to last month's value. In this example, DV 2 is compared to last month's value of 1,000. This shows a 3% change from 1,000 to 1,030. Month to Month measures change against the same point last month, no matter how many data versions you publish in between.

With DV to DV, the alert compares the new value to the previous data version value. In this example, DV 2 is compared to DV 1. This shows a 2% change from 1,010 to 1,030. DV to DV measures change against whichever data version came right before it. Use DV to DV if you publish more than one data version per month to catch changes between each one.