All Products
Search
Document Center

Platform For AI:Create an experiment metric

Last Updated:Jul 17, 2026

Track the performance of your experiments by defining experiment metrics, such as click-through rate (CTR) and conversion rate (CVR). The basic fields required to calculate these metrics are already registered in the source data table. You only need to define the computing logic for each metric.

Prerequisites

You must have already registered a source data table. For more information, see Register a source data table.

Create a metric

  1. Go to the ABTest page.

    1. Log on to the PAI console.

    2. In the left-side navigation pane, click Workspaces. On the Workspaces page, click the name of the workspace that you want to manage.

    3. In the left-side navigation pane, choose Model Deployment > PAI-ABTest.

  2. In the left-side navigation pane of the PAI-ABTest page, choose Metric Management > Experiments.

  3. On the Experiments page, click Create Metric.

  4. On the Create Metric page, configure the following parameters.

    Parameter

    Description

    Metric Group

    Metric Group Name

    Enter a custom name for the metric group. A metric group is a collection of related metrics.

    Metric group description

    Enter a description for the metric group.

    Sub-metric

    Sub-metric name

    Enter a custom name for the sub-metric.

    Sub-metric description

    Enter a description for the sub-metric.

    Source data table

    Select a registered source data table.

    Metric template

    The following built-in metric templates are available:

    • Page Views

    • Unique Visitors

    • Cumulative Value

    • Total Number of Occurrences

    • Statistics by Condition

    The fields you need to configure depend on the selected metric template:

    • If you select Page Views, Cumulative Value, or Total Number of Occurrences as the metric template, you must select a computing field.

    • If you select Statistics by Condition as the metric template, you must select a condition field and a computing field.

    Metric computing logic

    After you select a template and its fields, the metric computing logic is automatically generated. You can modify this logic as needed.

  5. After you configure the parameters, click OK.

Manage experiment metrics

You can perform the following operations on existing metrics:

  • View experiment metric details

    To view the details of a metric, click the image icon next to it. These details include the source data table, the number of associated experiments, and the metric computing logic.

  • Add a sub-metric

    In the Actions column of a metric group, click Add Sub-metric to define a new sub-metric. For information about how to configure sub-metric parameters, see Create a metric.

  • Delete an experiment metric

    To delete a sub-metric, click Delete in its Actions column. To delete an entire metric group, click Delete in the group's Actions column.

Associate experiment metrics with an experiment

To associate an existing metric with an experiment:

  1. In the left-side navigation pane of the PAI-ABTest page, choose Experiment Management > Experiments.

  2. In the experiment list, click the name of an experiment to open its Experiment details page.

  3. On the Basic Information tab, in the Experiment metrics section, click Edit.

  4. In the Experiment metrics panel, select a Core metric and any followed metrics, and then click OK.

Related topics

You can also associate an experiment metric during experiment creation. See Create an experiment for details.