---
title: Create a custom metric in AI Admin Center
description: Create a custom metric by specifying event-based condition logic and testing it against live data before saving.
locale: en-US
canonical_url: https://www.servicenow.com/docs/r/australia/intelligent-experiences/ai-admin-center-create-custom-metric.html
release: australia
topic_type: task
last_updated: "2026-10-01"
reading_time_minutes: 1
keywords: [AI Admin Center, Now Assist Center, custom metric, deflection, AI analytics, assistants]
breadcrumb: [Custom metrics, View AI assets usage and performance \(Lux UI\), Monitor, AI Admin Center, Enable AI experiences]
---

# Create a custom metric in AI Admin Center

Create a custom metric by specifying event-based condition logic and testing it against live data before saving.

## Before you begin

Role required: `ai_engmt_admin`

## About this task

Custom metrics let you specify the event criteria that count as a deflected assistant interaction in your organization. After creating a metric, edit it to map it to one or more AI assistants and activate it.

## Procedure

1.  Navigate to **All** > **AI Admin Center** > **AI Admin Center \(Legacy\)**.

2.  Select **Analytics** in the side navigation bar.

3.  Select the **Custom Metric** tab.

4.  Select **New metric**.

    The Custom Metric editor opens.

    \[Omitted image "ai-admin-center-custom-metric-new-screen.png"\] Alt text: New metric page showing the Metric Details, Condition Groups, and Test Against Live Data sections.

5.  Enter a **Name** for the metric.

6.  Enter a **Description**.

7.  Build the condition logic in the condition builder.

    Add events and connect them with AND, OR, or NOT operators. Groups can be nested to form recursive logic. Each condition specifies an event type, a reference point, and an optional time window.

    Select from the following event types when building conditions: Response provided, Feedback, Event type, Conversation transferred to Live Agent, Request, and Response.

    For time-based conditions, select the unit \(Seconds, Minutes, or Hours\) from the duration unit picker next to the duration field. The picker defaults to the largest unit that the stored duration value divides evenly into.

    Hover over a column header in the condition builder for a description of that field.

8.  Select **Test against live data** to validate the condition logic before saving.

    Select a date and an assistant, then select **Run test**. The results show the number of requests that matched the metric's criteria for the selected day \(matching count\). The results also show the number of those requests counted as deflected under the metric \(deflected count\).

9.  Select **Save**.


## Result

The custom metric is added to the Custom Metrics list. To map the metric to AI assistants and activate it, see [Edit a custom metric in AI Admin Center](https://raw.githubusercontent.com/ServiceNow/ServiceNowDocs/australia/markdown/australia/intelligent-experiences/ai-admin-center-edit-custom-metric.md).

**Parent Topic:**[Custom metrics in AI Admin Center](https://raw.githubusercontent.com/ServiceNow/ServiceNowDocs/australia/markdown/australia/intelligent-experiences/ai-admin-center-custom-metric.md)