> For the complete documentation index, see [llms.txt](https://docs.radar.citadel-ai.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.radar.citadel-ai.com/en/application-owner-guide/evaluation-library/custom-metrics.md). # Create Custom Metrics Custom Metrics are Evaluation Metrics that can be created to match company-specific evaluation criteria, guidelines, rules, and similar requirements. AI applications have different risks to consider depending on their purpose of use, business workflow, data handled, and how output results are used. Therefore, in addition to Built-In Metrics based on common risk domains, it is important to define evaluation criteria according to the application's characteristics and use case. By creating Custom Metrics, you can continuously evaluate the inputs and outputs of an AI application with criteria aligned with your company's business and operational rules. ### How-to create a Custom Metric #### 1. Select Add metric Select `Add metric` from `Evaluation Library` in the sidebar. #### 2. Select a creation method Custom Metrics can be created using one of the following methods.

Creation method	Use case
Add a new Evaluation Category	When creating an Evaluation Metric for a new risk domain
Select an existing Evaluation Category	When adding new evaluation criteria or a target metric to an existing evaluation category

When adding to an existing Evaluation Category, metrics created in the past are not overwritten and are managed as versions. {% hint style="info" %} For the relationship between Evaluation Categories and Evaluation Metrics, see [Manage the Evaluation Library](/en/application-owner-guide/evaluation-library.md). {% endhint %} **When adding a new Evaluation Category** Create a new Evaluation Category and the Evaluation Metric linked to that category.

Setting	Overview
Evaluation Category	Enter the Evaluation Category name. As the category represents a risk or threat domain, set a name that clearly describes the risk or threat you wish to monitor.
Description	Enter the Evaluation Category description. This does not affect the evaluation criteria.
Target	For the evaluation target, select either `Input (prompt/user text)` or `Output (model response)`.
Evaluation criteria	Use natural language to describe the conditions that trigger the evaluation metric.

{% hint style="success" %} When writing the evaluation criteria, refer to the definitions of existing built-in metrics. Using `AI Autofill` can automatically generate the Evaluation criteria based on the category name and description. Review and revise the generated content as needed. Evaluation results tend to be more stable when the evaluation criteria specifically describe the conditions that should be detected and the conditions that should be excluded. {% endhint %} Check the entered details, then click `Save` if there are no issues. **When selecting an existing Evaluation Category** Create a new evaluation metric linked to an existing evaluation category.

Setting	Overview
Evaluation category	Select the existing Evaluation Category to which the metric will be added.
Target	For the evaluation target, select either `Input (prompt/user text)` or `Output (model response)`. If a metric already exists in the selected category, a new version will be created. Otherwise, a completely new metric will be created.
Current latest evaluation criteria	If a metric already exists, the current evaluation criteria are displayed.
Evaluation criteria	Use natural language to describe the conditions that trigger the evaluation metric.

{% hint style="success" %} When writing the evaluation criteria, refer to the definitions of existing Built-In Metrics. Using `AI Autofill` can automatically generate the Evaluation criteria based on the category name and description. Review and revise the generated content as needed. Evaluation results tend to be more stable when the evaluation criteria specifically describe the conditions that should be detected and the conditions that should be excluded. {% endhint %} Check the entered details, then click `Save` if there are no issues. #### 3. Review the created custom Evaluation Category and Metrics You can review the created custom Evaluation Categories and Metrics using the same procedure as [Review Built-In Metrics](/en/application-owner-guide/evaluation-library/built-in-metrics.md). Custom made Evaluation Categories and Metrics havve a `Custom` badge.