> For the complete documentation index, see [llms.txt](https://docs.umbraco.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.umbraco.com/ai-in-umbraco/19.latest/using-the-api/decision/score.md).

# Score Questions

Score content against labelled levels with the experimental Decision API.

{% hint style="warning" %}
Decision is experimental and disabled by default. See [Enabling Decision](/ai-in-umbraco/19.latest/using-the-api/decision.md#enabling-decision).
{% endhint %}

Use `AIScoreDecisionQuestion` to rate content against an ordered list of labelled levels. The answer is an `AIScoreDecisionAnswer`, which returns a score and a probability for each level.

## Question Properties

| Property       | Type                                  | Required | Description                                                                             |
| -------------- | ------------------------------------- | -------- | --------------------------------------------------------------------------------------- |
| `Instructions` | `string`                              | Yes      | What to score.                                                                          |
| `Levels`       | `IReadOnlyList<AIDecisionScoreLevel>` | Yes      | The levels, **lowest first**. Between 2 and 10 entries, none with an empty description. |

Each level is an `AIDecisionScoreLevel` with a `Description`:

{% code title="AIDecisionScoreLevel.cs" %}

```csharp
public sealed record AIDecisionScoreLevel(string Description);
```

{% endcode %}

A level is identified by its position in `Levels`, starting at 0 for the lowest level. Two levels can share the same description.

Pass the content to judge as the `state` argument of `AskAsync`, or as the request's `State`. Questions also have an optional `Id`. See [Requests, Questions, and Answers](/ai-in-umbraco/19.latest/using-the-api/decision.md#requests-questions-and-answers).

## Answer Properties

| Property        | Type                               | Description                                                                                                                                                  |
| --------------- | ---------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| `Score`         | `double`                           | Where the answer landed, as a 0-based index into `Levels`. The value can be fractional, such as `1.8`.                                                       |
| `Probabilities` | `IReadOnlyDictionary<int, double>` | The estimated probability for every level, keyed by its 0-based index into `Levels`. The probabilities add up to 1.                                          |
| `Confidence`    | `double?`                          | The provider's own confidence in `Score`, from 0.0 to 1.0. `null` when the provider does not report one. Values from different providers are not comparable. |

`AskAsync` returns an `AIDecisionResponse<AIScoreDecisionAnswer>`. Read the answer from its `Answer` property. The response also has `ModelId` and `Usage`.

The answer does not include a level label. To get one, round `Score` to the nearest index and look it up in `Levels`. With the levels `poor`, `ok`, and `good`, a `Score` of `1.8` sits between `ok` (1) and `good` (2). The nearest level is `good`.

## Example

The following service rates how readable a page summary is:

{% code title="ReadabilityRater.cs" %}

```csharp
#pragma warning disable UMBRACOAI_DECISION

using Umbraco.AI.Core.Decision;

namespace MyProject.Services;

public class ReadabilityRater
{
    private readonly IAIDecisionService _decisionService;

    public ReadabilityRater(IAIDecisionService decisionService)
    {
        _decisionService = decisionService;
    }

    public async Task<string> RateAsync(string summary, CancellationToken cancellationToken)
    {
        var question = new AIScoreDecisionQuestion
        {
            Instructions = "How easy is this summary to read for a general audience?",
            Levels =
            [
                new AIDecisionScoreLevel("poor"),
                new AIDecisionScoreLevel("ok"),
                new AIDecisionScoreLevel("good")
            ]
        };

        var response = await _decisionService.AskAsync(
            decision => decision.WithAlias("readability-rating"),
            question,
            summary,
            cancellationToken);

        // Round the score to the nearest level
        var nearest = Math.Clamp((int)Math.Round(response.Answer.Score), 0, question.Levels.Count - 1);

        return question.Levels[nearest].Description; // "poor", "ok", or "good"
    }
}
```

{% endcode %}

Use `Score` when you need a finer value than the label, such as for sorting or averaging. Use `Probabilities` to see how the model spread its answer across the levels. For example, `response.Answer.Probabilities[0]` is the probability of `poor`.

## Related

* [Decision](/ai-in-umbraco/19.latest/using-the-api/decision.md) - Overview, enabling, and profile selection
* [Yes/No Questions](/ai-in-umbraco/19.latest/using-the-api/decision/binary.md)
* [Pick-One Questions](/ai-in-umbraco/19.latest/using-the-api/decision/choice.md)


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation by asking a question.

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://docs.umbraco.com/ai-in-umbraco/19.latest/using-the-api/decision/score.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.