> ## Documentation Index
> Fetch the complete documentation index at: https://docs.raily.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Evals

> See how well search answers real queries against your data

## Overview

Evals score how well Raily answers real queries against your data. Every search is rated, so you can see whether the right content is surfacing, spot weak queries, and catch regressions after a config change. See them in the dashboard at [app.raily.ai/dashboard/evals](https://app.raily.ai/dashboard/evals).

<Frame caption="Evals">
  <img src="https://mintcdn.com/railyai/wzVHZ5n0Rff2YZxy/images/evals.png?fit=max&auto=format&n=wzVHZ5n0Rff2YZxy&q=85&s=0797b5faa9385ef39e7d1f8450082e15" alt="Evals scores for recent searches" width="1457" height="772" data-path="images/evals.png" />
</Frame>

## What gets scored

Each search is evaluated across three stages:

```mermaid theme={null}
graph LR
    A["Query<br/>Understanding"] --> B["Retrieval"] --> C["Output"]

    A1["Does search understand<br/>what the user wants?"] -.-> A
    B1["Does it find the<br/>right content?"] -.-> B
    C1["Is the result<br/>good?"] -.-> C

    style A fill:#E0E7FF,stroke:#4F46E5,stroke-width:2px
    style B fill:#DBEAFE,stroke:#4F46E5,stroke-width:2px
    style C fill:#BFDBFE,stroke:#4F46E5,stroke-width:2px
    style A1 fill:#F9FAFB,stroke:#D1D5DB,stroke-width:1px
    style B1 fill:#F9FAFB,stroke:#D1D5DB,stroke-width:1px
    style C1 fill:#F9FAFB,stroke:#D1D5DB,stroke-width:1px
```

* **Query understanding** — does Raily correctly interpret what the user is asking?
* **Retrieval** — does it find the right content for that query?
* **Output** — is the final answer good?