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

# Review tests and apply winners

> Read experiment evidence and make a clean decision about an opening.

<div className="darwin-location"><strong>Find it in Darwin</strong> Experiments → Overview, Pages, or History <a href="https://app.meetdarwin.ai/admin/experiments">Open this screen ↗</a></div>

<Frame caption="Experiments shows automation state, work needing attention, and active tests.">
  <img src="https://mintcdn.com/meet-darwin/j9QNY_OzVc2Uv_0E/images/screenshots/experiments.webp?fit=max&auto=format&n=j9QNY_OzVc2Uv_0E&q=85&s=541ed01513d563c6b3c55598cc27f2e3" alt="Experiments shows automation state, work needing attention, and active tests." width="1280" height="720" data-path="images/screenshots/experiments.webp" />
</Frame>

<p className="darwin-screenshot-note">Screenshots use a demonstration workspace. Your data, plan, and activation status may differ.</p>

Open an experiment to compare its original opening with challengers. Review the target page, hypothesis, copy, impressions, chats, meetings, and the decision reason together.

## Read the numbers

| Measure                     | How to interpret it                                                                                    |
| --------------------------- | ------------------------------------------------------------------------------------------------------ |
| **Impressions**             | Exposure recorded for that variant in this test.                                                       |
| **Chats / meetings**        | Recorded outcomes attributed to that opening.                                                          |
| **Chance to beat original** | The estimated probability the challenger’s expected value exceeds the control.                         |
| **Lift**                    | Estimated relative expected-value improvement, not a promise of the same lift in meeting count.        |
| **— / gathering evidence**  | There is not enough comparable data to show a useful estimate.                                         |
| **Prior impressions**       | Discounted history carried by a control that has already run. This is separate from fresh impressions. |

## Available actions

* **Check results** evaluates the available evidence without waiting for the next scheduled check.
* **Pause** stops a running test; **Resume** makes a paused test eligible to run again.
* **Apply winner** publishes a selected winner for its target. If automatic promotion is on, Darwin can do this for qualifying results.
* **Archive** releases a non-running test’s place so another test can be considered.

Use **History** for closed outcomes. Read why an experiment ended; a closed or failed test is not necessarily an improvement.

<Warning>Editing copy after a variant has impressions mixes old and new wording in the same record. For a clean comparison, prefer a new variant or test. Replacing baselines also resets their evidence.</Warning>

A promoted winner becomes the page’s starting point for future tests. [Learn how the loop works →](/optimization/how-it-works)