> For the complete documentation index, see [llms.txt](https://docs.humblytics.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.humblytics.com/understanding-your-data/understanding-experiments.md).

# Understanding Experiments

Plan, run, and review A/B tests in one place, from first idea to final result.

An A/B test (also called a split test) shows two versions of a page to different visitors at the same time, so you can see which one gets more conversions.

Go to **Optimize → Experiments**.

## The stat strip

The top of the page sums up your testing program: **running tests**, **finished tests that found a winner** with the **average lift** across them, and **visitors in running tests**.

The header has two buttons: **Plan a test** opens the [Test planner](https://docs.humblytics.com/split-testing-overview/using-the-humblytics-a-b-sample-size-calculator.md), and **New test** starts a test.

## Table and Board views

**Table** lists every test with its status, goal, page, versions, visitors, and result. Filter it with the chips: **All tests**, **Needs decision** (tests with enough data to call), **Awaiting approval** (tests created by an AI agent that you haven't activated yet), **Running**, **Planned**, **Ideas**, and **Done**.

**Board** organizes the same tests by stage:

| Column      | What goes here                                                      |
| ----------- | ------------------------------------------------------------------- |
| **Ideas**   | Test ideas and learnings you've captured but not planned yet.       |
| **Planned** | Drafts you've started. Click **Continue** to finish setting one up. |
| **Running** | Live tests, with visitors and the chance to beat Control so far.    |
| **Decided** | Finished tests.                                                     |

## The idea backlog

You don't have to build a test the moment you think of it. Click **Capture** to save an idea with a sentence or two about what you'd change and why. When you're ready, click **Plan it** on the idea to turn it into a draft test with the idea carried into its notes. Finished tests can be turned into new ideas the same way, so learnings don't get lost.

## Test statuses

| Status                | What it means                                                                                                             |
| --------------------- | ------------------------------------------------------------------------------------------------------------------------- |
| **Idea**              | Captured, not planned.                                                                                                    |
| **Planned**           | A draft in setup.                                                                                                         |
| **Awaiting approval** | Created by an AI agent through the API or MCP. It isn't served to visitors until you open it and click **Activate test**. |
| **Running**           | Live and collecting data.                                                                                                 |
| **Needs decision**    | Running, with enough data to call.                                                                                        |
| **Done**              | Stopped, called, or auto-stopped.                                                                                         |

There is no pause. A running test can be **stopped** (it keeps its data and can't be restarted) or **archived**.

Tests stop on their own after **30 days** by default. You can change that limit when you create the test. Tests never stop automatically on reaching significance; you decide when to call a winner.

## Test health

Every results page starts with a health check so you know whether to trust the numbers:

* **Healthy:** traffic, split, and tracking look normal.
* **Check:** something looks off, such as an uneven split or no conversions. The card says what.
* **Needs traffic:** too few daily visitors to reach a result in a reasonable time.

## Settings

The **Settings** disclosure on a test's page shows how it was configured. These can't be changed after launch:

**Goal at launch**, **Destination** (for destination goals), **Versions**, **Split**, **Page**, **Audience** (device, traffic source, and country restrictions in one row, or *All visitors*), **Counting** (**Per visit** or **Per person**), **Minimum run**, **Maximum run**, **Started** and **Ended**, your notes, and the **Experiment ID**.

## Results page

Click any test to open its results: the health card, the goals list with the **chance that each version beats Control**, the run timeline, and the actions **Call test** and **Ship**. The header has **Share** (a view-only link) and **Export CSV** (every version and goal as a spreadsheet). See [How to analyze split test data](https://docs.humblytics.com/split-testing-overview/how-to-analyze-split-test-data.md).

## Create a test

Click **New test**. A page picker opens: choose the page to test, then edit it in the visual editor (headlines, buttons, images, visibility, order) with no code or redeploy. To compare two different live URLs instead, choose **Split by URL**.

Setup has three steps: **Setup**, **Variants**, and **Review**. See [How to set up a split test](https://docs.humblytics.com/split-testing-overview/how-to-setup-a-split-test.md).

## Plan limits

Active tests are counted account-wide: 1 on Business, 5 on Scale, and unlimited on Agents plans. Only running tests count; ideas, drafts, and finished tests don't. At your limit, **New test** changes to **Upgrade**, and agent-created tests stay **Awaiting approval** until you stop another test or upgrade. See [Plan limits & upgrades](https://docs.humblytics.com/understanding-your-data/plan-limits-and-upgrades.md).

## Learn more

* [How to set up a split test](https://docs.humblytics.com/split-testing-overview/how-to-setup-a-split-test.md)
* [How to analyze split test data](https://docs.humblytics.com/split-testing-overview/how-to-analyze-split-test-data.md)
* [Deciding how long to run a test](https://docs.humblytics.com/split-testing-overview/deciding-how-long-to-run-an-a-b-test.md)
