> For the complete documentation index, see [llms.txt](https://docs.elevateab.com/elevate-helpcenter/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.elevateab.com/elevate-helpcenter/theme-testing/create-a-theme-experiment.md).

# Create a Theme Experiment

Step by step walkthrough for building and launching a theme experiment in Elevate A/B Testing.

***

Five setup steps, then QA, then launch.

### Before you start

* Your alternative theme is in **Online Store > Themes** and unpublished.
* It is fully customized across every page type.
* Apps that inject into your storefront are configured and working on it.
* You have previewed it in Shopify and it looks right.

### Step 1: General info and goal

1. Open **Experiments** and select **Create Experiment** on the **Theme Experiment** card.
2. Enter an **Experiment name**. Something you will recognize later, such as "New Dawn theme vs current".
3. Optionally add a **Description** and a **Hypothesis**. Record what you expect and why before you launch.
4. Optionally attach **Assets**: designs, briefs, notes.
5. Choose your goal metric under **What do you want to measure?**

**Revenue per visitor** is recommended and is usually right for theme tests. A theme changes the entire shopping experience, so a single narrow metric rarely captures whether it worked.

Select **Continue**.

### Step 2: Set variations

1. Under **Number of Variations**, choose how many themes to compare, up to five including your control.
2. Your published theme is set as the **Control** automatically. Its name and ID are shown, with a live preview.
3. For each variation, choose how to supply the theme.

| Option                         | Use it when                                                                                                                                 |
| ------------------------------ | ------------------------------------------------------------------------------------------------------------------------------------------- |
| **Select**                     | The theme already exists in your library. Pick it from the list.                                                                            |
| **Duplicate theme in Shopify** | You want to test a modified version of your current theme. Elevate opens Shopify so you can duplicate it, make your changes, and come back. |
| **Book a help call**           | You would like help from the Elevate team setting the experiment up.                                                                        |

The theme picker shows each theme with its creation and last updated dates, so you can tell versions apart. Your live theme is marked **Theme In Use**. Choose a theme other than that one.

4. Name each variation something meaningful, such as "Minimalist" rather than "Variation 1".
5. Use **Preview** on any card to open that theme, and **Edit in Shopify** to make changes to it.

Select **Continue**.

#### Duplicating a theme

If you chose **Duplicate theme in Shopify**:

1. Select **Open Shopify Themes**.
2. Duplicate your current theme.
3. Make your changes to the duplicate.
4. Save, and leave it unpublished.
5. Return to Elevate and select the duplicate from the theme picker.

### Step 3: Traffic allocation

The default is an even split. Leave it there unless you have a reason not to, because an even split reaches a result fastest.

If you want to limit exposure while you build confidence in a new theme, weight the control higher, for example 70/30. The trade is a longer run.

**Isolate traffic for this experiment** is optional and off by default. See [Traffic Isolation](https://docs.elevateab.com/elevate-helpcenter/experiment-setup/traffic-isolation).

Select **Continue**.

### Step 4: Audiences

Every visitor is eligible by default.

Useful narrowing for theme tests:

* **Mobile only**, if the redesign is mostly about mobile layout
* **New visitors only**, for unbiased first impressions from people with no habits on your store
* **One country**, if the redesign targets a specific market

Every filter shrinks the pool and lengthens the run. See [Audience Targeting](https://docs.elevateab.com/elevate-helpcenter/experiment-setup/audience-targeting).

Select **Continue**.

### Step 5: Review

Check the summary:

* The right theme is assigned to each variation
* Traffic split is what you intended
* Audience targeting is what you intended, and no narrower

The experiment is created in **Preview**. It is not live and no visitors are affected.

### QA before you launch

Do not launch straight from here. A theme experiment changes every page for every assigned visitor, so QA matters more than for any other type.

Work through [Preview and QA a Theme Experiment](https://claude.ai/chat/preview-and-qa.md) first.

### Launch

When QA passes, select **Launch Experiment**. The experiment moves to **Running**.

Watch it closely for the first 24 hours. Theme tests affect every visitor, so a problem you missed in QA shows up quickly and costs more than it would on a single page test.

Do not publish a variation theme while the experiment is running.