> 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/theme-testing-overview.md).

# Theme Testing Overview

What theme testing does in Elevate A/B Testing, when to use it, what you need before you start, and what to know before comparing two themes on live traffic.

***

Theme testing compares complete themes against each other on live traffic. One group of visitors sees your published theme, another sees an alternative, and Elevate A/B Testing reports which performed better.

Instead of publishing a new theme and hoping, you test it first and switch only when the data says to.

### How it works

Your published theme is always the control. Each variation is a different theme from your Shopify theme library.

Visitors are split between them and each one browses the entire store through their assigned theme: homepage, collections, product pages, cart, and checkout. Assignment is sticky, so a visitor who returns sees the same theme as before and nobody watches your store change shape between visits.

From a visitor's point of view the store is completely normal. Cart and checkout work exactly as they always do.

You can test up to five themes at once, including your control.

### When to use it

| Use theme testing when                                                            | Use something else when                                                                     |
| --------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------- |
| You bought or built a new theme and want proof before publishing                  | You want to test one page template. Use a Page Experiment.                                  |
| You have done a full redesign, too large to test page by page                     | You want to test a headline, image, or style. Use a Content Experiment.                     |
| You are comparing design approaches, such as minimal against feature rich         | You want to know which specific change caused the difference. A theme test cannot tell you. |
| You want to try a seasonal layout during peak season                              |                                                                                             |
| You upgraded a theme to a new major version and want to confirm nothing regressed |                                                                                             |

A theme test answers one question: is this whole theme better than that one. If the new theme wins, you know to switch. If it loses, you will not know which of a hundred differences caused it.

Use it as a decision tool before a migration, not as an optimization tool.

### What you need before you start

**A finished alternative theme.** Fully customized, with every section and setting configured. Half built themes lose on their own merits and teach you nothing about the design.

**Apps working on that theme.** Review widgets, chat, upsells, subscriptions, and anything else that injects into your storefront needs configuring on the variation theme too. Apps do not carry across automatically.

**Comparable content on both.** Sections, menus, and settings do not transfer between themes. If your published theme has a testimonial block and the variation does not, you are testing a testimonial block as much as a theme.

### Requirements

* Elevate A/B Testing installed on your store.
* Your control theme published, and each variation theme present in **Online Store > Themes**.
* Every variation theme left unpublished for the duration of the experiment.

Theme testing works on all Shopify plans.

### Two things to know

**The variation theme must stay unpublished while the experiment runs.** The experiment depends on your control being the published theme. Publishing a variation mid test breaks it.

**Some third party analytics may not track a variation theme.** Tools that only fire on your published theme can under report. Elevate's own tracking covers both, but check anything you rely on separately.

### What to expect

Theme tests usually need longer runs than other experiment types. They affect every page, so the effect on any single metric is smaller and noisier than a focused test.

Run for at least a full business cycle, and expect to wait rather than seeing a clear answer in the first days.

### Next steps

* [Create a Theme Experiment](https://claude.ai/chat/create.md)
* [Preview and QA a Theme Experiment](https://claude.ai/chat/preview-and-qa.md), including how to preview themes without being redirected
* [Theme Experiment Results and Ending](https://claude.ai/chat/results-and-ending.md)