> 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-experiment-results-and-ending.md).

# Theme Experiment Results and Ending

How to read theme experiment results in Elevate A/B Testing, when to trust them, how to end the experiment, and how to publish a winning theme.

***

### Reading the results

**Revenue per visitor** is the metric to read for a theme test. A theme changes the whole shopping experience, so a narrower metric can move for reasons unconnected to whether the theme is better.

Conversion rate, add to cart rate, and average order value are worth looking at as secondary signals. They tell you where in the journey a theme helped or hurt, which is useful even when the headline result is flat.

Elevate uses a Bayesian model and tells you when a result is trustworthy. Until it does, the numbers are provisional however large the gap looks.

### Give it longer than other experiments

Theme tests usually need more time. The change affects every page, so its effect on any single metric is diluted and noisier than a focused test.

Run for at least a full business cycle. Weekday and weekend visitors behave differently, and a result built on a few weekdays does not describe your store.

### What a flat result means

Often that the two themes are equally good, which is genuinely useful: you can now choose on maintainability, speed, or how much you like working in it, without worrying you are losing money.

Before concluding that, check the two themes actually differed in ways a shopper would notice. A redesign that looks dramatic to you can be invisible to someone buying a single product.

### What to do when a theme wins but you do not know why

You will not learn which change caused it. That is the nature of a theme test.

If you want to know, follow up with focused Page or Content experiments on the specific differences that matter most. The theme test tells you where to switch. Focused tests tell you what to keep.

### Ending the experiment

Ending stops the experiment. Visitors return to your published theme, no new visitors are assigned, and no further data is collected. Your results remain available.

The experiment moves to **Done**.

### Publishing a winning theme

If a variation won and you want to keep it, publish that theme in Shopify: **Online Store > Themes**, then publish the winning theme.

Elevate does not publish it for you. Ending the experiment and publishing the theme are two separate actions.

Before you publish:

* Confirm the theme is still in the state you tested. If you edited it mid experiment, you are publishing something different from the thing that won.
* Check anything that was only configured on your old published theme, such as apps or custom code added during the run.

### After ending

Anyone who used `eabUserPreview=true` during development should clear their cookies for the store domain, or use an incognito window, so they see the store as a normal visitor again.

Record what you learned, including a flat result. Knowing two themes perform the same is worth as much as knowing one wins, and it saves you re-running the same test next year.