Running an A/B test

An A/B test compares two whole versions of one page on a single public URL. Visitors are split, and you keep the version that converts better.

Before you start. The page must be published. If it is not, the test screen tells you so and nothing else is offered.

Creating the test. On the pages list, open the "…" menu on the published page and choose Create A/B Test. That opens the test screen for that page, where you pick one of two starting points:

• Start from this page. Variant B is created as an editable copy of the current page.
• Test two existing pages. Variant B is another published page you pick from a dropdown.

Then click Create A/B test. The test starts in Draft, and nothing is split yet.

Configuring it. The Test settings card holds:

• Test title.
• Winning metric: Clicks, Opt-ins, Conversions, or Revenue (GMV). If the page has no form or checkout, use Clicks.
• Test duration (hours), anywhere from 1 hour to 720 hours (30 days).
• Fallback variant, served to remaining visitors if the result stays inconclusive.
• Traffic split, shown as A 50% and B 50%. This is fixed and cannot be changed.

The Variants card links to Edit this version for each side, so edit B until it genuinely differs from A.

Going live. Click Launch test. Settings lock at that point. While it is still a draft you can click Cancel test instead, which deletes the test and, if B was a clone, the clone with it. A page you picked as an existing variant B is left alone.

Reading the result. The Results card lists Views, Clicks, Opt-ins, Conversions and GMV per variant plus an overall row, with your chosen metric highlighted. A banner above it says either "Variant X is leading" with a confidence percentage, or "Too close to call yet, keep the test running to gather more data". The chart below can be switched between Views, Conversions and GMV. Counts are approximate, because cookie blockers hide some visitors.

Finishing. Click Declare winner and choose Auto (the metric leader), A or B. The chosen version's content is copied into the live page, so 100% of traffic gets the winner, the test flips to Completed, and the decision is frozen. This cannot be undone. Picking Auto while the result is still inconclusive is refused, so choose a side or keep running.

The list. Website → A/B Test shows every test with its status (Draft, In progress, Completed), the leading or winning variant, view count and dates. If that entry is missing from your sidebar, A/B testing is not part of your current plan.