A/B Testing Gets Real-Site Preview and Guidance on Visitors Needed for a Winner

We've added two features to A/B testing. Real-site preview lets you check how a variant looks on your live site before it goes public, and the results page now shows guidance on how many visitors are still needed for a winner decision.

Real-site preview

In addition to the existing preview in the editor, you can now check each variant directly on your own site.

A "Real-site preview" section has been added to the experiment detail page, with a "Copy preview URL" button for each variant. Open the copied URL in a browser to see the page with that variant's changes applied.

• Works for draft and paused experiments too. Check how copy fits and whether the layout breaks on the real page before the test goes live.
• Nothing is tracked during preview. It's never recorded in views, conversions, or heatmaps, so opening it as many times as you like won't affect your results.
• Redirect tests are supported too. You're taken to the destination page, which then renders in preview mode.
• Share the URL with a teammate or agency to show them the exact same look.

Preview only applies to the first page you open from the URL. Navigating to another page on the site returns to the normal display.

Visitors remaining until a winner decision

A guidance message like the following now appears at the top of the results page.

About 160 more visitors are needed to reach the minimum for a winner decision (about 4 days at the current pace)

• The visitor count is the total shortfall across all variants until each one reaches the minimum for a decision (100 visitors per variant).
• The day estimate is based on the actual visit pace since the test started, and shows how long it will take for the slowest variant to reach the minimum. Even if traffic is split unevenly, it never underestimates.
• The day estimate appears only after one day has passed since the start and visits have come in. Until then, only the visitor count is shown.
• Once every variant reaches the minimum, this guidance disappears and the display switches back to showing the leading variant, as before.

The minimum is the lower bound at which a decision can begin. If the difference between variants is small, more visits will still be needed.

For tests that were already running before this feature launched, the start date wasn't recorded, so only the visitor count is shown — no day estimate.

Pricing

Both features are available on all plans, including Free.

Scope of impact

Uses your existing tracking tag as-is. No configuration changes are needed on your site. The tracking tag (tracker.js) is cached by browsers for up to 24 hours, so right after release some environments may not pick up "Real-site preview" until the following day.