External Links in a New Tab: Return to Site A/B Test

Analytics sessions that end on an external-link click, or support tickets from people who could not find their way back, show visitors who click resource or partner links not returning. The behavior you hope for is a switch back to the first tab and a new action on the site within the same window. If the external page simply met their need, a new tab will not change that; first check that the site offers a next step worth returning to. Limit the test to a fixed link set, and never open payment or form submission in a new tab.

Active return to the original site

Among all eligible assigned visitors, compare those who take a predefined active action on the original site after using a tested external link; do not count passive background time.

Also track

• External-link click-through rateCompare clicks on the tested external links among assigned visitors.
• Active time on the original siteMeasure active interactions and exclude time when the tab is open in the background.
• Return-path useTrack returns via the Back button or tab switching; neither alone proves confusion.

• DestinationUse the same predefined external-link set and compare current-tab with clearly announced new-tab behavior.
• ScopeKeep internal links unchanged.
• ExpectationClearly indicate that a link opens a new tab without changing its label or destination.
• DeviceReview return behavior separately on mobile and desktop.
• AccessibilityMake the new-tab behavior clear to keyboard and screen-reader users.

1. 1A tab left open in the background is not proof of interest, so do not count passive time as success. Even if external-link clicks or active time on the original site go up, active return remains the deciding metric. Returns through the Back button or a tab switch prove neither confusion nor success on their own.
2. 2Test the new-tab behavior in every supported browser to make sure it works the same; an arm that behaves inconsistently skews the result. In both arms, visitors should keep the option to right-click a link and open a tab themselves.
3. 3Apply the result only to the set of external links you tested. Keep mobile and desktop apart in the analysis, since switching tabs may work differently on each, and validate again before extending the behavior to other kinds of links.
4. 4An active return can only happen after an external-link click, so these events may be few. The sample size calculator gives the visitors needed per variant; divided by each arm's daily eligible traffic, that number sets your run length.