Adding a Second CTA at the End of the Page A/B Test

Move this test up when homepage visitors scroll all the way down and leave without acting, such as high exits among sessions that reach the end of the page, or session recordings of visitors scrolling back up from the bottom to find the button. When the idea works, some of the visitors who reach the end move on from the lower button, and total target-action completions rise in the variant arm. If few visitors reach the end, the issue may be length or content, so look there first, and build the lower CTA as a single button in the page flow, not a fixed bar visitors cannot dismiss.

Primary action completion rate

Compare the share of assigned visitors who complete the target action, not total button clicks.

Also track

• Share of clicks on the lower CTAReview button use by click distribution; lower-button clicks alone do not prove incremental value.
• Page-end reach rateMake sure fewer visitors reach the end of the page.
• Page exit rateTrack whether the repeated CTA increases exits.
• Post-click completion rateMake sure visitors who click the lower CTA complete the next step.

• Repeat CTAAdd one copy of the primary CTA at the end of the page.
• Keep constantKeep button wording, style, destination and page sections unchanged.
• IncrementalityCompare total target-action completions across arms rather than counting every lower-button click as a gain.
• DeviceReview mobile and desktop results.
• Separate testsTest alternate wording, intermediate placements or more repetitions separately.

1. 1Clicks on the lower button are not an added gain by themselves; if clicks on the upper CTA fall at the same time, visitors may simply have switched buttons. Count primary action completions across everyone assigned to each arm, not clicks on either button; also weigh any rise in page exits or drop in post-click completion.
2. 2Do not compare people who click the lower button with those who click the upper one as if they were equivalent groups; they chose their own path. This page length and a single added copy define what the finding covers. Review mobile and desktop separately; intermediate placements or different wording need their own tests.
3. 3Because the added button can only affect visitors who reach the end of the page, the overall difference may stay small. Check whether the gap between arms stands out from random variation with the A/B test significance calculator, and do not report an added contribution if it does not.