Run an A/B test for popup campaigns

Prepare two complete campaigns for the same exact website, with reviewed designs and targeting. Change one meaningful element when possible so the result is interpretable. You need editing access and A/B testing access in your workspace. Starting a test publishes its participating campaigns and freezes their tested revisions and traffic split.

1. Open A/B tests from the popup area.
2. Enter a test name, choose Campaign A and Campaign B, then set the traffic percentage for A. B receives the remaining share.
3. Select Create test and check the selected campaigns.
4. Choose Start and publish only when both versions are ready for visitors.
5. Review assigned units, shown units and submitted units. Keep Consented visitor and Page only results separate.

The experiment’s submission rate uses assigned units, including assignments whose trigger never fired. It is not the same denominator as the Reports page’s impression-based conversion rate. Results do not claim statistical significance or automatically select a winner. Use Stop test to pause all participating campaigns, or Keep [campaign] to stop the test and leave that tested campaign published. Stop the run before changing participating publications.

Verify both campaign names and traffic allocation before starting. After stopping, inspect each campaign’s actual published/paused status. Compare like-for-like measurement units and enough representative traffic for your own decision. See Reports and publication states.