Two versions, one audience, and a winner only when the result is real. Available from Growth.
Split a campaign into two versions and send a share of your traffic to each. They can differ in anything — the rewards, the colours, the copy, even which game runs. No other popup app can A/B test a game, because none of them have one. A/B testing is available from the Growth plan, and the two variants together still count as a single running campaign.
Start a Test
Never under 100 visitors per version, and never below 95% statistical confidence. Below those thresholds the difference between two versions is noise, and reporting it as a result would be worse than reporting nothing. It will also tell you outright when two versions are genuinely the same, because that is a real answer and not a failure.
Reading the Result
Because the result so far cannot be distinguished from chance. Rather than leaving you to guess how long to wait, the app tells you the size of the gap — for example, “you need about 1,900 more visitors per version before this means anything”.
Not Enough Data Yet
Promote it. The winning version becomes the campaign, and the test ends. You can do this by hand whenever you are satisfied with the result.
Promote the Winner
Yes, if you ask for it. With auto-promote on, the winning version is promoted for you once the result clears the same thresholds a manual promotion would have waited for — 100 visitors per version and 95% confidence.
Auto-Promote
These topics are covered in the help centre inside Lucky You, where they sit next to the settings they describe.
