A/B Testing
Showing two versions of a popup to different visitors at random, then keeping whichever one actually converts better.
A/B testing splits your traffic (usually 50/50) between variant A and variant B of the same popup, changing exactly one thing, the headline, the button color, the timing, so any difference in conversion rate can be attributed to that one change.
The trap is testing too many things at once, or calling a winner too early. A 3% lift over 40 visitors is noise, not a result; wait for a sample size large enough that the difference could not plausibly be chance.
Not every popup needs a formal test. For low-traffic pages, a well-reasoned single version beats an underpowered test that never reaches significance. Save A/B testing for the popups your highest-traffic pages actually run.
See it in practice