Free tool
Did your test really win?
Enter each version's visitors and conversions: see how big the difference is, and how likely it is to be real rather than luck.
…
How to read it
Very likely real and likely real mean a difference this big would rarely come from luck alone (under 1 in 100, and under 1 in 20). Might be real means it's leaning, but not proven: keep the test running. Could be chance means the numbers can't tell the versions apart yet.
The range is where the true difference most likely lies (95%). If it spans zero, B could be better or worse.
Decide how many visitors you need before you start, and check once you have them: looking every day and stopping at the first good number finds winners that aren't. The sample size calculator tells you how many. Peekbird's own results are made to be checked as often as you like: how sure its calls are.
Peekbird works this out for every goal of every test, with revenue and order value too, and says when it's too early to tell. Start free.
