Split traffic 50/50

A way to describe that half of traffic – or 50% – saw one version (Version A) in A/B test, while the other half of traffic was directed to the other version (Version B).

Note, traffic does not have to be split 50/50. Traffic can be divided anyway you like. For example, if you’re very uncertain about the test results and would like to minimize risk, you might want to allocate only 20% of traffic to the variant and 80% to the original version.

However, traffic is typically split 50/50 since it is a fast, accurate, and reasonably low risk way to conduct an A/B test.