3

I have an A/B testing experiment running with 3 variants Control, Var A and Var B at 33.3% distribution for each. Now I want to stop that experiment and run it again with just 2 variants: Control and Var B. That way they will have 50% distribution for each. My question is: will the 33.3% of users who got variant B during the first run of the same experiment get a different variant when I run this experiment again (by keeping the same remote config parameter name) with just 2 variants (Control and Var B)?

Frank van Puffelen
  • 565,676
  • 79
  • 828
  • 807
iamronak
  • 626
  • 6
  • 14

1 Answers1

1

Yes. When you create a new A/B test (even if it's a duplicate of a previous one), you're essentially "re-rolling" all of your users to determine what variant they go into.

Todd Kerpelman
  • 16,875
  • 4
  • 42
  • 40