Sample Size Calculator for NON 50/50 split | Community
Skip to main content
joshd5166100
joshd5166100
Level 1
November 8, 2019
Solved

Sample Size Calculator for NON 50/50 split

  • November 8, 2019
  • 1 reply
  • 3324 views

We are prepping an A/B test that will use a 80/20 traffic split.

Can someone direct me to a sample size calculator that can provide an estimated duration for this non 50/50 approach?

    This post is no longer active and is closed to new replies. Need help? Start a new post to ask your question.
    Best answer by Shruti_Premkumar

    joshd5166100​ There is no separate sample size calculator for a non even split. However a few things to keep in mind are Your data is only as good as the smallest group, so 80-10-10 is really 10-10-10.Instead of using your full population, you could base it on the 'daily visitor' number you would get for your smaller experience. Alternatively if you reduced the total entrants and used an equal split, it would take the same amount of time and the normal traffic size calculator would work fine.

    i.e. if 100 visitors came into your test, and the 80/10/10 split showed non-default to 20 people, you could allow only 30% of traffic into the test and then split it 33/33/33? This would still result in 20 people seeing non-default  as an example.

    If you have any further questions on the same, you could reach out to an Adobe business consultant on the same and they can provide more guidance on the same.

    Hope this helps.

    1 reply

    S
    Adobe Employee
    Shruti_PremkumarAdobe EmployeeAccepted solution
    Adobe Employee
    November 11, 2019

    joshd5166100​ There is no separate sample size calculator for a non even split. However a few things to keep in mind are Your data is only as good as the smallest group, so 80-10-10 is really 10-10-10.Instead of using your full population, you could base it on the 'daily visitor' number you would get for your smaller experience. Alternatively if you reduced the total entrants and used an equal split, it would take the same amount of time and the normal traffic size calculator would work fine.

    i.e. if 100 visitors came into your test, and the 80/10/10 split showed non-default to 20 people, you could allow only 30% of traffic into the test and then split it 33/33/33? This would still result in 20 people seeing non-default  as an example.

    If you have any further questions on the same, you could reach out to an Adobe business consultant on the same and they can provide more guidance on the same.

    Hope this helps.

      Terms & ConditionsAccessibility statement

      Loading footer...