Datalad calculator
Sample Size for Revenue and Other Continuous Metrics
Sizing a test on revenue per user is not the same problem as sizing one on conversion rate, and it needs a number most teams cannot look up: the standard deviation. This works it out from figures you already have.
1
What you are measuring
The first needs no database query
Per cent of users who order
Among people who order
Among people who order
Across everyone, including non-buyers
Across everyone. This is the number that matters
2
What you want to detect
Per cent, relative
Across all variants
Trims the tail, shrinks the spread
3
The answer
0
users per variant
0
users in total
0
weeks to run
Uses the standard two-sample comparison of means. The result is the sample size at which a test of this metric would reach the stated power.
[…] power, tails and multiple comparisons Cap outliers, and turn weekly users into a duration Open the calculator Free · runs in your browser · no […]