Hey, can anyone explain what chance to win means i...
# ask-questions
i
Hey, can anyone explain what chance to win means in an experiment result? How do i interpret this for individual metrics?
w
The amount of certainty you have that this metric indeed performs better for the variation you're looking at than for the other variation(s). There's always some risk involved, because your data is only a sample of all the data you could possibly collect. You can determine for yourself which chance is high enough. I use 90%.
i
But how is this chance being determined? I can see very less chance of risk - 43% which i believe is not recommended to bring confidence in the metric. Not able to understand how it is calculated
w
I don't completely understand what you mean, but the chance is dependent on the data that is collected. If variant B has a higher conversion rate (for example) than variant A, then the chance is higher than 50% that variant B is better for the chosen metric. The bigger the difference in conversion rate, the higher the chance (in general).