Statistical Confidence Level and Minimum Conversions
Set the Confidence Level for the Results
There are several options for adjusting the statistical evaluations of the experiments to your needs.
Setting the Confidence Level
Confidence indicates how certain one wants to be that the uplift or downcast reported as significant is actually one. Confidence is the reciprocal of the p-value and is calculated as follows:
Confidence = 100% - Significance (p-value).
The p-value indicates the probability with which we would have measured at least as high an uplift if we had again performed an A/A test (i.e., control vs. control) with the same number of visitors.
It is possible to set a value between 50% and 99.9%.
- Choose a low value if you need fast results
- Choose a high value if you need a statistically most reliable result
See here for more details:
To prevent a winner from being determined too soon after the start of an experiment, a minimum number of conversions should be set.
Meaningful values can be 100 or 1000, for example.