A/B/C Conversion Optimisation

#1
Hi All,

I'm trying to understand the process behind testing both A/B site changes for statistically significant increases in conversion rate, and also how then to expand this to testing multiple variables.

My understanding is that this is a binomial distribution (n=visits, p=conversion rate), and using z-scores, significance can be calculated like this.

What do I do if I have multiple variables though? I have previously compared each variation to the average performance of all other variations, but this feels open to misrepresentation of actual significance.

Would it be better to set one variation as the control and compare each new variation to the control only? Or should they be all be tested individually against each other?

Hopefully I've explained this clearly enough!

Thanks