As we all know, people behave differently – so your Split Test needs to collect enough data to average out all the different emotions and personalities. Over time, enough data is collected to reach a point of Statistical Significance. This is the point when you can feel confident that it’s a real result and not just an unlucky, odd sample. Your AB Split Tests are monitored and analysed to give you real time likelihoods of each variation being the winner. When a variation has over a 95% likelihood of being the winner, and has run for over 1 week, then when we declare a winner.
TL:DR At least a week, sometimes much longer for low traffic sites.
