Hey there! Here is a screenshot of our AA test (Bayesian). D28 RLTV is Revenue per user on 28th Day and D28 Conversion is percentage of users converted on 28th Day. We run current phase for 90 days. Did anyone face such situations that AA test shows significant change? Does it mean that we did something wrong? Appreciate any comments on our situation.

Hey Andrey,we land the data from segment to redshift and then also calculate the results. I prefer because I see raw level info and I can debug. You can try to do it in internal db.