The overall rating reflects the current state of the app. It accounts for all app reviews but prioritizes the most recent ones.
It is mathematically impossible for the data returned from this app to be correct. Switching out different versions of a product over a fixed time period won't tell you anything because both unit sales and conversions naturally change over time. The only way to get statistically significant A/B testing values is to run two variations simultaneously, which this app can't do.
We're very sorry to hear that you've run into an issue with our app, and we would really like to look into this further for you.
Whenever convenient for you, please respond to our email and let us know when we can connect with you over the phone in order to solve this issue. Again, we are very sorry for any inconvenience this has caused and would like to rectify the issue immediately.