Yes but in your A/B test you shouldn't be picking first half vs second half. Each visitor should be randomly assigned, so it should mitigate the problem you men

Yes but in your A/B test you shouldn't be picking first half vs second half. Each visitor should be randomly assigned, so it should mitigate the problem you mentioned.