The analysis is helping you uncover how frequently people come back and at what likelihood. I start by defining “coming back” with at least one identified session with the user in the second time period. But you really want to organize this by the number of different days they engage and then use this to predict likelihood of returning. And then to understand why.