I look at retention in a similar way (for an online university). As for the growth model related here, I would suggest using a traditional survival analysis in

I look at retention in a similar way (for an online university). As for the growth model related here, I would suggest using a traditional survival analysis instead, which is conceptually simpler and accounts for whether or not some users are still active (right censoring). There are also many packages ('survival' in R, for example) which have the algorithms programmed (both simple Kaplan-Meier and multivariate Cox PH) and decent visualizations.