Except that dplyr is horribly slow and memory inefficient, and (if you were minded to actually put R code into production) it more or less forces you to do data prep in a database anyway.