The second part of your analysis (number of users) may look a bit weird to some.
I decided to swap actives/inactives to reduce the effect of the log10 scale which tended to reduce the share size of the active users in the graph.
Unfortunately, removing the log 10 scale would make the graph unreadable.
