A dozen questions, then the calendar. Rough numbers are fine — the model shows its assumptions, so nothing has to be exact.