Month-over-month change, days until the next order, the gap since the last visit: all of them compare a row with its neighbor. lag(x) shifts a column down one row, so each row sees the value before it; lead(x) shifts it up, so each row sees the next one. The first row has nothing before it, so its lag() is NA.
Two things decide whether the answer is right. Order: lag() uses the rows as they are, so sort with arrange() first when the data isn't already in date order. Groups: with several stores in one table, group first with group_by(store) or .by = store, so January of one store is never compared with December of another.
Percent growth is (this / previous - 1) * 100: divide by the earlier value, not the later one.
The previous month, the change and the percent growth, each cafe compared with itself.