Alberto Ferrari lets us compare up to specific dates between years:
Unfortunately, the calculation is not perfect. At the year level, it compares the full previous year against an incomplete current year – in this example there are no sales after September 5th in the current year.
Besides, the problem appears not only at the year level, but also at the month level. Indeed, in September the Previous Year measure returns sales for the entire month of September in the previous year. The comparison is unfair, as there are only five days’ worth of sales in September of the current year.
Read on for a better technique.