D/D, W/W, M/M, Y/Y... OMG/WTF? A Guide to Comparing Time Periods

Business owners, managers, execs, and leaders love talking about growth. When folks start talking about growth, one of the most important things to ask is, "What are you comparing it to?" The most common comparisons are day over day (D/D), week over week (W/W), month over month (M/M), and year over year (Y/Y). The period you choose to compare it to is absolutely critical because choosing the wrong period to compare can give you an artificially inflated or deflated picture of your business's performance. We've developed a guide for choosing the optimal comparison period.

Let's start with a day/day comparison – which period would you choose to compare today to (if today was 5/8/2013)?

1. The period immediately before (5/8)*

2. The period 1 week before (5/2)

3. The period 1 month before (4/8)

4. The period 1 year before (5/8/2012)

*Examples are given for Day/Day comparisons.

Day over day:

1. Yesterday is not comparable to Today. Weekly seasonality in the e-commerce industry follows certain patterns as does internet traffic. In general, stores tend to have more traffic during the weekdays than the weekends. Mondays are usually the strongest day and Sundays the weakest in terms of sales. Comparing Monday to Sunday would show a massive uptick in revenue not because your store is on an astronomical trajectory, but simply because Mondays are usually always better than Sundays.

2. Correct! Accounts for seasonality and minimizes daily fluctuation.

3. 5/8 is not like 6/8. What if 5/8 was a holiday or 6/8 was a Friday and not a Monday? This may be consistent with monthly seasonality (e.g. commission based sites will often hit peak sales toward the end of the month to meet quotas or make bonus), however, the daily seasonality issue previously mentioned still makes this a far from optimal comparison.

4. 5/8/2012 isn't at all like 5/8/2013. For starters, 5/8/2012 was a Tuesday and 5/8/2013 was a Wednesday. Weekly seasonality issue.

Week over week:

#1 and #2 are similar options here and either of them are correct. If the last 7 days rolling is taken into account, weekly seasonality is minimized. You do run into some snags if you're comparing a week with a holiday to a week without, but by and large this methodology is commonly accepted as the most accurate week over week comparison.

Month over month:

Since #1, #2, and #3 are the same option for monthly comparisons, we can only compare a month to a prior month or the same month a year ago. Comparing November 2011 to November 2010 or option #4 is actually the optimal choice since monthly seasonality like Thanksgiving and holiday shopping is accounted for. Comparing November to October will skew perceived growth since most e-commerce stores see an uptick during the holidays and a slump shortly after in January. Note that some folks call this a year over year comparison (comparing one month in a year to a month in the prior year).

Year over Year:

The only option here is really #4. We recommend performing a prior year to date versus current year to date (YTD) comparison. This takes care the seasonality issue and prevents holiday bump at the end of each year from influencing the numbers early on in the calendar year.