A couple of months ago, I posted on daily historical spam trends wherein I examined whether or not some days of the week experience higher volumes of spam than others.
I have supplemented that study with some more data. To begin with, are some days of the week more likely than others to experience a surge in spam? To check this, I calculated the weekly average in volume going back to the beginning of February 2006 (N=49) excluding weekends, since our weekends are lower volume due to a drop in legitimate mail traffic. One day I may include an adjustment factor to estimate spam only. The results below are the number of times the day's traffic was greater and lower than the weekly average:
Monday Tuesday Wednesday Thursday FridayAbove avg 21 28 33 26 23Below avg 28 21 16 23 26By excluding weekends, I have obviously skewed my data set somewhat but the results are still surprising, Wednesdays are clearly more likely to experience a surge in daily volume (and spam) than any other day of the week, whereas Mondays are more likely to be lower than average than any other day of the week.
Next, I calculated the daily volume as a percentage of the weekly average. I then calculated the averages for occurrences when the day was above average, and when the day was below average (in other words, when the day is above average, how much is it above average, and vice versa):
Monday Tuesday Wednesday Thursday FridayAverage - all 98.39% 100.15% 101.94% 100.10% 99.42%Above average 104.03% 103.85% 104.66% 103.32% 104.00%Below average 94.61% 95.76% 96.34% 96.45% 95.36%
When it comes to the average, both Monday and Wednesday differentiate themselves from the others, whereas Tuesday, Thursday and Friday are both hovering around the weekly average. However, surges in volume occur, Monday's experience the sharpest increase. When volume subsides, Wednesdays see the biggest drop.
It looks like the middle of the week is the worst day of the week when it comes to spam, while the first day of the week is typically the best. In my next post, I will attempt to normalize my results a little more and see if the results are the same.