Averages Are Deceptive
Averages are an unfortunately deceptive metric, especially if you have a website that has had spiky traffic on occasion. Keep this in mind when you calculate averages for your pageviews, earnings, and other metrics. For example, if you regularly get 200 pageviews per day and suddenly get linked to from some social bookmarking site (or an A- or B-list blogger), you might get a traffic spike and “enjoy” several hundred or even thousands of extra pageviews. These extra pageviews will skew your “average daily pageviews” metric for a period of time.
Here’s a specific example. Let’s say that you have 30 days of traffic data that shows that your average daily pageviews is 125. That means that your total pageviews in that time is 30 x 125 = 3750. Obviously, your actual daily pageviews are not necessarily 125 every day. One day might be 75, another 135, etc.
Now let’s say that in the next month, you’ve started to build some traffic and you average 150 pageviews daily for days 1-30. On the 31st day, you enjoy a traffic spike and get 2500 pageviews. (You’ll probably have higher pageviews for the next several days as well.) That means your average daily pageviews for month 2 is ((150×30) + 2500)/31 = 7000/31 =~ 225. There’s a fair bit of difference (relatively speaking) between 150 and 225. (This variation of values is a study in itself, but beyond this discussion.)
Similarly, when you first start your site(s), you may not have a great deal of traffic. For example, my main website, chameleonintegration.com, was set in late Mar or early Apr of 2005. Aside from my test views and a trickle of other traffic, I essentially get very little traffic to that site even a year later. One, I don’t have a lot of content there, until recently, when I set up a technology weblog. Two, my focus is on my other journals, which I didn’t set up until June or July 2005.
That means that I had several months of virtually no traffic. When I average total traffic, those months of no traffic significantly affect my metrics. One thing that I learned in a college statistics “bird” course called “The Uses and Abuses of Statistics” is that stats are only really useful when taken in YOUR context. If it means more to you to ignore months where you were unable to write and post fresh articles, fine. If you want a more general measure of your productivity and prefer to include traffic-less days, also fine.
The value of a metric is dependent on the person viewing it. For a similar perspective, check Seth Godin’s brief discussion of averages and a followup.
Technorati Tags: mathblog, math blog, averages, variance
[…] Web Metrics: Mean vs Median For those blogger entrepreneurs that don’t relish the thought of browsing web statistics, there may be confusion about the value of various web metrics. Two such metrics are Mean vs Median, which can be applied to number of visitors, number of pageviews, etc. There is a distinct difference between the two, especially on websites that have signficant variance in day to day statistics. In particular, averages are deceptive. Seth Godin explains the difference between Mean and Median. […]