Custom Search

Monday, November 24, 2008

Statistical Analysis of a Blog’s Traffic

Today blogger and MBA student Eric Rosenberg fromThe Israel Situation and Narrow Bridge Adventures talks us through some statistical analysis of the traffic to his blog.

As a hobby blogger trying to increase readers to my Israel blog, The Israel Situation, I have always been glued to my site traffic. What does a spike mean? How can I bring more readers? Is there anything I can do to help my blog grow?

I am currently an MBA student, and I took a statistics class this past quarter. For a class project, I tried to use statistics to prove what you can do to help your blog grow. I used a lot of technical tests including regression, chi-squared, and correlations to prove my theories. Darren gave me the opportunity to share my findings with all of you, and I have translated the findings into plain English to help you grow your blog readership.

I first tested to find a relationship between posts per day and site visitors. I started my blogging believing that the key to more visitors was more posts. Not true! A two sample test comparing site visitors to page loads proved that there is no association between the two. If you are working your tail off to post ten times a day, it might be better to focus your efforts on fewer, better quality posts.

Next up I tried to find a way to drive subscriber growth. I tested the relationship between page views and subscriber count. A two sample t-test (statistics talk for two variable relationship test) proved that again, there is no relationship. At this point, it seemed that no numbers could predict a successful blogger, but I continued my testing.

-1.png

Next, I used a test called chi-squared to find if day of the week had any big impact in visitors. The test is designed to test if proportions are equal across periods. I did find a difference here. I found that there were constantly more visitors on Sunday and fewer on Friday and Saturday. This, I believe, is specific to my target audience. Shabbat, the Jewish day of rest, is from Friday night to Saturday night. As a blog that focuses on Israel, many of my visitors are Jewish. This explains the difference. If you have a blog with a niche audience, you might find this variation as well. A business blog, for example, may have a spike in readers on Monday morning at the beginning of the week and less visits on the weekend.


stats-1.png


My next test is the one that applies to virtually every blogger. I tested page views against time. I found that time since starting the blog does correlate to an increase in page views. At first I had trouble substantiating this data. I removed outliers, or non-regular data, and was able to then prove my hypothesis. As you can see from my chart, I was able to demonstrate that days since blog inception closely related to an increase in regular visits. My spikes in traffic, caused by Reddit and hosting the Haveil Havalim blog carnival, did not cause an increase in traffic. Just to reiterate, social bookmarking and other spikes in traffic did not cause a statistically significant increase in blog viewers. You can see the five dots that stick out in the chart below. Those are my big days of social bookmarking and blog carnival hosting.


View Original Article

No comments: