Understanding Day of Week Effects with Google Analytics

The amount of variance in your user behavior across days of the week can easily exceed the size of differences that matter for your business.  In addition, understanding your day of week effects can offer insight into when to launch promotions, etc.

In general, the beginning of the week is the point of the most dramatic activity.  The weekends bring a different type of user mindset with typically a little bit more time than during the week.  Note, this availability of time may work against you if the user chooses to try a different site instead of working harder on your site.

http://alwaysbetesting.com/abtest/images/posts/ga_date_widget_mod.gif
You're likely familiar with the core date widget (to the right) if you spend much time in Google Analytics. The widget circled in red is the date selector and is the bread and butter of data manipulations. The blue circled widget is the date comparer and splits a selected data range into two segments, altering the data grids to show whether the metrics of the current report are trending up or down.

http://alwaysbetesting.com/abtest/images/posts/ga_compare_date_mod.gif
Clicking the blue widget in the core in the default Date Range view brings up the view shown at left.


There's an undocumented feature of the date compare widget that faciliates day of week analyses. Simply select a Day of Week header to view only data from that day across the date range.



If Tuesday is the day you make the most money, it may be that this is the day that is most sensitive to your split testing for certain features. On the other hand, a split test that's not showing significant results may be working for you on the weekend but failing on the weekdays. Try it out, at worst, you'll learn about the strong day of week effects present on every site.

Related Blog Entries

Comments (Comment Moderation is enabled. Your comment will not appear until approved.)
ivan's Gravatar Controlling for seasonality (I saw a great X-mas effect for mobile phone searching in 2006 : +30% query volume, likely due to a new phone from Santa) can also be important. Holidays and Fridays off for a 3 day weekend can add serious skew to expected metrics. Adding to the mix is that most businesses don't have enough data to predict the specifics of "the St. Patrick's Day Effect" (feel free to substitute your favorite holiday here). The take home is to be careful, plan ahead, and consider weighted trailing averages to predict likely ranges of expected behavior. Flight early and often.
# Posted By ivan | 4/3/07 10:53 PM
Built with BlogCFC, version 5.9. Contact Andy Edmonds or read more at Free IQ or SurfMind. © 2007.