Go to Top

Increase Reporting Accuracy for Google Analytics by comparing weeks, not months

As Google Analytics continues to improve its excellent product, it’s important to remember the fundamental nature of traffic patterns for your business and how they can affect your reporting. Even with the introduction of advanced segmentation, motion charts, and other awesome features, there are real-time traffic patterns that can create major errors in your analytics reporting if you don’t account for them.

One of the biggest ones is highly cyclical web traffic, which can have a huge impact on trend comparisons across months.

Why does this matter? Well, let’s look at a real example from one of our clients. If we try to do a traffic comparison month to month, say, between October 2008 and October, 2007, and we didn’t check the cyclical patterns, we would think the increase in traffic in 2008 was 0nly about 0.57%, a pretty small climb for a company that is working hard to improve a site’s traffic.

However, if we look at the month to month comparison to see what the cyclical patterns are, we see that things are a little off:

In this case, the 2008 sample is has two fewer high-traffic days than the 2007 sample. This can have a strong distorting effect.

The fix is simple, however: do your time sample from week to week instead of month to month! In most cases a five week slice, starting and ending on the same day, that includes the entire sample month is all you need, although for some periods you will need a six week slice. In this case, we took a five week slice starting on Sunday and ending on Saturday. The results were pretty dramatic: Once the cyclical traffic had been synchronized, the five week period containing October 2008 had 3.75% more traffic than the five week period containing October 2007.

Google Analytics makes this comparison extremely easy with the use of the calendar function; basically choosing five weeks is nearly as easy as choosing a single month, so week-to-week synchronization is a quick and easy adjustment.

We have heard some concerns that week to week comparison are harder to explain to clients. We appreciate that this makes the time periods a little less fixed than month to month comparisons. But in fact we have created a more formal version of this blog post that basically walks them through the process, and we’ve found that with the pictures shown here, it’s generally very easy for them to understand. One of two things will happen: either they will accept the fact that month-to-month reporting may be inaccurate, but not hold you accountable for trend data at that time period, or they will embrace the week-to-week synchronization.

facebooktwittergoogle_plusredditpinterestlinkedinmailby feather

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>