Sep 12, 2014

Solving The Mystery of Misattributed Traffic in Google Analytics

by:


One of the more enjoyable aspects of my job as a research and analytics specialist (I know, I know, with a job like that, everything must be a barrel of fun) involves making sense of anomalies in data. While sorting through Google analytics data for a recent project I came across one such discrepancy.

At first glance it appeared as though there was a 38% increase in month over month direct traffic. Upon further inspection the majority of direct traffic had been directed to a very specific URL, a page containing a case study. In most cases, direct traffic for a website is aimed at the homepage, the URL that most visitors will remember or bookmark. Think about it, most of us would go to google.com and then click the “about” link in the footer, not type the direct URL (https://www.google.com/intl/en/about/) into the browser.

Nearly half of the direct traffic landed on the case study. My[1] first thought was that the client must have distributed literature in the form of pamphlets or flyers that contained links to the case study URL. A quick call to the project manager nixed this theory.

Looking at the distribution of pageviews across the month, an obvious pattern emerged, with the bulk of traffic occurring over a four day period. By isolating all digital activity over the four day period, it was clear that the traffic spiked immediately following the posting of a link to the clinical study on Google+. This Google+ content had been posted via hootsuite and the resultant link used the owl.ly domain (Spoiler Alert: this becomes important later).

By definition, traffic referred to a domain from an external site should be considered referral traffic. Those visitors directed to the case study page from Google+ should have appeared in Google Analytics as referral or social traffic. Using the real-time feature, I was able to confirm that the owl.ly Google+ links were being misattributed to direct traffic on most occasions (there were a few instances of correct attribution). This anomaly was tested with other link shorteners on Google+ but the opposite was found. Those links shortened using hlt.li and bit.ly were correctly attributed to referral or social traffic. Similarly, using direct links to the URL on Google+ allowed for the correct attribution of traffic; those visits were counted as social traffic.

So what does this mean to you, the savvy reader? I think there are two major takeaways. At the micro-level you should consider, whenever possible, the placement of direct links to destination URLs. Not only do direct links track appropriately (of critical importance to analytics), but they allow for added brand visibility in the anchor text.

At the macro-level, this scenario stresses the importance of good data analytics. Most anyone can pull data from a dashboard but it takes time and effort to get at the real story being told. Taking these results at face value, it appeared that there was a significant increase in direct traffic. But by digging into the data a story unfolded in which the link shortened Google+ traffic was being masked as direct traffic, thus minimizing social/referral contribution to overall web traffic.


[1] At this point in the process I went to my superior and we worked through the mystery together but given that I am writing this article I will be taking full credit.