@mikecane on Twitter called my attention to this eye-opening explanation of social media referral traffic on the awe.sm Blog. I have been aware of this issue for some time, and Thomas Baekdal has previously written about it here. Both articles are well worth reading.
The most recent article, entitled “Twitter Drives 4 Times As Much Traffic As You Think It Does,” starts with a nice explanation of how referrer analysis works and why this model needs to be updated to take into account traffic from dynamic web applications such as desktop or mobile clients as well as apps and other services that use APIs to access web data. Traffic coming from such applications aren’t accounted for in the current referrer mechanism. Worse, traffic coming from intermediate sites aggregating this information through use of APIs may mislead referrer data tracking systems by attributing the intermediate system as the source referrer instead of the site on which the data was originally published.
Awe.sm is a tracking service that offers social media analytics for professional publishers and marketers. To illustrate these points, they used their data from the first half of 2011, to analyze referral traffic driven by Twitter with some surprising results.
…the referral traffic one sees from Twitter.com is less than 25% of the traffic actually driven by Twitter.
…more than 1 in 8 visits driven by Twitter sharing are actually referred from other sites.
Their findings are consistent with my own observations based on traffic analysis from the blogs I manage, including this one, and views of my webdocs on webdoc.com.*
Over the past few months I have noticed discrepancies in the referral information compared with actual traffic to my posts, and I have been performing a number of informal experiments with items I have posted in these places using different ways of publicizing them and comparing the results. WordPress provides referral data including Twitter referrals. For webdocs, I use bit.ly, which I have used for link shortening since long before Twitter had a link shortening service of its own. In both cases, I have consistently found that when I publicize my posts exclusively on Twitter I receive far more traffic than either WordPress or bit.ly is able to account for. The number typically ranges between 6-8 times the traffic attributed to Twitter, and in a few cases it has been at least 10 times or more.
I had been hesitant to report my experiments, since I had not yet been able to perform any controlled tests that would confirm the extent to which the efficiency of Twitter in publicizing my posts was being underestimated. The latest analysis by awe.sm and the examples they provide of syndicated Tweets leading to signficant sources of traffic corroborate my own findings. However, the discrepancy between actual traffic volume and bit.ly clicks history remains a mystery, as it would not seem to be vulnerable to the same problems as the referral data. I hope I will be able to do some controlled tests in the near future to better quantify the magnitude of these discrepancies.
*Disclaimer: Although I am currently Community Ambassador for webdoc.com, I have used the information from my personal account to perform these tests. This information, which consists exclusively of the number of view of my own webdocs, would be available to any webdoc.com user.