How StumbleUpon Traffic reached the top
Statistics, it is said, are like bikinis. What they reveal is suggestive, but what they conceal is vital. Last week TNW reported, quoting Statcounter Global stats that StumbleUpon had surpassed Facebook in the US as the site that created the most social media traffic. It was also the second most popular traffic-generating site around the world after Facebook.
Statcounter’s analysis covered seven social media sites linking to over three million websites with about 15 billion monthly pageviews in 2010. It showed that during the year, Facebook and StumbleUpon are engaged in a merry dance to become the top social media site for link referrers. Twitter came in third place with Reddit, YouTube, Myspace and Digg bringing up the rear. This, of course, was surprising. How could a site with one of the smallest profiles have passed Twitter and Facebook to become one of the leading sites for link referring online?
Our Analysis of Stumbleupons traffic
Our analysis shows that Statcounter’s results area broadly accurate - Facebook and StumbleUpon commonly sent the most number of impressions to our sites. But this is only the begnning of the story, we were interested in the value of these users - were StumbleUpon’s users more engaged than those from Facebook or Twitter. We examined a number of sites over a four month period, October 15th 2010 to January 15 2011, to see how social media users engaged with the sites that the visited. To do this we looked at the most common social media referrers as suggested by Statcounter; Facebook, Twitter and StumbleUpon. Our results show that StumbleUpon users spend, on average, nine seconds on a website, compared with 78 seconds (1:18 minutes) for Facebook users and a massive 126 seconds (2:06 minutes) for Twitter users. Looking at the bounce rate, the number of users who visit the site and then quickly leave, 81% of StumbleUpon users bounce, while 77% of Facebook users do the same. Twitter recorded the lowest bounce rate of 70% in our sample survey. So what happens if we removed StumbleUpon from our analytics, well we found that our bounce rate was reduced by just over 1%.
So what does this mean for your site?
Getting down to brass tax - revenue - which users are worth the most? You probably won’t be surprised to hear that Facebook users are with the most, on average they are five times more valuable than StumbleUpon and Twitter users. In short StumbleUpon will send more users to your site than Twitter and Facebook but these users will be far less valuable to you. This is due to the browsing model that StumbleUpon employes, Stumblers, the term used for StumbleUpon users, are directed to websites through a “Stumble” on the site. The button randomly picks websites based on personal information that user provides. There is no decerning choice by the user to visit your site - they simply happen upon it.
The social web is not the full story
A final note on referrers, one segment of social media referrers was overlooked by Statcounter, that of email referrers. Our analysis showed us that users from Yahoo! mail, Gmail, Live/hotmail and other email systems, spent the most amount of time on each site, 290 seconds (4:50 minutes), they returned half as much revenue as the highest earning site, Facebook. Users referred by email clicked a similar number of on-site links as Twitter users. Our results suggest something which many social media gurus have known for some time, Facebook and Twitter users are the ones to target - these sites will generate a larger number of users and a greater number of engaged users. It is worth noting that Twitter sent the lowest number of impressions to the sites, about one fifth of those from StumbleUpon but this low number is in part due to the verity of sites, such as Hootsuite, Tweetdeck as well as site embeds, which can host user tweets. For consistency with Statcounter’s results we did not include such sites in our analysis. Also in my personal opinion the statcounter statistics are a little bit too "Good" to be reliable as shown in this impossible mirror image of statistic provided by statcounter