Viral Tweet Test Results Part 1: Trending Topics and Forking URLs

Yesterday morning I launched a little off-the-cuff experiment in viral marketing via Twitter. I made a simple page on my site that asked people to retweet the link (and I specified a shortened URL from tinyurl to aid in tracking, or so I thought). I also asked people to comment on the page with their Twitter username and the name of the person who’s tweet led them here. And then I tweeted it, at first once, asking for retweets, and throughout the day I posted it maybe 4 or 5 times.

The response was bigger than I could have imagined. As of this writing there were over 200 total tweets by other users that used the words “viral tweet test.” I wanted to study not what makes things go viral in Twitter, but the infrastructure. I wanted to start answering questions like how do you track a viral meme or message’s organic growth and spread on Twitter (without hashtags)? What is a baseline exposure-to-retweet transmission rate? What would a mapped social graph of such a viral tweet look like? I’ll admit it was a pretty unscientific test, for a lot of reasons, but I did learn a lot.

The first thing I learned was that merely specifying one shortened URL ( does not mean that is what everyone will use, so using a tinyURL as a tracking method is flawed. Pretty early on that URL forked into a Twurl URL that some folks were using ( and some people even managed to get the page’s whole URL in their tweets.

I named the page “Viral Tweet Test” just as a quick label and it stuck, so using those words as a search worked pretty well, but even this method did not catch everything as some people called it a viral Twitter test, or used other words entirely.

The repeated use of this name in the tweets did produce one unexpected result: shortly after the launch of the test “Viral Tweet Test” was listed as a “trending topic” on Twitter Search. This actually drove less interest and traffic than one might assume, but my analytics system reports that it sent just over 100 visitors to that page (this number is artificially low because some URL shorteners do not carry HTTP referrer values. It appears that the phrase became a “trending topic” after about 35 tweets containing the phrase from approximately 30 different users. Once the stream of tweets started to slow, it fell off the list of topics.

The majority of momentum around this died off about 5 or 6 hours after the first few tweets, but even as of this morning, there is still a slow trickle of comment and retweets coming in. I’ll be analyzing the data more deeply in the next few days and will have some more quantitative stuff coming (the charts and graphs I’m so fond of). What data points would you like to see from this little experiment?