This is a guest post by Dan Zarrella who has observed the art of Retweeting..
Being interested in the science of social and viral marketing, the phenomenon of ReTweeting represents a wealth of insight into how links and ideas spread in social media, so I began by collecting ReTweets to understand their characteristics. When I collated the numbers below, my database had just over 84,000 ReTweets in it, and it now has over 130,000. I was working towards a system to visually map and analyze ReTweet streams (tree-like pattern a meme takes through Twitter, from person to person via ReTweeting). I also just released a beta of this visual map. I wrote a post on the data on my blog, but I’ll present some of that data here as well.
I graphed the length of the ReTweets in my database against a random sampling of other Tweets and found that ReTweets tend to be longer than other tweets, a condition that is probably partially due to the structure of a ReTweet. ReTweets generally contain some variation of the word “ReTweet” or its abbreviation “RT”. They generally also contain one or more @ links indicating the user they’re ReTweeting from.
“RT” is used more commonly than the full “ReTweet” to indicate that the tweet is a ReTweet. I’ll be doing further research into other variations of the word.
The word “please” occurs in ReTweets much more often than in other tweets, indicating that many tweets contain a call to action, explicitly requesting the ReTweet. Many times users will actually ask for user to ReTweet using the phrase “please ReTweet”.
The average rate of ReTweets per hour increases during the EST business day, and peaks between 10am and 4pm, showing that this may be the sweet spot during with to publish tweets designed to get ReTweeted.
Nearly 70% of ReTweets contain a link. This may suggest that ReTweets are a preferred tactic to spread external content on Twitter.
I’ve released a beta of the system I’m developing to analyze and map ReTweets. More granular and advanced analysis will be possible once I’ve finished developing the entire mapping system. Currently there is a search feature, which links to a collapsible tree-view of ReTweet streams like in the screenshot below. There is also a Most ReTweeted page that lists the most ReTweeted users for the last hour, day and week.
Dan Zarrella is a self-proclaimed social media and viral marketing scientist, check out his viral marketing blog for more of his posts, research and tools.
{ 11 trackbacks }
{ 8 comments… read them below or add one }
Great analysis. As twitter still seems fairly new, I’ve been interested in research about it. This fits the bill. Thanks.
- Marshall Jones, Jr.
(@MarshallJonesJr)
Are you considering e.g. “puppies are great pls retweet” to be a retweet? To me that’s just a tweet, whereas “RT @puppylover puppies are great pls retweet” is a retweet.
Very interesting. Helpful info for folks like us who sometimes want to spread info quickly.
nice piece of analysis. I’m quite fascinated by the Retweet Ripple effect – the potential for ultra-viral is implicit there.
1) RT for help or assistance
2) RT because of ‘cool’
I think these would be the 2 basic types. (although there may be others)
Wow – nice work! Looking forward to more. (@timhaines)
Don’t forget advertising. I once made a comment about a product I enjoyed using and discovered that someone working for the company responsible for that product retweeted my comment, plainly using it for advertising. That was upsetting to me.
This is a beautifully illustrated post and a great example of informative analysis. I am not the world’s greatest speller myself (just read my blog to see) yet I cannot help but feel that your third and fifth graphic will have more credibility if occurrence was spelt correctly in both. Sincerely, John.
intriguing