On Valentine's Day this year, HubSpot's resident Social Media Scientist Dan Zarrella delivered one of his most controversial blog posts to date: New Data: “Engage in the Conversation” May Not Actually Work. It was a great piece that prompted a ton of conversation since it goes against many of Social Media's core beliefs. While the conversation was brewing, I decided that we needed a better way to dive into the data to pull out more substance.
Since there are several data points that are of interest simultaneously to best paint the picture, I found myself immediately drawn to one of my favorite data visualization tools for large data sets: The Bubble Chart. What I found when when going through the data was that while there were no easy quadrants to identify groups or types of users, there was a "sweet spot" that has more or less a defacto engagement rate to target.
So let's break down this chart. The data is based upon a random subset of twitter users identified by sucking in a days worth of the Twitter "Firehose" stream (in my case just looking at the "Sprinkler" or 1% of all twitter traffic). For each user, I obtained up to the last 200 tweets as well as some overall user stats (i.e. followers, # of all time tweets, account creation date, etc.). From this raw data, I distilled some core stats that I thought would be of most interest. I examined each of the individual tweets per user and placed them into buckets of @Replies, Retweets and Other. I also calculated the average # of tweets per day for each user, as well as their percentage of @Replies and Retweets. Using this data, I sorted the list by # of followers descending (to ensure all data points are visible) and plotted the data points with the follower count representing the size of the bubble.
While there is a lot going on in this chart, what you can see is a bunching of data points (the "sweet spot") between 40-60% @Replies matched with 1-30 tweets/day. If you actually think about it, this makes a lot of sense. If you're replying more than that, you're most likely at risk of loosing followers since the tweet's relevance may be lesser to the greater audience. If you're not engaging enough, you lose the interactiveness of the medium and thus become more of a one-way communicator. Match this with the tweets rate (or tweets/day). If you never tweet, obviously you're not engaging. If you tweet too much, your followers may not be able to keep up. Thus, staying within the "sweet spot", you're best equipped for engagement success.
I'd be remiss if I didn't also look at the same effects of Retweets. The chart below shows that the "sweet spot" for Retweets is more like 0-10% at the same 1-30 tweets/day rate.
And here's the scatter chart to show the same picture without the bubble obstruction.
What do you think about these "sweet spots"? Do you personally find yourself falling within or are you a complete outlier?
Wednesday, October 5, 2011 at 11:01PM 
;)
;)
;)
;)
;)
;)
;)
;)
;)
