This might not be news, but still it is shockingÂ that a tweet half-life is only a few minutes.Â According to Betaworks, itâ€™s 5 minutes, on the contrary, bitly has it at 2.8hrs. (see below for explanation of the discrepancy)
Our data perspective
At Wiselytics, we had a chance to research in detail a million tweets and in continuation to aÂ previous blog post which had sparked some interest, we decided to go a step further to graph the shelf-life of a tweet vs. a Facebook post.
On the graph below, we present median engagement Â over timeÂ for Facebook posts and tweets, with a confidence interval of +/-5% (to show variance among accounts). Engagement for Tweets was measured with retweets.
tweet and FB post first 12hrs shelf-life Â Â Â Â Â Â
Half-life of Â a tweet is 24 minutes vs. 90 minutes for a Facebook post, hence close to 4 times shorter.
For Facebook, a post reaches 75% of its potential engagement in 5 hours. A median tweet reaches this 75% mark in less than 3hrs.
Something else interesting came out of the comparison:Â Twitter starts off much faster than Facebook, but then their shelf-life crosses at 87%. The few last retweets come much later on Twitter than the last engagement for a Facebook post. This is probably due to virality which is much more prominent on Twitter than on Facebook.
Even though it’s interesting, your tweet persistence won’t make it huge: less than 10% of RT will happen after 1 day. What matters the most is really the first couple of hours.
Why 24min and not 5min nor 2.8Hrs
We limited the analysis to Â tweets with at least 10 retweets. Reason is that for instance, a tweet with 2 retweets reaches its half-life as soon as it get its first RTâ€¦ which is statistically unstable.. Without this limitation, we estimated that the half-life is 6 min (1 billionth the half-life of Carbon14). Even though this is the Â«Â realÂ Â» estimation of half-life of a tweet, it doesnâ€™t show anything about the progression of a tweet.
This might also explain how bitly ends up with an estimated half-life of 2.8 hours. They had sampled Â«Â most popular bitly linksÂ Â» (and they measured bitly clicks, not retweets), hence focusing only on the tweets which do propagate a lot. In our study, we are interested in everyoneâ€™s tweet, not just the most popular ones.
And for our statistically savvy readers, we did try to model the data with exponential decay to estimate a true half-life, but it wouldn’t fit (because of the virality effect) without too strong hypotheses which would introduce more bias.
Donâ€™t miss the right timing or your tweet will be forgottenâ€¦ forever. Sure you could just retweet the same information, but you know itâ€™s forbidden do you?
To really optimize the distribution of your tweets, you could try to tweet every time of the day, every day, for a while and run some statistics.
Otherwise you could use our algorithms which know right away when your audience is the most receptive to your very own content. Our solution does it all for you: it optimizes the timing, posts your tweets and sequence of tweets Â and reaches a wider audience for you.
Our Facebook shelf-life study was run last August, and since then, many changes have occurred with Facebook ranking. We are refreshing our data and will present a comparison of Facebook shelf-life between last summer and now.
Â bitly for this post:Â Â http://bit.ly/1h9sFcM