Monday, 12 December 2011

Content decay and timely delivery

We live our lives in a constant state of information overload. We are constantly bombarded by messages, images, symbols both real and virtual. Facebook twitter, YouTube accounts notify us of new events, updates, subscription feeds etc.  An endless stream of digital information. This makes getting our attention hard.  It begs the question, what are the chances of one item of information standing out and getting our attention?  
From bitly’s research on link half life (

Bitly, the URL shortening service,  selected 1000 popular bitly links and analyised the interest in that link over time. The above graph plots the half life for four different methods of referral. We can see that a link shared via facebook, twitter or directly by email or IM all follow a similar curve. There is an initial peak followed by a slow decay. The mean half lives are:

2.8 hours on twitter
3.2 for facebook
3.4 for email and IM

Only YouTube bucks the trend and has a half life of 7.4 hours. So it would appear that a link shared via YouTube will capture viewers interest for longer.

But this research is very biased. It only considered 1000 “popular” links. The exclusion of “unpopular” links and averagely popular links means these results are properly showing half lives longer than one would expect with normal content. In addition the research did not look the time of day of the posts, the content type etc. But given these factors we can still see the marked similarities across Twitter, Facebook, email and IM compared to YouTube.

Sysomos  has looked at a larger sample of 1.2 billion tweets over a 2 month period.

They found that only 29% of tweets got a retweet or a reply. 71% got no response. When they looked at the time span for the 29% they found an even more drastic drop off in interest.

96.9% or replies happened in the first hour. So the half life is drastically shorter than the bitly research. Assuming you can apply the matching found by Bitly we would expect Facebook, IM and email messages to follow a similar pattern.

Visibli studydone in 2010 found that the half life for likes on a brand’s facebook posting as 1 hour  20min.

Streaming information services such as Facebook, Twitter, IM and to an extent email are transient media. The old adage that today’s news is what you what you wrap tomorrows fish in, holds true but now the time span is measured in minutes rather than hours.

Email is a special case because of all the strategies and filtering options that users deploy. From a brand’s perspective this may make the potential of an email getting seen at all greatly reduced.

When it comes to video, Tubemogal have discovered that the half life of video content across multipul video services has dramatically declined over the last few years. In 2008 the half life was 2 weeks to get half its views within a 90 day period. Today in 2011 it takes 6 days. This is not shared content, as in the bitly research, it just represents the total views within a set period regardless of how the viewer discovered the video.

The picture does not look good. So how can a brand combat the ever increasing odds stacked against them in this attention economy?

So the evidence seems to indicate that video is the most long lasting medium. I think there are two fundamental reasons for this.
Firstly is the fact that Facebook, Twitter email, IM are information streams. They are like the news wire, or tickertape machine. If you are not constantly watching the stream you will miss information. The impetus to wade down stream to see old news is likely to be over weighted by the interest in the current stream.

However this still does not explain how the user comes to find out about the video content, as one would expect a large number of links to stem from these very media streams.

The way video is presents to the user is in an environment of related videos. One facebook element on my wall is only related to others in the stream by time and by a social graph. Not by other forms of metadata, such as similar content or subject matter. Each item in the stream stands alone and has no back of other items which may generate additional interest.

In the case of video, users browse and move through related subjects and versions or interpretations of the same subject.

Secondly is a deeper aspect of our reaction to the printed word as opposed to the image. The ability to represent meaning through the use of printed representations of words is a human invention. Our brains do not come with an innate ability to read and recognize words. We have to learn and apply a process to achieve this. Contrast this with our response to visual stimuli, where we have an innate ability to recognize and interpret what we see.

While video is linear and may take time to get to the point, it is more immediate and can bypass the need for the translation from symbolism to meaning required for reading. Of course YouTube videos live in symbiosis with their printed descriptions and meta data. But even if these were removed one could browse without any text references.

The downside of video is that it requires more time to digest and you are asking users to give you more of their precious time. The reward has to be worth it and the ability to judge this has to be made as early as possible into the video.

When using social media the aim must be to get the information in the most effective and appropriate medium in front of users at the most productive time. Get them when they come to drink at the stream.

A study by Virtue ( ) looked at the most effective times in terms of user engagement. They found 3 spikes at 11am, 3pm and 8pm, with 3pm being the highest activity.

They found that morning posts tended to be more effective. Most comments and posts appear in the afternoon, around 3pm, but posts in the morning performed 39.7% better in terms of user engagement than those posted during the afternoon spike of activity.

Posting in the first 15 minutes of the hour got greater interaction than posting at the end of the hour. They also discovered that posts with an image attachment gained more interaction than those with a video or just text. Simple text posts performed the worst.

While this is general data and any brand should also look at the technographics of their target user groups to see if their activitiy times have any specific patterns.

No comments:

Post a Comment