The Science of Viral Video

Recently YouTube published their most watched videos of 2009. This gives us an opportunity to look at these videos and to see if we can divine some rules of probability as to whether or not any given video has “what it takes” to go viral.

What Attributes do These Videos Have?

The first thing we are going to do is to watch each of the videos and come up with some words that distil the essence of the video. Of course each person is going to have their own opinion in this regard, but we can probably agree on at least the main theme. Having watched these videos I came up with the following themes: switch, cute, humour, children, music, BB hook. Of course these themes are not mutually exclusive and some videos share multiple themes. So what do these themes mean? Well, I think cute, humour, children and music speak for themselves. BB hook, is what I’ve termed a hook into a blockbuster, whether that be a movie or music release etc, if you have a video that somehow hooks into that then that’s a BB hook. A switch is when the audience is set up to think in one direction and then the producer suddenly “switches” your thinking to another direction. A good example of this is the Susan Boyle video. When she comes on stage you are set up to think that she is a frumpy, awkward, middle aged person and will probably have limited talent. This mindset is radically “switched” when she begins to sing.

How Can We Get From Themes to Probability?

Having decided what the main themes are, we have to work out a way to use them to measure how likely, or not, it is that our own videos will go viral. To do this I calculated each video’s share of the total views for those top 5 and expressed that as a percentage. For themes that were used in multiple videos I took an average. Since the sum of these percentages leads to a “whole” (or almost in the case of the averages) then we can use this as a crude probability of this theme going viral. Doing that yields this result:


If I Use Multiple Themes Does My Probability Increase?

The sample is too small to tell for sure. Intuitively you would think that if your video had both a good musical theme and combined that with some humour then you’d have a 22% probability that your video would go viral. This video, with over 1 million views, would seem to support that. But I believe that this is not the case. Instead, I think that it is the “size” of the theme that determines the probability of your video going viral. For example, the “size” (or impact may be a better word) is much greater in the Susan Boyle video than it is in this sports related video where the audience is set up to think this will be an interview with a new football signing and then the switch hits and we find out that the insurmountable language barrier makes for a funny few minutes. I believe it’s this impact that makes the former go viral and the latter to be merely popular.

What Lessons Can We Learn From This Year’s Top 5 Videos?

If you want a video to go viral, my advice to you is this. Use the “switch” theme, come up with the most powerful switch that you can, then combine it with another theme (switch/humour is powerful for example) and go with that.

Crowd Sourcing Research is Required.

Although useful for giving the budding viral marketeer some pointers, this research of mine is hardly conclusive. Statisticians may argue that a sample size of 5 videos is not significant, though I’d counter argue that combined views of 248 million probably means that it is. What is required is more research. We need to examine a sizeable volume of the corpus of YouTube videos and break them down into their themes and study the probability of those themes going viral. Of course the problem there is that there is a vast corpus of videos and there is no way that this can be fully automated, a human has to watch the video and decide what are the top 3 (say) themes of that video. The solution to this is crowd sourcing. I already have a method in mind for such a crowd sourcing experiment I just don’t have the infrastructure to host such a project. If any social scientists out there do have access to such infrastructure and want to collaborate on a project like this then leave a comment below.


Leave a comment

Filed under Social Media

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s