Editor’s Intro: Given the importance, social media marketing has taken for many brands, it is surprising how little we actually know about what will make something go viral or not. Ranjana Gupta describes work she and her team have done on how to understand what constitutes “virality”, and how to predict it. This is really nice work to see and should stimulate further thinking and research.
From the colour of a dress to a #movement, it seems like anything can go viral. Often I find people discussing the triviality of something that has gone viral and yet they are unable to resist its pull. One checks the trend to see what it is about and makes it trend even more.
On the surface of it, ‘going viral’ seems to defy any predictability, any pattern. Digging deeper, a different story emerges. Things go viral-driven by two factors:
- Emotion: What is it that the content makes the individual feel. Here again, there are two things which matter: a) Valence – whether it is positive or negative in sentiment b) Arousal – the intensity of the sentiment. For example, both anger and sadness are negative in valence but anger has a higher intensity and therefore is more likely to move the individual to action, which in this case would be sharing or interacting with the content.
- Network: This is a composite measure of how many people react to the content and their circle of influence (people who follow them or are friends with them); it directly affects the probable reach of the content and hence its virality. In social media parlance, influencers are the ones who are well connected and active on social media and evoke reactions on their posts. They play as much a role in making a content go viral as the quality of the content itself. If more influencers share something, the chances of it going viral are higher.
Viral dissemination itself follows either the Roadblock or Stagger Effect. The first is a top-down effect resulting in concentrated impressions in a short time period (Rebecca Black’s ‘Friday’ would fall in this category) while the second is a bottoms-up approach that results in an equal number of impressions as the Roadblock but over a longer period of time. An example would be Gangnam Style (and on a different medium, Game of Thrones HBO series).
Having broken down virality into its constituents, predicting ‘virality’ starts seeming possible and that can be a powerful tool in the hands of marketers. Every day, they spend a lot of money and time on own content that they hope will ‘go viral’, giving their brands (and I include all forms of media and artists here) much needed buzz. They create the content, launch it and then hope it would do its magic. Predicting virality takes away the need to cross your fingers; it can help the marketers (I include everyone – from CMOs at FMCG giants to movie studios to Netflix) to decide which content will bring in the eyes and get the fingers tapping away at the share and reaction buttons.
By no means is virality prediction an easy task. To begin with, there is no universal definition of virality; it changes from publisher to publisher and from time to time. Facebook calculates virality as % of people who have created a story from the source page post out of the total number of unique people who have seen it. A story can consist of “liking, commenting or sharing the post, answering a question or responding to an event.” Youtube’s definition is even harder to come by.
However, if we look at ‘virality’ as not an absolute, single metric but a relative, composite metric, the task becomes a little easier. Consider a matrix with two driving factors for virality forming the two axes – Emotion and Network. If the content is high on both, it has ‘definitely viral’ potential and if it is ‘low’ on both, it has ‘no-viral’ potential. For true success in the digital world, both the emotional connection and the connection through the right people are important.
To test this hypothesis, we did a small exercise using twitter #tags. Four hashtags were chosen:
Sentiment analysis of the tweets which used the above hashtags gave us the Emotion connect. A combination of retweet and follower count was taken as Network score.
When these were plotted on the 2X2 matrix, #Gameofthrones and #Euro2016 scored high on both the axes while #BacktoBack was low on both. #BacktoBack scored higher than average on one axis, but its low score on the other meant that it fell short of the ‘definitely viral’ zone. These results corresponded with what actually happened on the platform – #Gameofthrones and #Euro2016 went viral while the other two hashtags did not.
This small experiment proves that it is possible to predict virality and if done right, it can be a powerful predictor of digital content’s success. Virality, as Jonah Berger has claimed, is not magic. It is science and therefore, it can be studied, predicted and put to work. Maybe it is time that we turn virality into an understandable metric from an inscrutable phenomenon.