[Study] Where Does Viral Content Get Its Start?

For the last few years, viral marketing has been a reoccurring subject in the academic literature, the professional literature, and the blogosphere. Consumers are interested in consuming viral content. They eagerly share these YouTube videos, Internet memes, and web articles with their friends within social networks. Marketers are equally interested in creating the viral content. The Holy Grail of any marketer is to be able to get her customers to do her marketing for her. In a way, that’s exactly what viral marketing is.

This week, Jonah Berger (marketing professor at Wharton College of Business) has released a book called Contagious–a summation of the research he has done on viral marketing. Like Jonah’s book, much of the research in viral marketing has been focused on the content. What makes content worthy of being shared? What attributes does the content have that makes it compelling enough for people to proactively share it with their friends?

While these are certainly important questions to ask, there is something missing. Certainly, what kind of content you are creating makes a difference as to whether or not it will go viral. But could the initial audience to which the content is exposed also make a difference as to whether or not the content with go viral? After all, the point of viral marketing is to get a small number of people to spread the content to a larger number of people. So, that initial group that consumes the content is very important. Has there been any research done on how to select this initial group? It turns out there has…


YUPING, L. (2012). Seeding Viral Content: The Role of Message and Network FactorsJournal Of Advertising Research52(4), 465-478.

Yuping Liu-Thompkins of Old Dominion University has conducted a study that focuses on the effect the initial audience of a campaign has on its tendency to go viral. While previous research focuses on the content, Liu-Thompkins focuses on the distribution. Referred to by the author as “the viral diffusion process” and by industry professionals as “the seeding strategy,” this article reveals to marketers how to place content in front of people who are more likely to spread it and cause it to go viral.


The author uses insights from social-capital theory and social-network analysis to isolate four factors determining the appropriate group in which viral content should be seeded. These factors are:

  • The size of the initial group
  • The strength of the bond between the individuals and the marketer
  • The individuals’ levels of influence amongst their peers
  • The commonality of interests between individuals in the group

The author takes these factors into an account, comes up with a series of hypotheses, and then conducts an experiment using viral videos.


To assess the likelihood of content to go viral from various individuals, the author conducts an experiment with viral videos published to YouTube. When a video is first posted on YouTube, the creator’s subscribers are immediately alerted. Therefore, the author defines the initial group in this context as subscribers to the YouTube channels analyzed.

Over a period of 7 days, 101 videos were sampled (initially 105, but four were removed due to missing information) and information was collected about the publishers’ networks. Each of the videos was tracked daily over the course of 60 days, with each video’s views and ratings being tabulated.

At the end of the experiment, the author’s hypotheses were validated…

  • The larger the initial group is, the more viral the content will be. The study found that for every 100 additional subscribers, there was a 17% increase in the rate of diffusion.
  • A larger initial group will have a greater positive effect on the virality of the content when the quality of the content is low. The study revealed that, when the video had a lower rating, 100 additional subscribers contributed to a 30.72% increase in the diffusion rate. However, when the video had a higher rating, 100 additional subscribers only contributed to a 3.32% increase in the diffusion rate.
  • The stronger the bond between the content creator and the initial audience, the more viral the content will be. The study found that a 10% increase in the strength of bond between video creator and subscriber contributed to a 1.92% increase in the diffusion rate.
  • The larger the number of connections the individuals in the initial audience have, the less viral the content will be. (This one seemed counter-intuitive, but it was concluded from previous research that the greater number of connections one has, the weaker those connections are. Therefore, a negative relationship would be likely). The study indeed found a negative effect between the size of the individual subscribers’ networks and the virality of the content.
  • Moderate similarity in interests between individual members of the initial audience will lead to more viral content than low levels of commonality and high levels of commonality. The study found that low levels of interest lead to less viral content. Up to the point of 21.91% shared interests, commonality among subscribers lead to the content becoming more viral. However, after that threshold, the effect of increased commonality was negative. Therefore, the optimal level of shared interests among members of the initial audience is concluded to be about one-fifth.


The conclusion of the experiment can be re-stated as the following strategic recommendations for marketers developing a seeding strategy:

  1. Have as large of an initial audience as possible.
  2. Make up for a smaller initial audience by creating higher-quality content.
  3. Forge strong bonds between your brand and your initial audience.
  4. Focus on individuals who have a small number of strong connections.
  5. Find individuals for the initial audience who share a moderate amount of interests with each other.

The author concludes the paper with some commentary on previous research confirmed by this experiment: “To achieve successful diffusion, it is better to have a large number of easily influenced individuals than to have a few highly connected hubs in a social network.”


I suppose I find this study so fascinating, because it covers what often gets ignored–the customer. In creating content for marketing, we like to talk about the message and the medium. But what the about the audience? Other research operates under the somewhat arrogant assumption that marketers actually have control over the likelihood that their content will spread. It’s an “if you build it, they will come” mentality. This study flips that assumption and, instead, suggests that if you put it in front of the right people, then it will spread.

It’s not that businesses don’t have control. They do. But the control they really have has to do with who they decide to give control over to–that is, which people they will trust with their content. Gone are the days when consumers were primarily spectators. Viral marketing is the word-of-mouth of yesteryear on steroids. Now, when they tell their friends, consumers cover a lot more ground–reaching a lot more people in an increasingly shorter amount of time.

Though it isn’t explicitly stated as an implication of the article, I see this research as a wake-up call for marketers to realize that consumers are in control more than ever. The right strategy, then, isn’t to resist the momentum and try to recover the control. The right strategy is to try to influence which customers will have the control. The river will flow. You can’t stop it. But, if you’re smart, you just might be able to redirect it.


  1. How should the seeding strategy differ for various mediums of content (video, image, audio, text)?
  2. What is the correlation between content that goes viral and increased revenue for the content’s creator?
  3. On the social web, what amount of control do organizations really have over their true initial audiences?
  4. Does a stronger bond (more engagement) between brand and consumer translate into customer loyalty (repeat business)?

Leave a Reply