Skip to content

Article author: Essi Pöyry 5th February 2018

Anatomy of viral social media events – What explains event duration?

Social media scandals seem to have become increasingly more commonplace, but they also die at a faster pace. Both companies and individuals are often anxious of these scandals. Next time, could it be me in the eye of the storm? Then again, scandals are sometimes desirable if all publicity is thought to be good publicity.

Essi Pöyry

Researcher Essi Pöyry

In marketing, this kind of thinking falls to the category of viral marketing. Without a significant marketing investment, the discussion about one’s brand, for example, would spread like a virus. Social media is naturally important for viral marketing. Research has put a lot of effort in understanding what promotes the viral spreading of discussion topics. Three competing strategies are often discussed:

  • Seeding the topic to opinion leaders and influencers
  • Seeding the topic to a critical mass
  • Seeding the topic to individuals connecting otherwise unconnected communities (“bridge strategy”)

There is no consensus of the best strategy, but opinion leaders seem to play an important role in spreading the topic, simply because they create a lot of content to social media and their propensity to take part in marketing campaigns is high. On the other hand, the fact that social media belongs to the so called hybrid media system (meaning that social and legacy media are interlinked), celebrities might possess particular importance because the legacy media is interested in their doings.

Often research on virality focuses on the amount of discussion regarding a specific topic. One counts for example mentions, shares, downloads or views. It is less typical to study what factors explain the duration of a discussion topic. For example, discussion about the #metoo campaign has already lasted several months while the United Airlines scandal was hot only for a while.

We studied the topic in the Smarter Social Media Analytics research project by utilizing a service called Viral Watch developed by Futusome. The Viral Watch has been developed to identify discussion topics that start gaining unusually many mentions within a short period of time. The inspiration for the Viral Watch was the so called Mutti-gate that got started when chef Henri Alén in January 2014 wrote a praising tweet about the tomato sauce brand Mutti. As a consequence, Mutti cans were sold out in many stores.

The study analyzed different viral social media events identified between January 2015 and March 2017. Some events reached the same heights as the Mutti-gate but most of them quickly died after the algorithm identified it. We focused solely on hashtag and keyword-based events that received the highest ranking from the algorithm on the first day of its identification. This resulted in 1335 viral social media events.

We defined the duration of the event by counting the number of consecutive days the hashtag or the keyword was mentioned at least once. We ruled out events that lasted more than 30 days because they were considered continuous discussion topics. We were left with 960 viral social media events. We queried the Futusome API for different social media discussions that mentioned any of the hashtag or keyword. This included open Facebook discussions, Twitter, Instagram, blogs, news comments, etc. Finally, the data consisted almost 14 million distinct social media posts that mentioned at least one of the events we were interested in. The figure shows how the data was distributed across different channels.

The average duration of the events was three days for hashtag-based and nine days for keyword-based events. We also determined the average number of daily posts, authors, and sources, as well as the relative number of posts, authors and sources after the initiation of the event. By using different modeling methods, we found that there was one factor that clearly explained the duration of a viral social media event: The more sources the event was being discussed at, the longer the discussion lasted. The number of posts or authors did not significantly affect event duration.

Based on the results, it can be argued that discussion topics have a shorter life span within single social media channels and that new channels and communities are a prerequisite for their longer continuation. The result also supports the importance of individuals that connect otherwise unconnected communities from the perspective of viral social media event duration. It is also likely that when the legacy media joins the discussion, the discussion is spread to new channels in social media, as well.

Finally, here’s the key takeaway for those who are concerned of social media scandals: If the discussion about an unwanted topic is buzzing, don’t worry, it is not expected to last long – unless it spreads to new channels.

This blog post has originally been published in the blog of Rajapinta Association.

The study was presented in January 2018 at Hawaii International Conference on System Sciences conference with the title “Anatomy of Viral Social Media Events”, written by Essi Pöyry, Salla-Maaria Laaksonen, Arto Kekkonen and Juho Pääkkönen.



Read about use cases for web data

Check out our products