Article Preview
TopIntroduction
The Internet is entering a social media era. Many social media sites, such as Facebook, Twitter, and Pinterest, have emerged and become very popular with users in recent years. Users spend a large amount of time on social media, interacting with families and friends, reading and spreading news, playing online games, posting pictures etc. What makes social media different from other Internet services is that they are built upon relationships and interactions between real people. There is a complex social network embedded in each social media site (Yin, Li, & Niu, 2014).
Social media have brought great business potentials for companies. One way to harness such potentials is to conduct viral marketing campaigns. As an effective marketing approach to spread information about products or services from consumers to consumers, viral marketing has existed for a long time, probably as long as the commerce itself. However, social media can elevate it to a new level because it can be implemented on an unprecedented scale on social media.
A viral marketing campaign starts with selecting a group of “seed” users who will be the first to receive information about a product or a service, by letting them try the product, for example. After learning about the product, some of the seed users will, hopefully, share the information or their experiences with their families, friends, co-workers etc., who in turn will share the information with their families and friends. As a result, information about the product will reach more and more people as time passes. The success of a viral marketing campaign is determined by the number of people eventually reached by the marketing message and how fast those people are reached.
A critical success factor of viral marketing is the selection of initial seed users (Gruhl, 2004). In general, seed users need to have a large number of social connections and are trusted by other people in social networks. These people are considered influencers, i.e. people who exert great influence over others. As a result, selecting the “right” seed users for a viral marketing campaign boils down to identifying influencers on social media. The issue has attracted much attention from both practitioners and researchers. In fact, several companies, such as Klout, Tellagence, and Little Bird, built their businesses around calculating people’s influence on social media (Karif, 2012). In Klout’s case, the company calculates a person’s influence, on a scale of 1 to 100, based on her profile and activities on social media, such as the number of Twitter mentions and the number of connections on Facebook. Many companies have used Klout’s scores in their viral marketing campaigns. For instance, in one viral marketing campaign, General Motors identified 900 users who have Klout scores of 50 or higher and offered them 3-day rental of Chevy Volt for free. The campaign resulted in a large number of tweets on Twitter and blog posts about the vehicle, most of which are favorable.
Researchers have been focused on the effectiveness of using various centrality measures to identify influencers on social media. A centrality measure, such as in-degree, out-degree, betweenness, and sender rank, captures a user’s typological characteristics in a social network. For example, a user’s in-degree on a social media site is the number of connections that she has through which information flows from other users to the user under consideration. An example of this is the number of people that a user follows on Twitter. Out-degree is opposite to in-degree and is the number of connections through which information flows from the user under consideration to other users. An example of this is the number of people that follow the user on Twitter. On some social media sites, such as Facebook, information often follows in both directions in a connection. Such a connection will be counted in both in-degree and out-degree. Existing studies found that out-degree, betweenness, and sender-rank are quite effective for identifying influencers in social networks (Cha, 2010; Gosh, 2010; Kiss, 2008).