Furthermore, one should choose consumers who have strong ties with the advertiser and who also have strong influence on others, rather than simply wider reach. Among seed consumers, they should share a moderate amount of interest overlap instead of being too homogeneous or heterogeneous as a group. Introduction “Viral marketing” refers to the act of propagating marketing messages through the help and cooperation from individual consumers. It departs from traditional advertising in its reliance on consumer word of mouth (WOMB) instead of mass media as the message conveyance vehicle.
Compared with traditional advertising, viral marketing enjoys the benefits of lower cost, higher credibility, faster diffusion, and better targeting of consumers (Bamboo et al. , 2008; Double, Tolerant, and Overland, 2005). Furthermore, the emergence of online communities and social media in recent years have vastly extended individual consumers’ influence beyond their immediate circle of close friends to more casual acquaintances and sometimes even strangers (Dunn, Gu, and Winston, 2008). This significantly increased the scale of viral marketing, putting it into a more central position in company strategy (Ferguson, 2008).
Despite its increasing use, both marketing practitioners and researchers have pointed out the elusiveness of viral marketing success and a general lack of understanding of what drives the success of viral marketing efforts (Ferguson, 2008; Kalmia, McIntyre, and Masonic, 2007). Some see viral marketing as more of an art than a science (De Bruin DO’: 10. 2501 [JAR-52-4-465-478 and Linen, 2008). This view comes from the realization that viral marketing outcomes are affected by many factors that firms have limited control over.
Although these factors introduce a great deal of uncertainty into the viral-marketing process, it does to mean that firms cannot make informed decisions to maximize the possibility of success (Kalmia et al. , 2007). One such area that marketers may control is how to start the viral diffusion process-?what usually is referred to as the “seeding strategy. ” A seeding strategy involves determining how many initial consumers (“seeds”) are needed to disseminate a viral message to and what types of consumers to choose as seeds.
As these seed consumers are responsible for the initial dissemination of the viral message to other fellow consumers, selecting the right targets as seeds can eave a significant impact on later rounds of the viral diffusion process (Bamboo et al. , 2008; Watts and Preterit, 2007). Academic research on online viral marketing offers limited guidance on choosing a proper seeding strategy. Although existing studies have examined the impact of individual and content characteristics that affect the pass-along of viral information, few have explicitly addressed the proper choice of seed consumers.
December 2012 JOURNAL OF ADVERTISING RESEARCH 465 Seeding VIRAL coherent Addressing this gap in the literature, the current research draws from the sociological theory and social-network analysis to identify four key elements of the seeding strategy: ; the number of seeds to use, ; the strength of tie between seed individuals and the message originator, ; the level of influence of individual seeds, and ; the interest homogeneity among seed individuals.
Using viral videos from Youth as the backdrop, it empirically tests the relationship between these seeding decisions and viral-diffusion outcome. Because information for making these decisions easily can be obtained from observable online solicitations activities and internal customer data, the findings from this research can offer very practical guidance to optimally seeding a viral-marketing campaign. WHAT AFFECTS ONLINE VIRAL MARKETING SUCCESS? In the last 5 to 10 years, interest online viral marketing has increased among marketing and advertising scholars.
Studies in this area typically have focused on either intermediate actions/processes such as probability of opening and passing along viral information (e. G. , Ho and Dempsey, 2010; Phelps et al. , 2004), or end outcomes such as the eventual reach of a viral campaign and the adoption of the promoted product (e. . , Bamboo et al. , 2008; Acton, Cubes, and Scary, 2011). In answering the question of what affects viral marketing success, three types of factors have been suggested: ; message characteristics, ; individual sender or receiver characteristics, and ; social network characteristics.
Message characteristics relate to the content and creative design of a viral message, which are under the control of the advertiser (Ho and Dempsey, 2010; Kalmia et al. , 2007). An effective viral message should break through clutter and consumer indifference to encourage further pass-along of the message. Researchers have found, for instance, that humor and sex appeal are popular tactics used in viral messages (Goals and Stained, 2008) and that the social visibility of a viral message encourages its diffusion (Sugar et al. 2012; Slaking, Odds, and Watts, 2006). Besides message characteristics, individual consumers also play a critical role in the viral marketing process. This category of influence has received the most extensive examination in the literature. Findings in this area show that consumers’ personality traits (e. G. , Chic, Whish, Aka, and Lee, 2007; Sun, Young, Www, and Counterpart, 2006); anemographic (e. G. , Trusts, Baptist and Buckling, 2010); usage characteristics (e. G. , Indifference, Motto, Westfield, and Gordon, 2007; Sun et al. 2006); and motivation for sharing content (e. G. , Closest and Gripers, 2008; Phelps et al. , 2004) all can affect the success of viral messages. For example, researchers have found female and younger consumers tend to exert more influence on their targets and to be more susceptible to viral influences than male and older consumers (Acton et al. , 2011; Trusts et al. , 2010). Studies also have associated both extroversion and innovativeness with a higher tendency to pass along content (Chic et al. , 2007; Sun et al. 2006). From a motivational standpoint, research has consistently found altruism to drive message sharing (e. G. , Ho and Dempsey, 2010; Phelps et al. , 2004). Though individual characteristics focus on a single consumer, network characteristics describe the connection between consumers. The central thesis from this stream of research is that the structure of the social network through which a viral message spreads can affect the eventual reach and influence of the message (Bamboo et al. , 2008; De Bruin and Linen, 2008).
Furthermore, a consumer’s role in diffusion depends on his or her position in the social network as defined by the consumer’s relationship with others in the network, such as network centrality and tie strength (Goldenberg, Han Lehmann, and Hong, 2009; Kiss and Bicycle, 2008; Sugar et al. , 2012). Research in this area has often produced conflicting results. For instance, on the effect of network structure, some researchers have shown that a scale-free network, where only a few members have many connections, facilitates social contagion (BarbГis, 2002; Smith, Coyly, Lightproof, and Scott, 2007).
Others have found no difference, however, between a scale-free network and a random network, where most network members have a similar number of network connections (Kiss and Bicycle, 2008). Yet, a third study concluded that cascades were less likely to happen in a network where individual influence is highly unbalanced than in a random network (Watts and Odds, 2007). Gaps in the Literature Although academic researchers have started to construct a roadman of factors contributing to the success of online viral marketing, research in this area still has been very limited (Chic et al. 007; De Bruin and Linen, 2008) and has produced fragmented and sometimes conflicting results. More specifically, there are two important gaps in the literature that need to be addressed: ; Most existing studies have relied on computer simulations or consumer December 2012 surveys. Although simulation allows controlled experimentation with network properties that are difficult to implement in a field setting, results from such studies are constrained by parameter and model assumptions that often prove unrealistic in the real-world (Bamboo et al. , 2008).
As a result, the conclusion that a viral campaign an be more successful under a certain simulated condition may not mean it will happen in reality. Studies based on consumer surveys partly make up for this by offering a closer view of consumers’ attitudes and intentions. They suffer, however, from noise and bias often present in self-reported and retrospective data (De Bruin and Linen, 2008). Furthermore, previous survey studies often used a rather homogeneous sample such as college students to draw their conclusions and they were disproportionately focused on successful communications (De Bruin and Linen, 2008).
This limited the generalizations of the findings from these studies. Recognizing such limitations, other studies have issued a call for more research based on actual behavior of heterogeneous consumers in a natural setting (Bamboo et al. , 2008). ; Existing research often has failed to recognize the strategic nature of online viral marketing. Although online viral marketing often has been viewed more as an art than a science (De Bruin and Linen, 2008), a study of an online service provider showed that companies can tweak the inputs into a viral campaign to increase the chance of success (Kalmia et al. 007). To aid companies in such efforts, more research is needed to examine the decisions that advertisers can make in designing a viral-marketing campaign. Specifically, researchers have called for more analysis of a viral campaign’s seeding strategy (Bamboo et al. , 2008; Yang, Hay, Ma, and Chin, 2010), which defines the choice of consumers that companies should initially spread the viral message to. As these seeds will initiate viral propagation among fellow consumers, they can play a critical role in the eventual success of a viral campaign (Bamboo et al. , 2008; Watts and Preterit, 2007).
More research is needed to help identify ideal seed targets for viral- marketing campaigns. RESEARCH HYPOTHESES overview Addressing the gaps in the literature, the current research focused explicitly on the optimal seeding of viral- marketing campaigns. The author drew upon the socialistic theory and its paralleling social resources theory (Line, 1999; Porters, 1998) to identify important factors to consider when designing a seeding strategy. These theories state that one can derive significant tangible and intangible benefits from one’s social network and from the resources embedded in the network.
In other words, one’s social connections are a form of capital that can be utilized to attain one’s goals (Coleman, 1990). Although the relevance of social capital to brands is less obvious and less pervasive in the traditional advertising environment, the one-to-one interaction between businesses and consumers through social media has elevated the social plain for brands and has transformed brands into active participants in online social networks. From this view, a brand (or its company) can be considered an actor embedded in an extended network of consumers and other entities in the marketplace.
Its relationship with these consumers and other entities then comprise the social capital that it can draw upon for fulfilling its goals such as spreading a viral message or increasing brand awareness. An important benefit of social capital is the facilitation of information flow (Line, 2001; Van den Bullet and Witty, 2007), which, in essence, is what an advertiser aims for when it launches a viral campaign. To successfully propagate a viral message to a wide network of consumers, an advertiser needs to purposefully construct and mobile its social capital for optimal outcomes (Porters, 1998).
This process involves careful selection of the initial target consumers (I. E. , seeds) to maximize access to and manipulation of resources within the advertiser’s social network. In this respect, the social capital theory suggests three dimensions that should be considered (Line, 2001) ; : the extension of ties, which is captured by the size of the network (Broodier, 1986); ; relationship strength between the focal actor and its network connections (e. G. , Grandmother, 1973; Line, Ansell, and Vaughn, 1981); ; the resources embedded within the network as held by its members (Line, 1999).
The last dimension can be further broken down into the level and the diversity of resources possessed by network entities (Line, 2001; Van den Bullet and Witty, 2007). Based on these dimensions, the current research identified four critical aspects of a seeding strategy: ; seed network size, ; tie strength, ; seed influence, which signals resource level, and ; seed homogeneity, which serves as an indicator of seed resource diversity. 467 More specifically, when designing the seeding strategy of a viral campaign, an advertiser needs to answer four questions: ; How many seeds should be used? Should these seed consumers have strong or weak ties with the firm/ brand? ; Is it superior to use seed consumers who are social hubs with a large number of connections with other consumers? ; Should seed consumers be chosen from a heterogeneous or homogeneous population? By addressing these questions and empirically testing the effects of these decisions on actual diffusion outcomes, the current research aims to provide a systematic guide to choosing the best seed consumers for initializing a viral-marketing campaign.
Number of Seeds When picking the seed individuals to spread a viral message, a natural first consideration is owe many seeds to use. From a social-capital standpoint, the more network connections that are embroiled in a given situation, the more resources will be available for the focal actor to utilize in achieving its objectives (Burt, 1997; Line, 1999). This supports the use of a large number of seeds. From a mathematical perspective -?other things being equal-?the larger the number of seeds, the more opportunity there is for a message to reach other consumers and the more likely the message will create an impact.
This is the basic idea behind massed advertising and explains he popularity of advertising during major events such as the Super Bowl. The downside to using many seeds, however, is the high cost associated with the strategy. This partly defeats the cost-effective nature of viral marketing. Consequently, there is a tradeoff between using many seed individuals and maintaining a low viral campaign cost. To determine the right balance, some insight can be gleaned from epidemiology research.
A key concept in that literature is the basic reproductive ratio, defined as the expected number of secondary infections an infected individual will cause (Hoffmann, Smith, and Wahl, 2005). When this ratio is smaller than one, the network will show subtropical growth (Bamboo et al. , 2008), and the disease will wane out without saturating the population. When the ratio is greater than one, by contrast, a truly viral process is established, and exponential growth will be experienced through generations of the disease propagation process.
The larger the ratio, the more likely an epidemic will occur and affect the entire population (Hoffmann et al. , 2005). The relevance of this basic reproductive ratio to the seeding decision lies in its impact on how important the size of the initial seed roof is. When the ratio exceeds one and exponential growth ensues, having many initial seeds is not critical. The main consideration in this context is to have enough seeds to ensure that the propagation does not stop early (Bamboo et al. , 2008).
When the ratio is smaller than one, however, the size of the seed group becomes much more important and can determine the final reach of a disease or campaign. In such situations, there is research that advocates a “big-seed” approach, where a large number of seeds are used (Watts and Preterit, 2007). Applying the preceding to a viral racketing message, the importance of the initial seed group size may be contingent on the likelihood of seed consumers to pass along the viral message to other consumers.
Although various message and individual characteristics can affect this likelihood, the current research focuses on one factor: quality of the viral message. In this case, “quality’ broadly is defined as consumers’ general evaluation of the message. A high-quality message may be one that is creative, entertaining, informative, or socially valuable; a low- quality message, conversely, may be one that fails to pique interest among consumers.
In the former situation, the pass-along rate (hence basic reproductive ratio) is likely to be high, and the number of seed consumers becomes less important; in the latter situation, the number of seed consumers may determine the final outcome of a viral campaign. This leads to the first two hypotheses: HI: The number of seeds will have a positive effect on the diffusion of a viral message. H2O: The relationship in HI will be stronger when the viral message quality is low than when the message quality is high.
Strength of Tie with Seeds Besides choosing the right number of seeds to start a viral campaign, it is also important to consider the trench of connection a viral content creator has with seed consumers. For instance, in the case of viral brand messages, companies may want to consider tie strength as measured by brand loyalty or brand usage. According to the social-capital theory, strong ties provide ecological reasons for network members to lend resources to others, not necessarily for direct gain from the borrower but for reputation and other benefits that one can derive from the entire network (Burt, 1997; Line, 2001).
As a result, network members who are connected to the focal actor through a strong tie ill be more motivated to cooperate with the actor than those connected via a weak More specifically, in the context of an online viral-marketing campaign, a few advantages can result from having a strong tie with seed consumers: ; In today’s already-cluttered online environment, information shared through a strong tie is more likely to be noticed. Either due to higher interest or due to social pressure, consumers are more likely to open messages sent from an entity that they feel close to (De Bruin and Linen, 2008). For similar reasons, stronger ties can also increase he possibility that the message will be passed along to others (Chic et al. , 2007), which is crucial to starting later stages of the viral process. ; When persuasion is the goal, information shared through a strong tie tends to be more persuasive and, therefore, can have a larger influence on the recipient (Banal and Voyeur, 2000; Sun et al. , 2006). These advantages suggest the superiority of choosing seed consumers who have a strong tie with the firm.
It should be noted, however, that the social- capital literature also has suggested a critical role played by weak ties (Grandmother, 973). The argument is that weak ties connect network clusters that would otherwise be isolated from one another and, as a result, can increase the reach of a viral message (Goods and Manikin, 2004). A field experiment compared the effectiveness of spreading WOMB through a restaurant’s loyalty-program members (I. E. , regular customers) versus non-customers recruited from a thirdly panel by Byzantine (Goods and Manikin, 2009).
The authors concluded that WOMB initiated by non- customers (I. E. , weak company-consumer tie) created more incremental impact than that initiated by customers (I. E. Strong countermeasure tie). It should be noted, however, that the social-capital literature also has suggested a critical role played by weak ties. Although such results seem to suggest that one should select consumers with weak ties to the company as viral campaign seeds, a few issues undermine the strategic appropriateness of this decision.
First, as the authors pointed out, the results do not reflect the impact of overall WOMB but rather the incremental influence of WOMB by the sample consumers beyond existing WOMB (Goods and Manikin, 2009). As loyal customers likely already spread words about the assistant, their true impact is likely to have been underestimated. Second, non- customers have not experienced the company that is being promoted. As a result, their testimonials may be considered less credible and trustworthy than those conveyed by consumers who have a strong tie with the company.
Third, as consumers with no (or weak) ties to a company have low motivation to spread words about the business, the company may need to provide extra financial incentive to encourage these consumers’ participation in the viral process. In the case of the foregoing study, Byzantine compensated the non-customer sample for participation in he panel. This can make the use of weak ties a more costly strategy. For these reasons, the current research argues that, for the purpose of initially seeding viral content, it still is more effective to select consumers who have stronger ties to the content generator than consumers with weak ties.
This leads to the third hypothesis: HA: Seeding individuals who have strong ties with the message Seed Influence As seed consumers start to pass along a viral message to fellow consumers, the extent of influence each seed has can play an important role in further spreading of the message. This level of influence represents the social capital a seed consumer possesses that an advertiser can indirectly leverage in its viral campaign (Van den Bullet and Witty, 2007).
A frequently used proxy for influence is the number of connections an individual has (Wassermann and Faust, 1994), which has been shown to follow a power-law distribution in online networks (BarbГis, 2002). In a power-law distributed network, a small number of nodes have disproportionately large numbers of connections, whereas the rest have only a small number of connections. The former forms the hubs of the network. Given the large disparity between hubs and non-hubs, a key question is which type of these consumers is better suited for seeding viral content. The answer to this question is not exactly straightforward.
The well known two-step flow model of communication emphasizes the role of influential individuals in propagating information to a wider audience (Katz and Leasehold, 1955). This view was echoed in previous research on the viral diffusion of innovation (e. G. , Goldenberg et al. , 2009; Rogers, 1962). The basic argument is that the more individuals that a seed is creator will lead to more successful diffusion than individuals who have weak ties with the message creator. 469 connected to, the more people the seed can reach and potentially influence, creating what Van den Bullet and Josh (2007) call a “multiplier effect. This well-established argument has been challenged by recent studies, however. Specifically, one study contended that, instead of relying on influential, it was more important to have a large mass of easily influenced individuals for viral diffusion to succeed (Watts and Odds, 2007). The simulation in that research showed that the cascade window-?a egging in which large-scale diffusion was likely to occur-?resided at a rather low (or moderate) average number of connections. Several researchers have provided a theoretical explanation for why hubs are not necessarily better.
Due to the cost of maintaining a large network, individuals with many connections on average have weaker connections, which results in less impact on others that are connected to them (Acton et al. , 2011; Smith et al. , 2007). This weaker relationship can be especially detrimental in spreading viral marketing messages, due to the large number of messages circulating online. As one focus-group discussion revealed, individuals receiving a pass-along message often felt irritated or angry for their wasted time (Phelps et al. , 2004).
Therefore, when someone with a large number of connections tries to pass along a viral message to his or her weak connections, the message will be less likely to be relevant to the recipient and more likely to be ignored. For these reasons, the author expects a negative relationship between the number of connections a seed consumer has and the outcome of viral diffusion. This leads to the next hypothesis: HA: The number of connections seed consumers have ill have a negative effect on the diffusion of a viral message. Seed Homogeneity Network “homogeneity’ is the degree to which members of a network are similar to one another.
Just as birds of a feather flock together, researchers have found a tendency for humans to connect with others who are similar to them, a phenomenon called homophony (McPherson, Smithsonian, and Cook, 2001). In an online community, homophony is likely to occur as well. Users with similar backgrounds and tastes are likely to seek out and consume similar content and, as a result, are more likely to know and connect with each other. Network homogeneity can be assessed in multiple domains. Partly due to operational simplicity, most research has relied on demographic and socioeconomic variables to define similarity (e. G. De Bruin and Linen, 2008; Kalmia, 1998; Loach, 2000). In an online community, however, some of these demographic variables either become less relevant (e. G. , geographic distance) or are often unknown (e. G. , age and profession) to other individuals participating in the community. Instead, users tend to be guided by their mutual interest in certain topics such as sports, politics, or humor. Past research further has suggested that such deeper-level similarities are more important in a group setting than surveillance similarities characterized by factors such as race and gender (Phillips, Northeast, and Neal, 2006).
For this reason, the current research focuses on homogeneity as defined by the level of shared interest among seed consumers. This resembles the concept of perceptual affinity, which refers to individual similarity in personal values, experiences, and tastes (De Bruin and Linen, 2008). However, instead of using perceived affinity as reported by survey data in that duty, the current study derives homogeneity from interests manifested in actual content consumption behavior by seed consumers. This helps avoid potential recall error or bias that may be present in one-sided reports of perceptual affinity.
Regarding the consequence of network homogeneity, past research has produced inconsistent findings. Some studies have argued that similarities among individuals may facilitate information flow, as shared values and experiences among these individuals encourage more frequent and easier interaction with each other (McPherson et al. , 2001; Watts, 2003). In support of this theory, within the marketing literature, the homogeneity of a social system has been found to expedite diffusion and increase eventual market size (Agitation, Legalities, and Robertson, 1989).
More recent studies, however, have questioned this conclusion. Data from an online travel agency’s viral campaign demonstrated that diffusion speed was negatively affected by homogeneity (Lee, Lee, and Lee, 2009). In a cross-cultural setting, income homogeneity in a country was shown to lead to a lower diffusion rate (Van den Bullet and Stretchers, 2004). In line with these findings, another study found that encephalographic similar ties decreased the effectiveness of viral messages in terms of awareness, interest, and adoption (De Bruin and Linen, 2008).
This latter group of findings can be explained by the social-capital theory, where having diverse embedded resources in a network is considered to increase social capital and improve the chance of finding the right resources needed to achieving one’s goals (Burt, 2005; Line, 1999). Reconciling the disparate findings, the current research argues that the impact of seed homogeneity on diffusion success does not follow a linear relationship. Instead, it is an inverted U-shaped pattern, where both low and high levels of homogeneity can be detrimental to diffusion. 70 At very low levels of homogeneity, the network consists of an eclectic and somewhat random set of connections. For these heterogeneous individuals, group identification, tie strength, and stability are low (Lee et al. , 2009), and it is more difficult to effectively reinforce social norm (Alleghenies, Dollar, and Herrmann, 2005; Horn, 2008). Network members do not have strong incentives to pass along any particular message to others within or outside the network, which impedes the flow of information and the diffusion of viral messages.