www.elsevier.com/locate/intmar

Available online at www.sciencedirect.com

ScienceDirect
Journal of Interactive Marketing 28 (2014) 43–54

Consumer Decision-making Processes in Mobile Viral Marketing Campaigns

Christian Pescher & Philipp Reichhart & Martin Spann ⁎

Institute of Electronic Commerce and Digital Markets, Ludwig-Maximilians-Universität München, Geschwister-Scholl-Platz 1, D-80539 München, Germany

Available online 22 November 2013

Abstract

The high penetration of cell phones in today’s global environment offers a wide range of promising mobile marketing activities, including
mobile viral marketing campaigns. However, the success of these campaigns, which remains unexplored, depends on the consumers’ willingness to
actively forward the advertisements that they receive to acquaintances, e.g., to make mobile referrals. Therefore, it is important to identify and
understand the factors that influence consumer referral behavior via mobile devices. The authors analyze a three-stage model of consumer referral
behavior via mobile devices in a field study of a firm-created mobile viral marketing campaign. The findings suggest that consumers who place
high importance on the purposive value and entertainment value of a message are likely to enter the interest and referral stages. Accounting for
consumers’ egocentric social networks, we find that tie strength has a negative influence on the reading and decision to refer stages and that degree
centrality has no influence on the decision-making process.
© 2013 Direct Marketing Educational Foundation, Inc. Published by Elsevier Inc. All rights reserved.

Keywords: Mobile commerce; Referral behavior; Sociometric indicators; Mobile viral marketing

Introduction

The effectiveness of traditional marketing tools appears to be
diminishing as consumers often perceive advertising to be
irrelevant or simply overwhelming in quantity (Porter and
Golan 2006). Therefore, viral marketing campaigns may provide
an efficient alternative for transmitting advertising messages to
consumers, a claim supported by the increasing number of
successful viral marketing campaigns in recent years. One
famous example of a viral marketing campaign is Hotmail,
which acquired more than 12 million customers in less than
18 months via a small message attached at the end of each
outgoing mail from a Hotmail account informing consumers
about the free Hotmail service (Krishnamurthy 2001). In addition
to Hotmail, several other companies, such as the National
Broadcasting Company (NBC) and Proctor & Gamble, have
successfully launched viral marketing campaigns (Godes and
Mayzlin 2009).

In general, a viral marketing campaign is initiated by a firm that
actively sends a stimulus to selected or unselected consumers.
However, after this initial seeding, the viral marketing campaign

⁎ Corresponding author.
E-mail addresses: [email protected] (C. Pescher),

[email protected] (P. Reichhart), [email protected] (M. Spann).

1094-9968/$ -see front matter © 2013 Direct Marketing Educational Foundat
http://dx.doi.org/10.1016/j.intmar.2013.08.001

ion, In

relies on peer-to-peer communications for its successful
diffusion among potential customers. Therefore, viral market-
ing campaigns build on the idea that consumers attribute higher
credibility to information received from other consumers via
referrals than to information received via traditional advertising
(Godes and Mayzlin 2005). Thus, the success of viral marketing
campaigns requires that consumers value the message that they
receive and actively forward it to other consumers within their
social networks.

Mobile devices such as cell phones enhance consumers’
ability to quickly, easily and electronically exchange informa-
tion about products and to receive mobile advertisements
immediately at any time and in any location (e.g., using mobile
text message ads) (Drossos et al. 2007). As cell phones have the
potential to reach most consumers due to their high penetration
rate (cf., EITO 2010), they appear to be well suited for viral
marketing campaigns. As a result, an increasing number of
companies are using mobile devices for marketing activities.

Research on mobile marketing has thus far devoted limited
attention to viral marketing campaigns, particularly with respect to
the decision-making process of consumer referral behavior for
mobile viral marketing campaigns, e.g., via mobile text messages.
Thus, the factors that influence this process remain largely
unexplored. The literature on consumer decision-making suggests
that consumers undergo a multi-stage process after receiving a

c. Published by Elsevier Inc. All rights reserved.

44 C. Pescher et al. / Journal of Interactive Marketing 28 (2014) 43–54

stimulus (e.g., a mobile text message) and before taking action
(e.g., forwarding the text message to friends) (Bettman 1979; De
Bruyn and Lilien 2008). At different stages of the process, various
factors that influence consumer decision-making can be measured
using psychographic, sociometric, and demographic variables as
well as by consumer usage characteristics. Whereas previous
studies have mainly focused on selected dimensions, our study
considers variables from all categories.

De Bruyn and Lilien (2008) analyzed viral marketing in an
online environment and discussed relational indicators of
business students who had received unsolicited e-mails from
friends. This study provided an important contribution and
amplified our understanding about how viral campaigns work.
The present paper differs from the work of De Bruyn and Lilien
(2008) and goes beyond their findings in four important ways:
actor, medium, setting, and consumer characteristics. The first
difference is the actor involved. In viral campaigns, the
initiator, usually a company, sends the message to the seeding
points (first level). Next, the seeding points forward the
message to their contacts (second level), and so on. Whereas
De Bruyn and Lilien (2008) focused on the second-level actors,
the present study focuses on the first-level actors, e.g., the direct
contacts of the company. We believe that for the success of a
campaign, additional insights into the behavior of first-level
actors are very important because if they do not forward the
message, it will never reach the second-level actors. The second
difference is the medium used in the campaign. Although we
cannot explicitly rule out that participants of De Bruyn and
Lilien’s (2008) campaign used mobile devices, they conducted
their campaign at a time when the use of the Internet via mobile
devices was still very uncommon. Therefore, it is reasonable to
assume that at least the majority of their participants used a
desktop or a laptop computer when they participated in De
Bruyn and Lilien’s (2008) campaign. In contrast, the present
study explicitly uses only text messages to mobile devices. In
addition, mobile phones are a very personal media which is
used in a more active way compared to desktop or laptop
computers (Bacile, Ye, and Swilley 2014). The third difference is
the setting in which the viral campaign takes place. Whereas the
participants in the study by De Bruyn and Lilien (2008) were
business students from a northeastern US university, we conduct
a mobile marketing campaign in a field setting using randomly
selected customers. The fourth and most important difference is
that De Bruyn and Lilien (2008) focused exclusively on relational
characteristics. In addition to relational characteristics, this
paper also considers variables that describe demographic factors,
psychographic factors, and usage characteristics. As these
variables yield significant results, the study and its findings go
beyond the findings of De Bruyn and Lilien (2008).

The main goal and contribution of this work is, first, to
analyze consumers’ decision-making processes regarding their
forwarding behavior in response to mobile advertising via their
cell phone (i.e., text messages) in a mobile environment using a
real-world field study. To analyze consumers’ decision-making
processes, we use a three-stage sequential response model of
the consumer decision-making process. Additionally, we inte-
grate consumers’ egocentric social networks into a theoretical

framework to consider social relationships (e.g., tie strength,
degree centrality) when analyzing mobile viral marketing
campaigns. Thus, to understand referral behavior, we integrate
psychographic (e.g., usage intensity) and sociometric (e.g., tie
strength) indicators of consumer characteristics. We are then able
to determine the factors that influence a consumer’s decision to
refer a mobile stimulus and are able to identify the factors that
lead to reading the advertising message and to the decision to
learn more about the product.

Related Literature

Viral Marketing and Factors that Influence Consumer Referral
Behavior

Viral marketing campaigns focus on the information spread of
customers, that is, their referral behavior regarding information or
an advertisement. Companies are interested in cost-effective
marketing campaigns that perform well. Viral marketing cam-
paigns aim to meet these two goals and can, accordingly, have a
positive influence on company performance (Godes and Mayzlin
2009). Companies can spread a marketing message with the
objective of encouraging customers to forward the message to their
contacts (e.g., friends or acquaintances) (Van der Lans et al. 2010).
In this way, the company then benefits from referrals among
consumers (Porter and Golan 2006). Referrals that result from a
viral marketing campaign attract new customers who are likely to
be more loyal and, therefore, more profitable than customers
acquired through regular marketing investments (Trusov, Bucklin,
and Pauwels 2009).

Two streams of research can be identified. The first is the
influence of viral marketing on consumers, and the second is
research that has analyzed the factors that lead to participating
in viral marketing campaigns. First, previous research identified
that viral marketing influences consumer preferences and pur-
chase decisions (East, Hammond, and Lomax 2008). Further, an
influence on the pre-purchase attitudes was identified by Herr,
Kardes, and Kim (1991). In addition, viral marketing also
influences the post-usage perceptions of products (Bone 1995).

Second, previous research has identified satisfaction,
customer commitment and product-related aspects as the most
important reasons for participating in viral marketing campaigns
(cf., Bowman and Narayandas 2001; De Matos and Rossi 2008;
Maxham and Netemeyer 2002; Moldovan, Goldenberg, and
Chattopadhyay 2011). With respect to psychological motives,
self-enhancement was identified as a motive for consumers to
generate referrals (De Angelis et al. 2012; Wojnicki and Godes
2008). The importance of self-enhancement in addition to social
benefits, economic incentives and concern for others was identified
as a motive behind making online referrals (Hennig-Thurau et al.
2004). Referrals can be differentiated into positive and negative
referrals. Anxiety reduction, advice seeking and vengeance are
factors that contribute to negative referrals (Sundaram, Mitra, and
Webster 1998).

Within the referral process, the relationships and social
network position of the consumer are also influential. For
example, Bampo et al. (2008) found that network structure is

45C. Pescher et al. / Journal of Interactive Marketing 28 (2014) 43–54

important in viral marketing campaigns. Furthermore, it has
been determined that consumers are more likely to activate
strong ties than weak ties when actively searching for
information (Brown and Reingen 1987) because strong ties
tend to be high-quality relationships (Bian 1997; Portes 1998).
In addition, targeting consumers who have a high degree
centrality (e.g., quantity of relationships) leads to a higher
number of visible actions, such as page visits, than do random
seeding strategies (Hinz et al. 2011). Kleijnen et al. (2009)
analyzed the intention to use mobile services using sociometric
variables and evaluated how consumers’ network positions
influence their intentions to use mobile services. However, the
previous study contributes to the literature by analyzing a different
research question than is examined in our paper. Specifically,
Kleijnen et al. (2009) focused on the intention to use services,
while our study focuses on consumers’ decision-making processes
until they make a referral. In summary, previous research focused
on the consumers’ psychographic constructs or relationships and
social networks to explore why consumers participate in viral
marketing campaigns and why they make referrals, two constructs
that are rarely analyzed together. Iyengar, Van den Bulte, and
Valente (2011) used both constructs jointly and found that
correlations between the two are low. However, this study did
not take place in an online or mobile context but rather in the
context of referrals for new prescription drugs between specialists.
In contrast, our study analyzes both aspects together within a
mobile viral marketing campaign.

In addition to offline- or online-based viral marketing activities,
an increasing number of companies are conducting marketing
campaigns using mobile phones, and promising approaches
include mobile viral marketing campaigns. Research on mobile
viral marketing is relatively unexplored because most research in
the field of mobile marketing analyzes marketing activities such as
mobile couponing (Dickinger and Kleijnen 2008; Reichhart,
Pescher, and Spann 2013), the acceptance of advertising text
messages (Tsang, Ho, and Liang 2004) or the attitudes toward
(Tsang, Ho, and Liang 2004) and the acceptance of mobile
marketing (Sultan, Rohm, and Gao 2009). In the context of mobile
viral marketing research, Hinz et al. (2011) studied mobile viral
marketing for a mobile phone service provider and determined that
the most effective seeding strategy for customer acquisition is to
focus on well-connected individuals. In contrast to our study, their
referrals were conducted via the Internet (i.e., the companies’
online referral system) rather than via a mobile device (i.e.,
forwarding the text message immediately). Nevertheless, generat-
ing referrals using a mobile device can affect referral behavior.
Palka, Pousttchi, and Wiedemann (2009) postulated a grounded
theory of mobile viral marketing campaigns and found that trust
and perceived risk are important factors in the viral marketing
process. In comparison to our study, they used qualitative methods
and did not conduct a real-world field study. Okazaki (2008)
identified, for Japanese adolescents, consumer characteristics such
as purposive value and entertainment value are the main factors
in mobile viral marketing campaigns and that these factors
significantly influence the adolescents’ attitudes toward viral
marketing campaigns. Furthermore, both purposive value
and entertainment value are influenced by the antecedents’

group-person connectivity, commitment to the brand, and
relationship with the mobile device. In contrast to our study,
Okazaki (2008) did not analyze whether referrals were made,
nor did he analyze the referrals that were directly made via a
mobile device by forwarding the mobile text message. Instead, he
analyzed the general viral effect in the form of telling or
recommending the mobile advertising campaign. Further, our
field study analyzes the entire consumer decision-making process
for a mobile viral marketing campaign via text messages across
the three stages: from stage one, reading, to stage two, interest, to
stage three, decision to refer.

To summarize, in contrast to the existing studies in the field
of mobile viral marketing, we analyze consumers’ egocentric
networks via measures such as tie strength and degree
centrality. These sociometric factors are analyzed jointly
with psychographic constructs across the three stages in the
decision-making process. Thus, our study uses a real-world
mobile viral marketing campaign and enables us to test the
relative importance of social embeddedness and consumer
characteristics with respect to consumers’ decision to forward
mobile messages.

Decision-making Process and Specifics of the Mobile Environment

Consumer decision-making is a multiple-stage process
(Bettman 1979; De Bruyn and Lilien 2008; Lavidge and
Steiner 1961). In a viral marketing campaign, the final goal is to
generate a high number of referrals. Therefore, our model of
consumer forwarding behavior is designed for the specific
situation of mobile viral marketing campaigns.

The process and first stage begin with the consumer reading
a mobile advertising text message on his or her mobile phone.
If this text message sparks the consumer’s interest and the
consumer wants to learn more about the offered product, he/she
enters the interest stage, which is the second stage of the model.
If the consumer finds the product interesting after learning
about it, he or she makes a referral, which is the third stage of
our model (decision to refer).

In this study, we analyze the stages of the consumer
decision-making processes within a mobile environment, i.e.,
within a mobile viral marketing campaign. There are several
differences between mobile viral marketing and online or
offline viral marketing. A mobile text message is more
intrusive than an e-mail because it appears immediately on
the full screen. Consumers usually carry their mobile phone
with them and a mobile message may also reach them in a
private moment. Contrary, consumers may need to purposely
look into their e-mail accounts to receive e-mails. Therefore a
mobile message can be more personal compared to an e-mail.
In comparison to offline face-to-face referrals, mobile referrals
do not possess this personal aspect and can be transmitted
digitally within a few minutes to several friends in different
places simultaneously. This is not possible in the offline world.
Additionally, a mobile referral can reach the recipient faster
than an e-mail or an offline referral. Thus, the mobile device
may influence the referral behavior due to its faster digital
transmission of information.

46 C. Pescher et al. / Journal of Interactive Marketing 28 (2014) 43–54

Development of Hypotheses

While the factors that influence the stages of the decision-
making process can be divided into two groups, we analyze
them jointly in this study. The first group consists of the
psychographic indicators of consumer characteristics, thus
focusing on each consumer’s motivation to participate in the
campaign and his or her usage behavior. The second group of
factors includes sociometric indicators of consumer character-
istics, thus providing information about the type of relationship
that the consumer has with his or her contacts and his or her
resulting social network.

Psychographic Indicators of Consumer Characteristics

As mentioned in the related literature section, according to
Okazaki (2008), in viral marketing campaigns, purposive value
and entertainment value are the primary value dimensions for
consumers. This insight is based on the finding that consumers
gain two types of benefits from participating in sales promotions:
hedonic and utilitarian benefits (Chandon, Wansink, and Laurent
2000). Hedonic benefits are primarily intrinsic and can be
associated with entertainment value. Consumers participate
voluntarily and derive value from the fun of interacting with
peers by forwarding a referral (e.g., an ad might be of interest to
peer recipients) (Dholakia, Bagozzi, and Pearo 2004). A previous
study found that the entertainment factor influences intended use
in mobile campaigns (Palka, Pousttchi, and Wiedemann 2009).
Okazaki (2008) found that in a mobile viral marketing campaign,
the entertainment value directly influences the recipient’s attitude
toward the campaign, which, in turn, influences the recipient’s
intention to participate in a mobile viral campaign. Phelps et al.
(2004) showed that the entertainment value is a factor that
increases consumers’ forwarding behavior in viral marketing
campaigns conducted via e-mail. Thus, we may presume that
consumers who place high importance on the entertainment value
of exchanging messages are more likely to enter the reading and
interest stages than consumers who do not value entertainment to
the same degree. Additionally, the entertainment value can
also influence the decision to refer (i.e., forwarding) behavior
because a text message that addresses consumers who place
high importance on entertainment value causes the recipient to
think about forwarding the text message and motivates them
to forward the mobile advertisement to friends (i.e., decision
to refer stage).

H1. Consumers who place high importance on the entertainment
value of a message are more likely to a) enter the reading stage,
b) enter the interest stage and c) enter the decision to refer stage.

As utilitarian benefits are instrumental and functional, they
can be associated with purposive value (Okazaki 2008).
Dholakia, Bagozzi, and Pearo (2004) analyzed the influence
of purposive value in network-based virtual communities and
found that purposive value is a predictor of social identity and
a key motive for an individual to participate in virtual
communities. With respect to the mobile context, previous
research found that purposive value has a direct, significant

influence on a consumer’s attitude toward a mobile viral marketing
campaign and that this attitude significantly influences the
intention to participate in mobile marketing campaigns (Okazaki
2008). For some consumers, forwarding a (mobile) advertisement
in a viral marketing campaign can have a personal and a social
meaning (e.g., doing something good for friends by forwarding the
ad). Thus, we hypothesize that consumers who place high
importance on the purposive value of exchanging messages will
display a greater likelihood to enter the reading and interest stages.
We also hypothesize that consumers who place high importance on
the purposive value of a message are more likely to make the
decision to forward the message.

H2. Consumers who place high importance on the purposive
value of a message are more likely to a) enter the reading stage,
b) enter the interest stage and c) enter the decision to refer stage.

The intensity of usage (e.g., a high quantity of written text
messages) positively influences the probability of trial and
adoption (Steenkamp and Gielens 2003). Thus, consumers with
high usage intensities are more likely to actively participate in a
mobile viral marketing campaign. As mobile viral marketing
campaigns are a fairly new form of advertising, consumers with
high usage intensities are more likely to participate in mobile
viral marketing campaigns and are more likely to forward
messages than consumers with low usage intensities. Therefore,
we propose that usage intensity has an effect on the decision to
forward a mobile advertising text message. The likelihood of
deciding to forward the mobile advertisement increases with
the usage intensity of mobile text messages. This proposition is
consistent with Neslin, Henderson, and Quelch (1985), who
found that the promotional acceleration effect is stronger for
heavy users than it is for other consumers. Godes and Mayzlin
(2009) analyzed the effectiveness of referral activities and
argued that the sales impact from less loyal customers is
greater, but they also highlighted that this greater sales impact
does not mean that the overall referrals by less loyal customers
have a greater impact than those by highly loyal customers.
They concluded that companies who want to implement an
exogenous referral program to drive sales should focus on both
less loyal and highly loyal customers because focusing only on
highly loyal or less loyal customers is not necessarily the
cornerstone of a successful viral marketing campaign. In the
online context, a previous study found that experience with the
Internet influences channel usage behavior (Frambach, Roest,
and Krishnan 2007). Thus, as consumers with high usage
intensity are used to communicating with mobile phones, they
know how to write, read and forward mobile text messages.
Accordingly, it is likely that the threshold to forward a text
message is lower for consumers with high usage intensity than
it is for other consumers and that such consumers are thus more
inclined to refer. Further, the minimal effort required to directly
forward a mobile text message via a cell phone increases the
decision to refer. Thus, we hypothesize that heavy mobile users
will be more likely to refer than will light users.

H3. The usage intensity of the referral medium has a positive
influence on the likelihood of making the decision to refer.

47C. Pescher et al. / Journal of Interactive Marketing 28 (2014) 43–54

Sociometric Indicators of Consumer Characteristics

Sociometric indicators describe the interaction structure of
an individual consumer with his or her surroundings. When
consumers receive an interesting mobile advertising message, it
is likely that they want to find out more about it. Once the
consumer has visited the product homepage, he or she then
considers not only whether the message is worth forwarding but
also to whom it should be forwarded.

Sociometric indicators provide information about the social
network of each individual consumer. This individual network
influences the likelihood of knowing someone who may be
interested in the offered product. Thus, social networks have a
significant impact on the decision-making process in a viral
marketing campaign. The decision to forward the mobile
advertising message depends on two factors: the quality and the
quantity of relations, i.e., the tie strength and the degree centrality.

Tie strength is an important factor in viral marketing and
increases with the amount of time spent with the potential
recipient and with the degree of emotional intensity between
the sender and the potential recipient (Marsden and Campbell
1984). Consumers perceive strong ties to be more influential
than weak ties (Brown and Reingen 1987) because the strong
ties seem more trustworthy (Rogers 1995). Therefore, because
consumers are more motivated to provide high-value informa-
tion to strong ties (Frenzen and Nakamoto 1993), tie strength is
an indicator of the quality of the relationship.

Reagans and McEvily (2003) studied how social network
factors influence knowledge transfer at an R&D firm. To
measure the tie strength, they used two items that are analogous
to those that we used (Burt 1984). Their results indicated that
tie strength positively influences the ease of knowledge
transfer. Thus, network ties increase a person’s capability to
send complex ideas to heterogeneous persons. Overall, they
highlighted the importance of tie strength with respect to the
knowledge transfer process, and they postulate that tie strength
holds a privileged position. Other studies found that weak ties
make non-redundant information available (Levin and Cross
2004). In an online setting, participants were more likely to
share information with strong ties than with weak ties (Norman
and Russell 2006). With respect to viral marketing conducted
via e-mail, previous research has found that tie strength has a
significant influence on whether the recipient examines an
e-mail message sent from a friend (i.e., opens and reads the
message) (De Bruyn and Lilien 2008). Tie strength was also
determined to be less relevant in an online setting compared to an
offline setting (Brown, Broderick, and Lee 2007). In a non-mobile
or non-online context, stronger ties are more likely to activate the
referral flow (Reingen and Kernan 1986). Furthermore, tie
strength is positively related to the amount of time spent receiving
positive referrals (van Hoye and Lievens 1994).

As previously mentioned, research on word-of-mouth behav-
ior has shown that people engage in word-of-mouth for reasons
such as altruism (Sundaram, Mitra, and Webster 1998). However,
Sundaram, Mitra, and Webster (1998) did not control for the
quality of a relationship between sender and recipient. Research
concerning referral reward programs has identified that offering a

reward increases the referral intensity and has a particular impact
on weak ties (Ryu and Feick 2007). Brown and Reingen 1987
found that while strongly tied individuals exchange more
information and communicate more frequently, weak ties play
an important bridging role. Additionally, Granovetter (1973)
stated that one is significantly more likely to be a bridge in the
case of weak ties than in strong ties. In job search, when using
personal networks, it was found that weak ties have a higher rate
of effectiveness when addressing specialists for jobs compared to
strong ties (Bian 1997) and that the income of people using weak
ties was greater than those who used strong ties (Lin, Ensel, and
Vaughn 1981). At the information level, consumers who are
connected via strong ties tend to share the same information
that is rarely new to them, while consumers obtain important
information from weak ties who tend to possess information that
is “new” to them (Granovetter 1973). Consistent with this
finding, Levin …

Forster 17

Advertising on Social Network Sites:
A Structural Equation Modelling Approach

Anant Saxena
Uday Khanna

Abstract
Social networking sites (SNSs) emerged as one of the most powerful media for advertising across the globe. Globally, companies
are shifting a larger pie of their advertising budgets towards social networking sites for better reach and interactive platform. The
companies are also looking at it as a low-cost model, which could reap results in minimum time possible for the targeted ‘Facebook
generation’. These very facts motivate researchers to study the value of advertisements on social networking sites like Facebook,
LinkedIn, Twitter and others. The article is an empirical study to understand the implications of different variables in advertisements
on the delivery of advertising value to the respondents. Confirmatory factor analysis (CFA) has been conducted to test the reliability
of instrument being used for data collection. Further, a model has been proposed for measuring advertising value through structural
equation modelling. The predicted results confirm the roles of different variables, namely, information, entertainment and irritation, in
accessing value of advertisements displayed on social networking sites.

Key Words
Advertising Value, Social Networking Sites, Structural Equation Modelling

Introduction
Social networking websites (SNSs) have emerged as the
‘need of an hour’. Their journey started with the launch of
sixdegrees.com in the year 1997, which attracted millions
of users at that time. The site allowed the users to create
profiles listing their friends with the ability to surf the
friends list (Boyd and Ellison, 2007). This has been
followed by an array of SNSs like Facebook, Orkut,
Linkedin and MySpace in the year 2003–2004. Within a
short span of time, these websites become an addiction for
youngsters as these give them opportunity and platform to
express their feelings and emotions in the society. Websites
like Facebook, Orkut, Twitter and MySpace have become
household names and an integral part of people’s life so
much that it has become tough for regular users to imagine
a life without them. Globally, Internet users spend more
than four and a half hours per week on SNSs, more time
than they spend on e-mail (Anderson et al., 2011). As
more and more of what people think and do ends up getting
expressed on SNSs, it is expected that SNSs affect the
buying decisions greatly. In addition, the huge viewer’s
base of these websites makes them a favourable media for
advertisements by different companies. According to a
study done by comScore, Inc., a market research firm,
SNSs accounted for more than 20 per cent, that is, one in

five, display ads of all display ads viewed online, with
Facebook and MySpace combining to deliver more than 80
per cent of ads among sites in the social networking
category (comScore, 2009). According to Rizavi et al.
(2011) social networking websites act as a good platform
for advertising that attract millions of users from different
countries, speaking multiple languages belonging to
different demographics. According to Trusov et al. (2009)
referrals and recommendations on SNSs have a significant
impact on new customer acquisition and retention. This
fact led marketers to turn to Internet platforms like SNSs,
blogs and other social media as an avenue for cost-effective
marketing, employing e-mail campaigns, website adver-
tisements and viral marketing. Also from a marketing
perspective, these websites give potential customers the
opportunity to virtually explore a business, encourage
them to visit and at last share their views and experiences
with their friends (Phillips et al., 2010). Understanding
the effectiveness of SNSs in promoting product and
services through advertisements, companies across the
globe have increased their advertising budget for SNSs
which has led to increase in revenue generation for social
networking website companies. According to a report
released by Interactive Advertising Bureau (IAB),
Internet advertising revenues totaled $14.9 billion in 2011,
up 23 per cent from the $12.1 billion reported in 2010

Vision
17(1) 17–25
© 2013 MDI

SAGE Publications
Los Angeles, London,

New Delhi, Singapore,
Washington DC

DOI: 10.1177/0972262912469560
http://vision.sagepub.com

Article

18 Advertising on Social Network Sites

Vision, 17, 1 (2013): 17–25

(PricewaterhouseCoopers LLP., 2011). India shares the
same story in terms of Internet advertising revenues.

According to a report, the size of Internet advertising
industry was INR 7.7 billion in 2010 registering a
growth of 28.3 per cent over INR 6 billion in 2009
(PricewaterhouseCoopers Private Limited, 2011). The
same report highlights that in India SNSs have shown a
remarkable growth of 43 per cent in 2010 over 2009,
with a 54 per cent growth in advertising on SNSs in 2010–
2011 (PricewaterhouseCoopers Private Limited, 2011).
Considering the fact that advertising on SNSs is on a new
high, this research focus on studying the value of
advertisements being displayed on SNSs.

Literature Review and Hypothesis
Web advertising continues to be a major area of advertising
research from a long time. A number of studies have been
done discussing advertisements on the Web and their
effects. Berthon et al. (1996) have discussed the role of
World Wide Web as an advertising medium in the mar-
keting communication mix and proved that World Wide
Web is a new medium for advertising characterized by
ease-of-entry, relatively low set-up costs, globalness, time
independence and interactivity. In spite of the acceptance
of World Wide Web as an effective media for advertising,
few studies have focused on the value of advertisements
displayed on this medium. R.H. Ducoffe introduced the
concept of advertisement value in 1995. According to
Ducoffe (1995) advertising value is defined as the utility or
worth of the advertisement. Ducoffe (1996), in his another
study on World Wide Web, proved the significant impact
(either +ve or –ve) of entertainment, information and
irritation on advertisement value. Brackett and Carr (2001)
in their study on cyberspace advertising reports that
information, entertainment, irritation and credibility
significantly affect advertisement value which in turn
affects attitude towards advertisements. Discussion on
different predictors of advertisement value with reference
to SNSs advertisements is hereby illustrated:

1. Information: Information content is an important
determinant of advertisement effectiveness. Comp-
anies advertise for one main reason—providing
information about their product, services and brand
to consumers. Consumers reported that supplying
information is the primary reason why they approve
advertising (Bauer et al., 1968). According to Norris
(1984) information in advertisements enables the
customers to evaluate the products more rationally
leading to improved markets with low prices and
high quality of the product. Information content on
Internet can be delivered better in comparison to

television medium, reason being short time span of
television advertisements. Yoon and Kim (2001)
mentioned that Internet advertising differs from tra-
ditional advertising as it delivers unlimited informa-
tion beyond time and space and it gives unlimited
amount and sources of information. Web advertise-
ments provide information and generate awareness
without interactive involvement (Berthon et al.,
1996). On the contrary, information delivered
through SNSs advertisements is different from tra-
ditional Web advertisements because SNSs provide
a medium that is interactive in nature. A person
could scan and share information with online friends
and followers, thus making the advertisement infor-
mation viral in nature. Large media companies have
realized the potential of SNSs to reach and deepen
relationships with the ‘subscribed’ audience (Jhih-
Syuan and Pena, 2011). This specialty of SNSs
advertisement makes it the most competitive plat-
form for sharing information about products and
services. As the delivery and importance of infor-
mation for SNSs advertisements is different from
other forms of advertisements, it is important to note
its effect on advertisement value. Based on this
rationale, the hypothesis tested is:

H1: There is a significant positive impact of infor-
mation content of advertisements on the value of
advertisements displayed on social networking
websites.

2. Entertainment: An advertisement that is full of
information but nil in entertainment content is not
worthy. According to McQuail (1994) an advertise-
ment entertains when it fulfils the audience needs
for escapism, diversion, aesthetic enjoyment or
emotional release. The ability of advertising to
entertain can enhance the experience of advertising.
In addition, an advertisement could be information
for one and entertainment for other person at the
same time (Alwitt and Prabhaker, 1992). Consumers
who found advertising to be entertaining also evalu-
ate it as informative (Ducoffe, 1995). This shows
that entertainment and information are interrelated
concepts when talking about advertisements.
SNSs platform is interactive in nature and display
banner advertisements of different brands at the
same platform and same time; they have the power
to entertain the audience. Kim and Lee (2010)
noted that college students use SNSs for six
main reasons: entertainment, passing time, social
interaction, information seeking, information provi-
ding, and professional advancement. According to

Anant Saxena and Uday Khanna 19

Vision, 17, 1 (2013): 17–25

Taylor et al. (2011) SNSs advertisements provide
entertainment value to the audience. The same study
reported that entertainment exhibits almost four
times more strength of influence on favourable con-
sumers’ attitude towards advertisements than infor-
mation. With reference to the existing literature, it is
important to find the impact of entertainment on
advertisement value of SNSs advertisements. In the
same vein, the hypothesis tested is:

H2: There is a significant positive impact of enter-
tainment content of advertisements on the value of
advertisements displayed on social networking
websites.

3. Irritation: Irritation from advertisements arises
when we feel discomfort in watching advertisement
due to any reason. The reason can be personal or
social. A personal reason could be distraction while
focusing on a particular task on World Wide Web.
According to Wells et al. (1971) irritation is one
amongst six dimensions of personal reactions
towards advertising. It is the degree to which the
viewer disliked the contents that he had seen. The
words that came into the mind of the viewer at time
of getting irritated from an advertisement are
‘terrible’, ‘stupid’, ‘ridiculous’, ‘irritating’ and
‘phony’. An advertisement can be rewarding for
some viewers and yet be an irritant and unrewarding
for others (Alwitt and Prabhaker, 1992). According
to Aaker and Bruzzone (1985), increase in irritation
can lead to general reduction in the effectiveness of
advertisement. In case of Internet advertising, it also
generates considerable irritation (Schlosser et al.,
1999). As online behaviour including use of SNSs
is highly goal oriented, advertisements on SNSs
might irritate the user (Taylor et al., 2011). The lit-
erature suggested that irritation has a negative effect
on the effectiveness of advertisement irrespective of
the media. Based on this rationale the hypothesis
tested is:

H3: There is a significant negative impact of irrita-
tion content of advertisements on the value of adver-
tisements displayed on social networking websites.

A considerable amount of research on determinants of
Web advertising effectiveness and value has been done
(Berthon et al., 1996; Brown et al., 2007; Ducoffe, 1995;
Lei, 2000; Schlosser et al., 1999; Yoon and Kim, 2001);
however, these studies were more focused on traditional
websites rather than SNSs. Advertising through SNSs is
different from traditional websites due to several reasons.

First, advertisements on SNSs are different not only in form
and substance but also in delivery method. Some of the
messages are ‘pushed’ upon consumers while others rely on
consumers to ‘pull’ content; some generate revenue whereas
some are non-paid content delivered through media content
(Taylor et al., 2011). Second, SNSs have their own unique
user-to-user interface (Safko and Brake, 2009). Third, SNSs
users are increasing day by day all over the world, which
makes this medium suitable for advertising. As SNSs
advertising is different from traditional Web advertising and
a little is known about value of SNSs advertisements, this
study tries to fill this research gap by providing a model,
which tests the interrelationships between different
determinants of advertisement value.

Model Testing
The importance of advertisements displayed on SNSs is
increasing day by day. According to Stelzner (2011) 88 per cent
of the marketers have reported that their social media
advertisements have generated more exposure for their
businesses. This leads the authors to test a model for accessing
the value of advertisements displayed on SNSs by employing
structural equation modelling (SEM) approach. Use of SEM
technique gives us the opportunity to examine multiple
dependence techniques simultaneously. SEM approach is a
statistical methodology that combines the strength of factor
analysis and path analysis. According to Singh (2009) SEM is
considered as a more advanced technique than other multivariate
techniques because it can estimate a series of interrelated
dependence relationship simultaneously. According to Byrne
(1998) SEM technique is better because:

1. It accounts for measurement errors in course of
model testing.

2. It can incorporate observed (indicator) variables as
well as latent (unobserved) variables at same time
during model testing,

3. It tests a priori relationships rather than allowing the
technique or data to define the nature of relationship
between the variables.

In the present study, SEM analysis is conducted in
two major steps; first, to test the measurement model
and second, a structural model. Measurement model
provides the series of relationships that suggests how
observed variables represent latent variables (Figure 1),
tested by means of confirmatory factor analysis (CFA).
Structural model tests the conceptual representation of the
relationships between the latent variables. It tells whether
the proposed model is eligible to represent a proposed
concept and conceptual relationships between the variables
or not (Figure 2).

20 Advertising on Social Network Sites

Vision, 17, 1 (2013): 17–25

Figure 1. Measurement Model

ENTERTAINMENT

INFORMATION

IRRITATION

ADD.VALUE

ENTERTAIN 3

ENTERTAIN 2

ENTERTAIN 1

INFO 1

INFO 2

INFO 3

IRRITATION 1

IRRITATION 2

IRRITATION 3

ADDVALUE 1

ADDVALUE 2

ADDVALUE 3

0.57

0.17

0.19

0.47

0.40

0.51

0.45

0.80

0.39

0.80

0.86

0.52

0.23

0.75

0.65

0.76

0.77

0.73

Anant Saxena and Uday Khanna 21

Vision, 17, 1 (2013): 17–25

Figure 2. Structural Model

Method
Sample Design

The research focuses on social networking websites with
college students as sample respondents. The college students
were selected as sample for two basic reasons. First,
student sample is more homogeneous (less variable) in
terms of socio-economic background, demographics and
education (Peterson, 2001). Second, a number of studies
have reported that students are the main users of social
networking websites (Dwyer et al., 2007; Pempek et al.,
2009; Subrahmanyam et al., 2008). With this rationale,
present study sample includes postgraduate management
students of a reputed college based in India. 276 students
have responded to an online questionnaire mailed to 300
students. The questionnaires were mailed with Google
documents facility to form and mail online forms/
questionnaires. After removing incomplete questionnaires,
only 189 questionnaires were found to be useable for analysis
and further study. Resulting sample consists of 71 per cent
males and 29 per cent females. Subjects were asked to report
their reactions to instrument statements by considering their

perceptions of advertisements on SNSs in general, not a
single advertisement or advertisement for any particular
product. The objective of this generalization is to assess the
value of advertisement on social networking websites across
different advertisements of product and service categories.

Sample Size and SEM Analysis

Sample size is a key issue when performing SEM analysis.
According to Bentler and Bonett (1980) and Hair et al.
(2007) chi-square value is sensitive to increase in sample
size, while it lacks power to discriminate between good fit
and poor fit models with small sample size (Kenny and
McCoach, 2003). Hair et al. (2007) mentioned that 15 res-
ponses per parameter is an appropriate ratio for sample
size. Going on with this approach a sample size of 189 res-
pondents for measuring 12 parameters was appropriate.

Research Instrument

For measuring the advertisement value of advertisements
displayed on social media, a 12 item scale developed by

ADD. VALUE

ADDVALUE 3

ADDVALUE 2

ADDVALUE 1

IN1 IN2 IN3

e3 e2 e1

IRR1 IRR2 IRR3

e9 e8 e7

EN1 EN2 EN3

e6 e5 e4

INFORMATION

IRRITATION

ENTERTAINMENT

e10

e11

e13

e12

0.55 0.84 0.38

0.27

0.15

0.87 0.58

0.77 0.90 0.51

0.40

0.36
0.38

0.75

0.76

0.72

0.25

22 Advertising on Social Network Sites

Vision, 17, 1 (2013): 17–25

Ducoffe (1995) was used. The instrument was modified as
per the need of the study. A five-item Likert scale was used as
a response scale, from strongly disagrees to strongly agree.

Measurement Model
Measurement model is a specification of the measurement
theory that shows how constructs are operationalized by
sets of measured items. Confirmatory factor analysis is
used to test the reliability of a measurement model. Unlike
exploratory factor analysis, CFA allows the researcher to
tell the SEM programme which variable belongs to which
factor before the analysis (Hair et al., 2007). According to
Salisbury et al. (2001) CFA allows the researcher to specify
the actual relationship between the items and factors as
well as linkages between them.

Construct Validity

According to Hair et al. (2007) construct validity is the
extent to which a set of measured items actually represents
theoretical latent construct; those items are designed to
measure. The reliability of advertisement value scale was
examined by specifying a model in CFA using AMOS 19.
Reliability of an instrument can also be calculated by
Cronbach’s alpha, but use of SEM technique makes
such a practice unnecessary and redundant (Bagozzi
and Yi, 2012). The results (see Table 1) confirm the
overall fit of a measurement model when employed to
CFA.

According to Hair et al. (2007) one incremental fit index
(CFI), one goodness of fit index (GFI), one absolute fit
index (GFI, SRMR) and one badness of fit index (SRMR),
with chi-square statistic should be used to assess a model’s
goodness of fit. Our study results show all the different
types of indices in the acceptable range.

Convergent and Discriminant Validity

Convergent validity exists when the items that are
indicators of a specific construct converge or share a high
proportion of variance in common. In general, ‘factor
loading’ and ‘variance extracted’ measures are used to
measure convergent validity. We have used factor loading
measure in our study to measure convergent validity (Hair
et al., 2007; Salisbury et al., 2001). All the factor loadings
are statistically significant, a minimum requirement for
convergence (Hair et al., 2007). Furthermore, except items
‘Info 3’ and ‘Irritation 1’ all factor loadings are in the range
of 0.50 to 0.80, which is more than acceptable value of
0.50 (Hair et al., 2007) (see Figure 1). According to Chin
et al. (1997) discriminant validity exists if the correlation
between the constructs is not equal to 1. Following the rule,
our study shows the discriminant validity between the
constructs (see Figure 1).

Structural Model
After assessing the eligibility of scale for measuring
different variables in the study, the next step is to test the
hypothesized relationships in a structural model. Ducoffe
(1996) has proved the respective role of information,
entertainment and irritation on advertisement value for the
advertisements on the Web. In our study, we try to explore
the impact of these respective variables on advertisement
value vis-à-vis SNSs.

Performance of the Model

Hypothesized relationships are supported by the overall
model fit indices obtained. All of the fit indices are above
the recommended values. The c2/df value 2.31 met the
recommended value of less than 3 (Carmines and McIver,
1981). Hair et al. (2007) argues that chi-square value is
sensitive to sample size and number of variables; therefore,
c2/df value is not taken as a sole indicator of model fit.
Other model fit indicators taken are also within the
recommended range (see Table 2). In sum, various model
fit indices indicates that the proposed model fitted well
with the present data set.

Table 1. Model Fit Indices for Measurement Model

Statistic
Recommended

Value Obtained Value

Chi-square c2 92.616
Df 48
c2/df (Hinkin, 1995),

(Carmines and
McIver, 1981)

< 3.00 1.93

GFI (Hooper et al.,
2008), (Hair et al.,
2007)

> 0.90 0.92

AGFI (Muenjohn and
Armstrong, 2008)

> 0.80 0.88

SRMR (Hu and Bentler,
1999)

< 0.08 0.06

CFI (Watchravesringkan
et al., 2008)

> 0.80 0.92

Note: AGFI: Adjusted goodness of fit index; SRMR: Standardized root
mean square residual; CFI: Comparative fit index

Anant Saxena and Uday Khanna 23

Vision, 17, 1 (2013): 17–25

Table 2. Model Fit Indices for Structural Model

Statistic
Recommended

Value Obtained Value

Chi-square c2 115.539
Df 50
c2/df (Hinkin, 1995),

(Carmines and McIver,
1981)

< 3.00 2.31

GFI (Hooper et al., 2008),
(Hair et al., 2007)

> 0.90 0.91

AGFI (Muenjohn and
Armstrong, 2008)

> 0.80 0.86

RMSEA (MacCallum et al.,
1996)

< 0.10 0.08

CFI (Watchravesringkan
et al., 2008)

> 0.80 0.88

Note: SMSEA: Root Mean Square Error of Approximation

Estimated Standardized Path Coefficients

Figure 2 shows the standardized path coefficients of
the four constructs under investigation. All the path
coefficients were significant at the level of 0.01 with the
direction of influence as hypothesized (+ or −). Information
and entertainment were positively associated with
advertisement value whereas irritation is negatively asso-
ciated with advertisement value; thus all the hypotheses
framed are statistically supported. A significant correlation
between information and entertainment also indicates that
the consumers who find advertisement to be entertaining
are more likely to evaluate it as informative. These results
are consistent with another study (Ducoffe, 1995). Finally,
the squared multiple correlations (R2) indicates that the
present model explains 38 per cent of the variance in
advertisement value.

Discussion and Implication
The study yielded important new insights about a topic that
is important for both industry practitioners and aca-
demicians. The concept of advertisement value and factors
affecting it had been widely tested for various types
of advertisements in a number of studies but lack of work
for advertisements displayed on social networking websites
was the motivating factor to do research in the particular
domain. The study tests the model to assess advertisement
value by employing SEM approach. SEM combines the
strength of factor analysis and path analysis. It enables us
to test whether observed variables completely describes
latent variables or not. In addition, SEM is a more
successful technique than other multivariate techniques as
it can estimate a series of interrelated dependence
relationship simultaneously. It tells whether the proposed

model is eligible to represent a proposed concept and
conceptual relationships between the variables or not. The
results of CFA suggest that the observed variables are
suitable enough to represent different latent variables, that
is, information, entertainment, irritation and advertisement
value in the particular domain of social networking
advertising.

The findings of structural model analysis suggest that
the proposed model for accessing the value of
advertisements displayed on SNSs fits well. In addition,
the proposed hypotheses assessing the relationships
between the variables are statistically supported. The
findings suggest that when advertisements displayed on
SNSs provide entertainment and information content or
impressions, it increases the worth of the advertisement.
On the one hand, as has been proved true for other types of
media advertising, consumers derive utility from
advertisements that provide some useful or functional
information and increase hedonic value by entertaining
them. On the other hand, irritation decreases the net worth
of the advertisements displayed on SNSs. This suggests
that the companies using SNSs media for advertising their
products and services should reduce the contents, which
irritate the viewers’ base.

It is worth noting that ‘information’ exhibited around
1.6 times more strength of influence on advertisement
value than entertainment. This suggests that companies
should firstly focus on providing information content in
their advertisements to make their advertisements worth
for consumers. In addition, it is interesting to note that
findings of this study show a significant correlation
between information and entertainment, which indicates
that consumers who find advertisement to be entertaining
are more likely to evaluate it as informative.

Limitations
Although the study has been done taking into account the
methodological rigour, some limitations remain. First, the
sampling used is convenience sampling. Second,
exploration of other variables that affects the value of
advertisement is needed.

References
Aaker, D.A., & Bruzzone, D.E. (1985). Causes of irritation in

advertising. The Journal of Marketing, 49(2), 47–57.
Alwitt, L.F., & Prabhaker, P.R. (1992). Functional and belief

dimensions of attitudes to television advertising: Implications
for copytesting. Journal of Advertising Research, 32(5),
30–42.

Anderson, M., Brusa, J., Price, J., & Sims, J. (2011). Turning
‘Like’ to ‘Buy’ social media emerges as a commerce

24 Advertising on Social Network Sites

Vision, 17, 1 (2013): 17–25

channel (pp. 1–9). Booz & Company. Available at http://
www.booz.com/media/uploads/BaC-Turning_Like_to_Buy.
pdf

Bagozzi, R., & Yi, Y. (2012). Specification, evaluation, and
interpretation of structural equation models. Journal of the
Academy of Marketing Science, 40(1), 8–34.

Bauer, R.A., Greyser, S.A., Kanter, D.L., & Weilbacher, W.M.
(1968). Advertising in America: The consumer view. Division
of Research, Graduate School of Business Administration,
Harvard University Boston.

Bentler, P.M., & Bonett, D.G. (1980). Significance tests and
goodness of fit in the analysis of covariance structures.
Psychological Bulletin, 88(3), 588–606.

Berthon, P., Pitt, L.F., & Watson, R.T. (1996). The World Wide
Web as an advertising medium: Toward an understanding
of conversion efficiency. Journal of Advertising Research,
36(1), 43–54.

Boyd, Danah M., & Ellison, N.B. (2007). Social network sites:
Definition, history, and scholarship. Journal of Computer-
Mediated Communication, 13(1), 210–230.

Brackett, L.K., & Carr, B.N. (2001). Cyberspace advertising vs.
other media: Consumer vs. mature student attitudes. Journal
of Advertising Research, 41(5), 23–32.

…

Journal of Marketing Research
Vol. XLIX (April 2012), 192 –205

*Jonah Berger is Joseph G. Campbell Assistant Professor of Marketing
(e-mail: [email protected]), and Katherine L. Milkman is Assistant
Professor of Operations and Information Management (e-mail: [email protected]
wharton.upenn.edu), the Wharton School, University of Pennsylvania.
Michael Buckley, Jason Chen, Michael Durkheimer, Henning Krohnstad,
Heidi Liu, Lauren McDevitt, Areeb Pirani, Jason Pollack, and Ronnie
Wang all provided helpful research assistance. Hector Castro and Premal
Vora created the web crawler that made this project possible, and Roger
Booth and James W. Pennebaker provided access to LIWC. Devin Pope
and Bill Simpson provided helpful suggestions on our analysis strategy.
Thanks to Max Bazerman, John Beshears, Jonathan Haidt, Chip Heath,
Yoshi Kashima, Dacher Keltner, Kim Peters, Mark Schaller, Deborah
Small, and Andrew Stephen for helpful comments on prior versions of the
article. The Dean’s Research Initiative and the Wharton Interactive Media
Initiative helped fund this research. Ravi Dhar served as associate editor
for this article.

JONAH BERGER and KATHERINE L. MILKMAN*

Why are certain pieces of online content (e.g., advertisements, videos,
news articles) more viral than others? This article takes a psychological
approach to understanding diffusion. Using a unique data set of all the
New York Times articles published over a three-month period, the authors
examine how emotion shapes virality. The results indicate that positive
content is more viral than negative content, but the relationship between
emotion and social transmission is more complex than valence alone.
Virality is partially driven by physiological arousal. Content that evokes
high-arousal positive (awe) or negative (anger or anxiety) emotions is
more viral. Content that evokes low-arousal, or deactivating, emotions
(e.g., sadness) is less viral. These results hold even when the authors
control for how surprising, interesting, or practically useful content is (all
of which are positively linked to virality), as well as external drivers of
attention (e.g., how prominently content was featured). Experimental
results further demonstrate the causal impact of specific emotion on
transmission and illustrate that it is driven by the level of activation
induced. Taken together, these findings shed light on why people share
content and how to design more effective viral marketing campaigns.

Keywords: word of mouth, viral marketing, social transmission, online
content

What Makes Online Content Viral?

© 2012, American Marketing Association
ISSN: 0022-2437 (print), 1547-7193 (electronic) 192

Sharing online content is an integral part of modern life.
People forward newspaper articles to their friends, pass
YouTube videos to their relatives, and send restaurant
reviews to their neighbors. Indeed, 59% of people report that
they frequently share online content with others (Allsop,
Bassett, and Hoskins 2007), and someone tweets a link to a
New York Times story once every four seconds (Harris 2010).
Such social transmission also has an important impact on

both consumers and brands. Decades of research suggest

that interpersonal communication affects attitudes and deci-
sion making (Asch 1956; Katz and Lazarsfeld 1955), and
recent work has demonstrated the causal impact of word of
mouth on product adoption and sales (Chevalier and Mayz –
lin 2006; Godes and Mayzlin 2009).
Although it is clear that social transmission is both fre-

quent and important, less is known about why certain pieces
of online content are more viral than others. Some customer
service experiences spread throughout the blogosphere,
while others are never shared. Some newspaper articles earn
a position on their website’s “most e-mailed list,” while oth-
ers languish. Companies often create online ad campaigns
or encourage consumer-generated content in the hope that
people will share this content with others, but some of these
efforts take off while others fail. Is virality just random, as
some argue (e.g., Cashmore 2009), or might certain charac-
teristics predict whether content will be highly shared?
This article examines how content characteristics affect

virality. In particular, we focus on how emotion shapes
social transmission. We do so in two ways. First, we analyze
a unique data set of nearly 7000 New York Times articles to
examine which articles make the newspaper’s “most e-
mailed list.” Controlling for external drivers of attention,
such as where an article was featured online and for how
long, we examine how content’s valence (i.e., whether an

What Makes Online Content Viral? 193

article is positive or negative) and the specific emotions it
evokes (e.g., anger, sadness, awe) affect whether it is highly
shared. Second, we experimentally manipulate the specific
emotion evoked by content to directly test the causal impact
of arousal on social transmission.
This research makes several important contributions. First,

research on word of mouth and viral marketing has focused
on its impact (i.e., on diffusion and sales; Godes and May-
zlin 2004, 2009; Goldenberg et al. 2009). However, there has
been less attention to its causes or what drives people to share
content with others and what type of content is more likely
to be shared. By combining a large-scale examination of real
transmission in the field with tightly controlled experiments,
we both demonstrate characteristics of viral online content
and shed light on the underlying processes that drive people
to share. Second, our findings provide insight into how to
design successful viral marketing campaigns. Word of mouth
and social media are viewed as cheaper and more effective
than traditional media, but their utility hinges on people
transmitting content that helps the brand. If no one shares a
company’s content or if consumers share content that por-
trays the company negatively, the benefit of social transmis-
sion is lost. Consequently, understanding what drives peo-
ple to share can help organizations and policy makers avoid
consumer backlash and craft contagious content.

CONTENT CHARACTERISTICS AND SOCIAL
TRANSMISSION

One reason people may share stories, news, and informa-
tion is because they contain useful information. Coupons or
articles about good restaurants help people save money and
eat better. Consumers may share such practically useful
content for altruistic reasons (e.g., to help others) or for self-
enhancement purposes (e.g., to appear knowledgeable, see
Wojnicki and Godes 2008). Practically useful content also has
social exchange value (Homans 1958), and people may share
it to generate reciprocity (Fehr, Kirchsteiger, and Riedl 1998).
Emotional aspects of content may also affect whether it is

shared (Heath, Bell, and Sternberg 2001). People report dis-
cussing many of their emotional experiences with others,
and customers report greater word of mouth at the extremes
of satisfaction (i.e., highly satisfied or highly dissatisfied;
Anderson 1998). People may share emotionally charged con-
tent to make sense of their experiences, reduce dissonance, or
deepen social connections (Festinger, Riecken, and Schachter
1956; Peters and Kashima 2007; Rime et al. 1991).
Emotional Valence and Social Transmission
These observations imply that emotionally evocative

content may be particularly viral, but which is more likely
to be shared—positive or negative content? While there is a
lay belief that people are more likely to pass along negative
news (Godes et al. 2005), this has never been tested. Fur-
thermore, the study on which this notion is based actually
focused on understanding what types of news people
encounter, not what they transmit (see Goodman 1999).
Consequently, researchers have noted that “more rigorous
research into the relative probabilities of transmission of
positive and negative information would be valuable to both
academics and managers” (Godes et al. 2005, p. 419).
We hypothesize that more positive content will be more

viral. Consumers often share content for self-presentation

purposes (Wojnicki and Godes 2008) or to communicate
identity, and consequently, positive content may be shared
more because it reflects positively on the sender. Most peo-
ple would prefer to be known as someone who shares
upbeat stories or makes others feel good rather than some-
one who shares things that makes others sad or upset. Shar-
ing positive content may also help boost others’ mood or
provide information about potential rewards (e.g., this
restaurant is worth trying).
The Role of Activation in Social Transmission
Importantly, however, the social transmission of emo-

tional content may be driven by more than just valence. In
addition to being positive or negative, emotions also differ
on the level of physiological arousal or activation they
evoke (Smith and Ellsworth 1985). Anger, anxiety, and sad-
ness are all negative emotions, for example, but while anger
and anxiety are characterized by states of heightened
arousal or activation, sadness is characterized by low
arousal or deactivation (Barrett and Russell 1998).
We suggest that these differences in arousal shape social

transmission (see also Berger 2011). Arousal is a state of
mobilization. While low arousal or deactivation is charac-
terized by relaxation, high arousal or activation is character-
ized by activity (for a review, see Heilman 1997). Indeed,
this excitatory state has been shown to increase action-
related behaviors such as getting up to help others (Gaertner
and Dovidio 1977) and responding faster to offers in nego-
tiations (Brooks and Schweitzer 2011). Given that sharing
information requires action, we suggest that activation
should have similar effects on social transmission and boost
the likelihood that content is highly shared.
If this is the case, even two emotions of the same valence

may have different effects on sharing if they induce differ-
ent levels of activation. Consider something that makes peo-
ple sad versus something that makes people angry. Both
emotions are negative, so a simple valence-based perspec-
tive would suggest that content that induces either emotion
should be less viral (e.g., people want to make their friends
feel good rather than bad). An arousal- or activation-based
analysis, however, provides a more nuanced perspective.
Although both emotions are negative, anger might increase
transmission (because it is characterized by high activation),
while sadness might actually decrease transmission
(because it is characterized by deactivation or inaction).

THE CURRENT RESEARCH
We examine how content characteristics drive social

transmission and virality. In particular, we not only examine
whether positive content is more viral than negative content
but go beyond mere valence to examine how specific emo-
tions evoked by content, and the activation they induce,
drive social transmission.
We study transmission in two ways. First, we investigate

the virality of almost 7000 articles from one of the world’s
most popular newspapers: the New York Times (Study 1).
Controlling for a host of factors (e.g., where articles are fea-
tured, how much interest they evoke), we examine how the
emotionality, valence, and specific emotions evoked by an
article affect its likelihood of making the New York Times’
most e-mailed list. Second, we conduct a series of lab
experiments (Studies 2a, 2b, and 3) to test the underlying

process we believe to be responsible for the observed
effects. By directly manipulating specific emotions and
measuring the activation they induce, we test our hypothe-
sis that content that evokes high-arousal emotion is more
likely to be shared.

STUDY 1: A FIELD STUDY OF EMOTIONS AND
VIRALITY

Our first study investigates what types of New York Times
articles are highly shared. The New York Times covers a
wide range of topics (e.g., world news, sports, travel), and
its articles are shared with a mix of friends (42%), relatives
(40%), colleagues (10%), and others (7%),1 making it an
ideal venue for examining the link between content charac-
teristics and virality. The New York Times continually
reports which articles from its website have been the most
e-mailed in the past 24 hours, and we examine how (1) an
article’s valence and (2) the extent to which it evokes vari-
ous specific emotions (e.g., anger or sadness) affect whether
it makes the New York Times’ most e-mailed list.
Negative emotions have been much better distinguished

from one another than positive emotions (Keltner and
Lerner 2010). Consequently, when considering specific
emotions, our archival analysis focuses on negative emo-
tions because they are straightforward to differentiate and
classify. Anger, anxiety, and sadness are often described as
basic, or universal, emotions (Ekman, Friesen, and Ellsworth
1982), and on the basis of our previous theorizing about
activation, we predict that negative emotions characterized
by activation (i.e., anger and anxiety) will be positively linked
to virality, while negative emotions characterized by deacti-
vation (i.e., sadness) will be negatively linked to virality.
We also examine whether awe, a high-arousal positive

emotion, is linked to virality. Awe is characterized by a feeling
of admiration and elevation in the face of something greater
than oneself (e.g., a new scientific discovery, someone over-
coming adversity; see Keltner and Haidt 2003). It is gener-
ated by stimuli that open the mind to unconsidered possibil-
ities, and the arousal it induces may promote transmission.
Importantly, our empirical analyses control for several

potentially confounding variables. First, as we noted previ-
ously, practically useful content may be more viral because it
provides information. Self-presentation motives also shape
transmission (Wojnicki and Godes 2008), and people may
share interesting or surprising content because it is entertain-
ing and reflects positively on them (i.e., suggests that they
know interesting or entertaining things). Consequently, we
control for these factors to examine the link between emotion
and virality beyond them (though their relationships with
virality may be of interest to some scholars and practitioners).
Second, our analyses also control for things beyond the

content itself. Articles that appear on the front page of the
newspaper or spend more time in prominent positions on
the New York Times’ home page may receive more attention
and thus mechanically have a better chance of making the
most e-mailed list. Consequently, we control for these and
other potential external drivers of attention.2 Including these

controls also enables us to compare the relative impact of
placement versus content characteristics in shaping virality.
While being heavily advertised, or in this case prominently
featured, should likely increase the chance content makes
the most e-mailed list, we examine whether content charac-
teristics (e.g., whether an article is positive or awe-inspiring)
are of similar importance.
Data
We collected information about all New York Times arti-

cles that appeared on the newspaper’s home page (www.
nytimes. com) between August 30 and November 30, 2008
(6956 articles). We captured data using a web crawler that
visited the New York Times’ home page every 15 minutes
during the period in question. It recorded information about
every article on the home page and each article on the most
e-mailed list (updated every 15 minutes). We captured each
article’s title, full text, author(s), topic area (e.g., opinion,
sports), and two-sentence summary created by the New York
Times. We also captured each article’s section, page, and
publication date if it appeared in the print paper, as well as
the dates, times, locations, and durations of all appearances
it made on the New York Times’ home page. Of the articles
in our data set, 20% earned a position on the most e-mailed
list.
Article Coding
We coded the articles on several dimensions. First, we

used automated sentiment analysis to quantify the positivity
(i.e., valence) and emotionality (i.e., affect ladenness) of
each article. These methods are well established (Pang and
Lee 2008) and increase coding ease and objectivity. Auto-
mated ratings were also significantly positively correlated
with manual coders’ ratings of a subset of articles. A com-
puter program (LIWC) counted the number of positive and
negative words in each article using a list of 7630 words clas-
sified as positive or negative by human readers (Pennebaker,
Booth, and Francis 2007). We quantified positivity as the
difference between the percentage of positive and negative
words in an article. We quantified emotionality as the per-
centage of words that were classified as either positive or
negative.
Second, we relied on human coders to classify the extent

to which content exhibited other, more specific characteris-
tics (e.g., evoked anger) because automated coding systems
were not available for these variables. In addition to coding
whether articles contained practically useful information or
evoked interest or surprise (control variables), coders quan-
tified the extent to which each article evoked anxiety, anger,
awe, or sadness.3 Coders were blind to our hypotheses.
They received the title and summary of each article, a web
link to the article’s full text, and detailed coding instructions
(see the Web Appendix at www.marketingpower.com/jmr_
webappendix). Given the overwhelming number of articles
in our data set, we selected a random subsample for coding

194 JOURNAL OF MARKETING RESEARCH, APRIL 2012

1These figures are based on 343 New York Times readers who were asked
with whom they had most recently shared an article.
2Discussion with newspaper staff indicated that editorial decisions about

how to feature articles on the home page are made independently of (and
well before) their appearance on the most e-mailed list.

3Given that prior work has examined how the emotion of disgust might
affect the transmission of urban legends (Heath, Bell, and Sternberg 2001),
we also include disgust in our analysis. (The rest of the results remain
unchanged regardless of whether it is in the model.) While we do not find
any significant relationship between disgust and virality, this may be due
in part to the notion that in general, New York Times articles elicit little of
this emotion.

What Makes Online Content Viral? 195

(n = 2566). For each dimension (awe, anger, anxiety, sad-
ness, surprise, practical utility, and interest), a separate
group of three independent raters rated each article on a
five-point Likert scale according to the extent to which it
was characterized by the construct in question (1 = “not at
all,” and 5 = “extremely”). We gave raters feedback on their
coding of a test set of articles until it was clear that they
understood the relevant construct. Interrater reliability was
high on all dimensions (all ’s > .70), indicating that con-
tent tends to evoke similar emotions across people. We
averaged scores across coders and standardized them (for
sample articles that scored highly on the different dimen-
sions, see Table 1; for summary statistics, see Table 2; and
for correlations between variables, see the Appendix). We
assigned all uncoded articles a score of zero on each dimen-
sion after standardization (i.e., we assigned uncoded articles
the mean value), and we included a dummy in regression
analyses to control for uncoded stories (for a discussion of
this standard imputation methodology, see Cohen and
Cohen 1983). This enabled us to use the full set of articles
collected to analyze the relationship between other content
characteristics (that did not require manual coding) and
virality. Using only the coded subset of articles provides
similar results.

Additional Controls
As we discussed previously, external factors (separate

from content characteristics) may affect an article’s virality
by functioning like advertising. Consequently, we rigor-
ously control for such factors in our analyses (for a list of
all independent variables including controls, see Table 3).
Appearance in the physical newspaper. To characterize

where an article appeared in the physical newspaper, we
created dummy variables to control for the article’s section
(e.g., Section A). We also created indicator variables to
quantify the page in a given section (e.g., A1) where an arti-
cle appeared in print to control for the possibility that
appearing earlier in some sections has a different effect than
appearing earlier in others.
Appearance on the home page. To characterize how much

time an article spent in prominent positions on the home
page, we created variables that indicated where, when, and
for how long every article was featured on the New York
Times home page. The home page layout remained the same
throughout the period of data collection. Articles could
appear in several dozen positions on the home page, so we
aggregated positions into seven general regions based on
locations that likely receive similar amounts of attention
(Figure 1). We included variables indicating the amount of
time an article spent in each of these seven regions as
controls after Winsorization of the top 1% of outliers (to
prevent extreme outliers from exerting undue influence on
our results; for summary statistics, see Tables WA1 and
WA2 in the Web Appendix at www.marketingpower. com/
jmr_ webappendix).
Release timing and author fame. To control for the possi-

bility that articles released at different times of day receive
different amounts of attention, we created controls for the
time of day (6 A.M.–6 P.M. or 6 P.M.–6 A.M. eastern standard
time) when an article first appeared online. We control for
author fame to ensure that our results are not driven by the
tastes of particularly popular writers whose stories may be
more likely to be shared. To quantify author fame, we cap-
ture the number of Google hits returned by a search for each
first author’s full name (as of February 15, 2009). Because

Table 1
ARTICLES THAT SCORED HIGHLY ON DIFFERENT DIMENSIONS

Primary Predictors
Emotionality

•“Redefining Depression as Mere Sadness”
•“When All Else Fails, Blaming the Patient Often Comes Next”

Positivity
•“Wide-Eyed New Arrivals Falling in Love with the City”
•“Tony Award for Philanthropy”

(Low Scoring)
•“Web Rumors Tied to Korean Actress’s Suicide”
•“Germany: Baby Polar Bear’s Feeder Dies”

Awe
•“Rare Treatment Is Reported to Cure AIDS Patient”
•“The Promise and Power of RNA”

Anger
•“What Red Ink? Wall Street Paid Hefty Bonuses”
•“Loan Titans Paid McCain Adviser Nearly $2 Million”

Anxiety
•“For Stocks, Worst Single-Day Drop in Two Decades”
•“Home Prices Seem Far from Bottom”

Sadness
•“Maimed on 9/11, Trying to Be Whole Again”
•“Obama Pays Tribute to His Grandmother After She Dies”

Control Variables
Practical Utility

•“Voter Resources”
•“It Comes in Beige or Black, but You Make It Green” (a story
about being environmentally friendly when disposing of old
computers)

Interest
•“Love, Sex and the Changing Landscape of Infidelity”
•“Teams Prepare for the Courtship of LeBron James”

Surprise
•“Passion for Food Adjusts to Fit Passion for Running” (a story
about a restaurateur who runs marathons)
•“Pecking, but No Order, on Streets of East Harlem” (a story about
chickens in Harlem)

Table 2
PREDICTOR VARIABLE SUMMARY STATISTICS

M SD
Primary Predictor Variables
Emotionalitya 7.43% 1.92%
Positivitya .98% 1.84%
Awea 1.81 .71
Angera 1.47 .51
Anxietya 1.55 .64
Sadnessa 1.31 .41

Other Control Variables
Practical utilitya 1.66 1.01
Interesta 2.71 .85
Surprisea 2.25 .87
Word count 1021.35 668.94
Complexitya 11.08 1.54
Author fame 9.13 2.54
Author female .29 .45
Author male .66 .48
aThese summary statistics pertain to the variable in question before

standardization.

of its skew, we use the logarithm of this variable as a con-
trol in our analyses. We also control for variables that might
both influence transmission and the likelihood that an arti-
cle possesses certain characteristics (e.g., evokes anger).
Writing complexity. We control for how difficult a piece

of writing is to read using the SMOG Complexity Index
(McLaughlin 1969). This widely used index variable essen-
tially measures the grade-level appropriateness of the writ-
ing. Alternate complexity measures yield similar results.
Author gender. Because male and female authors have

different writing styles (Koppel, Argamon, and Shimoni
2002; Milkman, Carmona, and Gleason 2007), we control
for the gender of an article’s first author (male, female, or
unknown due to a missing byline). We classify gender using
a first name mapping list from prior research (Morton,
Zettelmeyer, and Silva-Risso 2003). For names that were
classified as gender neutral or did not appear on this list,
research assistants determined author gender by finding the
authors online.
Article length and day dummies. We also control for an

article’s length in words. Longer articles may be more likely
to go into enough detail to inspire awe or evoke anger but
may simply be more viral because they contain more infor-

mation. Finally, we use day dummies to control for both
competition among articles to make the most e-mailed list
(i.e., other content that came out the same day) as well as
any other time-specific effects (e.g., world events that might
affect article characteristics and reader interest).
Analysis Strategy
Almost all articles that make the most e-mailed list do so

only once (i.e., they do not leave the list and reappear), so
we model list making as a single event (for further discus-
sion, see the Web Appendix at www.marketingpower.com/
jmr_ webappendix). We rely on the following logistic
regression specification:

where makes_itat is a variable that takes a value of 1 when
an article a released online on day t earns a position on the
most e-mailed list and 0 otherwise, and t is an unobserved
day-specific effect. Our primary predictor variables quantify
the extent to which article a published on day t was coded as
positive, emotional, awe inspiring, anger inducing, anxiety
inducing, or sadness inducing. The term Xat is a vector of
the other control variables described previously (see Table
3). We estimate the equation with fixed effects for the day
of an article’s release, clustering standard errors by day of
release. (Results are similar if fixed effects are not included.)
Results
Is positive or negative content more viral? First, we

examine content valence. The results indicate that content is
more likely to become viral the more positive it is (Table 4,
Model 1). Model 2 shows that more affect-laden content,
regardless of valence, is more likely to make the most e-
mailed list, but the returns to increased positivity persist
even controlling for emotionality more generally. From a
different perspective, when we include both the percentage
of positive and negative words in an article as separate pre-
dictors (instead of emotionality and valence), both are posi-
tively associated with making the most e-mailed list. How-
ever, the coefficient on positive words is considerably larger
than that on negative words. This indicates that while more
positive or more negative content is more viral than content
that does not evoke emotion, positive content is more viral
than negative content.
The nature of our data set is particularly useful here

because it enables us to disentangle preferential transmis-
sion from mere base rates (see Godes et al. 2005). For
example, if it were observed that there was more positive
than negative word of mouth overall, it would be unclear
whether this outcome was driven by (1) what people
encounter (e.g., people may come across more positive
events than negative ones) or (2) what people prefer to pass
on (i.e., positive or negative content). Thus, without know-
ing what people could have shared, it is difficult to infer
much about what they prefer to share. Access to the full cor-

=

+ −

α + β ×
+ β ×
+ β × + β ×
+ β ×
+ β × + ′θ ×















































(1)makes_it 1

1 exp

z-emotionality
z-positivity
z-awe z-anger
z-anxiety
z-sadness X

,at
t 1 at
2 at
3 at 4 at
5 at
6 at at

Table 3
PREDICTOR VARIABLES

Variable Where It Came from
Main Independent Variables
Emotionality Coded through textual analysis

(LIWC)
Positivity Coded through textual analysis

(LIWC)
Awe Manually coded
Anger Manually coded
Anxiety Manually coded
Sadness Manually coded

Content Controls
Practical utility Manually coded
Interest Manually coded
Surprise Manually coded

Other Control Variables
Word count Coded through textual analysis

(LIWC)
Author fame …

https://doi.org/10.1177/2056305119847516

Creative Commons Non Commercial CC BY-NC: This article is distributed under the terms of the Creative Commons Attribution-
NonCommercial 4.0 License (http://www.creativecommons.org/licenses/by-nc/4.0/) which permits non-commercial use, reproduction

and distribution of the work without further permission provided the original work is attributed as specified on the SAGE and Open Access pages
(https://us.sagepub.com/en-us/nam/open-access-at-sage).

Social Media + Society
July-September 2019: 1 –13
© The Author(s) 2019
Article reuse guidelines:
sagepub.com/journals-permissions
DOI: 10.1177/2056305119847516
journals.sagepub.com/home/sms

Original Article

Introduction: The Disruption of
Traditional Advertising

For decades, the advertising industry was based on an asym-
metrical communication model, where marketers would
engage audiences via paid media channels. The advent of
social media platforms completely transformed the general
media landscape, along with the advertising model, as audi-
ences shifted from the role of content receivers to content
creators, distributors, and commentators (Keller, 2009; Scott,
2015). Simply put, the empowerment of audiences from
mere viewers to active content distributors effectively flipped
the advertising model on its head. Where paid media (in this
case, advertising) was once supported by earned and owned
media, the modern advertising model uses owned, shared,
and earned media as the key media planning strategy, sup-
ported by paid media (Pearson, 2016). Recognizing the
increased potential for free content distribution, marketers
realized that creating highly engaging advertising content
could expand potential reach, a cheaper and more credible
tactic than traditional paid advertising (Cho, Huh, & Faber,
2014; Golan & Zaidner, 2008). This fundamental disruption
of the advertising and marketing world led to growing
interest in content creation, co-creation, and distribution.

Generally defined, advertising refers to the “paid non-
personal communication from an identified sponsor using
mass media to persuade or influence an audience” (Wells,
Moriarty, & Burnett, 2000, p. 6). Consistent with most, but
not all, of these requirements, Porter and Golan (2006)
defined viral advertising as “unpaid peer-to-peer communi-
cation of provocative content originating from an identified
sponsor using the Internet to persuade or influence an audi-
ence to pass along the content to others” (p. 33).

The expanding literature on viral advertising recognizes
the ways in which peer-to-peer distribution of advertising
content are redefining the industry. When examined holisti-
cally, the literature has several limitations. First, existing
viral advertising research is limited primarily to advertising
spread within one step of the original source (e.g., predicting
the number of message shares), while information on social

847516 SMSXXX10.1177/2056305119847516Social Media <span class=”symbol” cstyle=”Mathematical”>+</span> SocietyHimelboim and Golan
research-article20192019

1University of Georgia, USA
2University of South Florida, USA

Corresponding Author:
Itai Himelboim, Department of Advertising and Public Relations, Grady
College of Journalism and Mass Communication, University of Georgia,
Athens, GA 30602-3018, USA.
Email: [email protected]

A Social Networks Approach to Viral
Advertising: The Role of Primary,
Contextual, and Low Influencers

Itai Himelboim1 and Guy J. Golan2

Abstract
The diffusion of social networking platforms ushered in a new age of peer-to-peer distributed online advertising content,
widely referred to as viral advertising. The current study proposes a social networks approach to the study of viral advertising
and identifying influencers. Expanding beyond the conventional retweets metrics to include Twitter mentions as connection
in the network, this study identifies three groups of influencers, based on their connectivity in their networks: Hubs, or
highly retweeted users, are Primary Influencers; Bridges, or highly mentioned users who associate connect users who would
otherwise be disconnected, are Contextual Influencers, and Isolates are the Low Influence users. Each of these users’ roles
in viral advertising is discussed and illustrated through the Heineken’s Worlds Apart campaign as a case study. Providing a
unique examination of viral advertising from a network paradigm, our study advances scholarship on social media influencers
and their contribution to content virality on digital platforms.

Keywords
viral advertising, social networks, Twitter, viral marketing, social media influencers

2 Social Media + Society

media often spreads beyond a single step from the original
source. Second, in focusing on the characteristics of shared
content or sharing users, researchers make the assumption
that all shares are equal in terms of their impact. However,
sharing-impact varies among users, based on their connectiv-
ity. Third, the metaphor of virality, the idea that content is
spread gradually among individuals and their immediate
contacts, may not fully capture what is often a complex
multi-actor process of content distribution. Cascades of con-
tent distribution were found to be centered on a small num-
ber of distributors, creating a hierarchical, rather than
egalitarian, pattern of content distribution (Baños, Borge-
Holthoefer, & Moreno, 2013).

This study proposes a social networks approach to address
these limitations, using Heineken’s Worlds Apart campaign
as a case study. Data are collected for all Twitter users post-
ing links to the original advertisement on YouTube, and the
subsequent retweets and mention relationships. While a
growing body of scholarship examines the potential impact
of social media influencers in online marketing campaigns,
they often treat all influencers as one and the same (Evans,
Phua, Lim, & Jun, 2017; Phua & Kim, 2018).

We argue that different types of influencers impact social
networks in different degrees and ways. Informed by a body
of scholarship in social networks, we propose that there are
three types of influencers: primary, contextual, and low
influencers. Primary influencers are hubs, users who attract
large and disproportionate retweets from other users in the
network. Contextual influencers play a role of bridges in the
network by providing context regarding the overall discus-
sion and thus help to understand the distribution of content
beyond the quantity of retweets. Low influencers are users
who shared a link to online content; however, these users
were neither retweeted nor mentioned by anyone else in the
network. While low influencers have limited individual con-
tributions to content distribution, their aggregate influence is
substantial.

Social Media Influencers

An emergent body of scholarship in the field of marketing,
advertising, and public relations examines the intermediary
function of influencers between brands and consumers, orga-
nizations, and stakeholders in social media engagement (De
Veirman, Cauberghe, & Hudders, 2017; Freberg, Graham,
McGaughey, & Freberg, 2011; Phua, Jin, & Kim, 2016). At
the most basic level, influencer is identified by their number
of followers and their ability to impact social media conver-
sation regarding brands or topics (Watts & Dodds, 2007).
While the term social media influencer is ubiquitously
applied, there are few formal definitions of what an influ-
encer actually is. Brown and Hayes (2008) defined influenc-
ers broadly as individuals who hold influence over potential
buyers of a brand or product to aid in the marketing activities
of the brand. Others narrow the definition of an influencer to

reflect on the latest marketing trend in which social media
celebrities are paid by advertisers to promote products
(Abidin, 2016; Evans et al., 2017; Senft, 2008).

Moving beyond definitions, scholars attempt to theorize
why it is that some social media users grow more influential
than others via relationship building. To explain the influ-
ence of influencers, media scholars often depend on the
parasocial relationship explanation (Daniel, Crawford, &
Westerman, 2018; Lou & Yuan, 2018; Rasmussen, 2018).
Moving beyond a temporary parasocial interaction (as origi-
nally conceptualized by Horton & Wohl, 1956), parasocial
relationships between audience members and mediated
characters are formed over a period of time and provide
audience members with a sense of engagement with on-
screen characters (Klimmt, Hartmann, & Schramm, 2006;
Tukachinsky, 2010). In the context of social media, such
parasocial relationships provide influencers with unique
social capital that leads to audience trust (Tsai & Men, 2017;
Tsiotsou, 2015).

Indeed, the central role of trust in parasocial relation-
ships may provide a plausible explanation for the influencer
phenomenon and the rise of influencer marketing (Audrezet,
De Kerviler, & Moulard, 2018). Trust has been identified as
a key predictor of several advertising consequences includ-
ing recall, attitude, and likelihood to share (Cho et al.,
2014; Lou & Yuan, 2018; Okazaki, Katsukura, & Nishiyama,
2007). Abidin (2016), building on the concept of parasocial
relations, identified four ways that influencers appropriated
and mobilized intimacies: commercial, interactive, recipro-
cal, and disclosive. Influencers are identified not only based
on their sheer number of such parasocial relationships, such
as subscribers or followers on social media, but primarily
based on their ability to impact social media conversation
and subsequent behavior regarding brands or topics (Watts
& Dodds, 2007).

We propose to complement existing conceptualization of
influencers by shifting the focus from influencers’ engage-
ment or the nature of individual connections with them, to
their ability to reach large, unique, and relevant audiences
and to shape the conversation about brands and topics. It is
the distribution of content that allows influencers to influ-
ence, and therefore provides a key theoretical framework for
identifying social media influencers. We next discuss viral
advertising as a theoretical framework for content reach, fol-
lowed by its limitations. We then take a social networks
approach to theorize social media influencers, bridging both
bodies of literature.

Viral Advertising

As explained by Golan and Zaidner (2008), there are several
key differences between viral and traditional advertising.
First, viral advertising earns audience eyeballs, as opposed to
paying for them. This is a major departure from the tradi-
tional advertising exchange, where brands purchase media

Himelboim and Golan 3

space and interrupt an audience’s media consumption with
advertisements. Second, viral advertisements provide such
increased value to audiences that they transform audiences
from passive content receivers to active social distributors
who play a key role in advertisement distribution. Third,
although there are limited studies speaking to this point, it is
worth noting that information sharing has been shown to
increase a user’s followers on Twitter, which is a long-term
benefit for marketers (Hemsley, 2016).

What Makes Advertising Go Viral?

Why do some advertisements receive wide-scale viewership
via audience distribution, while others do not? Scholars offer
different approaches to this question, one focusing on con-
tent characteristics (Brown, Bhadury, & Pope, 2010; Golan
& Zaidner, 2008; Petrescu, 2014) and another examining
virality attribute factors such as brand relationships (Hayes
& King, 2014; Ketelaar et al., 2016; Shan & King, 2015).

Porter and Golan (2006) specifically identify provocative
content as contributing to advertising virality. Other studies
identify appeals to sexuality, as well as shock, violence, and
other inflammatory content as key elements of message viral-
ity (Brown et al., 2010; Golan & Zaidner, 2008; Petrescu,
2014). Eckler and Bolls (2011) argue that the emotional tone
of advertisement is directly related to audience intention to
forward ads to others. Yet advertising content, tone, and emo-
tion cannot fully account for ad virality. Scholars point to a
variety of other variables significantly related to advertising
virality including brand relationship (Hayes & King, 2014;
Ketelaar et al., 2016; Shan & King, 2015), attitude toward the
ad (Hsieh, Hsieh, & Tang, 2012; Huang, Su, Zhou, & Liu,
2013), and credibility of the sender/referrer (Cho et al., 2014;
Phelps, Lewis, Mobilio, Perry, & Raman, 2004).

Hayes, King, and Ramirez (2016) advanced research on
viral advertising by illustrating the importance of interper-
sonal relationship strength in referral acceptance. Their study
suggested that individuals are motivated to share advertising
content based on reputational enhancement and reciprocal
altruism. Alhabash and McAlister (2015) conceptualized
virality based on three key components: viral reach, affective
evaluation, and message deliberation. The authors linked
virality and online audience behaviors in what they refer to
as viral behavioral intentions (VBI). This linkage is sup-
ported by later research indicating that the virality of digital
advertising is often related to several VBIs motivated by a
variety of audience-based characteristics (Alhabash, Baek,
Cunningham, & Hagerstrom, 2015; Alhabash et al., 2013).

Limitations of Viral Advertising Research

In essence, viral advertising represents a “peer-to-peer com-
munication” strategy that depends on distribution of content
(Petrescu & Korgaonkar, 2011; Porter & Golan, 2006).
Despite the fact that most peer-to-peer social media shares

include multiple distribution phases (e.g., from user A to user
B to user C), existing viral advertising research is mostly
limited to one-step advertisement spread (e.g., predicting
number of message shares). Studies suggest that while con-
tent may be shared by many users, most viral content is
spread beyond this single step (Bakshy, Hofman, Mason, &
Watts, 2011). The body of literature concerning viral adver-
tising does not examine advertising spread beyond a user’s
immediate set of connections.

Second, the literature conceptualizes virality based on
such sharing metrics as shares or retweets. In doing so, schol-
ars fail to account for the possibility that the overall impact
of such user actions may not result in equal content distribu-
tion outcomes. In fact, studies on virality of content and cas-
cades of information flow highlight that “popularity is largely
driven by the size of the largest broadcast” (Goel, Anderson,
Hofman, & Watts, 2015, p. 180). In other words, it is not only
the number of consumer-to-consumer interactions but the
connectivity of these consumers with others that determines
the impact of viral advertising. One user’s retweet may count
more than another user.

A third limitation is the more subtle assumption of virality
as metaphor. The idea that content is spread gradually from
one source to that source’s immediate small group of connec-
tions, to their neighbors, and so on is a powerful metaphor
that resonates well with many scholars (Miles, 2014; Porter
& Golan, 2006). However, research shows no foundation for
such an egalitarian assumption. Connections are distributed
in a skewed manner across individuals, a phenomenon
referred to in ways that vary by discipline:

in economics it goes by the name “fat tails,” in physics it is
referred to as “critical fluctuations,” in computer science and
biology it is “the edge of chaos,” and in demographics and
linguistics it is called “Zipf’s law.” (Newman, 2000)

At the end of the day, most pieces of shared content are not
re-shared by others, and thus are spread by very few.
Similarly, from an advertisement and social media perspec-
tive, Nielsen (2006) presented the “1-9-90 rule,” suggesting
that content is created by 1% of users and distributed by 9%
to the remaining 90% of content receivers. Baños et al.
(2013) showed that only a small minority of content dis-
tributors will account for content virality. In addition, Pei,
Muchnik, Andrade, Zheng, and Makse (2014) suggested
that “due to the lack of data and severe privacy restrictions
that limit access to behavioral data required to directly infer
performance of each user, it is important to develop and
validate social network topological measures capable to
identify superspreaders” (p. 8).

To address these key gaps in the literature of viral adver-
tising and subsequently our ability to theorize influential
users in terms of their content diffusion, we take a social net-
works approach, which focuses on patterns of connectivity
among users. We propose that social media influencers are

4 Social Media + Society

ultimately determined by their position in an issue or brand-
specific conversation network, allowing their posted content
to be distributed in a strategic manner. As such, these influ-
encers play key roles in the virality of any advertising cam-
paign on social media. A social networks approach, as
illustrated by Himelboim, Golan, Moon, and Suto (2014)
provides for a macro-understanding of social media relation-
ships, content flow, and the role of social media influencers
within the network.

The Social Networks Approach

The social networks conceptual framework shifts the focus
from individual traits to patterns of social relationships
(Wasserman & Faust, 1994). Applying a social networks
approach to social media activity allows researchers to cap-
ture content virality and identify key social media influenc-
ers that affect the conversation about a brand and reach key
groups of consumers. A social network is formed when con-
nections (“links”) are created among social actors (“nodes”),
such as individuals and organizations. The collections of
these connections aggregate into emergent patterns or net-
work structures. On Twitter, social networks are composed
of users and the connections they form with other users when
they retweet, mention, and reply to (Hansen, Shneiderman,
& Smith, 2011).

The network approach can bridge the viral advertising
and social media influencer’s bodies of literature. As dis-
cussed earlier, social media platforms allow individuals to
maintain parasocial relationships with influencers (Abidin,
2016). In the case of Twitter, such engagement is manifested
in the form of mentions, likes, and retweets. In social net-
works research, these relationships are conceptualized as
links in a network.

The social networks approach allows us to capture the
distribution of a specific piece of content (i.e., an advertise-
ment) and identify users in key positions in the network that
are responsible for the distribution of ads, as social media
influencers. It should be noted that even in studies on infor-
mation diffusion in related disciplines, it is quite rare to track
the virality of a single piece of content, rather than the over-
all diffusion of messages in a broader conversation.

Viral advertising research often focuses on the most visi-
ble type of content that is spread, shared, or retweeted on
Twitter. Social media influencers are often examined by their
number of connections in a social media platform (De
Veirman et al., 2017). However, a link to a video advertise-
ment, or any other source of paid advertising content, may be
posted by more than a single user who contributes to its dif-
fusion. In other words, while the advertisement itself may
have a single point of origin (e.g., a YouTube video page),
this advertisement may have multiple users who may account
for multiple points of origin for distribution on Twitter. While
a particular video may have gained many views and shares
on its platform of origin (“gone viral”), not all shares on

Twitter contributed equally to its virality. We therefore ini-
tialize our understanding of content distribution patterns by
asking,

RQ1: What is the distribution structure of a viral adver-
tisement on Twitter?

A single network can have different types of links, or ties,
that connect its users. On Twitter, users can be connected,
among others, by relationships of retweets and mentions. A
network of advertising virality captures users who posted
content with a hyperlink to a given ad. Such Twitter users
share a link to a given advertisement via a tweet, expanding
its reach one step away from the source (YouTube). Some
studies have examined the overall network structure to
explain virality. Pei et al. (2014) used social network analysis
on LiveJournal, Twitter, Facebook, and APS journals and
found that users who spread the most content were located in
the K-Core (a metrics of subgroup cohesiveness in the net-
work). At the node-level, a few users are expected to contrib-
ute further to the virality by having their tweets shared, or
retweeted, by many additional users. Such users capture
virality beyond a single step away from the source. Users
with many connections in the network are known as social
hubs (Goldenberg, Libai, & Muller, 2001) or simply Hubs.
Using computer simulations, Hinz, Skiera, Barrot, and
Becker (2011) found that seeding messages to hubs outper-
formed a random seeding strategy and seeding to low-degree
users, in terms of number of referrals. Kaplan and Haenlein
(2011) also illustrated the role that hubs play in integrative
social media and viral marketing campaigns.

Recognizing that the emergent literature on social media
influencers is somewhat undermined by the various uses of
the term influence to reflect different functions of influence,
we recommend the categorization of influencers into three
different types, based on the type of relationships, links in the
network, that makes them central in a network.

Social networks literature repeatedly shows that given the
opportunity to interact freely, connections among users will
be distributed unequally, as a few will enjoy large and dis-
proportionate number of relationships initiated with them,
while most will have very few ties. On Twitter, content
posted by a few users will enjoy major distribution via
retweeting, while the rest will gain little shares, if any.
Indeed, Araujo, Neijens, and Vliegenthart (2017), define
influentials as “users with above average ability to stimulate
retweets to their own messages” (p. 503), consistent with
conceptualization of influencers based on impact on content
distribution (Cha, Haddadi, Benevenuto, & Gummadi, 2010;
Kwak, Lee, Park, & Moon, 2010). Hubs as conceptualized in
social networks literature, therefore, are one type of social
media influencers as conceptualized in social media scholar-
ship, as each one makes a major contribution to content dis-
tribution. One type of influencer, from a social networks
conceptualization, is therefore the Primary Influencer, as it

Himelboim and Golan 5

is one of few members responsible for the distribution of
content in the network. We therefore present the following
research question:

RQ2: Which users serve as Primary Influencers in a viral
advertising network?

On Twitter, retweets are attributed to the original tweet;
therefore, operationalizing links in this network only as
retweets fails to capture information flow beyond one step
away from a user who shared a link to an ad. In other words,
since users are unlikely to share the same link more than
once, the network of retweets will create distinct subsets of
users, each retweeting a single tweet. These subsets are com-
pletely, or almost completely, disconnected from one another.
As discussed earlier, a key limitation of viral advertising lit-
erature is that studies are limited to the extent they measure
diffusion from a single source. In order to maximize insights
from the social networks approach to viral advertising, other
types of ties should be considered.

The practice of mentioning users on Twitter, using the @
symbol, serves two main purposes. First, it associates a post
with another user (e.g., an individual, an organization, a
brand), serving as metadata for that tweet. Second, it serves
as a secondary route of content distribution. When a tweet
mentions a given user, that tweet will appear on the recipi-
ent’s Notifications tabs and Home timeline view if the author
of the tweet follows the sender. Conceptualizing mentions on
Twitter as links in a social network captures the context of
the virality of advertisements by connecting users beyond
immediate retweeting of a single source. In other words, this
practice bridges the otherwise disconnected subsets of
retweeting users. In social network literature, bridging is a
concept that can advance the understanding of advertisement
virality and the key users who play a key role in it.

Bridges and Structural Holes

Burt’s (1992, 2001) theory of structural holes examines
social actors (e.g., individuals and organizations) in unique
positions in a social network, where they connect other actors
that otherwise would be less connected, if connected at all. In
Burt’s (2005) words, “A bridge is a (strong or weak) relation-
ship for which there is no effective indirect connection
through third parties. In other words, a bridge is a relation-
ship that spans a structural hole” (p. 24). A lack of relation-
ships among social actors, or groups of actors, in a network
gives those positioned in structural holes strategic benefits,
such as control, access to novel information, and resource
brokerage (Burt, 1992, 2001). Actors that fill structural holes
are viewed as attractive relationship partners precisely
because of their structural position and related advantages
(Burt, 1992, 2001).

The nature of Twitter retweets, however, rarely allows
bridges to form as retweets that are associated with an

original tweet (unless modified retweets are used). In other
words, the spread of retweets remains within a single step
away from the author who posted that message. Therefore,
this additional type of structural characteristic is not enough
to characterize a new type of influential user in viral adver-
tising. Conceptualizing a second type of parasocial relation-
ship on the network—mentions (the inclusion of a reference
to another Twitter user in a post)—as links in a network
allows bridges to form as they provide an additional connec-
tion among users. While mentions do not represent primary
stages in content distribution, they do provide meaningful
points of context that allow researchers to better understand
the overall virality of an advertisement.

Since content distribution or virality on Twitter does not
take place in a vacuum but rather is often responsive to the
broader online conversion, the distribution of any specific
tweet may be impacted by contextual factors. For example,
the distribution of a tweet about a pharmaceutical company
may be impacted by related actors linked to the industry in
news coverage. On Twitter, users often provide context to
their posted content, among others, by mentioning related
users via their handles (@). While such users do not take an
active role in the conversation, they are nominated, so to
speak, as influencers in the network, as they provide addi-
tional explanation for content virality. In other words, they
allow researchers and practitioners to understand that the vast
distribution of an ad on Twitter is driven by a larger context.

We therefore define a second type of social networks-
driven influencer type as Contextual Influencer—highly
mentioned users who bridge otherwise separated groups of
retweeting users.

RQ3: Which users serve as Contextual Influencers in viral
advertising networks?

Beyond a few users in key positions—Hubs or Bridges—
many users’ content sharing is more limited. Each user con-
tributes little to advertisement virality, as they reach only
their immediate Twitter followers. However, as such users
are often the majority of distribution agents, they ultimately
make a major contribution to ad virality. We call users who
are isolated in the network (defined as incurring no retweets
for their shared video tweets) Low Influence users.

RQ4: What percentage do Low Influencers make of all
users in the network?

Proof of Concept: Heineken’s “Worlds
Apart” Viral Advertisement

To illustrate the conceptual framework proposed in the cur-
rent study, we selected a popular Heineken advertisement on
YouTube, titled “Heineken | Worlds Apart | #OpenYourWorld.”
Heineken described the ad as, “Heineken presents ‘Worlds
Apart’ An Experiment. Can two strangers with opposing

6 Social Media + Society

views prove that there’s more that unites than divides us?” In
this ad, Heineken harnesses a social issue, political and social
polarization, and the importance of a constructive conversa-
tion across opinions …

Annotated Bibliography|FAQs

What is an annotated bibliography?
An annotated bibliography is an analytical summary (or annotation) of the books, scholarly articles, book chapters, or
web documents that you consulted during your research. Writing an annotation encourages you to critically read and
reflect on the points of view and arguments presented in each of your sources and relate them to your research topic.

What do I include in my annotation?

For each source that you consulted:

 RETELL: Briefly summarize the main points*
 what are the author’s arguments (thesis)?
 what research methods were used (survey, questionnaire, interviews,

experiments, self‐assessments)?
 what are the findings or conclusion?

 REFLECT: Comment on the strengths and weaknesses
 RELATE: Assess the relevance of each source

*Note: The length of an annotation can vary (1 paragraph or more). Sometimes you are asked to simply summarize and
other times you may have to compare one source to another. Ask your instructor about their requirements.

If you want to comment on the… Suggested phrases you can use
author’s argument or thesis The researcher argues that….

According to the author, there are two reasons for…
This study suggests that…

strengths and weaknesses The main strength of the article was that….
One limitation of the study was that….

relevance/usefulness The article was useful since it evaluates ….
This book was somewhat relevant to my study because…

conclusions or findings The author concludes that…
The data indicates that . . .
A study exploring…… revealed that…..

How do I format the document?
 List your sources in the appropriate referencing style (APA, MLA, AMA, etc.)

 Alphabetize the list by Author’s last name

 Write an annotation below (or sometimes right after) each reference

 Double‐space the text

 The second and following lines of each reference entry and the annotation are indented so
that only the author’s last names are along the left margin

 In‐text citations are not needed in an annotated bibliography

Sample Annotated Bibliography

Question: What are the benefits of school‐based fitness programs and the challenges of establishing such programs?

Assignment: Submit an annotated bibliography with 10 scholarly sources in APA format. (Final essay due 3 weeks later).

Running head: FITNESS PROGRAMS 1

Title Page

(Ask your instructor for requirements)

School-Based Fitness Programs

Mina Sharma

Sheridan College

References
are listed
alphabetically

Second and
following lines
are indented

FITNESS PROGRAMS 2

Annotated Bibliography

Abudayya, A., Shi, Z., Abed, Y., & Holmboe-Ottesen, G. (2011). Diet,

nutritional status and school performance among adolescents in
Gaza Strip. Eastern Mediterranean Health Journal, 17 (3), 218-
225. Retrieved from
http://www.emro.who.int/publications/emhj/index.asp
Researchers in this study used a “food frequency” questionnaire
to determine the relationship between food intake and school
performance. Results from the 932 surveyed students indicate
that diet and nutrition had a significant impact on school
performance and students’ health. This suggests that……

Racette, S. B., Cade, W. T., & Beckmann, L. R. (2010). School-based

physical activity and fitness promotion. Physical Therapy, 90 (9),
1214-1218. Retrieved from http://ptjournal.apta.org/
The authors conducted a 3 year intervention and assessment of a
school based physical activity program for students in grades K-
12. This report was useful because it gave suggestions on how to
set up a school fitness program. Researchers convinced school
administrators of the program benefits by using evidence from
Cochrane systematic reviews to highlight the health benefits of
physical activity as well as the improved focus that students
would have on their academics. The study found that there was a
6% increase in the number of students who met the physical
fitness standards. One weakness in this study was that it did not
take into account the impact of diet….

Siega-Riz, A.M., El Ghormli, L. Mobley, C. Gillis, B. Stadler, D.

Hartstein, J., …Bridgman, J. (2011). The effects of the
HEALTHY study intervention on middle school student dietary
intakes. The International Journal of Behavioral Nutrition and
Physical Activity 8 (7). doi: 10.1186/1479-5868-8-7

Note:
Writing
should be
double‐
spaced

Describe the
research
methods and
results

Comment on the
usefulness of the
report

Comment on the
results and any
weaknesses of
the study

Adapted from Successful College Writing, 4th edition. Call # PE 14098.M397 2010, Purdue OWL: Annotated Bibliographies retrieved from
https://owl.english.purdue.edu/owl/resource/614/03/ and Deborah Knott’s Writing an Annotated Bibliography retrieved from
http://www.writing.utoronto.ca/advice/specific‐types‐of‐writing/annotated‐bibliography Page 2