# 5 WOM / Virality

To understand virality, we first need to comprehend its root, which is WOM. Acknowledging that there are subtle differences among WOM, virality, and sharing, we use the three terms interchangeably in this paper. Previous authors provide strong supports for the similarity between virality and WOM

Formally, defined WOM as “informational communication directed at other consumers about the ownership, usage, or characteristics of particular goods and services or their sellers.” In the age of the Internet, WOM evolves to “word of mouse” (or e-WOM). Both virality and WOM take advantages of the network effect to reach wide audiences , but virality does it better due to its competitive cost, rapid diffusion, and better targeting . Hence, virality is more synonymous with e-WOM.

Virality comes up as a by-product of the Internet, social media platforms, and network theory developments. In network theory, researchers from the domain of sociology or computer science study the structural characteristics/properties of a node (an individual) or a network, such as a node’s position in a network or network’s cluster. While epidemiologists, management, and marketing researcher study the characteristics of the entity being transmitted, such as virus or message itself. Thus, virality can be driven by content or context of the message (Berger & Milkman, 2012), nature of the spreader, and audience (Hennig-Thurau, Gwinner, Walsh, & Gremler, 2004; Iyengar, van den Bulte, & Valente, 2011), seeding strategies (Kalish, Mahajan, & Muller, 1995; Libai, Muller, & Peres, 2005), and social network structure (Bampo et al., 2008). In this conceptual development, we focus on the content and context drivers of virality.

Virality is defined as the probability of an entity being passed along (Hansen, Arvidsson, Nielsen, Colleoni, & Etter, 2011). The entity can be a message, content, video, knowledge, or disease. Alternatively, (Tucker, 2015) defines virality as “a process whereby an ad is successively shared by viewers.” These two definitions focus on virality as a process; instead, we focus on viral as a construct. Hence, we are in line with (Chandler & Munday, 2016) defining virality (social media virality, social shareability, social media shareability) as “the potential for spreadability of any given content; or particular qualities which are considered to have led to content ‘going viral’.”

For WOM to work, people must share brand-related content, and attention needs to be translated into sales. With these positive associations towards a brand, marketers can increase a customer’s positive brand attitude and later purchase behavior (Faircloth, Capella, & Alford, 2001).

81 percent of a telecom firm indicated that they are likely to refer other to the company; however, only 33 percent actually did. And of those who are referred, only 8 percent became profitable customer.

Even though suffering from the same fate as brand equity, WOM has made it come back under a different alias - virality. The attention that virality has garnered in recent years can be attributed to several reasons:

1. the dawn of the Internet helps propel marketing into a new paradigm - digital marketing - where earned media is the new king
2. the introduction of social network platforms such as Facebook, Twitter that facilitate the transmission of information and messages seamlessly and provide an unlimited source of data
3. The advancement in network theory both theoretically, and empirically.

One platform that can be representative of this digital revolution is Twitter. According to a recent report, Twitter has 145 million monetizable daily active users. Twelve percent of Americans get their news from Twitter. And Twitter has 2 billion video impressions per day (25 Twitter Stats All Marketers Need to Know in 2020. There are more than 500 million Tweets are sent per day. (The 2014 #YearOnTwitter, 2014). One cannot deny the power of online WOM as well as its potential reach; hence, viral marketing has been born with these platforms.

The age group from 18-24 can be more challenging to target since they reported fewer positive responses compared to other age groups . Since their threshold is higher due to the overwhelming digital content they encounter, their emotions are harder to activate.

Wide gaps between two opposing opinions can increase future discussion (reviews); however, fluctuation around one opinion does not. Higher informational content moderate positively the relationship between opposing opinions and future reviews.

WOM should be taken under the context of the evolution of conversation.

Contagion/ Virality used to be thought of in term of S-shaped curve, but there are problems with this notion:

1. S-shaped could result from different processes that are not contagion

1. Population heterogeneity

2. Marketing efforts

2. Selection bias (or survival bias)

Questions:

• When diffusion is “viral”?

• What does viral diffusion look like?

• 90% events have no adoption

• less than 99% events has less than 2 viral adoptions

• Can we predict it?

Viral is different from Popular

• needs diversity (i.e., differences/ distinction)
• Total number of retweet over time is not a good approximate of virality

“Big data” help understand rare events.

Valuable virality

Establishes causality of appeals and brand integralness to virality.

• Compared to informative appeal ads (focus on product features), emotional appeal ads (drama, mood, music, emotion-eliciting strategies) are more likely to be shared.
• Informative appeals boost brand evaluations and purchase because the brand is an integral part of the ad content.
• Hence, to combine the best of both worlds, emotional brand-integral ads boost sharing while also bolstering brand-related outcomes (embedded the brand into the stories).
• The mechanism that ads affect brand evaluation through two simultaneous mechanisms: persuasive attempt and brand knowledge.

“keeping viewers involved depends in large part on two emotions: joy and surprise”(Teixeira, 2012) - “Research: The Emotions that Make Marketing Campaigns Go Viral”

“ads that produce stable emotional states generally aren’t effective at engaging viewers for very long.” And “shock may get people to watch an ad privately, it often works against their desire to share the spot. Like Clothing Drive that try to mimic”Swear Jar” by Bud Light” (Teixeira, 2012)

(Westbrook, 1987, p. 261) defined WOM as “informational communication directed at other consumers about the ownership, usage, or characteristics of particular goods and services or their sellers.” WOM evolves to “word of mouse” (or e-WOM).

:

WOM is goal-driven and serves five key functions:

The motivation behind this process is self -serving and even subconscious level. Hence, we can predict the types of news and information people are most likely to discuss.

• Over 70% of everyday speech is about the self, such as personal experiences or relationships (Dunbar et al., 1997). Consistent with previous research, online behaviors exhibit the same pattern: over 70% of social media posts are about self (Naaman et al., 2010).
• At any given point in time, multiple drivers can be present.

## 5.1 Structural Virality

$$S = f(Q) + E$$

where

• Q = skill

• E = luck

• S = successful viral process

Measure fraction of variance remaining after conditioning on Q

$F = \frac{E(var(S|Q))}{var(S)} = 1 - R^2$

Hence, if

• $$R^2 \to 1$$, you can succeed in the world we created based on “pure skill”

• $$R^2 \to 0$$ you can’t succeed in the world we created (because it’s based on luck)

The more variations in the initial seed, the more likely there is a larger variation in the prediction, which means lower R-squared

They show there is a theoretical limit to the predictive power of predicting cascade event due to its intrinsic dynamic properties.

### 5.1.1 Network Structure and position

• Network management on social network

• Networking activities and ego network structure on an actor’s online success

• Related literature:

• Seeding viral marketing campaigns:

• Contagion and product diffusion: network position and network structure

• How do network structures impact online success?

• Method: stochastic actor-based models

• Ego Network measures

• Ego network density

• Degree centrality

• Betweenness centrality

• Closeness centrality (not applicable to ego network)

• Endogeneity

• Time fixed effects

• Control function: observed time-varying and firm-specific instrumental variables

• Instrumental Variables are those associated with the networking activities: independent US social platform on which those focal European firms also main their profile pages. (activities in the US is correlated with activities in Europe on social media, but not directly to the success of European market).

### 5.1.2 Seeding Strategies

Hinz et al. (2011)

• Despite analytical models and simulations that support seeding strategies don’t work, this study finds evidence in small-scale field experiments and real campaign that seeding to well-connected people can lead to 8 times more successful because they are more active in using their greater reach (even though they don’t have more influence on their peers than less connected people)

• seeding strategy can help increase online WOM, and decrease conversation about competing products (in the same category) but also focal brand products in other categories.

• data: cosmetics and beauty products in Korea.

## 5.2 Mechanisms/ Processes

### 5.2.1 Impression Management

Consumers buy to signal their derived identities to achieve desired impressions (J. Berger & Heath, 2007)

Interpersonal communication (WOM) influences impression management in 3 ways: (1) self-enhancement, (2) identity -signaling, and (3) filling conversational space.

• Self-enhancement: human has a tendency to self-enhance (Markus, 1987). Hence, people like to share things to help them look good (Chung & Darke, 2006; Hennig-Thurau et al., 2004). Since bragging too much about oneself might have an opposite effect. People sometimes engage in” humblebragging”: brag but self-deprecating at the same time (Sezer et al., 2018). Wojnicki and Godes (2017) found people share content for self-enhancement.
• Identity-signalling: they talk about themselves to signal they have certain traits or expertise (G. Packard & Wooten, 2013), such as opinion leaders talk to signal their identity (share to show their knowledge).
• Filling conversational space: (small talk) people engage in small talk to avoid pauses and silence in between conversations so that the other party would not make lousy inferences about the person.

Hence impression management induces people to share things that are

• Entertaining (i.e., interesting, surprising, funny or extreme) so that they shared look more interesting, and in-the-know. Interesting products get more word of mouth . Moderate controversy is a catalyst for word of mouth to spread (Chen & Berger, 2013). And sometimes, people even exaggerate stories to make things more interesting (C. Heath, 1996); around 60% of the stories are distorted . Things that can be considered interesting are those that are either novel, exciting, or violate previous expectations (i.e., surprise) (Berlyne, 2006; Silvia, 2008). Nevertheless, comprehensibility can be a moderate of whether a novel thing is interesting: novel and comprehensible equals interesting, while novel and incomprehensible equal confusing (Silvia, 2008). Novelty refers to things that are new, surprising, exciting, or complex . Comprehensibility refers to the fact that the novelty must be understandable.

• found that false news are spread faster, farther, deeper and more broadly because they are more novel where true or false based on 6 independent fact-checking organizations. This effect is greater about terrorism, natural disasters, science, urban legends, or financial information. False stories inspire fear, disgust and surprise replies, where as true stories inspire anticipation, sadness, joy and trust. Even after controlling for bots account, their results still hold, which means that most of false stories are driven by humans, not bots.
• Useful: because it makes the sharer smart and helpful. Empirical evidence show that useful stories(J. Berger & Milkman, 2012) and higher quality brands (Lovett et al., 2013) are more likely to be shared.

• Self-concept relevant: different people see different domain as the optimal signal amplifier such as politics, economics, etc. hence symbolic products are more likely to be shared than utilitarian ones (Chung & Darke, 2006) (even though useful information is helpful, but the signal to be smart and helpful are not appreciated as being interesting and entertaining).

• Status related: premium brands are talked about more (Lovett et al., 2013). Knowledge is can also be considered as a status symbol, and people share to signal that they re in the know (Ritson & Elliott, 1999)

• Unique: unique products are more likely to be shared, but people with a high need for uniqueness will be less likely to share positive WOM because they do not want people to adopt the product and reduces their uniqueness (Cheema & Kaikati, 2010). Sometimes they even scare others by citing product complexity (Moldovan et al., 2015)

• Common ground: sharing common things induces listener to convey that there is interpersonal similarity and facilitates conversation.

• Emotional valence: people prefer to share positive things than negative news . However, in some special cases, negative WOM can be perceived as a desired component such as reviewers are seen as more intelligent . Which is moderated by whether people are refereeing to themselves or others will affect WOM valence (positive vs. negative) (Kamins et al., 1997). Positive WOM when talking oneself conveys a positive image, while negative WOM about others convey that they are relatively better than the other party.

• Incidental arousal: unrelated arousal (e.g., running in place) can increase sharing in general Berger and Schwartz (2011) because of the increase in arousal levels.

• Accessible things: products with more cued or triggered will be discussed more (J. Berger & Schwartz, 2011). Hence, more advertised products generate more WOM . “top of mind” and “tip of the tongue.” Availability bias means that the easier we can recall something, the more likely we are going to talk bout it. Publicly visible products also have more WOM (J. Berger & Schwartz, 2011)

### 5.2.2 Emotion regulation

WOM help consumers regulate their emotions. Emotion regulation refers to “a person’s ability to effectively manage and respond to an emotional experience” (Rolston & Lloyd-Richardson, 2017). and sharing with other (WOM) can facilitate emotion regulation by:

• Generating social support: sharing can give you comfort and consolation (Rimé, 2009). For example, after a negative emotional experience, sharing can help increase well-being through perceived social support
• Venting: WOM (sharing) helps people achieve catharsis when experiencing negative experiences(Alicke et al., 1992): consumers express when they are angry (Wetzer et al., 2007) or dissatisfied (Anderson, 1998).
• Sensemaking: “talking can help people understand what they feel and why” (J. Berger, 2014)
• Reducing dissonance: even after making a decision, we want to make sure we make the right choice by talking to others tot reduce feelings of doubt (Rosnow, 1980)
• Taking vengeance: to punish the company (different from venting, which makes you feel better), consumers share negative consumption experience (Ward & Ostrom, 2006)
• Rehearsal: people talk to relive positive experiences (Rimé, 2009)

Emotion regulation drives people to share (1) emotional content (2) influence the valence of the content shared, (3) lead people to share more emotionally arousing content:

• Emotionality: more emotional intensity, more sharing (J. Berger & Milkman, 2012), people share more emotional social anecdotes (Peters et al., 2009). But not all emotions increase sharing: for example, shame and guilt decreases sharing (Finkenauer & Rimé, 1998) since sharing those things makes people look bad.

• Valence: “Emotion regulation tends to focus on the management of negative emotions” (J. Berger, 2014), but a counter vailing effect is that under impression management, people avoid sharing negative stories since it might reflect on the shares. Sharing negative things can decrease willingness to share (Chen & Berger, 2013)

• Emotional arousal:

• Negative side: low arousal emotion (sadness), high arousal (anger or anxiety) should increase the need to vent.
• Positive side: low arousal (contentment), high arousal (awe, excitement or amusement) increase desires for rehearsal.

### 5.2.3 Information acquisition

WOM helps acquire information.

• Seeking advice: (Hennig-Thurau et al., 2004; Rimé, 2009). Rather than trial and error, or direct observation, gossip serves as another form of learning (Baumeister et al., 2004)
• Resolving problems: online forums or other online opinion platforms help customers find a solution quickly from others (Hennig-Thurau et al., 2004)

Hence the process of information acquisition would induce to talk more about

• Risky, important, complex, or uncertainty-ridden decision: talking to others can reduce risk, simplify complexity, and increase consumer’s confidence (Hennig-Thurau et al., 2004)
• Lack of (trustworthy) information:

### 5.2.4 Social bonding

“people have a fundamental desire for social relationships (Baumeister & Leary, 1995), and interpersonal communication fills that need (Hennig-Thurau et al., 2004)”. Empirically, people in brand communities because want to connect with others like them (Muniz & O’Guinn, 2001). Sharing facilitates social bonding through

• Reinforce shared views: group memberships, and one’s place in the social hierarchy. WOM (sharing) allows people to fit in as a group member (J. Berger & Heath, 2007; Ritson & Elliott, 1999)
• Reducing loneliness and social exclusion: sharing decrease interpersonal distance and decreases loneliness and exclusion (Maner et al., 2007)

Social bonding drives what people share: people share things are :

• Common ground: social bonding drives people to talk about things they have in common because it makes them feel socially connected (Clark & Kashima, 2007)
• Emotionality: “sharing an emotional story or narrative increases the chance that others will feel similarly” (J. Berger, 2014). Emotional similarity increase group cohesiveness (Barsade & Gibson, 2007). And emotion sharing and social bonding can be a simultaneous process since previous works found that emotion sharing bonds people (Peters & Kashima, 2007) and high arousal emotion increase social bonding needs (Chen & Berger, 2013)

people use WOM to persuade others to affect their satisfaction or choices (J. Berger, 2014). Persuasive drive people to share things that are

• Emotionally polarized (polarized valence): since the goal is to convince something is good or bad (i.e., people share extremely rather than moderately positive(negative) information
• Arousing content: “people who want to persuade others may find shared arousing content to incite others to take desired actions” (J. Berger, 2014)

## 5.3 Moderators

• Two key moderators when different WOM functions have a greater impact:

• Communication audience:

• Communication Channel

• Written vs. oral
• Identifiability: whether communicators are identifiable. (anonymously)
• Audience salience: whether the audience is salient during communication (online communication, the audience is less salient).

Social motivation: why we share :(why_some_videos_go_viral_2015)

• Impression management:

• authority (e.g., “I want to demonstrate my knowledge”)
• Social utility (e.g., this could be useful to my friends)
• coolhunter (e.g., I want to be the first o tell my friends)
• Zeitgeist (e.g., It’s about a current trend or event)
• Conversation starting (e.g., I want to start an online conversation)
• Self-expression (e.g., it says something about me)
• Social good (e.g., It’s for a good cause, and I want to help”).
• Information acquisition:

• Opinion seeking (e.g., “I want to see what my friends think)
• Social bonding:

• Shared passion (e.g, “it lets me connect with my friends bout a shared interest)
• Social in real life (e.g., it will help me socialize with my friends offline).

More frequent exposure to perceptually or conceptually related cues increases product accessibility and makes the product easier to process. In turn, this increased accessibility influences product evaluation and choice, which are found to vary directly with the frequency of exposure to conceptually related cues. These results support the hypothesis that conceptual priming effects can have a strong impact on real-world consumer judgment”

### 5.3.1 Tie Strength

• the authors establish that a receiver is more likely to share content from a sender with whom they share more common followees, common followers, or common mutual followers even after accounting for other measures.

• Assess the impact of network overlap across dyads on the level of content sharing in social platforms.

• Network overlap is defined as:

• Common followees

• Common followers

• Common mutual followers

• Network overlap (similarly embeddedness or social cohesion ) can influence the sharing propensity :

• A high number of common followers means that senders and receivers have similar interests and may have a similar propensity to share a common content

• More common followers and common mutual followers meant that their followers share a similar interest. Hence a receiver thinks content is more suitable for her audience and has a higher propensity to share

• A user may be less likely to share popular content because many others have already shared it

• Used data on Twitter and Digg

J. K. Lee and Kronrod (2020)

• Consensus language: Words and expressions that imply widespread agreement among a group of people about a viewpoint, a product, or a conduct

• The strength tie between the communicator and the receiver of WOM affects the interpretation and persuasiveness of consensus language.

• When employing consensus language, weak links (e.g., distant friends, acquaintances) have more influence than strong ties, because weak relationships inspire views of a broader and more diversified group in agreement, signaling higher validity for the topic at hand.

• For tie strength operationalization in the literature, see table 1 (p. 355)

• Descriptive norms provide social proof 5.4.1 (i.e., social validation)

• For the first experiment, the language of the message was too formal (p. 360). Hence, it’s more likely that the strong ties will become suspicious of the news, while weak ties might think it’s normal and click on it.

## 5.4 Drivers of Virality

### 5.4.1 Social Currency

found that (self) image-related utility is a dominant driver (vs. intrinsic utility) of posting activity

Different people see different domain as the optimal signal amplifier, such as politics, economics

Even though useful information is helpful, the signal of being smart and helpful is not appreciated as being interesting and entertaining

People talk more about status-related products because they enhance their impression.

In extreme cases, they even scare others by citing product complexity (Moldovan, Steinhart, & Ofen, 2015).

Interesting things are discussed more in online settings (J. Berger & Milkman, 2012), but this effect fails to replicate in the case of face-to-face settings (Berger & Schwartz, 2011). Hence, research has found that written communication (e.g., texting, messaging, posting online) leads people to talk more about interesting products or brands than oral communication because they have more time to think and polish what to say, leading to more prevalent self-enhancement motives (Berger & Iyengar, 2013). In other words, not because people do not have self-enhancement motives, but because synchronous conversations do not allow for much time to think and polish ideas, people might talk about mundane things to fill the gaps between conversational turns (Berger, 2014).

### 5.4.2 Accessibility

Accessibility drives ongoing WOM (i.e., publicly visible products are cued more in the environment – top of mind), regardless of how mundane it might be.

From the theory of semantic process (Collins & Loftus, 1975), priming can help activate related perceptual or conceptual constructs in memory spontaneously (Berger & Fitzsimons, 2008). For example, prior exposure affects the brand choice of both target brand and competing brands (Nedungadi, 1990). In other words, the activation of one brand can lead to others more accessible, which increases the chance of remembered brands to be included in the consideration set. (Lee & Labroo, 2004) argues that ease of processing drives our positive evaluation, which is also consistent with the discrepancy -attribution hypothesis. The Discrepancy-attribution hypothesis states that if people cannot easily attribute the ease of processing to a source, they will likely attribute it to the positive quality of the target. Additionally, (Nedungadi, 1990) found that recently exposed brands are more likely to be included and chosen from consumers’ consideration set. Repeated exposure to an object increases favorable ratings (e.g., an object, or person) (Zajonc, 1968), which also apply to attractive people (Moreland & Beach, 1992), and brand (Baker, 1999). Behavioral can be influenced by incidental exposure to stimuli because the primes can activate relevant constructs associated in memory (Christian Wheeler & Petty, 2001; Dijksterhuis & Bargh, 2001; Wheeler & Berger, 2007)

### 5.4.3 Emotions

#### 5.4.3.1 Valence

Milkman and Berger (2014) found that contents that are (1) emotional aroused, (2) useful, or (3) interesting have a higher chance of being shared. Consistently, Milkman and Berger (2014) also found that women produce articles that are more likely to be shared because female authors write more comprehensible, useful, and interesting articles.

People share their extreme emotional experience (i.e., either highly satisfied or highly dissatisfied) (Anderson, 1998)

#### 5.4.3.2 Arousal

Lastly, people use WOM to persuade others to affect their satisfaction or choice . Thus, emotionally polarized (polarized valance) are more likely to be shared due to persuasion reasons. Since the goal is to convince something is good or bad, people share extremely rather moderately positive (negative) information.

Virality is partially driven by physiological arousal Berger and Milkman (2012). These results hold even when the authors control how surprising, interesting, or practically useful content is (all of which are positively linked to virality) and external drivers of attention (e.g., how prominently content was featured) . Different emotion evokes different levels of physiological arousal (i.e., activation) . Arousal is an excitatory state that increases action-related behavior, such as helping others . Hence, we see contents that are physiological arousal get shared more. Specifically, Negative high arousal emotions (e.g., anger or anxiety) are shared more than negative low arousal (e.g., sadness). Moreover, sharing negative experiences can help people vent to regulate their emotions Berger (2014). On the other hand, positive emotion high arousal emotions (e.g., awe, excitement, or amusement) are shared more than positive low arousal (e.g., contentment). Furthermore, sharing for the purpose of rehearsal (i.e., people talk to relive positive experiences) helps people regulate their emotions .

Akpinar and Berger (2017) posits that valuable virality is virality contents that are beneficial to the brand.

• Informative appeals are benefit to brands, but less likely to be shared

• Emotional are more likely o be shared, but less beneficial to the brand

• Combining both into emotional-brand-integral ads boost sharing and brand-desirable outcomes.

• Research is based on Unruly data and lab experiments.

Operationalization via lanaguge: D. Yin, Bond, and Zhang (2017)

### 5.4.4 Usefulness

Wojnicki and Godes (2017) found that sharing useful content can help sharers appear more knowledgeable

Fehr, Kirchsteiger, and Riedl (1998) found that people share useful content to generate reciprocity

Homans (1958) posits that sharing useful content can have social exchange value.

### 5.4.5 Narratives

• Storytelling is important to branding.

• In the social media space, brand coc-reate brand pserofrmances with cosnuemrs.

• The process of improvisation is more critical than the output

• Managing brands on social media is about keeping the brand performance alive

## 5.5 Other variables

### 5.5.1 Popularity

• Difference between views (popularity) and shares (virality)

• Popularity usually contaminated by promotion (advertising)

### 5.5.2 Contractuality

defined contractuality as ” the extend to which a perk is perceived to be conditional on specific behaviors and contingencies dictated by a company.”

• Consumers like to have perks with less salient contractuality.
• The existing relationship with the company (e.g., dislike, distrusted) may giving consumers the wrong intention, or let them interpret the park as a manipulative act with ulterior motives.
• Low contractuality perks can foster “real” authentic WOM, but you have to trade off with effectiveness and efficiency (e.g., as in the case of filling out satisfaction survey).

### 5.5.3 Locus of Control

• An individual personality construct is locus of control

• Locus of control is defined as “individuals’ general and daily expectancies about the causes of their reward and punishment , consisting of internal (i.e., one can control his or her lives) and external (i.e., external factors can control their lives such as luck, or fate).

• Internals tended to be more educated , higher household incomes , men and those in senior positions . Internals are more action-oriented, risk-taking.

• Externals perform avoidance behavior, greater needs for affiliation

• Individuals with high internal locus of control are more likely to engage in WOM communication with their out-groups.

• Individuals with high external locus of control are more likely to engage in WOM with their in-group.

### 5.5.4 Horizontal/Vertical Individualism

• Antecedents of WOM: homophily, tie strength, trust.

• Culture affects individuals’ communication attitudes and styles.

• Vertical (i.e., emphasizing hierarchy) and horizontal (i.e., emphasizing equality) dimensions of individualism and collectivism (HVIC)

• opinion leadership: “the tendency that an individual attempts to influence the decisions of others by giving his or her opinion about products, services, or firms”

• socially appropriate (horizontal orientation)

• show off knowledge (vertical orientation)

• opinion seeking “the tendency that an individual seeks information or opinions from more knowledgeable others to find out about and/or evaluate products or services”

• well-informed decisions (individualism)

• find values and beliefs of the reference group (collectivism)

• Horizontal individualism is “a cultural pattern where an autonomous self is postulated but the individual is more or less equal in status with others”

• Vertical individualism is “a cultural pattern in which an autonomous self is postulated, but individuals see each other as different, and inequality is expected” .

• Horizontal collectivism is “a cultural pattern in which the individual sees the self as an aspect of an in-group” .

• Vertical collectivism is “a cultural pattern in which the individual sees the self as an aspect of an in-group, but the members of the in-group are different from each other, some having more status than others” .

• Collectivist culture will have greater levels of opinion seeking, whereas individualist culture has greater information-giving.

• Results:

• “horizontal individualism to opinion leadership, vertical individualism to opinion leadership and opinion seeking”

• “horizontal collectivism to opinion leadership, and vertical collectivism to opinion leadership and opinion seeking”

### 5.5.5 Linguistic style

• Moore (2012) found how different explaining language can influence the storytellers. Linguistic content in nontraditional WOM (e.g., online reviews, or other online channels) can influence the storyteller.
• language abstraction
• explaining language
• figurative language
• markers of modesty or boastful
• dispreferred markers
• personal pronouns
• linguistic mimicry

## 5.6 Negative Virality

Herhausen et al. (2019) offer a framework for drivers of negative eWOM with strategies that usually employ by firms to militate negative eWOM. Level of arousal, structural tie strength, and linguistic style match can all affect the firestorm. Optimal strategy: the response must match the intensity of arousal in the negative eWOM

## 5.7 Articles

• Empirical evidence

• contagion in consumer packaged goods
• Interesting stat:

• “Articles by business academics, psychologists, and economists, for example, are more likely to be shared than articles by physicists, geneticists, and biochemists”

• “Specifically, men see the same scientific summaries as more comprehensible (P < 0.001), interesting (P < 0.001), and useful” hence they share more than women

• “though intentionally outrageous videos command attention (Tellis 2004), an ad design of this type ultimately detracts from the ad’s persuasiveness” (Tellis, 2004)

• the impact of dispersion declines over time; hence should measure WOM early in a product’s life. Main finding: higher WOM dispersion leads to higher future ratings. WOM is both precursors and consequences of consumer behavior.

• Promotional giveaways increase WOM

Berger and Milkman (2012)

• Positive emotional valence content is more likely to be shared

• Activation = physiological arousal induces action.

• Low arousal = deactivation = relaxation

• HIgh arousal = activation = activity

1. Examine 7000 articles from thew New York Times

1. Examine emotionality, prominent features, interest evoked can affect likelihood to make the most email list. (controlling for practically content usefulness, interestingness, surprise, release timing and author fame (using hits for first author’s full name from the number of Google hits), writing complexity, author gender, article length and day dummies).

2. Robustness: control for article’s general topic.

2. Lab experiments

1. Amusement case (fictitious): high arousal

2. Anger case (real): high arousal

3. Sadness (real vs. fictitious): low arousal

All hypotheses are confirmed

Potential confounders: structural virality

How likely they would share a story? (no social risk involved - risk to other weak ties, hence the effect might be inflated, the same thing with the New York Times study )

All experiments have low participant numbers

Tellis et al. (2019)

Two field studies

Information-focused content is less likely to be shared (exception risky contexts)

Positive emotions (e.g., amusement, excitement, inspiration, warmth) are more likely to be shared

Drama elements (e.g., surprise, plot, characters, babies, animals, celebrities) increase arousal, which in turn increases sharing.

Prominent placement of brand name (brand prominence)

Emotional ads are shared more on general platforms (Facebook, Twitter) as compared to professional one (e.g., LinkedIn), while informational ads are more likely to be shared on professional ones.

Optimal length is 1.2 to 1.7 min ads.

Third study: identifies predictors of sharing

• How to determine Influential users in online social networks