iBranding The Impact of Social Media on Corporate Brands
David Eberle | 336553 | September 2010
Global Business & Stakeholder Management | RSM Erasmus University
Coach: Mr. Guido Berens | Co-Reader: Ms. Ting Li
IBRANDING: THE IMPACT OF SOCIAL MEDIA ON CORPORATE BRANDS
Preface The author declares that the text and work presented in this Master thesis is original and that no other sources other than those mentioned in the text and its references have been used in creating the Master thesis. The copyright of the Master thesis rests with the author. The author is responsible for its contents. RSM Erasmus University is only responsible for the educational coaching and beyond that cannot be held responsible for the content.
Acknowledgements I wish to thank my thesis coach, Guido Berens, for his time, his guidance and his strong commitment to support my research. Without his help, this project would not have gotten as far as it is now. I also would like to express my gratitude to my co-reader, Ting Li, whose knowledge and suggestions led the research into a clearer direction. I am also very thankful for the support of my girlfriend Marisol and my friends Lorenzo, Pablo, John, Hendra and Anil, who supported my study and who listened to my thoughts, questions and ideas during the entire process. Without such people, I would have been less persistent and the fun of doing research would have been taken away. Last but not least, I am grateful for the time of the 205 participants of my study.
1 | David Eberle | 2010, RSM Ersasmus University IBRANDING: THE IMPACT OF SOCIAL MEDIA ON CORPORATE BRANDS
iBranding: The Impact of Social Media on Corporate Brands
As spending on traditional advertising is slipping, companies display a rising interest in social media and how corporate brands are affected by interactive online communication. In this paper, the author tries to determine whether using social media in a company’s marketing mix is beneficial to its brand. Drawing on theories of corporate communications, marketing, sociology and psychology, the author proposes a novel social media branding model, where the relationship between corporate reputation and interactivity and pre-exposure word-of-mouth is mediated by credibility and embeddedness. The author finds that although perceived interactivity leads to higher credibility and embeddedness, which, in turn, boost corporate reputation and post-exposure word-of-mouth, the existence of negative comments are harmful to the brand.
Keywords: online corporate branding, message credibility, website interactivity, social media, electronic word-of-mouth, internet marketing, web 2.0 marketing
DAVID EBERLE 2010, RSM Erasmus University
1. Introduction Together with the rise of online communication channels spending on “The advent of the Internet has created traditional advertising is plummeting (The new avenues for reaching people,” Economist, 2010). Since customers can proclaimed Charles E. Brymer, chairman interact more with each other their market of Brandchannel, an advertising power has magnified. A survey of The consultancy, in 2001 (The New York Economist reveals that customers finally Times, 2001). His statement has been “are kings” due to the possibility to access proven right. Barack Obama’s marketing desired information at their fingertips (The campaign heavily relied on Internet Economist, 2005). channels, especially online social This changing market environment networks, to stimulate funding (TIME, certainly has an impact on the business 2007). Companies, too, are embracing strategy of many companies. Since social media channels to reach their traditional marketing strategies are customers. Coca-Cola’s Facebook page, unlikely to work in the future, businesses for example, boasts more than 5m fans. need to develop new methods about how
2 | David Eberle | 2010, RSM Ersasmus University IBRANDING: THE IMPACT OF SOCIAL MEDIA ON CORPORATE BRANDS
to effectively communicate and reach 2. Social Media existing and potential customers. The term ‘social media’, sometimes also Many firms are currently experimenting referred to as ‘consumer-generated media’ with various innovative strategies. In April (CGM), “describes a variety of new and 2008 Nike launched a video clip showing emerging sources of online information the basketball celebrity Kobe Bryant that are created, initiated, circulated and performing a stunt in Nike sneakers. The used by consumers intent on educating video has been watched nearly 5m times each other about products, brands, on Youtube and probably many times more services, personalities and issues” through other websites. This clip literally (Blackshaw & Nazzaro, 2006). Social went ‘viral’. media is an umbrella term for different Yet, it is of utter importance to firms to types of online communication channels. know how such online campaigns The most prominent forms can be divided ultimately affect corporate brands. Some into four categories2 (Constantinides & Youtube users believe that Kobe Bryant’s Fountain, 2007; Kaplan & Haenlein, stunt is a “fake”1. Many of the other 2010): thousands of comments express a similar opinion. Consequently, the effect of this § Collaborative projects (e.g. Wikipedia, ‘viral’ video on Nike’s reputation is comparison shopping sites) unclear and the company probably would § Blogs like to have a better knowledge on whether § Content communities (e.g. Youtube) using such media add value to their § Social networking sites (e.g. Facebook) communication strategy. Since little literature exists on this topic as Social media allow consumers to share the coming sections show, it is the aim of their opinions, criticisms and suggestions this thesis to close gaps in the knowledge in public. On Amazon, for instance, about social media and corporate brands. customers can write reviews and rate The next chapter introduces the term social products. Another example is Pricewatch, media and is followed by a chapter which lists the cheapest seller of any reviewing and categorizing existing electronic good. Such services increase research about online corporate branding. market transparency, which consequently The fourth chapter defines the research magnifies the bargaining power of objectives and also shows the relevance of customers (Constantinides & Fountain, such a study. The theoretical framework 2007; Ind & Riondion, 2001). Companies, that forms the basis of this study is in turn, lose part of their dominance over elaborated on in chapter five. Chapter six information flows (Bunting & Lipski, explains the research design and chapter 2000) and are thus less able to exert seven presents the outcome of the pilot control over the communication among experiment. The results of the main study consumers (Mangold & Faulds, 2009). are discussed in chapter eight. This is These online channels can, in contrast, followed by a summary, a discussion also be useful to businesses. They offer about the results, suggestions for further new ways of talking to customers through research, as well as managerial and company-sponsored content (e.g. online theoretical implications. advertisements or customer feedback pages). However, such content is less
2 Kaplan and Haenlein (2010) suggest two 1 Youtube.com. (9. April 2008). Laker Kobe Bryant additional categories, ʻvirtual game worldsʼ and attempts massive stunt...and succeeds! Real? ʻvirtual social worldsʼ. Both categories, however, Retrieved 26. March 2010 from Youtube: attract a highly specific audience and are therefore http://www.youtube.com/watch?v=yURa9T0-Rjk not of relevance to this thesis.
3 | David Eberle | 2010, RSM Erasmus University IBRANDING: THE IMPACT OF SOCIAL MEDIA ON CORPORATE BRANDS likely to be as effective as truly ‘viral’ (2009) introduce a new paradigm where content, i.e. messages transmitted via social media is “a hybrid element of the word-of-mouth (Bunting & Lipski, 2000). promotion mix in that it combines some of A message that is sent by another the characteristics of traditional IMC tools consumer is perceived as more authentic with a highly magnified form of word-of- than company-sponsored content (Parise & mouth communications in which Guinan, 2008). marketing managers cannot control the These facts put companies in a dilemma. content and frequency of such information On the one hand, firms would like to exert (Mangold & Faulds, 2009). control over what information about their Constantinides and Fountain (2008) brands is being published online. On the suggest that, in addition to traditional other hand, meddling in social media marketing stimuli, ‘web 2.0 experience’ might even exacerbate the prevailing also influences consumer decisions. Such opinion about the brand. online marketing factors are, in contrast to traditional online channels, uncontrollable and are subject to the input of other 3. Online Corporate Branding consumers. Consumers thus influence each other’s decisions through the use of social Various articles have been published about media channels. online corporate branding, Internet These findings outline a general agreement marketing, social media in corporate on the importance of social media in communications, and online word-of- corporate communications. Yet, there exist mouth. These papers can roughly be no broadly accepted theory. Researchers divided into three categories. First, theory each introduce their own concepts and the building research; second, empirical absence of a common theoretical base studies; and third, papers addressing complicates an accepted notion of the managerial implications and advisory. impact of social media on corporate brands
and reputation. 3.1. Theoretical Research
Literature generally agrees that the 3.2. Empirical Research traditional paradigm of marketing Empirical evidence on the relation communications has lost its validity in between social media and branding is today’s marketplace. Buntin and Lipski rather precarious as well. Most of these (2000) suggest a weaker relationship studies exclusively focus on online between corporate communications and companies. reputation, where power is shifted away The probably largest study in this field has from corporations towards consumers. been conducted by Alwi and Da Silva Yet, they assure that “[PR professionals] (2007). The authors compare the should welcome the emergence of the ‘corporate character’ of two bookstores in Internet as a valuable new the United Kingdom. They conclude that communications and reputation-building “[the] corporate brand image may vary tool” (Bunting & Lipski, 2000). depending on which context it operates in Ind and Riondino (2001) propose the (online or offline),” but that a addition of ‘unplanned communications’ generalization of the results to other channels into an updated version of industries is not possible (Alwi & Da Stuart’s brand management model (Stuart, Silva, 2007). In another article these same 1999). Such channels influence the authors find that the trait approach of corporate reputation and are “[an] human personality developed by Davies et important phenomenon” (Ind & Riondion, al. (2004) is applicable in a strictly online 2001). Similiary, Mangold and Faulds
4 | David Eberle | 2010, RSM Erasmus University IBRANDING: THE IMPACT OF SOCIAL MEDIA ON CORPORATE BRANDS setting and suggest further research using limited, the magnitude of ‘viral’ marketing their approach. is recognized. An analysis of the effects of A framework for “dimensionalizing, word-of-mouth (WOM) at Friendster, a measuring and predicting etail quality” is social networking site, shows that “WOM described by Wolfinbarger and Gilly referrals have a strong impact on new (2003). They find that the quality of an customer acquisition” (Trusov et al., online retailer is determined by reliability, 2009). The authors add that WOM website design, privacy and security, as stimulated by companies, however, is well as customer service (Wolfinbarger & probably substantially less effective than Gillin, 2003). Similarly, a study among true referrals (Trusov et al., 2009). 312 undergraduate students demonstrates that online purchase intention is 3.3. Practical Research significantly influenced by the online A third type of literature focuses on store’s usefulness, enjoyment, practical implications of social media and trustworthiness and the settlement managerial advice. Chen (2001) states performance (Van der Heijden & “online communities are…a double-edged Verhagen, 2004). A study conducted in sword,” since less well-known brands can Australia proposes a new structural model quickly become popular and pose a threat of ‘e-tailer’ branding where the brand is to more established brands. Rubinstein and influenced by navigability, trust, fun and Griffiths (2001) thus emphasize the interactivity (Merrilees & Fry, 2002). Da importance of online branding; yet, they Silva and Alwi (2008) propose that online agree that the “customer is seizing corporate brand images are determined by control”. the ease of use, security, personalization, Therefore, companies need to establish an customer care, and reliability of corporate emotional bond with their customers websites. (Clauser, 2001). Due to its interactive Stuart and Jones (2004) stress the technology the Internet is “ideal for importance of distinguishing between the allowing businesses to create an emotional use of the Internet as a communication connection between their consumer and tool or as a business. Most of the studies their brand” (Clauser, 2001). Online cited above analyze online retailers and communities are of particular interest to thus fall into the second category. marketers; however, due to the openness Studies examining the effectiveness of of these networks, ‘undesirable Internet marketing are often technically participants’ pose a threat to a company’s outdated and are thus only of partial use to brand. managerial decision-making. Phelps et al., In order to mitigate negative impacts on for instance, examine how consumers brands Catalano (2007) proposes certain respond to and pass along email (Phelps et guidelines and a monitoring-and-analysis al., 2004). Today, however, attention has program for companies to capitalize on shifted to truly social media rather than blogs. Blogging in general can be a highly one-way communication methods. effective tool for public relations efforts. Porter and Golan (2006) compare the Kent (2008) points out that “blogs can content of television commercials with provide an entrée for organizations into ‘viral’ advertisements and do find previously ‘private’ realms” and that they differences regarding the advertising thus allow a firm to “reach the choir”. appeals (e.g. sex appeals). Yet, the authors According to Rogers (2003), reaching the do not investigate the effectiveness of the choir helps when companies seek to two different marketing forms. communicate with opinion leaders, Although empirical evidence of the effects innovators, and early adopters. of social media on corporate brands is
5 | David Eberle | 2010, RSM Erasmus University IBRANDING: THE IMPACT OF SOCIAL MEDIA ON CORPORATE BRANDS
Stuart and Jones (2004) assert “the ability regarded as an umbrella term that of the internet to build communities needs incorporates the features of corporate to be embraced by organizations rather reputation (Knox & Bickerton, 2003). than feared”. The authors urge firms to Existing theoretical, empirical, and ensure consistency between the corporate practical studies outline a link between brand and online communications (Stuart social media and corporate branding. & Jones, 2004). Today companies cannot neglect that Managers, however, could benefit from online tools are becoming an increasingly more concrete advice on the usefulness of powerful tool in managing the corporate social media in the corporate brand. However, past research has not communications mix. The ‘social media clearly identified the relationship between metrics definitions’ published by the social media and corporate brands and Interactive Advertising Bureau (IAB) are reputation. The knowledge about social supposed to fill this gap. These media and corporate branding thus needs benchmarks allow measuring the success to advance in order to provide both of social media (e.g. a company’s academia and management with a better Facebook page) using different understanding of the topic. measurement dimensions such as unique The purpose of this thesis is to determine visitors, page views and time spent (IAB, whether the use of social media in a 2009). Yet, these criteria completely omit company’s online communication mix qualitative aspects of social media (Fisher, positively influences the corporate brand. 2009) and should thus used with caution. An online communication mix contains all For brand managers it is ultimately corporate messages distributed through interesting to know whether views of, for online channels, such as advertising, press instance, a video on Youtube changes the releases, and product information. consumer’s perception of the corporate Conducting research on this subject is of brand. value to academia and stakeholders in multiple ways. According to Van Tulder and Van der Zwart (2006), stakeholders 4. Research Objectives can be grouped into the spheres government, market, and civil society. Corporate branding is an essential part of With social networking sites alone an organization’s communication strategy. boasting over 300m active users (Nielsen Argenti and Druckenmiller (2004) outline Wire, 2010), social media is gaining the importance of corporate branding, importance for all three stakeholder which “has come into the business spheres. Today people spend more than spotlight in recent years”. A firm’s 5.5 hours per month networking and financial performance is determined by twittering, an increase of 157% over the both the company’s business activities as past two years (Nielsen Wire, 2010). well as its communication efforts, which Social media not only seizes more time in turn determine the corporate reputation from consumers but also is the place (Van Riel & Fombrun, 2007). Corporate where surfers are exposed to reputation can be defined as “subjective, advertisements. According to comScore, a collective assessments of the research firm, 21.1% of all online ads are trustworthiness and reliability of firms [by displayed on social networking sites their stakeholders]” (Fombrun & Van Riel, (comScore, 2009). 1997). The term corporate branding A report published by Netpop, a think- additionally considers whether the promise tank, argues that using the Internet for made by the organization is actually information and communication is delivered (Balmer, 2001), and can thus be becoming more important to consumers
6 | David Eberle | 2010, RSM Erasmus University IBRANDING: THE IMPACT OF SOCIAL MEDIA ON CORPORATE BRANDS
(Netpop, 2009). Since users rely more This thesis also adds value to academic heavily on information available online theory. As previous research has not and use social media more frequently to focused on a direct link between social access information, the stake of civil media and corporate reputation, this study society in social media seems evident. offers relevant insight for corporate The market is equally tangled with online communications theorists. A further aim of communication channels. A study among this paper is to provide academia with a 297 communication professionals finds starting point for future research regarding that of the organizations questioned “78% this field. use blogs, 63% use online video, 56% use social networks and 49% use podcasts in their organization’s communications 5. Conceptual Model initiatives” (Carrabis et al., 2008). The central assertion of this thesis is that Moreover, a survey by McKinsey reveals the use of social media in a company’s that at least 37% of the 2,847 managers communication mix positively influences questioned use or plan to use social the corporate brand. This paper suggests networking sites in their public relations that ‘credibility’ and ‘embeddedness’ (McKinsey, 2007). Follow-up research mediate the influence of social media on discloses that of 956 respondents 52% the corporate brand. think that new web technologies “increase A major attribute of social media channels the effectiveness of marketing” is that it allows consumers to interact with (McKinsey, 2009). companies and with each other beyond the Although a significant share of companies firm’s control (Thorson & Rodgers, 2006). reports measurable benefits, the mediators There is a significantly strong connection why ‘it works’ are largely unclear. The between perceived interactivity and fact that many managers also find no attitude towards websites (Jee & Lee, benefits when using certain web 2002; Thorson & Rodgers, 2006). technologies (McKinsey, 2009) shows that Consequently, this thesis argues that the a lot of firms are still experimenting with use of social media in company initiated social media because established know- corporate communication is beneficial in how is not available yet. regards to the corporate brand. This Managers largely rely on quantitative statement is based on the findings of measures (the importance of search engine various scholars that interactive media ranking is mentioned first) (Carrabis et al., increases message credibility and feelings 2008); yet, such benchmarks are heavily of embeddedness, which, in turn, exert a criticized. Since, at the same time, firms positive impact on corporate brands. using social media seek to enhance relationships with stakeholders and to 5.1. Embeddedness improve the corporate reputation (Carrabis et al., 2008), this thesis is of particular People actively chose organizations they importance to management and public can identify with, even if they are not relations decision-making. formal members (Bhattacharya & Sen, Also the government has an interest in a 2003). The authors refer to ‘embedded better understanding of social media and relationships’ between customers and reputation. In the United Kingdom, for companies. These relationships are likely example, the Labour Party recently to be “strong, intricate, and trusting, launched a large-scale social media resulting in consumers feeling more like campaign to boost electoral support insiders than outsiders” (Bhattacharya & (Gordon MacMillan, 2010). Sen, 2003). Such feelings of embeddedness may occur through the use
7 | David Eberle | 2010, RSM Erasmus University IBRANDING: THE IMPACT OF SOCIAL MEDIA ON CORPORATE BRANDS of “company-sponsored forums”, the channels is more credible than company- interaction with other company controlled messages. stakeholders through “on- and offline In a study using low and high interactive communities”, and the involvement of versions of a political candidate’s website consumers into the corporate decision Thorson and Rodgers (2006) find that making process (Bhattacharya & Sen, websites with a high level of perceived 2003). interactivity positively influence the Consumers that interact with companies attitude towards the website, impressions feel closer to those organizations and get a of the candidate as well as voting feeling of embeddedness. Thorson and intentions. A major mediating factor Rodgers (2006) confirm that the ability to between perceived interactivity and interact induces consumers to engage in attitude is trust (Thorson & Rodgers, positive word-of-mouth and fosters the 2006), which in turn is a part of credibility creation of an intimate relationship (Newell & Goldsmith, 2001). between organization and customer, i.e. These findings suggest that social media causing feelings of embeddedness. channels are more credible than traditional Because social media channels are highly marketing channels. Since the credibility interactive, they serve as an ideal of online channels matter in regards of candidate for a platform that promotes message credibility, this thesis proposes such consumer identification. This paper the following hypothesis. thus introduces the following hypothesis. H1b: Corporate communication messages H1a: Corporate communication messages appearing on social media channels are appearing on social media channels evoke more credible than messages appearing on higher feelings of embeddedness than traditional online channels. messages appearing on traditional online channels. 5.3. Impact on Corporate Brands Credibility is an important aspect of 5.2. Credibility corporate reputation (Fombrun, 1996). Scholarly research suggests that Various research show that a higher interactivity in media may also increase corporate credibility enhances the the credibility of the message (Beninger, customer’s attitude towards 1987). Bickart and Schindler (2001) advertisements and brands, as well as suggest that information presented in purchase intentions (Goldsmith et al., social media has a higher credibility than 2000; Goldsmith & Lafferty, 1999; marketer-generated information. Gruen et Shamdasani et al., 2001). al. (2006) find a significant correlation The credibility of advertising messages between the knowledge exchange between seems to have a positive influence on the consumers and the intentions to engage in attitude towards the ad as well as the brand word-of-mouth. Electronic word-of- (Choi & Rifon, 2002; MacKenzie & Lutz, mouth, in turn, is more credible if found 1989). This is due to the consumer’s on an interactive website, where judgment of advertisements and brands consumers exchange information and based on whether the source is deemed to opinions. be trustworthy. MacKenzie and Lutz Social media allows consumers to publish (1989) cite a study3 saying that credibility criticism on the presented facts, thereby can act as a ‘peripheral cue’, which causes quickly exposing fraudulent claims. Therefore, information about or from a 3 Petty, R., & Cacioppo, J. (1981). Attitudes and company published through social media Persuasion: Classic and Contemporary Approaches. Dubuque: AC Brown.
8 | David Eberle | 2010, RSM Erasmus University IBRANDING: THE IMPACT OF SOCIAL MEDIA ON CORPORATE BRANDS higher persuasion even when the recipient the brand”, and which ultimately is not actively processing the message reinforces brand equity. Consequently, the content. following hypothesis is proposed. Consequently, there exists a two-way relationship between credibility and H2b: Feelings of embeddedness among brands. On the one hand, corporate consumers positively influences corporate credibility, which results from a brands. company’s reputation, increases the attitude towards the message. Message credibility, on the other hand, positively 5.4. User Interaction influences the attitude towards the brand Social media channels typically display and the company in general. not only the original message written by The latter construct is of special interest to the author of the page but also comments this research since it has been identified as by various visitors. Such opinions can be a possible mediator between social media very critical about the statement made on and corporate brands. Yet, the findings that particular page and sometimes also presented above suggest that both include ratings. Especially product related relationships are intertwined and that they information is often complemented with influence each other. This study thus user ratings to give potential customers a focuses on companies with no initial more complete picture of the marketplace. reputation; the section ‘research design’ On Apple’s online store, for instance, introduces a fictitious company, where customers can review and rate any item on consumers impossibly possess any prior sale; the ‘magic mouse’ has been rated opinion. 1812 times and has an average of four This thesis proposes the following stars4. hypothesis. In literature there is a “general agreement that critics play a role” (Basuroy et al., H2a: Message credibility has a positive 2003). Cameron (1995) likens criticism to influence on the corporate brand. advertising since it “provides information that can be used by consumers in forming Also consumers’ feelings of the hedonic price of cultural demand”. embeddedness play a vital role in Obermiller and Spangenberg (2000) find corporate branding, since the creation of that consumers are less skeptical toward so-called ‘brand communities’ can have a information presented in consumer reports large impact on brand equity (Kotler & or told by a friend than advertisements or Keller, 2008). Such communities occur sales persons. also online (Muniz & O'Guinn, 2001), With corporate communication messages, particularly through social media. As however, the situation looks slightly consumers interact with each other and the different. Rating a corporate responsibility company, they feel closer to the report, for example, is first less straight organization and the brand (Bhattacharya forward and second less meaningful than & Sen, 2003). Various research suggest handing out stars to a product, where the that such proximity is beneficial to brands reviewer can be more concrete about the (Kotler & Keller, 2008; Muniz & pros and cons. O’Guinn, 2001). According to Kotler and Nevertheless, as companies publish their Keller (2008), getting consumers more communication messages on social media actively involved with a brand increases
‘brand resonance’, which is characterized 4 Apple. (2010). Apple Magic Mouse. Retrieved 11. by “the intensity or depth of the April 2010 from Apple Store (U.S.): psychological bond customers have with http://store.apple.com/us/product/MB829LL/A?fnod e=MTY1NDA1Mg&mco=MTc0MjYxMDA
9 | David Eberle | 2010, RSM Erasmus University IBRANDING: THE IMPACT OF SOCIAL MEDIA ON CORPORATE BRANDS channels, readers often make use of the positive or negative) does not have a possibility to state their opinion about the significant impact on consumer attitudes. company in general or that message in Duan et al. (2008) reach a similar particular. General Electric’s wind energy conclusion and suggest that online reviews commercial5 on Youtube has been viewed increase consumer awareness instead of more than 150,000 times and features directly influencing public opinion. They nearly 200 comments. A six-minute find that the rating of reviews has “no speech by Néstle’s CEO Peter Brabeck persuasive effect on consumer purchase about genetically modified food6 has decisions” (Duan et al., 2008). attracted about 14,000 viewers and caused Likewise, a recent study reveals that while close to 80 opinion statements. negative reviews hurt the book sales of User interaction clearly plays an important well-known authors, they increase the role when companies communicate purchase likelihood of previously through social media channels. unknown books (Berger et al., Consequently, not only the consumers’ forthcoming). The authors also attribute reaction to the corporate message but also these findings to an awareness effect, i.e. their response to other consumers’ opinion that any kind of word-of-mouth increases needs to be taken into account when product awareness. determining the influence of social media Basuroy et al. (2003), however, find that on corporate brands. during the first week of a film’s run Yet, not all comments are positive and negative publicity hurt sales more than some executives wonder whether allowing positive reviews increase revenues. They the public to read previous criticism does pin their findings on the existence of not harm the corporate brand. Mr. Dzubay, influencers, a person who is “regarded by publisher of the Hudson Star-Observer, a group or by other people as a person wondered whether negative comments put having expertise or knowledge in a the credibility of his paper at risk (Post, particular subject” (Basuroy et al., 2003). 2007). As a consequence, his paper took This thesis aims at investigating to what the option to comment down in 2007; in degree allowing consumers to read the meantime, however, commenting on comments of other consumers is beneficial the Star-Observer was made possible to the corporate brand. Clearly, largely again. positive comments are ought to enhance Scholars, too, find mixed results the brand image. Of special interest, concerning the impact of negative reviews however, is whether publicity with a on brands and sales. Liu (2006) discovers substantial amount of negative comments that while the volume of word-of-mouth is superior to no publicity at all. Previous significantly increases box office findings suggest that this could be the revenues, there is no considerable case, especially if the company is difference between positive or negative previously unknown (Berger et al., reviews. The author argues that while forthcoming). consumer awareness is determined by the This paper suggests that pre-exposure volume of word-of-mouth, the valence of word-of-mouth (preWOM), i.e. comments word-of-mouth (i.e. whether a review is that already exist when the consumer sees the corporate message, influences the corporate brand through credibility. Smith 5 Youtube. (17. August 2007). GE General Electric and Vogt (1995) discovered a link between commercial - Wind Energy. Retrieved 11. April 2010 from Youtube: word-of-mouth and the perceived http://www.youtube.com/watch?v=fViObqGvIjM credibility of advertisement. Their study, 6 Youtube. (24. March 2008b). Nestlé CEO Peter however, only tests the impact of purely Brabeck. Retrieved 11. April 2010 from Youtube: http://www.youtube.com/watch?v=qyAzxmN2s0w
10 | David Eberle | 2010, RSM Erasmus University IBRANDING: THE IMPACT OF SOCIAL MEDIA ON CORPORATE BRANDS negative word-of-mouth and says thus published by a consumer after having seen little about mixed settings. the message, is largely created by East et al. (2008) find that positive word- consumers who feel involved with the of-mouth has a stronger impact on brand organization or community. purchase probability than negative Hennig-Thurau and Walsh (2003) indicate publicity. They argue that the general that consumers engage in word-of-mouth understanding of marketers that negative because it causes feelings of community word-of-mouth has a higher impact than membership. Likewise, Wang and positive word-of-mouth is mistaken (East Fesenmaier (2003) identify involvement as et al., 2008). A mixed setting with a the main motivational factor of online similar amount of positive and negative community participation. comments could therefore still be PostWOM, consequently, is also beneficial to a company’s brand. determined by the degree of Similarly, Doh and Hwang (2009) notice embeddedness, which includes feelings of that the credibility of electronic word-of- closeness and inclusiveness. This paper mouth was the highest when there were thus proposes the following hypothesis: also some negative comments about a product. H4: Feelings of embeddedness foster the In summary, word-of-mouth in general creation of post-exposure word-of-mouth. boosts consumer awareness of brands. While positive comments clearly increase the attitude towards the brand, the strength 5.5. Social Media Branding Model of the impact of negative word-of-mouth is The theoretical frameworks presented in likely to be less than for positive this section can be integrated into the comments. A setting where both positive ‘social media branding model’. The use of and negative comments exist thus social media channels in corporate probably leads to a more positive brand communication increases message attitude and a higher degree of credibility credibility and evokes feelings of than a setting without any word-of-mouth. embeddedness among customers. These two effects, in turn, have a positive impact H3: Pre-exposure word-of-mouth, even on the corporate brand. Furthermore, when partially negative, has a positive allowing comments to be visible to other impact on the credibility of the message. customers (pre-exposure word-of-mouth) increases the credibility of the message. In social media channels comments and Lastly, feelings of embeddedness foster reviews are not simply existent but are the creation of post-exposure word-of- actively published by consumers. This mouth. The entire model is presented in thesis also aims at identifying the impact Figure 1. of social media on the generation of new word-of-mouth. Post-exposure word-of- mouth (postWOM), i.e. comments
11 | David Eberle | 2010, RSM Erasmus University IBRANDING: THE IMPACT OF SOCIAL MEDIA ON CORPORATE BRANDS
Figure 1: The ‘social media branding model’
6. Research Design information on which they are later asked to judge the company regarding its The study of this thesis was planned to credibility, embeddedness, reputation, and examine the relationships between the whether they would engage in (post- different elements of the previously exposure) word-of-mouth. developed ‘social media branding model’.
The experiment should reflect a 6.1. Scale Development naturalistic exposure setting in order to measure truthful consumer perceptions. The first variable, credibility, is measured Since the participants of the study should by using Newell and Goldsmith’s be unfamiliar with the material used in the corporate credibility scale (Newell & experiment, a fictitious company named Goldsmith, 2001). Based on past research HappyBev was created to mimic a real and corporate data of IBM and Exxon, the company as much as possible. two authors developed a scale consisting The company has a logo and a website, of two dimensions, trustworthiness and which presents a brief history of the firm, expertise, of which each contains four an introduction to its product, bottled individual items. water called ‘Aqua Montoé’, as well as a To quantify embeddedness this paper text about its water management practices makes use of the connection and self- to the reader (see Appendix B). While the categorization scale developed by first paragraphs have been written by the Einwiller et al. (2006). author for the purpose of this study, the The third variable, reputation, is story about its water management was determined by using the Reputation taken from Nestlé Waters’ website7. The Quotient introduced by Fombrun et al. story claims that the company treats its (2001). This scale consists of 20 items; yet water sources in a sustainable and since not all of them are applicable to the environmentally friendly way. These survey material, some adaptations were paragraphs provide the readers with necessary. Out of the six dimensions only ‘emotional appeal’ and ‘social and environmental responsibility’ were 7 http://www.nestle-waters.com/environment/water- selected and latter was adapted further to care/local-sustainable-water-management.html
12 | David Eberle | 2010, RSM Erasmus University IBRANDING: THE IMPACT OF SOCIAL MEDIA ON CORPORATE BRANDS match with HappyBev’s water scenario four has a ratio of 5:5 while management story. scenario five has a ratio of 2:8 of positive Finally, to compute post-exposure word- to negative comments (see Figure 2). of-mouth two questions were taken from a scale developed by Maxham (2001) while the author of this thesis created two additional questions specifically for this study. These four question compounds are followed by a series of manipulation check and control variables. See Appendix A for Figure 2: Scenario overview a detailed overview of the scales before and after the alterations including the Since Doh and Hwang (2009) only tested additional questions. positive allocations with a ratio of 6:4 or above, the settings examined in this study 6.2. Manipulating Interactivity and are of special interest for this field of Word-of-Mouth research, since negative publicity online Interactivity is manipulated by creating can exert a lot of damage to businesses two versions of HappyBev’s website, one (eReleases, 2009). with a low and one with a high degree of Similar to Doh and Hwang’s study, the interactivity. Similarly to Thorson and comments were gathered from existing Rodger’s (2006) study, the first version social media, such as Blogs or Youtube. In displays only HappyBev’s corporate order to match with HappyBev’s material message without any possibility to interact the comments have been slightly adapted. with the website, while the second version provides the visitor with the possibility to write a comment about the corporate 7. Pilot Results message. The website makes the visitor The effectiveness of the manipulation of believe that his message will be published interactivity and preWOM is crucial to within the following 24 hours on the main obtain valid results. A qualitative pilot webpage. study with five Master students of the The second version of the website is Rotterdam School of Management further split up into four different groups revealed that there is a significant in order to manipulate (pre-exposure) difference between scenario one and the word-of-mouth. In addition to the two other scenarios in regards to the perceived versions described above, scenarios three, interactivity of the website. The average four and five each display ten existing difference of the perceived interactivity comments, while scenarios one and two do was 5 on a scale from 1 to 10. not show any comments. Since this paper Moreover, the tone of the comments was wants to examine the full impact of interpreted correctly: scenario three was preWOM scenarios with a different ratio perceived to have mostly positive of positive to negative comments are comments, while the comments of essential. scenario five were regarded as mainly According to Doh and Hwang (2009) who negative. Yet, two out of five participants analyzed the impact of online product did not actively read the comments reviews on product attitude, purchase because they were “not interested in what intent and credibility, a ratio of eight others had to say”. In order to capture this positive to two negative reviews yields the potentially interesting finding in the highest credibility. Similarly, scenario following study, the question how many of three also has a ratio of 8:2. In contrast,
13 | David Eberle | 2010, RSM Erasmus University IBRANDING: THE IMPACT OF SOCIAL MEDIA ON CORPORATE BRANDS the comments the respondents have read from the Rotterdam School of has been included. Management and the University of Zurich. After some improvements, HappyBev’s Seven participants do not possess a website and its corporate story were university degree, however. The chi- perceived as very credible by all square test shows that differences among participants. Also the comments were seen the five groups do exist (see Table 1); yet, as authentic, after the length of the since five cells have an expected count comments and the style of the avatars were less than ‘5’ the results might not be varied stronger. meaningful. Indeed, by excluding the respondents with a high school diploma, the chi-square test shows no more 8. Study Results disparities. Nevertheless, further analyses were Data was gathered through an online conducted both with and without these experiment. This study aimed at gathering seven samples to check whether there are at least 195 complete responses, which is significant differences between the two sufficient for measuring medium effect data sets. As no significant differences sizes in ANOVAs with five groups at could be found it is thus assumed that the α=0.05 (Cohen, 1992). Of 339 persons distribution of education among the five participating 205 complete samples could groups is consistent. be used (the others did not complete the survey). Pearsons Chi-Square The first step in evaluating the results is to (asymp. sig. 2-sided) check the consistency of the distribution of Gender 3.007, df=4 (0.557) participants’ gender, nationality, education Nationality 1.189, df=4 (0.880) and age as well as the control questions Education 3.061, df=4 (0.548) across the five groups. This is followed by [15.782, df=4 (0.046) an analysis of the reliability and factor using original data set] analyses of the scales used in the experiment. The third step involves Table 1: Chi-square tests of participant statistics examining the manipulations of interactivity and tone of comments. Lastly, The mean age of all participants is 24.98 the hypotheses are tested using regression years. An ANOVA test shows that there analyses in step four. are no significant differences in the mean age across the five groups (significance 8.1. Descriptive Statistics between the groups = 0.613). A total of 99 females and 106 males A total of four control questions were participated in the survey. A chi-square asked. On average, participants think of test reveals no significant differences of themselves as being very familiar with the gender distribution among the five groups. internet (6.62 on a 7-point Likert scale). Due to the high level of internationality Respondents like using social media a lot among the participants, nationality was (6.06 on a 7-point Likert scale). 97% of divided into ‘western’ and ‘non-western’ the survey respondents say to have a countries, using Huntington’s profile on a social networking site whereas classification (Huntington, 1997). There only 73% use social media to retrieve are no significant differences among the information about companies or products. five groups in regards to the total 117 The distribution of the answers is not western and 88 non-western respondents. significantly different across the five Most of the interviewees are either groups (see Table 2). Bachelor (103) or Master (95) students
14 | David Eberle | 2010, RSM Erasmus University IBRANDING: THE IMPACT OF SOCIAL MEDIA ON CORPORATE BRANDS
ANOVA F-Values To further examine the validity of the four (sig. between scales a factor analysis is conducted. The groups) ‘elbow bend’ of the scree plot of a I am familiar with 0.366, df=4, 199 principal component analysis reveals that using the Internet (0.833) I like using social 1.155, df=4, 196 there are indeed four different components media (0.332) across the items of all four scales (see I have a profile on a 0.562, df=4, 200 Figure 3). social networking site (0.691) I use social media to 1.148, df=4, 199 retrieve information (0.335) about companies or products
Table 2: ANOVA of control questions
8.2. Scale Validities The reliability of the four scales, credibility, embeddedness, reputation, and word-of-mouth, was analyzed using Cronbach’s Alpha. The two negatively Figure 3: Scree plot of principal component formulated questions in the credibility analysis scales (‘HappyBev does not have much experience’ and ‘I do not believe what The principal component analysis was HappyBev tells me’) were inverted before therefore run again with four factors and calculating Cronbach’s Alpha. rotated using the Varimax procedure. The All scales have an Alpha >0.80 and can rotated component matrix can be thus be used as a reliable scale (Garson, interpreted as component one being 2010) (see Table 3). Three items would reputation, component two being slightly improve the Alpha of their scale if embeddedness, component three being removed. ‘HappyBev does not have much credibility and component four being experience’ would increase the Alpha of word-of-mouth (see Appendix C). the credibility scale by 0.008 while ‘I do Five of the total 22 items are not not believe what HappyBev tells me’ associated with the corresponding would increase it by 0.001. The Alpha of component, i.e. they correlate best with a the embeddedness scale would by component linked to the scale other than heightened by 0.015 if the item they belong to. The four items of the ‘employees of HappyBev are probably trustworthiness dimension of the similar to me’ were removed. Since the credibility scale seem to correspond best improvements are extremely small and due with the component associated with to the high Alphas the scales are not reputation. This finding is probably not too altered in any way. surprising, since the questions of trustworthiness items and reputation items Cronbach’s Alpha are very alike; two questions are even Credibility 0.856 Embeddedness 0.925 identical (‘I trust HappyBev’). The thought Reputation 0.944 that trustworthiness and expertise could be Word-of-Mouth 0.889 divided into two components has been
proven wrong: a second principal Table 3: Cronbach’s Alphas of scales
15 | David Eberle | 2010, RSM Erasmus University IBRANDING: THE IMPACT OF SOCIAL MEDIA ON CORPORATE BRANDS
component analysis with five factors did Group 1 not change the picture. Mean Differences The fifth item, ‘I will talk positively about (Significance) HappyBev’, belongs to the word-of-mouth Group 2 0.418 (0.220) Group 3 0.575 (0.100) scale. Although it is best associated with Group 4 0.173 (0.617) the reputation component (factor loading = Group 5 0.350 (0.315) 0.581) the correlation with the word-of-
mouth scale has a similar strength (factor Table 5: ANOVA post-hoc of interactivity loading = 0.503). For this study, the scales mean scores are used in their primary format despite these two irregularities. One reason for the weak differences could be an invalid scale used to measure 8.3. Manipulation Checks interactivity. This study asked participants The experiment conducted in this study to rate “the interactivity of HappyBev” on employs two manipulations, website a 7-point Likert scale. Liu (2003), for interactivity and tone of comments. The instance, developed a scale to measure experiment was designed in a way that the interactivity consisting of 15 items divided first group (‘traditional website’) sees a into 3 dimensions. One dimension focuses website with little possibility to interact, on ‘two-way communication’, which is while the other four groups (‘social exactly the factor that is being manipulated media’) are presented a much more in this study. Such a scale would most interactive website. They can post likely capture perceived interactivity better comments, which the website of the first than the one-dimensional item used in this group does not allow. study. An ANOVA reveals that although the Another possible reason is that the mean interactivity score (on a 7-point manipulations were not strong enough to Likert scale) is slightly higher for the create two significantly distinguishable social media groups than for the traditional websites. This experiment used the website, the difference is not significant on possibility to post a comment on the α=0.10 (significance between the groups = company’s website to evoke a sense of 0.167) (see Table 4). interactivity. In many cases, however, social media is more complex than posting Mean single comments. Users can also interact Traditional Website 2.90 with each other, i.e. a user can comment (group 1) on another user’s comment and so on. Social Media 3.25 Participants of the pilot study did not have (group 2,3,4,5) any troubles understanding the question of
how interactive the website is. A most Figure 4: Means of interactivity scores likely reason why there was a large difference between the interactive and A Fisher’s least significant difference test non-interactive website regarding their with each group treated separately shows interactivity score is that participants were that although group one’s mean shown both versions. This comparison interactivity score is lower than the mean allowed them to see that one website was of all other four groups, the differences are indeed more interactive. The final study, again not significant on α=0.10 except for however, did only present one version to the difference with group three (see Table the participants. 5). Consequently, although interactivity is measured correctly, the possibility to post comments is not strong enough to make
16 | David Eberle | 2010, RSM Erasmus University IBRANDING: THE IMPACT OF SOCIAL MEDIA ON CORPORATE BRANDS the website being perceived as very interactive and non-interactive groups. The interactive. When compared to a website low R square (R square = 0.034) of the without such a possibility, however, the adjusted model strongly suggests that there contrast becomes stark. Therefore, for the are other variables influencing perceived validation of the social media branding interactivity, which have not been captured model the score of perceived interactivity within this experiment. is used instead of the five different groups The second manipulation included the to capture the level of interactivity. display of comments in a different tone, Now it certainly is interesting what mostly positive (group three), neutral influenced the participants’ perceptions of (group four), and mostly negative (group interactivity. A regression analysis with five). The differences in the mean of the perceived interactivity as the dependent tone of the comments across the three variable and the control questions as groups are obvious and highly significant independent variables reveals that actual (significance between groups = 0.000) (see interactivity as manipulated in the study Table 7). and region have an impact on perceptions (see Table 6). A second regression (model Mean B) with insignificant variables removed Positive Comments 5.31 shows that while actual interactivity (group 3) slightly increases perceptions, coming Neutral Comments 3.76 (group 4) from a Western country significantly Negative Comments 2.87 decreases perceptions of interactivity by (group 5) about 0.5 points.
Table 7: Tone of comments scores Model A Beta (Significance) Constant 2.483 (0.110) Participants were also asked to rate the Interactivity 0.424 (0.135) influence that the comments had on their I am familiar 0.133 (0.499) perception of HappyBev on a 7-point with using the Internet Likert scale. Interestingly, an ANOVA I like using -0.033 (0.766) shows that the influence is not the same social media across the three groups (significance Age 0.003 (0.899) between the groups = 0.001). Negative Gender 0.026 (0.908) comments seem to have a greater influence Education -0.075 (0.715) on people than positive comments (see Region -0.482 (0.041) Table 8). A Fisher’s least significant Model B Beta (Significance) difference test shows that the mean of Constant 3.158 (0.000) group 5 is significantly different Interactivity 0.407 (0.134) (significance < 0.01) from the means of Region -0.492 (0.024) group 3 and 4, while the divergence between group 3 and 4 is not significant Model B Model Summary (significance = 0.226). R Square 0.034 R Square 0.025 (Adjusted) Mean Positive Comments 3.33
Table 6: Regression analysis of perceived (group 3) interactivity Neutral Comments 3.83 (group 4) Negative Comments 4.92 Apparently the manipulations of (group 5) interactivity did work out but were too
weak to draw a clear line between Table 8: Influence of comments scores
17 | David Eberle | 2010, RSM Erasmus University IBRANDING: THE IMPACT OF SOCIAL MEDIA ON CORPORATE BRANDS
One way of possibly explaining this confirm the exact causality between the finding is to look at how many comments different variables regarding the tone and people say they have read (on a scale of 0 influence of comments. to 10). People again state that they have read more comments in groups 5 and 4 8.4. Model Testing than in group 3 (significance between The social media branding model groups = 0.054) (see Table 9). A Fisher’s previously introduced in chapter five least significant difference test shows that contains four hypotheses that are analyzed only the difference between positive in this section. Hypothesis 1a and 1b look against neutral and negative comments is at the impact of the setting in which the significant (significance = 0.056 and corporate message is published on 0.025, respectively). credibility and embeddedness. As the previous manipulation check revealed Mean perceived interactivity is used to capture Positive Comments 4.85 the online media setting. Regression (group 3) Neutral Comments 6.41 analysis is employed to test these two (group 4) hypotheses. Negative Comments 6.68 (group 5) Model 1a: Credibility = a + b1*Interactivity + e
Table 9: Means of number of comments read Model 1b: This finding, in turn, suggests that the Embeddedness = a + b1*Interactivity + e more comments people read, the higher the influence of the comments on their Regression analyses of both models perceptions. A regression of ‘comments confirm a positive link between perceived read’ as the independent variable and interactivity and the dependent variables ‘comments influence’ as the dependent credibility and embeddedness, respectively variable confirms this conjecture (beta = (see Tables 10 and 11). 0.257, significance = 0.000). A next conjecture is that the number of Model 1a comments read depends on the tone of the Beta (Significance) comments, i.e. the more negative the Constant 3.933 (0.000) comments the higher the interest of the Perceived 0.199 (0.000) reader to actually read them. A regression Interactivity 0.328 (standardized) of ‘comments tone’ as the independent variable and ‘comments read’ as the Model Summary R Square 0.107 dependent also supports this proposition R Square 0.103 (beta = -0.501, significance = 0.016). (Adjusted) A significant direct relationship between comments tone and comments influence Table 10: Regression analysis of model 1a also exists, using influence as a dependent variable (beta = -0.334, significance = Model 1b 0.001). These findings suggest that Beta (Significance) negative comments cause a higher interest Constant 2.166 (0.000) in people to read more comments, while Perceived 0.372 (0.000) reading more comments increase the Interactivity 0.415 (standardized) influence that comments exert on the final perceptions. Yet a study designed to Model Summary measure these effects is needed in order to R Square 0.172
18 | David Eberle | 2010, RSM Erasmus University IBRANDING: THE IMPACT OF SOCIAL MEDIA ON CORPORATE BRANDS
R Square 0.168 significant and their betas did not change (Adjusted) much compared to the single models.
Table 11: Regression analysis of model 1b Model 1d
The coefficients of both models are highly Beta (Significance) significant, which suggests not rejecting Constant 3.076 (0.000) preWOM 0.182 (0.000) the hypotheses. Also the ANOVA F-tests 0.306 (standardized) confirm the validity of the models. Yet, Perceived 0.197 (0.000) especially model 1a seems not to be a very Interactivity 0.313 (standardized) good predictor of future outcomes (R square = 0.107). One major reason for this Model Summary outcome is that hypothesis 3 proposes an R Square 0.215 impact on credibility by pre-exposure R Square 0.201 (Adjusted) word-of-mouth, which can be combined
into a model jointly with perceived Table 13: Regression analysis of model 1d interactivity. Two additional models are used to test hypothesis 3 alone (model 1c) Hypotheses 1a and 1b are confirmed on and together with hypothesis 1a (model the grounds of highly significant 1d): coefficients and robust ANOVA results. Nevertheless, these findings need to be Model 1c: digested with some wariness. The low Credibility = a + b1*preWOM + e values of R square, although somewhat Model 1d: better in the combined model, might Credibility = a + b1*preWOM + indicate that either the variables are b2*Interactivity + e measured poorly or that other factors are being omitted. An analysis of hypothesis 3 independently In order to test hypothesis 3, an ANOVA of hypothesis 1a also shows highly least significant difference test is significant coefficients, a robust ANOVA conducted because the independent F-test but, again, a rather low R square (R variable preWOM only contains samples square = 0.118) (see Table 12). that have existing comments (i.e. samples of the groups 3, 4 and 5). The regression Model 1c analyses in models 1c and 1d thus estimate the influence of tone of comments on Beta (Significance) Constant 3.636 (0.000) credibility. Yet, it is also of interest to preWOM 0.204 (0.000) compare the means of credibility between 0.344 (standardized) the groups with comments (groups 3, 4 and 5) and the group without (group 2). Model Summary Since the interactivity in those four groups R Square 0.118 remains the same, the only manipulations R Square 0.111 conducted are the existence of comments (Adjusted) and the tone of comments.
Table 12: Regression analysis of model 1c While group three is slightly more credible than group two, which has no comments at The combined model definitely is a better all, groups four and five score significantly predictor of credibility than the two single lower on the credibility scale (see Table versions (R square = 0.215) (see Table 14). This implies that while the existence 13). Both coefficients are again highly of largely positive comments is slightly beneficial to the message credibility, as
19 | David Eberle | 2010, RSM Erasmus University IBRANDING: THE IMPACT OF SOCIAL MEDIA ON CORPORATE BRANDS soon as the amount of negative comments reputation. Model 2 combines the effects reaches a certain point message credibility of both credibility and embeddedness on is significantly decreased. reputation.
Group Difference Sig. Model 2: Group 2 3 -0.0428 0.833 Reputation = a + b1*Credibility + 4 0.3762 0.064 b2*Embeddedness + e 5 0.4291 0.036 Group 3 2 0.0428 0.833 Both coefficients are highly significant 3 0.4190 0.044 and the ANOVA F-tests suggests that the 4 0.4719 0.024 model is robust. Furthermore, model 2 Group 4 2 -0.3762 0.064 3 -0.4190 0.044 seems to be a good predictor (R square = 5 0.0529 0.798 0.673) (see Table 16). Group 5 2 -0.4291 0.036 3 -0.4719 0.024 Model 2 4 -0.0529 0.798 Beta (Significance)
Table 14: ANOVA least significant Constant -0.359 (0.188) differences of credibility Credibility 0.766 (0.000) 0.538 (standardized) Embeddedness 0.374 (0.000) Therefore, hypothesis 3 cannot be 0.383 (standardized) confirmed on the grounds that even in the neutral setting credibility is by far lower Model Summary than in the group without any comments. R Square 0.673 An additional ANOVA testing the impact R Square 0.670 of comments on embeddedness reveals a (Adjusted) similar outcome. Embeddedness too is Table 16: Regression analysis of model 2 negatively affected by the existence of negative comments (see Table 15). Even Based on the outcome of the regression of the group with mostly positive comments model 2 hypotheses 2a and 2b are has a slightly lower value of confirmed. Since the coefficients are embeddedness, although the difference is highly significant and the model is robust not significant. there are no reasons for concern regarding this result. Group Difference Sig. The last hypothesis, which suggests a Group 2 3 0.1505 0.616 4 0.5176 0.083 positive impact of embeddedness on post- 5 0.8342 0.006 exposure word-of-mouth, is tested using Group 3 2 -0.1505 0.616 the following model. 3 0.3672 0.229 4 0.6836 0.027 Model 3: Group 4 2 -0.5176 0.083 postWOM = a + b1*embeddedness +e 3 -0.3672 0.229 5 0.3166 0.299 This model is estimated twice, one time Group 5 2 -0.8342 0.006 using only the data gathered with the 3 -0.6837 0.027 4 -0.3166 0.299 word-of-mouth scale presented earlier in this paper (postWOM1) and a second time
Table 15: Regression analysis of model 2 with the additional question of whether the participant would like to share Hypotheses 2a and 2b propose a positive HappyBev’s website with his friends on causality between credibility and Facebook (provided he or she has a embeddedness and the dependent variable Facebook profile). For the second analysis
20 | David Eberle | 2010, RSM Erasmus University IBRANDING: THE IMPACT OF SOCIAL MEDIA ON CORPORATE BRANDS all scale items were standardized and Due to a higher ANOVA F value and a weighted equally to be ultimately greater R square postWOM1 is preferred combined into the dependent variable to postWOM2. Since the standardized postWOM2. In both versions the values of the coefficient are basically coefficient is highly significant and very identical, there is no reason of concern similar in value (the standardized when omitting the Facebook question in coefficient is nearly identical). The the regression. ANOVA F-tests reveal that the models are Hypothesis 4 is therefore confirmed due to very robust and the R square measures a highly significant coefficient, and a solid indicate that the models are fairly good model. predictors (see Tables 17 and 18). In summary all hypotheses except hypothesis 3 are confirmed while some Model 3 (postWOM1) reservations exist still. Although all regression coefficients are highly Beta (Significance) Constant 0.897 (0.000) significant, the models of hypotheses 1a, Embeddedness 0.709 (0.000) 1b and 3 do not produce high R square 0.715 (standardized) values. Figure 4 presents an overview of the relationships between the tested Model Summary variables using standardized beta R Square 0.511 coefficients. For hypotheses 1a and 3 the R Square 0.508 (Adjusted) model 1d is used, as its R square is superior to the models 1a and 1c.
Table 17: Regression analysis of model 3 PostWOM1 is preferred over PostWOM2 since the regression yields a higher R Model 3 (postWOM2) square while the coefficients are basically identical. Beta (Significance) Constant 0.654 (0.002) Embeddedness 0.810 (0.000) 0.708 (standardized)
Model Summary R Square 0.501 R Square 0.499 (Adjusted)
Table 18: Regression analysis of model 3
21 | David Eberle | 2010, RSM Erasmus University IBRANDING: THE IMPACT OF SOCIAL MEDIA ON CORPORATE BRANDS
Figure 4: Regression results of the ‘social media branding model’
9. Conclusions spreading around the globe, its management most likely is interested in 9.1. Summary the impact of such propaganda on its Companies are increasingly concerned corporate brand. about the impact of social media on their The two factors that make social media a brands. Greenpeace, for instance, attacked unique channel are the interactive Nestlé regarding its palm oil suppliers functionality, i.e. the ability to respond to whose practices endanger the orang-utan. what has been published, as well as the Its Youtube movie ‘have a break?’8, which possibility to read what others have shows an employee chewing an orang-utan commented on the message so far. This finger in the shape of a Kit-Kat, had been results not only in an interaction between watched 1.5m times and caused 200,000 consumers and the company but also protest emails (The Economist, 2010). between visitors themselves. After trying to “bury the nasty spoof” Scholarly theory (Beninger, 1987; Bickart Nestlé suspended all orders from the & Schindler, 2001; Thorson & Rodgers, accused supplier (The Economist, 2010). 2006) suggests that interactivity of Social media clearly are a powerful websites increases the credibility of channel, which can even break the biggest corporate messages. This is due to the fact conglomerates. In contrast to traditional that the possibility to directly respond to a websites, social media allow consumers to company’s claim makes the firm reluctant interact with each other. Greenpeace’s to publish claims that are quickly video, for example, prompted 488 denounced. This interactive link also comments so far; while some are functions as a bridge between consumers denouncing Nestlé, others are rather in and the company thus evoking feelings of favor of the food giant. Since the company embeddedness among consumers. was unable to stop the video from Although existing comments (pre- exposure word-of-mouth) boost consumer 8 Youtube. (17. March 2010). Have a Break? awareness, their impact on consumer Retrieved 18. July 2010 from Youtube: attitude is rather ambivalent (Basuroy et http://www.youtube.com/watch?v=VaJjPRwExO8
22 | David Eberle | 2010, RSM Erasmus University IBRANDING: THE IMPACT OF SOCIAL MEDIA ON CORPORATE BRANDS al., 2003; Berger et al., forthcoming; Duan together with positive reviews, harshly et al., 2008; Liu, 2006). While positive decrease message credibility. They also comments obviously increase the harm embeddedness. credibility of the corporate message, These findings, however, might not be that negative comments are thought to be less surprising after all. Since participants were beneficial. Nevertheless, mixed settings more strongly influenced and read more with partially negative and positive word- comments when the comments were more of-mouth should, according to previous negative stresses that although positive research (East et al., 2008; Doh & Hwang, comments would probably boost 2009), still have an elevating effect on credibility, people are simply not really credibility. interested in reading those comments. Corporate reputation, in turn, is positively Therefore, positive comments are rather affected by credibility, as outlined in skipped when reading a message on social contemporary corporate communication media and are thus unable to exert much literature (Fombrun, 1996; Goldsmith & influence on consumer attitude. Lafferty, 1999; Goldsmith et al., 2000; This is consistent with the category MacKenzie & Lutz, 1989). diagnosticity approach introduced by Finally, higher feelings of embeddedness Skowronski and Carlston (1989), which are also beneficial to corporate reputation states that in case of “equal but opposite since consumers feel more connected to cues” the negative information is assigned the firm and are thus more concerned more weight. This negativity bias is about the firm’s state, which is consistent explained by the fact that negative or with Bhattacharya and Sen (2003), Kotler extreme cues are more informative than and Keller (2008), and Muniz and positive or moderate ones. For instance, a O’Guinn (2001). Consumers that have person who robs a bank (negative cue) higher feelings of embeddedness are also might also pay taxes on his property more likely to engage in (post-exposure) (positive cue), but is still perceived as word-of-mouth, as shown by Hennig- being dishonest (Skowronski & Carlston, Thurau and Walsh (2003) and Wang and 1989). Baumeister et al. (2001) go further Fesenmaier (2003). by stating that “bad is stronger than good in a disappointingly relentless pattern”. 9.2. Discussion and Future Research They find that even when correcting for The experiment conducted in this paper explanations such as the category basically confirmed all hypotheses diagnosticity approach, “any reversals are developed on the basis of past research likely to remain mere exceptions” except hypothesis 3, implying that the (Baumeister et al., 2001). existence of negative comments hurt the Such negativity effects might also have an credibility of the message. Using five impact on the relationship between different versions (non-interactive, embeddedness and the tone of comments. interactive, and with different sets of Although interactive websites are likely to comments) of a website of a fictitious evoke higher feelings of embeddedness company called HappyBev, regression among consumers, the existence of analyses produce highly significant results. negative comments could reduce the The probably most unexpected finding is resulting boost in reputation. In fact, when that the existence of comments does not there are a lot of negative comments, the boost credibility and has even a slightly increase in credibility due to interactivity negative, although not significant, impact might be much lower than when comments on embeddedness. While positive are mostly positive. This suggests a comments do not harm the company, moderating role of the tone of comments negative comments, even when appearing on the impact of interactivity on
23 | David Eberle | 2010, RSM Erasmus University IBRANDING: THE IMPACT OF SOCIAL MEDIA ON CORPORATE BRANDS embeddedness. This implies that the tone leads to higher credibility and of comments ultimately has a greater embeddedness or the other way round. influence on reputation than interactivity Yet, evidence from Thorson and Rodger and that firms should pay greater attention (2006), who find that perceived to what is being written (message content) interactivity is a better predictor of the than where it is being published (message attitude towards the website than setting). The current additive model does interactivity itself, points towards a good not take into account a difference in direction. Indeed, a regression analysis influence of the message content on testing what elements captured in this reputation from the message setting. It is study might impact perceived interactivity thus the task of future research to analyze shows that interactivity does have an such a moderating role, especially since influence on perception. Interestingly, negative comments can have an immense participants coming from Western impact on company’s business practices, countries evaluate the website with a lower as Nestlé’s story shows. value of interactivity. Not all confirmed hypotheses, however, Yet, a very low R square value implies could be accepted without hesitation, alas. that some other major factor influenced The manipulation of interactivity did not interactivity perceptions. One possible work out as planned despite good results complication of the discrepancy between in the pilot study. Although the interactive interactivity and perceived interactivity is versions were perceived as being slightly that it is not clear why exactly participants more interactive than the non-interactive evaluated the same version of the website website, the differences are statistically not with a different score of perceived significant. interactivity. There might be elements at The manipulation of interactivity therefore play which are not influenced by whether was probably not strong enough. Instead a website is traditional or social. Then, of simply adding an area where visitors although perceived interactivity does can post their comments as done in influence credibility and embeddedness, Thorson and Rodger’s (2006) study, social media would not boost the interactivity probably needs to be more perception but some other, yet unknown, complex like on today’s major social factor. networking sites, where users can also The low R square values obtained from the comment on existing posts and where they regression analyses of hypotheses 1a and can recommend certain parts to their 1b point towards that direction, since low friends. R square values usually indicate that there Perhaps participants also have different is some other factor, which is not expectations what an interactive website regressed, influencing the dependent consists of; yet, since most participants variable. In order to obtain clearer results said to be very experienced internet users in regards to the impact of interactivity on this conjuncture cannot be verified at this credibility and embeddedness, a second point. experiment that uses stronger The major drawback of this study manipulations of interactivity should be therefore is that instead of using actual conducted. interactivity as manipulated in the An additional factor that has an impact on different versions of HappyBev’s website, credibility is pre-exposure word-of-mouth, perceived interactivity is used instead to which, when included in the regression, measure whether the website is interactive boosts R square to a more acceptable but or not. One limitation of this is the still rather low level. When pre-exposure possibility of reverse causality, i.e. that it word-of-mouth is used as the sole is not clear whether perceived interactivity estimator of credibility, R square is also
24 | David Eberle | 2010, RSM Erasmus University IBRANDING: THE IMPACT OF SOCIAL MEDIA ON CORPORATE BRANDS
very low. These findings might pinpoint 9.3. Managerial Implications credibility to be the scapegoat, i.e. to be Implications for management resulting badly measured. Yet, the scale for from this research are threefold. First, this credibility has been extensively validated study shows that credibility and (Newell & Goldsmith, 2001) and the embeddedness do have a highly Cronbach alpha is very high, proving the significant, positive influence on corporate robustness of the scale. Consequently, reputation. Therefore, companies should there is most likely a third (and perhaps be concerned about how to increase these even a fourth) factor influencing two factors. Moreover, consumers are credibility and embeddedness. more likely to engage in positive word-of- It is thus the task of future research to mouth if they feel closer to the company, examine such possible influencers. A first i.e. when embeddedness is high. step is to repeat the experiment conducted Second, the results of this paper reveal that in this thesis while adapting the way an increase in perceived interactivity leads interactivity is manipulated. By drawing a to a higher credibility and stronger feelings clearer line between the version that of embeddedness towards the company. represents the traditional website and the Managers clearly can build on this by version that is ought to be social, communicating with their customers interactivity and perceived interactivity through channels that are perceived to be should be more congruent with each other. more interactive. One factor of increasing That could minimize measurement errors interactivity is actual interactivity, and hopefully produce models that are meaning that very interactive media, such better estimators. A second step also as social media, increase the perceptions involves further study of past literature in of visitors. order to find possible other estimators of Third, admitting visitors to read existing credibility and embeddedness, which have comments on the website does only not been analyzed in this study. slightly heighten the credibility of the An additional explanation of the low R company when the comments are very square values could be that credibility and positive, while negative comments clearly embeddedness are ill chosen mediators. hurt the company. In settings with 5:5 and Yet, a regression analysis using reputation 2:8 positive to negative comments as the dependent variable and perceived credibility was substantially lower than for interactivity and pre-exposure word-of- a case where no comments were displayed. mouth as the independent variables does This is because consumers are more not yield a high R square either (R square interested in negative comments; they read = 0.316). When estimating reputation more comments when there is a higher using the mediators credibility and proportion of negative opinions and they embeddedness (model 2) the model is a claim that they are more influenced by the good predictor (R square = 0.673) and the comments as they become more negative. coefficients are highly significant. Positive comments, in contrast, are not Both credibility and embeddedness thus really considered and their impact on clearly have a positive impact on credibility is thus rather weak. reputation as predicted by theory. Consequently, when a company has to Furthermore, higher feelings of choose between allowing consumers to embeddedness lead to more post-exposure read comments that are substantially word-of-mouth (model 3). Both R square negative and not presenting any comments and coefficient are significant and at all, the findings of this study suggest therefore leave no room for concern. that latter option ultimately results in a higher credibility of the company.
25 | David Eberle | 2010, RSM Erasmus University IBRANDING: THE IMPACT OF SOCIAL MEDIA ON CORPORATE BRANDS
This study overall shows that while some looking for the answer whether the Kobe elements of social media are beneficial to Bryant video was beneficial to the corporate brands (e.g. the ability to post shoemaker’s brand; yet, the answers of comments), other characteristics are rather this study can give only preliminary cues negative (e.g. the presence of negative and have to be further examined. Future reviews). Managers who take such factors studies have to concentrate not only on the into account will be more able to make aspects of interactivity, but also on optimal use of online channels. strategies how a company can mitigate the impact of negative comments or spur the 9.4. Academic Implications publication of positive contributions. With For academics this thesis adds to their such insights, firms are best equipped for knowledge in two ways. First, it confirms the new age of corporate communication – the impact of social media on corporate iBranding. reputation as suggested by various empirical studies (Alwi & Da Silva, 2007; Merrilees & Fry, 2002; Van der Heijden & 10. Bibliography Verhagen, 2004; Wolfinbarger & Gilly, Alwi, S., & Da Silva, R. (2007). Online and Offline 2003). These studies examined mostly Corporate Brand Images: Do They Differ? online retailers whereas current situations Corporate Reputation Review , 10 (4), 217-244. often involve companies that publish Apple. (2010). Apple Magic Mouse. Retrieved 2010 11-April from Apple Store (U.S.): information online without directly selling http://store.apple.com/us/product/MB829LL/A?fno products. This paper shows that social de=MTY1NDA1Mg&mco=MTc0MjYxMDA media exerts a major influence on Argenti, P., & Druckenmiller, B. (2004). Reputation corporate reputation also in these settings and the Corporate Brand. Corporate Reputation and should therefore leverage academic Review , 6 (4), 368-374. Bunting, M., & Lipski, R. (2000). Drowned out? interest in this area of research. Rethinking corporate reputation management for Second, this study develops a the Internet. Journal of Communication comprehensive social media branding Management , 5 (2), 170-178. model with theory drawn from relevant Baumeister, R., Bratslavsky, E., Finkenauer, C., & literature in the field of corporate Vohs, K. (2001). Bad Is Stronger Than Good. Review of General Psychology , 5 (4), 323-370. communications, marketing, sociology and Balmer, J. (2001). Corporate identity, corporate psychology. This framework could act as a branding and corporate marketing: Seeing through basic notion for future research in this the fog. European Journal of Marketing , 35 (3-4), field since no common ground has been 248-291. reached so far, as outlined earlier in this Basuroy, S., Chatterjee, S., & Ravid, S. (2003). How Critical Are Critical Reviews? The Box Office paper. Effects of Film Critics, Star Power, and Budgets. Journal of Marketing , 67, 103-117. 9.5. Concluding Remarks Beninger, J. (1987). Personalization of Mass Media and the Growth of Pseudo-Community. With more people being active on social Communication Research , 14 (3), 352-371. media, these interactive online channels Berger, J., Sorensen, A., & Rasmussen, S. are certainly gaining importance also for (forthcoming). Positive Effects of Negative branding experts. This paper has shown Publicity: Can Negative Reviews Increase Sales? that companies should pay attention to Marketing Science . these new forms of communication since Bickart, B., & Schindler, R. (2001). Internet forums as influential sources of consumer information. their reputation certainly is or will be Journal of Interactive Marketing , 15 (3), 31-40. affected by it. The insights gained in this Bhattacharya, C., & Sen, S. (2003). Consumer- study can help decision makers capitalize Company Identification: A Framework for on the possibilities of online Understanding Consumers Relationships with communication. Nike is probably still Companies. Journal of Marketing , 67, 76-88.
26 | David Eberle | 2010, RSM Erasmus University IBRANDING: THE IMPACT OF SOCIAL MEDIA ON CORPORATE BRANDS
Blackshaw, P., & Nazzaro, M. (2006). Consumer- Doh, S., & Hwang, J. (2009). Rapid Generated Media (CGM) 101: Word-of-Mouth in the Communication: How Consumers Evaluate eWOM age of the Web-Fortified Consumer. New York: (Electronic Word-of-Mouth) Messages. Nielsen BuzzMetrics. CyberPsychology & Behavior , 12 (2), 193-197. Cameron, S. (1995). On the Role of Critics in the Fisher, T. (2009). ROI in social media: A look at the Culture Industry. Journal of Cultural Economics , arguments. Database Marketing & Customer 19, 321-331. Strategy Management , 16 (3), 189-195. Carrabis, J., Cass, J., Gillin, P., Natch, R., & Fombrun, C. (1996). Reputation: realizing value Peverill-Conti, G. (2008). New Media, New from the corporate image. Boston: Harvard Influencers and Implications for Public Relations. Business School Press. San Jose: SCNR Press. Fombrun, C., & Van Riel, C. (1997). The Catalano, C. (2007). Megaphones to the Internet Reputational Landscape. Corporate Reputation and the World: The role of blogs in corporate Review , 1 (1/2), 5-13. communications. International Journal of strategic Fombrun, C., Gardberg, N., & Sever, J. (2000). The communication , 1 (4), 247-262. Reputation Quotient: A multi-stakeholder measure Chen, S. (2001). Assessing the impact of the of corporate reputation. The Journal of Brand Internet on brands. Journal of Brand Management , Management , 7 (4), 241-255. 8 (4), 288-302. Garson, D. (2010 30-January). Reliability Analysis. Choi, S., & Rifon, N. (2002). Antecedents and Retrieved 2010 28-June from NC State University Consequences of Web Advertising Credibility: A CHASS: study of consumer response to banner ads. Journal http://faculty.chass.ncsu.edu/garson/PA765/reliab. of Interactive Advertising , 3 (1), 12-24. htm Clauser, R. (2001). Offline rules, online tools. Goldsmith, R., & Lafferty, B. (1999). Corporate Journal of Brand Management , 8, 270-287. Credibility's Role in Consumer's Attitudes and Cohen, J. (1992). A Power Primer. Psychological Purchase Intentions When a High versus a Low Bulletin , 112 (1), 155-159. Credibility Endorser Is Used in the Ad. Journal of Business Research , 44, 109-116. Constantinides, E., & Fountain, S. (2007). Web 2.0: Conceptual foundations and marketing issues. Goldsmith, R., Lafferty, B., & Newell, S. (2000). The Journal of Direct, Data and Digital Marketing Impact of Corporate Credibility and Celebrity Practice , 9 (3), 231-244. Credibility on Consumer Reaction to Advertisements and Brands. Journal of Advertising comScore. (2009). Social Networking Sites Account , 29 (3), 43-54. for More than 20 Percent of All U.S. Online Display Ad Impressions. comScore. Gordon MacMillan. (2010 25-January). Labour flips Obama message of change for social media East, R., Hammond, K., & Lomax, W. (2008). campaign. Retrieved 2010 27-February from Measuring the impact of positive and negative marketingmagazine.co.uk: word of mouth on brand purchase probability. http://www.marketingmagazine.co.uk/news/97955 International Journal of Research in Marketing , 25, 6/Labour-flips-Obama-message-change-social- 215-224. media-campaign/ Einwiller, S., Fedorikhin, A., Johnson, A., & Kamins, Gruen, T., Osmonbekov, T., & Czaplewski, A. M. (2006). Enough Is Enough! When Identification (2006). eWOM: The impact of customer-to- No Longer Prevents Negative Corporate customer online know-how exchange on customer Associations. Journal of the Academy of Marketing value and loyalty. Journal of Business Research , Science , 34 (2), 185-194. 59, 449-456. eReleases. (2009 27-November). The True Cost of IAB. (2009). Social Media Ad Metrics Definitions. Bad Publicity. Retrieved 2010 24-June from Interactive Advertising Bureau. eReleases: http://www.ereleases.com/prfuel/true- cost-bad-publicity/ Ind, N., & Riondion, M. (2001). Branding on the Web: A real revolution? Brand Management , 9 (1), Duan, W., Gu, B., & Whinston, A. (2008). Do online 8-19. reviews matter? — An empirical investigation of panel data. Decision Support Systems , 45, 1007- Huntington, S. (1997). The Clash of Civilizations and 1016. the Remaking of World Order. New York: Touchstone. Da Silva, R., & Alwi, S. (2008). Online Brand Attributes and Online Corporate Brand Images. Hennig-Thurau, T., & Walsh, G. (2003). Electronic European Journal of Marketing , 42 (9/10), 1039- Word of Mouth: Motives for and Consequences of 1058. Reading Customer Articulations on the Internet. International Journal of Electronic Commerce , 8 Davies, G., Chun, R., Da Silva, R., & Roper, S. (2), 51-74. (2004). A Corporate Character Scale to Assess Employee and Customer Views of Organization Jee, J., & Lee, W. (2002). Antecedents and Reputation. Corporate Reputation Review , 7 (2), consequences of perceived interactivity: an 125-146.
27 | David Eberle | 2010, RSM Erasmus University IBRANDING: THE IMPACT OF SOCIAL MEDIA ON CORPORATE BRANDS
exploratory study. Journal of Interactive Advertising Phelps, J., Lewis, R., Mobilio, L., Perry, D., & , 3 (1), 34-45. Raman, N. (2004). Viral Marketing or Electronic Kaplan, A., & Haenlein, M. (2010). Users of the Word-of-Mouth Advertising: Examining Consumer world, unite! The challenges and opportunities of Responses and Motivations to Pass Along Email. Social Media. Business Horizons , 53, 59-68. Journal of Advertising Research , 44 (4), 333-348. Kent, M. (2008). Critical analysis of blogging in Post, T. (2007 25-October). Newspapers struggle public relations. Public Relations Review , 34, 32- with online comments. Retrieved 2010 11-April 40. from Minnesota Public Radio NewsQ: http://minnesota.publicradio.org/display/web/2007/ Knox, S., & Bickerton, D. (2003). The six 10/25/onlinecomments/ conventions of corporate branding. European Journal of Marketing , 37 (7/8), 998-1016. Porter, L., & Golan, G. (2006). From Subservient Chickens to Brawny Men: A Comparison of Viral Kotler, P., & Keller, K. (2008). Marketing Advertising to Television Advertising. Journal of Management. Upper Saddle River: Prentice-Hall. Interactive Advertising , 6 (2), 26-33. Liu, Y. (2006). Word of Mouth for Movies: Ist Shamdasani, P., Stanaland, A., & Tan, J. (2001). Dynamics and Impact on Box Office Revenue. Location, Location, Location: Insights for Journal of Marketing , 70, 74-89. Advertising Piacement on tiie Web. Journal of Liu, Y. (2003). Developing a Scale to Measure the Advertising Research , 41 (4), 7-21. Interactivity of Web Sites. Journal of Advertising Skowronski, J., & Carlston, D. (1989). Negativity Research , 43 (2), 207-216. and Extremity Biases in Impression Formation: A Newell, S., & Goldsmith, R. (2001). The Review of Explanations. Psychological Bulletin , development of a scale to measure perceived 105 (1), 131-142. corporate credibility. Journal of Business Research Smith, R., & Vogt, C. (1995). The Effects of , 52, 235-247. Integrating Advertising and Negative Word-of- Netpop. (2009). Snapshot: United States 2009. Mouth Communications on Message Processing Netpop Research. and Response. Journal of Consumer Psychology , Nielsen Wire. (2010). Led by Facebook, Twitter, 4 (2), 133-151. Global Time Spent on Social Media Sites up 82% Stuart, H. (1999). Towards a definitive model of the Year over Year. The Nielsen Company. corporate identity management process. Corporate Muniz, A., & O'Guinn, T. (2001). Brand Community. Communications: An International Journal , 4 (4), Journal of Consumer Research , 27, 412-432. 200-207. Maxham, J. (2001). Service recovery’s influence on Stuart, H., & Jones, C. (2004). Corporate Branding consumer satisfaction, positive word-of-mouth, in Marketspace. Corporate Reputation Review , 7 and purchase intentions. Journal of Business (1), 84-93. Research , 54, 11-24. Rubinstein, H., & Griffiths, C. (2001). Branding MacKenzie, S., & Lutz, R. (1989). An Empirical matters more on the Internet. Brand Management , Examination of the Structural Antecedents of 8 (6), 394-404. Attitude Toward the Ad in an Advertising Pretesting Rogers, E. (2003). Diffusion of Innovations. New Context. Journal of Marketing , 53, 48-65. York: Free Press. Mangold, W., & Faulds, D. (2009). Social Media: TIME. (2007 5-July). Obama's Viral Marketing The new hybrid element of the promotion mix. Campaign. TIME . Business Horizons , 52 (4), 357-365. The Economist. (2005 31-March). Crowned at last. McKinsey. (2007). How businesses are using Web The Economist . 2.0: A McKinsey Global Survey. McKinsey & The Economist. (2010 14-January). Good news. Company. The Economist . McKinsey. (2009). How Companies are benefiting The Economist. (2010 24-June). The other oil spill. from Web 2.0. McKinsey & Company. The Economist . Merrilees, B., & Fry, M. (2002). Corporate Branding: The New York Times. (2001 6-April). A new Web A Framework for E-tailers. Corporate Reputation site offers everything you wanted to know about Review , 5 (2/3), 213-225. brands in the Internet age. The New York Times . Obermiller, C., & Spangenberg, E. (2000). On the Thorson, K., & Rodgers, S. (2006). Relationships Origin and Distinctness of Skepticism toward between Blogs as eWOM and Interactivity, Advertising. Marketing Letters , 11 (4), 311-322. Perceived Interactivity, and Parasocial Interaction. Parise, S., & Guinan, P. (2008). Marketing Using Journal of Interactive Advertising , 6 (2), 34-44. Web 2.0. Proceedings of the 41st Hawaii Trusov, M., Bucklin, R., & Pauwels, K. (2009). International Conference on System Sciences. Effects of Word-of-Mouth Versus Traditional Petty, R., & Cacioppo, J. (1981). Attitudes and Marketing: Findings from an Internet Social Persuasion: Classic and Contemporary Networking Site. Journal of Marketing , 73, 90-102. Approaches. Dubuque: AC Brown.
28 | David Eberle | 2010, RSM Erasmus University IBRANDING: THE IMPACT OF SOCIAL MEDIA ON CORPORATE BRANDS
Wang, Y., & Fesenmaier, D. (2003). Assessing Retrieved 2010 26-March from Youtube: Motivation of Contribution in Online Communities: http://www.youtube.com/watch?v=yURa9T0-Rjk An Empirical Investigation of an Online Travel Youtube. (2008b 24-March). Nestlé CEO Peter Community. Electronic Markets , 13 (1), 33-45. Brabeck. Retrieved 2010 11-April from Youtube: Wolfinbarger, M., & Gillin, M. (2003). eTailQ: http://www.youtube.com/watch?v=qyAzxmN2s0w dimensionalizing, measuring and predicting etail Van der Heijden, H., & Verhagen, T. (2004). Online quality. Journal of Retailing , 79, 183-198. store image: conceptual foundations and empirical Youtube. (2007 17-August). GE General Electric measurement. Information & Management , 41, commercial - Wind Energy. Retrieved 2010 11-April 609-617. from Youtube: Van Riel, C., & Fombrun, C. (2007). Essentials of http://www.youtube.com/watch?v=fViObqGvIjM Corporate Branding: Implementing practices for Youtube. (2010 17-March). Have a Break? effective reputation management. New York: Retrieved 2010 18-July from Youtube: Routledge. http://www.youtube.com/watch?v=VaJjPRwExO8 Van Tulder, R., & Van der Zwart, A. (2006). Youtube. (2008a 9-April). Laker Kobe Bryant International Business-Society Management: attempts massive stunt...and succeeds! Real? Linking corporate responsibility and globalization. New York: Routledge.
29 | David Eberle | 2010, RSM Erasmus University IBRANDING: THE IMPACT OF SOCIAL MEDIA ON CORPORATE BRANDS
11. Appendix A
Red color indicates that the item has been altered or added.
30 | David Eberle | 2010, RSM Erasmus University IBRANDING: THE IMPACT OF SOCIAL MEDIA ON CORPORATE BRANDS
12. Appendix B
HappyBev Main Website
31 | David Eberle | 2010, RSM Erasmus University IBRANDING: THE IMPACT OF SOCIAL MEDIA ON CORPORATE BRANDS
HappyBev Comment Posting Function
HappyBev Comments (Examples)
Direct link to survey introduction page: http://www.davideberle.com/survey/index.html Direct link to scenario 1: http://www.davideberle.com/survey/happybev1.html Direct link to scenario 2: http://www.davideberle.com/survey/happybev2.html Direct link to scenario 3: http://www.davideberle.com/survey/happybev3.html Direct link to scenario 4: http://www.davideberle.com/survey/happybev4.html Direct link to scenario 5: http://www.davideberle.com/survey/happybev5.html
32 | David Eberle | 2010, RSM Erasmus University IBRANDING: THE IMPACT OF SOCIAL MEDIA ON CORPORATE BRANDS
13. Appendix C
Rotated Component Matrixa Component
1 2 3 4 HappyBev has a great amount of .120 .126 .876 .156 experience. HappyBev is skilled in what they do. .361 .212 .764 .145 HappyBev has great expertise. .311 .294 .702 .237 HappyBev does not have much experience. .069 -.052 .806 -.076 I trust HappyBev. .791 .336 .179 .087 HappyBev makes truthful claims. .795 .232 .063 .125 HappyBev is honest. .802 .235 .049 .141 I do not believe what HappyBev tells me. .409 .157 .238 .232 I feel associated with HappyBev. .282 .819 .097 .265 I have a sense of connection with .324 .823 .064 .285 HappyBev. I consider myself as belonging to the group .438 .751 .124 .248 of people who are in favor of HappyBev. HappyBev is probably similar to me. .276 .798 .187 .258 Employees of HappyBev are probably .290 .655 .128 .175 similar to me. I have a good feeling about HappyBev. .786 .350 .207 .227 I admire and respect HappyBev. .687 .397 .254 .304 I trust HappyBev. .762 .409 .183 .171 HappyBev has a responsible approach to .782 .174 .280 .212 water management. HappyBev is an environmentally responsible .791 .185 .204 .247 company. I will talk positively about HappyBev. .581 .267 .186 .503 I will recommend HappyBev to my friends. .470 .302 .193 .679 I will talk about HappyBev on blogs and .184 .401 .046 .805 social networking sites. I will invite friends to learn more about .240 .294 .117 .823 HappyBev online.
Extraction Method: Principal Component Analysis. Rotation Method: Varimax with Kaiser Normalization. a. Rotation converged in 6 iterations.
33 | David Eberle | 2010, RSM Erasmus University