71 Journal of Marketing Vol. 74 (March 2010), 71���89 �� 2010, American Marketing Association ISSN: 0022-2429 (print), 1547-7185 (electronic) Robert V. Kozinets, Kristine de Valck, Andrea C. Wojnicki, & Sarah J.S. Wilner Netwo rked Narratives: Under standingWord-of-Mouh t Marketing in OnlineCommunitie s Word-of-mouth (WOM) marketing���firms��� intentional influencing of consumer-to-consumer communications���is an increasingly important technique. Reviewing and synthesizing extant WOM theory,this article shows how marketers employing social media marketing methods face a situation of networked coproduction of narratives. It then presents a study of a marketing campaign in which mobile phones were seeded with prominent bloggers. Eighty- three blogs were followed for six months.The findings indicate that this network of communications offers four social media communication strategies���evaluation, embracing, endorsement, and explanation. Each is influenced by character narrative, communications forum, communal norms, and the nature of the marketing promotion. This new narrative model shows that communal WOM does not simply increase or amplify marketing messages rather, marketing messages and meanings are systematically altered in the process of embedding them. The theory has definite, pragmatic implications for how marketers should plan, target, and leverage WOM and how scholars should understand WOM in a networked world. Keywords: advertising and promotions, consumer communication, online communities, online consumer behavior, Internet marketing, social media, word of mouth Robert V. Kozinets is Associate Professor of Marketing (e-mail: rkozinets@schulich.yorku.ca), and Sarah J.S. Wilner is a doctoral candi- date in Marketing (e-mail: swilner06@schulich.yorku.ca), Schulich School of Business, York University. Kristine de Valck is Assistant Professor of Marketing, HEC School of Management, Paris (e-mail: devalck@hec.fr). Andrea C.Wojnicki is Assistant Professor of Marketing, Rotman School of Management, University of Toronto (e-mail: andrea.wojnicki@rotman. utoronto.ca). This research was partially supported by the first author���s Social Sciences and Humanities Research Council of Canada standard research grant (No. 410-2008-2057). Special thanks to the great people at ���Buzzablog��� who have kindly and patiently shared their data, their time, and their valuable insights. Wmunolc-mouencbngrkeoifogsuWOlMmc)okeuimgritech- ord- f uth ma t n ( M is the nten- tiona infl i of c n mer-to- ns e com i ations y pr essiona ar t n - niques. Known also as social media marketing, viral marketing, buzz, and guerilla marketing, a plethora of popu- lar books on WOMM have recently been released (e.g., Jaffe 2007 Kelly 2007 Rosen 2009 Sernovitz 2006), and industry associations, such as the Word of Mouth Marketing Association, have grown rapidly and have advocated for the burgeoning new industry. According to recent research, marketers spent more than $1.54 billion on WOMM initia- tives in 2008, and that amount is expected to rise to $3 bil- lion in 2013 (PQ Media 2009). Marketers and sociologists have recognized the impor- tance of the phenomenon of word of mouth (WOM)��� conceptualized as a naturally occurring phenomenon���for more than half a century, proposing, for example, that WOM affects the majority of all purchase decisions (Brooks 1957 Dichter 1966). However, these theories and observations about informal, unsolicited WOM were con- structed in a marketing world untouched by the Internet (Brown, Broderick, and Lee 2007 Dellarocas 2003 Godes et al. 2005 Hennig-Thurau et al. 2004). The Internet���s accessibility, reach, and transparency have empowered marketers who are interested in influenc- ing and monitoring WOM as never before. This article is situated firmly within this new marketing reality. We offer theory that integrates these transformations into the world of WOM. We begin with an overview of the development of WOM theory and its assumptions, placing recent develop- ments into historical context. Building on this framework, we detail our empirical investigation of an online WOMM campaign that engaged consumers who were seeded with a new technology device to generate WOM in their personal blogs. Our findings specify how this marketing process unfolds in the realm of consumer-to-consumer WOM com- munications. We discuss both theoretical and managerial implications. We discover and demonstrate four distinct blogger com- munication strategies in response to the product seeding. These communication strategies are marked by the promi- nent tension between commercial and communal norms. In addition, they are shaped to fit the represented individual blogger narratives. These communication strategies have specific implications for how marketers should leverage WOMM campaigns, both online and offline. The Transformation of WOM Theory As markets change, marketing theories must also change to accommodate them. In this section, we provide an assump- tive frame for this article by briefly reviewing the develop-

ment of WOM theory and practice, as summarized in Fig- ure 1. Although the overview casts theoretical development as a series of three evolutionary shifts, considerable histori- cal overlap has occurred. All three models currently coexist, and each pertains to different circumstances. The Organic Interconsumer Influence Model Early scholarship established WOM as a significant social force, influencing early marketing thought and practice. For example, Ryan and Gross���s (1943) diffusion study sug- gested that conversations among buyers were more impor- tant than marketing communications in influencing adop- tion (see also Rogers 1962). We refer to the earliest and simplest understanding of consumer WOM as a model of organic interconsumer influence (see Figure 1, Panel A). These interconsumer communications pertain to the exchange of product and brand-related marketing messages and meanings. In this model, WOM is ���organic��� because it occurs between one consumer and another without direct prompting, influence, or measurement by marketers. It is motivated by a desire to help others, to warn others about poor service, and/or to communicate status (Arndt 1967 Engel, Kegerreis, and Blackwell 1969 Gatignon and Robertson 1986). Views of WOM in this model assume that 72 / Journal of Marketing, March 2010 WOM occurs naturally among consumers when marketers perform their job of developing market innovations and per- forming effective product notification through advertising and promotions (Bass 1969 Whyte 1954). The Linear Marketer Influence Model As marketing scholarship and practice advanced, theories of WOM began to emphasize the importance of particularly influential consumers in the WOM process (e.g., Feick and Price 1987 King and Summers 1976). Accordingly, it was in marketers��� interests to identify and attempt to influence these influential, respected, credible, WOM-spreading con- sumers. This understanding now incorporates an active attempt by the marketer to influence consumer WOM through the use of traditional means, such as advertising and promotions. Therefore, we refer to this stage as a model of linear influence (see Figure 1, Panel B). Occurring dur- ing the ���cultural engineering��� marketing practices of the post���World War II era, which were formed to overcome increasingly resistant buyers (Holt 2002), some consumers were viewed as potential ���opinion leaders��� who smart mar- keters could target and influence. Marketers would now be able to work through ���the friend who recommends a tried and trusted product��� rather than the ���salesman who tries to get rid of merchandise��� (Dichter 1966, p. 165). Accurate, ���realistic information��� in marketing was important in these early conceptions because the opinion leader was assumed to transmit marketing messages more or less faithfully, without substantially altering them or having them altered by ongoing communications with other consumers (Brooks 1957 Engel, Kegerreis, and Blackwell 1969 Katz and Lazarsfeld 1955). The Network Coproduction Model The next stage of understanding is the most recent, and though it coincides with the development and recognition of the importance of the Internet, it is not limited to this domain. Marketers have become interested in directly man- aging WOM activity through targeted one-to-one seeding and communication programs, with the Internet allowing unprecedented new levels of management and measurement of these campaigns and new professional organizations allowing the efficient development and diffusion of WOMM knowledge. Marketing scholarship has evolved from a transaction orientation to one based on relationships (Vargo and Lusch 2004), with increasing importance placed on the role of consumer networks, groups, and communities (Cova and Cova 2002 Hoffman and Novak 1996 Mu��iz and O���Guinn 2001). Consumers are regarded as active coproducers of value and meaning, whose WOM use of marketing commu- nications can be idiosyncratic, creative, and even resistant (Brown, Kozinets, and Sherry 2003 Kozinets 2001 Mu��iz and Schau 2005 Thompson and Sinha 2008). Thus, WOM communications are coproduced in consumer networks. There are two distinguishing characteristics of this new model of understanding (see Figure 1, Panel C). First is marketers��� use of new tactics and metrics to deliberately and directly target and influence the consumer or opinion FIGURE 1 The Evolution of WOM Theory A: The Organic Interconsumer Influence Model B: The Linear Marketer Influence Model C: The Network Coproduction Model Marketer Marketer Marketer Consumer Consumer Marketing- Mix Elements Marketing- Mix Elements Marketing-Mix Elements Consumer Opinion Leader Consumer Consumer Consumer Consumer Consumer Consumer Direct influence (seeding, one to one) Indirect influence (advertisements, promotions) Marketing message and meanings Marketing message and meanings Marketing message and meanings

leader. Second is the acknowledgment that market messages and meanings do not flow unidirectionally but rather are exchanged among members of the consumer network. Yet, despite awareness of the complexity of these com- munal relationships, marketers are just beginning to under- stand the formation, reaction, and effects of communally based marketing promotions. This article���s contribution is based on empirical inquiry that attempts to further develop the understanding already captured in the coproduction model and to answer the following three questions: How do communities respond to community-oriented WOMM? What patterns do WOM communicator strategies assume? and Why do they assume these patterns? Method We explore the lived phenomenon of WOMM in a natural- istic context. Our research approach is qualitative and aims to generate scientific propositions about this new phenome- non that can be subject to further testing and verification. Our investigation focuses on a blog-based campaign in six North American cities. The blog context is highly relevant to our study because blogs have been increasingly popular sites of WOMM campaigns (Kelly 2007 Rettberg 2008 Sernovitz 2006). Estimates by eMarketer placed U.S. blog advertising at $283 million in 2007 and suggest that 50% of all Internet users are regular blog readers, a figure the firm predicts to rise to 67% by 2012 (Kutchera 2008). Further- more, 2006 European surveys indicate that blogs ���are sec- ond only to newspapers as a trusted information source��� (Brown, Broderick, and Lee 2007, p. 16). We studied a ���seeding��� campaign���that is, a campaign in which the product is placed among influential consumers so that they can communicate favorably about it to other consumers (see Balter 2005). The campaign was conducted by one of the pioneer North American WOMM firms (���Buzzablog���) for a major global technology manufacturer (���MobiTech���).1 The campaign, designed to promote a new camera-equipped mobile phone (the ���MobiTech 3839���), seeded the phone as well as accessories and a usage tutorial with 90 bloggers who had previously been contacted by e-mail and telephone and screened for consideration. Influ- ential bloggers were chosen on the basis of the lifestyle- related relevance of their blog content and blog traffic of at least 400 unique visitors per day. Although participants were encouraged to blog about the phone, the campaign did not require them to do so. Most of the products were seeded in two large cities with a combined population of more than seven million people. The campaign yielded significant activity. Of the selected bloggers, 84% mentioned the phone in their blog. In an interview with the Buzzablog director who supervised the campaign, he remarked on its success: I���d say we were pleasantly surprised how many of the influencers that we engaged actually participated in some Networked Narratives / 73 way along the lines I was just describing, by creating con- tent online about the product [recommending it and inspiring others to product purchase]���. We achieved what we wanted to in terms of influencers talking about the MobiTech brand and about this product to a very high degree. On the whole, the commentary about the product was very positive. So people who were talking about the properties of the product were very positive on it. And we know from follow-up surveys with all of these influencers that we had a significant number of sales that were actu- ally influenced by these influencers, as a result of recom- mending the product. So, in all those ways, we got the results that we had hoped for and that MobiTech had hoped for. (personal communication, March 28, 2009) However, if marketers want to fully grasp the effects of their WOMM strategies, it is necessary to look beyond mea- sures of communication frequency or valence and consider its content. To do this, we conducted an online ethnography, or netnography (Kozinets 2002). We followed Kozinets���s (2007) recent recommendations for adapting the netno- graphic technique to blog content. Although we observed these blogs without posting to them, we participated in sev- eral discussions with Buzzablog about the campaign. We also met to discuss and analyze our ongoing individual par- ticipation in online communities, as well as our readership of, participation in, and (in one case) authorship of blogs. Although we did not participate directly in the focal blogs, our analysis reflects the participative component that is a hallmark of interpretive depth in both ethnography and netnography. Of the 90 bloggers who were seeded in the campaign, we collected data from the 83 whose blogs remained acces- sible throughoutthe entire length of this study.These bloggers ranged in age from 22 years to 45 years at the initial time of the campaign, and 59% were male. They also had a range of occupations, including photographers, designers, writers, programmers, consultants, and administrative personnel. To establish a baseline reading of the blogs��� content, we collected blog entries beginning three months before the product seeding. We collected data from the start of the campaign until immediately after it to establish initial reac- tions. We continued to follow the bloggers and their post- ings for a period of approximately three months after the WOMM campaign. This longitudinal approach allows for analytic depth and enriches the effort to develop relevant theory. Total downloaded data amounted to more than 4300 single-spaced pages of 12-point-font text, representing approximately 1,376,000 words and 6722 postings, as well as significant additional amounts of visual and audiovisual data. The total number of postings in which the bloggers wrote about the phone or the campaign amounts to 220. Our data set contains approximately 700 comments of blog readers who responded to the WOMM campaign. We sorted and classified all postings into individual blogger files, and we categorized the files as before, during, and after the WOMM campaign. We coded the data into ini- tial categories, analyzing them for themes relevant to our investigation of WOMM. Through group comparison and multiple rounds of in-person discussion, a grounded, metaphorical, and hermeneutic interpretation emerged. We identified recurrent social and cultural tendencies within the 1Although we realize the limitations of pseudonyms in the age of the Internet search engine, we follow convention and use pseu- donyms for the names of community participants, as well as the name of the technology company, its product, and the marketing agency.