Facebook webcare strategies:
How do webcare strategies differ between companies that offer goods and companies that offer services?
Elena Schröck Student ID: 10967788
Master's Thesis
University of Amsterdam
Graduate School of Communication
Master's Programme Communication Science Supervisor: Sandra Zwier
Abstract
More than 80% of the Fortune 500 companies have their own Facebook page nowadays. Apart from the positive effects that corporate social media can have for customers’ engagement with a company and its brands, social media can also become a platform for consumers to publicly share their dissatisfaction. Especially service companies may be vulnerable to negative consumer posts as consumers rely heavily on other peoples’ opinions when it comes to evaluating the mostly intangible attributes of services. In order to protect the corporate reputation, it is crucial how companies handle such negative posts. The present study investigated differences in the reaction time, complaint handling strategies, and the usage of conversational human voice by services and goods companies in their reactions to consumer posts on their corporate Facebook pages. In total, 600 consumer posts and the companies’ reactions were analysed by means of a content analysis, which were derived from the corporate Facebook pages of six services and six goods producing Fortune 500-listed companies. Findings showed that the larger share of consumer posts on the Facebook pages were negative in tone, and that companies tended to react faster to positive than negative consumer posts, whereby reactions were mostly marked by a high share of apology making and ‘closeting’ (request to remove critical discussion to a private channel such as telephone or email). Further, service companies indeed received a higher share of negative consumer posts than goods companies, to which they reacted faster and made more apologies than goods companies. Goods companies however made more use of conversational human voice in their reactions. Future studies could integrate qualitative data in the analysis to get a better idea of the significance and organisation of webcare departments in services versus goods companies. Keywords: Electronic word of mouth, webcare, complaint handling, conversational human voice
Facebook webcare strategies
“Am I too fat for your everyday range? Should I just accept that accessible and affordable high street and on-trend fashion isn't for people like me?” This is how the British student Ruth Clemens approached H&M on Facebook to vent her anger about the unrealistic jeans sizes of the Swedish fashion company (H&M Facebook, 2016). The community reacted promptly – and overwhelmingly: More than 88.000 Facebook-‘likes’ and 11.000 ‘shares’ demonstrated the support of the Facebook community for the concern of this young woman. Within only five days, Ruth Clemens did not only get the approval of so many other Facebook users, she also got the attention of inter alia the women magazine Cosmopolitan and the Internet media company BuzzFeed for her concern (Allan, 2016; Spary, 2016). This example shows the immense impact a single post on the Facebook page of a company can have.
With the growth of social media, consumers are as powerful as never before to raise their voices and to express their opinion about organisations (Hennig-Thurau, Wiertz & Feldhaus, 2015). By placing their opinions on social media, grateful as well as disappointed, consumers do not only share their feelings with the company, but also with other social network members. This means that consumers’ opinions about an organisation nowadays are no longer only determined by direct experience or traditional media such as newspapers, but also by reviews provided by other consumers through various social media channels. In this context, electronic word-of-mouth (eWOM), meaning any statement of a consumer towards an organisation via the Internet (Hennig-Thurau, Gwinner, Walsh & Gremler, 2004), gets increasingly popular. Consumers mainly engage in brand-related social media activities because they want to exchange information about products, brands and services (Muntinga, Moorman & Smit, 2011).
Although eWOM could spur a company’s reputation, it may also cause serious reputation damage for a company or brand. Particularly negative word-of-mouth (nWOM) can be a risk for organisations as consumers are enabled to share their complaints with a large audience (Gruen, Osmonbekov & Czaplewski, 2006). Thus, in response to the development of eWOM, organisations are challenged to make use of ‘webcare’, which can be defined as the “the act of engaging in online interactions with consumers, by actively seeking out the web to address consumer feedback” (Van Noort & Willemsen, 2012, p. 133).
Engagement on social media is relevant for most companies that want to keep pace with developments of the media landscape. Extant research has among others targeted the effect of webcare on brand-generated versus consumer-generated platforms on brand evaluations (Van Noort & Willemsen, 2012), the impact of personal webcare on consumer engagement (Schamari & Schaefers, 2015) and motives for consumers to engage in nWOM (Willemsen, Neijens & Bronner, 2013). However, it is not as yet known if webcare strategies differ with certain company characteristics. In order to gain more insight into how webcare meets current challenges, it is imperative to consider certain company aspects. Some companies offer tangible goods with observable attributes and qualities and some offer services that cannot be as easily observed or qualified. eWOM can be assumed to be more relevant for services than for goods companies, because consumers are more dependent on experiences of other consumers to gauge the quality of services (Willemsen, Neijens, Bronner & De Ridder, 2011). To our knowledge no study so far investigated webcare strategies of companies that offer services versus goods by analysing corporate Facebook comments to consumer posts. The present study aimed at giving answers to the following research question: How do the webcare strategies differ between companies that offer goods and companies that offer services?
By analysing the webcare strategies of a selection of Fortune Global 500 companies on Facebook, the present study strived to give insight into differences in webcare strategies employed by goods versus service companies, with a focus on: a) how fast organisations react to consumer posts, b) what they say, and c) how they formulate their reactions. By doing this, comparisons can be made between how some of the world’s largest goods companies versus service companies deal with current webcare challenges.
Theoretical framework
One cannot deny that social media have become an integral part of our lives in Western societies. Social media have not only influenced the way people communicate, but also parts of their daily routines. As numbers of network members show, many people worldwide have integrated social network sites (SNS) in their everyday life: The photo and video sharing platform Instagram for instance is used by 400 million people around the globe. Twitter has as many users as the USA inhabitants, namely around 320 million. With
worldwide around 1.55 billion monthly active users, Facebook is currently the largest SNS and is also a major platform for business communication (Lillqvist & Louhiala-Salminen, 2014). In the Netherlands, Facebook has around 9.4 million users with 6.6 million of them using SNS on a daily basis (Rietberg, 2015). Especially the fact that social media thrive on interaction is an important opportunity for companies. On social media, companies can communicate with their consumers fast, directly and on eye-level.
eWOM, nWOM and what this means for webcare strategies
As a consequence of the rise of SNS, consumers' opinions about an organisation are no longer shaped by direct contact with the company and via traditional mass media alone, but also by the social media activities of other consumers through several social media channels.
Statements of consumers towards an organisation via the internet are often referred to as ‘e word-of-mouth’ or eWOM (Hennig-Thurau et al., 2004). eWOM does not only have positive implications for companies: If consumers are not content with a product, brand or service, they are nowadays as powerful as never before to raise their voice and to express their opinion publicly (Hennig-Thurau et al., 2015). This presents a high risk for organisations: The fact that consumers can share their dissatisfaction with a high number of people in a very short time can cause serious reputation damage to a company or brand (Hornik, Satchi, Cesareo & Pastore, 2015). Therefore, companies are advised to be very attentive towards nWOM (Gruen et al., 2006) by applying an effective webcare strategy.
A study that investigated trends in consumer statements on Twitter, in the following referred to as ‘tweets’, comes from Jansen, Zhang, Sobel and Chowdury (2009). The researchers analysed 150,000 tweets that contain branding comments, sentiments and opinions. When it comes to the valence of tweets, they found that more than 60% of the sentiment tweets (tweets that contained an opinion about a brand, company, product or service) were positive in tone, while only around 22% were negative. Despite the fact that positive statements seem to play a major role in eWOM, it needs to be considered that positive and negative statements cannot necessarily be evaluated by the frequency of
occurrence alone: Apart from the fact that nWOM can cause a high risk for a reputation image as consumers disseminate negative content to more people in a more elaborate manner than they do that with positive information (Hornik et al., 2015), research has also found that nWOM carries more weight for consumers (Hennig-Thurau et al., 2015). This can be
explained by the so-called “negativity bias” (Skowronski & Carlston, 1989) that describes the tendency of people to give greater weight to negative things than to positive ones.
The underlying mechanism is best explained by the Prospect Theory of Kahneman and Tversky (1979) that states that in general people assign more importance to negative statements than to positive ones. In the context of consumer decisions, this means that consumers consider it more important not to suffer from a bad choice than to benefit from a good choice.
To summarize, it can be assumed that nWOM will carry more weight for consumers than positive word-of-mouth (pWOM) as “good news travels fast, bad news travels faster” (Hornik et al., 2015). Therefore, to avoid bad news to travel fast in the spheres of social media, it can be expected that organisations will react faster on negative than on positive Facebook posts from consumers:
H1: Companies react faster on negative Facebook posts than on positive posts. Webcare strategies: Company form
Could it be that there are differences in webcare strategies of companies depending on what they supply? As some companies offer tangible products with observable attributes and qualities and others provide services that cannot be as easily observed or qualified, the distinction between goods and services seems of relevance in the study of webcare strategies. In the following, it is described how and why the webcare strategies of goods versus service companies may differ.
A study that investigated different webcare strategies in response to consumer tweets about goods and services comes from Hornikx and Hendriks (2015). They conducted a quantitative content analysis of 1,920 tweets that Dutch consumers posted on their personal Twitter accounts, targeting 12 Dutch companies that offer services and 12 that offer goods. The first finding was that most of the investigated tweets (66.72%) were not sentiment tweets, meaning that they did not include an opinion towards the brand.
If they did so, there was no significant difference between the percentage of positive and negative tweets targeting the two company types. However, when comparing companies that offer services versus goods, the researchers found that Twitter is used more prominently to post negative opinions about services than about products. The present study aimed at investigating if the same holds true for Facebook posts, supposing that:
H2: There are more negative Facebook posts targeting companies that offer services than companies that offer goods.
As mentioned above, goods are tangible products with a rather stable quality while quality in service is more heterogeneous, making it more likely that people will rely on
reviews when they consider using a service than when purchasing a good. This is because one of the motivations for consumers for online opinion seeking is risk reduction (of making the wrong choice) (Goldsmith & Horowitz, 2006), and the risk of choosing the wrong service is higher than choosing the wrong product. A study by Willemsen, Neijens, Bronner and De Ridder (2011) showed earlier that nWOM has a stronger impact on the evaluation of experience products (products with intangible properties that need to be experienced to be evaluated, in the present study comparable to services) than on search products (products with more concrete properties, information can be gathered before using the product, in the present study also referred to as goods). When the quality is not directly observable, consumers have a greater chance of making the wrong decision so that they rely more heavily on reviews of other consumers. As a result, companies that offer services can be expected to want to react steadily and fast on Facebook posts when consumers express their disfavour. Therefore, the following hypothesis was formulated:
An additional question is what companies should say in response to negative consumer feedback on Facebook. Bansal and Zahedi (2014) found evidence for the relevance of apology in webcare. They investigated the effects of several webcare response types, namely apology, denial and no response, on rebuilding trust of customers. They identified apology as the most reparative of the three response types, followed by denial and no response, and emphasized that the more information a company gives to the customer during webcare, the higher the repaired trust will be. The researchers explained the positive effect of apology by the fact that apology provides a perception of fairness, sincerity and respect and thereby gives the recipient the feeling of being valued.
Gelbrich and Roschk (2010) conducted a meta-analysis of prior empirical findings regarding WOM and post-complaint handling strategies. The researchers could identify three main types of organisational responses to consumer complaints, namely compensation, favourable employee behaviour and organisational procedures. Thereby, ‘compensation’ can refer to both tangible benefits such as financial benefits but also to “intangible responses outcomes” (p. 26) such as an apology by a company to express its regret for a failure. The researchers stated that an apology can compensate for the social loss (like the loss of self-esteem) that a service failure often entails. ‘Favourable employee behaviour’ means empathic and friendly behaviour of the employee who is in charge of the webcare duty. Lastly,
‘organisational procedures’ refer to policies and structures an organisation applies when handling a complaint. For the present study, the insights of Gelbrich and Roschk (2010) regarding compensation are of special interest. As stated above, compensation does not always need to be an economic compensation, but can be also be psychological. The psychological loss is aimed to be compensated by an apology on the part of the company.
Tax, Brown and Chandrashekaran (1998) found a special relevance of apology when consumers evaluated the fairness of the outcomes of their complaint, whereby consumers were also found to include the psychological costs of complaining in their evaluation of an appropriate compensation. The researchers therefore argue that not only economic cost, but also emotional cost need to be considered in customer care.
To conclude, apology has been shown to be one of the most important components of a successful webcare strategy in previous research about service companies’ complaint handling strategies. It is predicted that these results can be transferred to webcare responses on
Facebook:
H4: Companies that offer services make more use of apology in their Facebook webcare responses than companies that offer goods.
After considering the ‘what to say’ in response to nWOM, it is finally also important to consider how to say it. Searls and Weinberger (2001) initially launched the concept of
conversational human voice (CHV) as part of their ‘markets as conversations approach’. Within this approach, there is an emphasis on building relationships via conversational communication with audiences rather than making audiences the target of promotional marketing messages. In 2009, Kelleher defined CHV as “an engaging natural style of organisational communication as perceived by an organisations public based on interactions between individuals in the organisation and individuals in publics” (p. 117). According to Van Noort, Willemsen, Kerkhof and Verhoeven (2014), CHV consists of three components, namely message personalisation, informal speech and invitational rhetoric.
Message personalisation as the first component of CHV means the degree to which a message addresses a specific individual, e.g. by using second-person pronouns such as 'you' and identifying the receiver by direct address.
Using personalised messages, e.g. by acting as a human-being that is responding on behalf of the company, is perceived as more human and leads to higher willingness to engage in
interpersonal communication both for Twitter (Rybalko & Seltzer, 2010) and for blogs (Lee, Hwang & Lee, 2006). The positive effect of message personalisation on building relationships with stakeholders was also shown by Pollach (2005). He claims that using first-person
pronouns such as 'I' or 'us' signals that the company is communicating personal beliefs and that will reduce the impersonal character of mass communication via the Internet.
The second component of CHV, informal speech, means using a casual and expressive language that resembles spoken language, as well as abbreviations and non-verbal cues such as emoticons. In contrast to formal messages that are restricted to a language associated with corporate language, informal messages remind more of human-to-human speech.
Lastly, invitational rhetoric describes a style of communication that is oriented to an exchange of ideas and opinions with a company's stakeholders, expressed for example by saying “Let us know what you think”. Although the effects of informal speech and invitational rhetoric have not been investigated so far, it can be assumed that a high level of informal speech and invitational rhetoric can make up for the impersonal computer-mediated character of a message and thereby reduce the impersonal character of the companies.
Generally speaking, a high degree of CHV in organisational communication has been shown to have a positive effect on the ability of a company to foster trust, commitment and satisfaction (Kelleher & Miller, 2006; Kelleher, 2009). Furthermore, Van Noort and
Willemsen (2012) stress that brands engender more positive brand attitudes and higher purchase intentions when they are perceived as using a high level of CHV in their webcare responses.
Also Sweetser and Metzgar (2007) stressed the positive effects of a human tone of voice by showing that applying human communication, especially in crisis situation, signals that the company cares and that they are open and willing to solve the problem.
As was stated above, services are characterized by non-tangible qualities so that the risk of choosing the wrong service is higher than choosing the wrong product. Therefore, creating a good relationship with the customers that shows them that the company cares in case of any dissatisfaction is crucial for organisations that offer services. Hence the following hypothesis was formulated:
H5: Companies that offer services apply a higher level of CHV than companies that offer goods.
Method
In order to gain insights about current webcare strategies of goods and service companies, the present study reflects a content-analysis of the consumer posts and the
company responses on the social media sites of six large companies that offer services and six companies that offer goods. The materials were derived from corporate Facebook pages of selected organisations in different industries.
Selection of social network site: Facebook
The platform Facebook was mainly used because of two reasons: First, Facebook is currently the largest social network site with estimated one billion users around the globe and around 40 million business pages in 2015 (Ha, 2015). In the US, around 85% of the
companies with more than 100 employees have their own Facebook page (Smith, 2016). But also small businesses do not want to forego the immense reach of Facebook: In 2011, a worldwide survey of 1,972 small businesses showed that 96% of them use Facebook as a marketing tool (Cohen, 2011).
Considering the Fortune 500 companies, a study of the University of Massachusetts Darmouth Center for Marketing Research revealed that in 2014, around 80% of the Fortune 500 companies were on Facebook, with an increase of 10% compared to 2013 (Shively, 2014). Therefore, it could be expected that the percentage of Fortune 500 companies with a corporate Facebook page would be even higher in 2016.
Furthermore, Facebook offers a high degree of interaction. Social network members can ‘like’ corporate pages in order to receive news and updates from the company in their own newsfeed. Additionally, they can actively communicate with the company by posting on the company’s Facebook 'wall' (Dekay, 2012) or they can comment on a company posts. If visitors initiate a post, it appears on the left hand side under 'visitor posts', as the timeline of the particular company can only be posted by the page administrator (Facebook Advertiser Help Centre, 2016). Moreover, most companies allow customers to send private message by using the 'message' button, although this function can also be removed by the site
administrator (Facebook Help Centre, 2016a). Selection of companies
The companies that were considered for the present study are all ranked in the global Fortune 500 list. This list provides an overview of the top 500 corporations in the world measured by revenue (Fortune, 2015). By choosing only Fortune 500 companies it was ensured that the companies under study are well known and have a large stakeholder base with a likely high level of interaction on the wall. 12 organisations were randomly selected from the Fortune 500 list version 2015, thereby ensuring that six were companies that primarily offer services, and six were companies that primarily offer goods.
After the first sampling of twelve companies from the Fortune 500 list, it was tested if the companies have a corporate Facebook page with posts in English.
By default, the official Facebook page was considered that is mostly for the country where the company was founded. If the original language of the company was not English, the site for United Kingdom (UK) was taken into consideration by choosing the option “switch region” and selecting UK. From the original sample, only the pages of E.ON and Vodafone were not exclusively in English. As E.ON is a German company, the official Facebook page 'E.ON SE' contains some German visitor and company posts, although the page is mainly in English. Therefore, the region was switched to UK. The same holds true for the Vodafone group that does not have an overall corporate page, but separate pages for every country where it is operating. Therefore, the page for UK was selected. UPS, McDonalds, Southwest Airlines, Coca Cola, Foot Locker, GAP and General Motors are all American companies, which is why their official Facebook page is in English. Barclays and Tesco are British, hence there was no need to switch the region. Although Nestlé was founded in Switzerland, the company’s official Facebook page is in English.
Another condition used for the Facebook site was that it is a verified company page, meaning that “Facebook confirmed that this is an authentic page for this public figure, media company or brand” (Facebook Help Centre, 2016b). Thereby, it was ensured that the site is organised by the company itself and not by consumers or other parties. All the chosen
companies had a certified corporate Facebook page. Other information such as the page traffic or the number of ‘likes’ that a company has, are summarized in Table 1 for the service
companies and in and Table 2 for the goods companies.1 The exact links to the Facebook pages of each company that was analysed can be found in Appendix A.
1
Note: Average number of Facebook posts and estimated number of visitor posts were collected for the time from the 26th of April 2016 to the 26th of May 2016. Number of Facebook fans as of 1st of June 2016.
Table 1
Selected companies that offer services and main characteristics of their Facebook (FB) page
Company Sector Creation
of FB page Ø number of company posts/week Ø number of visitor posts/week Facebook fans E.ON Energy 4/2012 1 117 59.139 Vodafone Telecommunicatio ns 3/2009 2 300 comments on company posts 888.204
UPS Freight Delivery 6/2011 2 56 1.608.312
Barclays Finance & Banking 9/2011 1 110 573.366
McDonalds Food Services 3/2010 3 185 64.840.788
Southwest Airlines
Airlines 7/2008 7 630 5.039.995
Table 2:
Selected companies that offer goods and main characteristics of their Facebook (FB) page Company Sector Creation of
FB page Ø number of company posts/week Ø number of visitor posts/week Facebook fans
Coca Cola Beverages 12/2008 1 per month 45 97.724.936
Tesco Food
Retailing
6/2011 3 840 2.058.929
GAP Apparel 8/2009 5 76 7.845.200
Nestle Food Processing 4/2008 2 25 8.402.375 General Motors Automotive 5/2009 6 128 701.730
Research design
To get insights into how companies deal with webcare challenges, a quantitative content analysis of consumer posts and the companies’ responses on their Facebook pages was conducted. As a content analysis is defined as a “research technique for the systematic classification and description of communication content according to certain usually predetermined categories” (Wright, 1986, p. 125), it can be seen as an optimal method to analyse the content of media messages.
Sample
50 consumer posts on the Facebook page of each of the 12 companies were subject to the investigation, making a sample size that consisted of 600 Facebook posts. For Vodafone, the section 'visitor posts' did not exist as the company does not allow consumers to initiate posts. Therefore, comments to posts that Vodafone published on its wall were taken into consideration.
Starting with the most recent consumer post on the company Facebook page, every third post was included in the sample. This method was followed in order to avoid a heavy influence of timely occurring events. The material was collected and preserved by making screenshots of the relevant posts to avoid possible modifications after the posts were published and analysed.
Observed variables
The observed variables were defined by considering the hypotheses of the present study. In the following, the main variables of the codebook are described. The full codebook can be found in Appendix B.
Valence of consumer post
Coding the valence of the consumer post was crucial because some hypotheses aim only at giving insights about reactions to negative consumer posts. A consumer post was coded as either 1 = positive, 2 = negative or 3 = neutral towards the company's service, products or employees. If the post was not connected to the company, the option
4 = unrelated was chosen. The consumer post was coded as positive if it included a positive opinion towards the company, e.g. expressed by words such as 'happy', 'content', 'good'. A negative post on the other hand contained critique or a negative opinion towards the company. Mostly, words such as 'disappointed' or 'bad' expressed consumer dissatisfaction. A neutral post could not be identified as clearly positive or clearly negative because it contained e.g. a question that aims at asking for information such as “How long is you store opened on a Sunday?”. An example for an unrelated post on the corporate page of Vodafone was the statement: “Leister city all the way” which was referring to the surprising victory of Leicester City in the English Premier League in the beginning of May 2016.
Company reaction to consumer post
The next coding variables concerned the company’s reaction to the consumer post: Company reaction to consumer post: Response of the company. The possible coding options were 1 = Company reacted on consumer post and 2 = No reaction of the company to the post. If the company did not react to the post, the coding was stopped.
Company reaction to consumer post: Date and time of the post. In order to be able to compare the response time, the date was coded by recording the date of the consumer post in the format dd.mm.yyyy.
Company reaction to consumer post: Date and time of the company response. By coding the same information for the company response, the reaction time could be derived by subtracting the time the post was made by the time of the response. The date and time of the company response was coded in the same manner and format as the date and the time of the consumer post.
Complaint handling
The complaint handling variables give information about how companies reacted to negative consumer posts and that is why this variable was only coded when the valence of the consumer post was identified as negative in the previous coding. Complaint handling was measured by counting the number of times a specific complaint handling strategy occurred in the company reaction. This way it was ensured that it was not only analysed if the complaint handling strategy was used, but also to what extent. E.g. if a company apologised twice in one response, the apology option was coded as 2. The different complaint handling strategies in the code book were:
Complaint handling: Denial. If a company gives the fault of the failure to another party or questions the accuracy of a complaint, it avoids taking the responsibility for the
dissatisfaction. Thus, if an organisation reacted by rejecting the blame, it was coded as denial. Complaint handling: Apology. Apology can be expressed in several ways. The company could express its regret for an incident by saying 'sorry', 'we apologize' or 'excuses'. As with the other complaint handling options, apology was coded by counting the number of apologies in one reaction as in some cases the company apologized more than once in a reaction.
Complaint handling: Economic compensation. Compensation includes the willingness of a company to replace the product or service a consumer is not content with.
Also, it can mean that it tries to make up for a service failure by sending a refund, voucher or a gift to the unhappy customer.
Complaint handling: Justification. If a company explains why it came to the incident that caused the dissatisfaction, it is identified as justification. The company gives reasons for the failure by e.g. stating that the corporate policy or other circumstances forced it to a certain action.
Complaint handling: Closeting. Sometimes, a company attempts to move the dialogue with the consumer to another private channel, presumably in order to avoid the public to follow the conversation or to save personal data. If the company asked the consumer to get in touch via a private channel such as telephone or email, it was considered as closeting. Also, asking to write a personal message within the same medium was considered a form of closeting the public conversation.
Conversational human voice (CHV)
Similar to complaint handling, the use of conversational human voice (CHV) was coded by counting the several elements in the company reaction. As described in the theoretical framework, CHV consist of message personalisation, informal speech and
invitational rhetoric. For a better understanding of the concept of CHV, Figure 1 illustrates the three elements of CHV and their components.
Figure 1. Composition of CHV.
Respectively, the coding elements of message personalisation concerned:
CHV: Use of second- and first-person pronouns. When it comes to the usage of pronouns, the number of first- and second pronouns in one comment was counted.
CHV: Identification of the company representative. The company representative can be identified either by the name or a picture, both or none. If a picture or the name was present, the value for the variable was 1, if a picture and a name of the representative was shown, it was coded as 2 and if none occurred, it was coded as 0.
CHV: Addressing the consumer. Addressing a consumer includes for instance using the consumers name to greet him or her. If the company addresses the consumer with 'Hey' plus the name, it was coded as 2. Only mentioning the name without greeting the consumer was coded as 1. If the company did not mention the name nor greeted the social network member, it was coded as 0.
CHV: Informal speech consists of the use of contractions, abbreviations, non-verbal cues and interjections. As before, the number of contractions (e.g. yr instead of your), abbreviations (pm for personal message), non-verbal cues (emoticons such as :)) and interjections (e.g. oh, wow) in the company post was counted. For instance, if the company answered a consumer posts by writing: “Hey Betty! Oh, that sounds awful! :( Please write us a pm so we can get back to you asap :)”, the response was coded as follows regarding informal speech elements: contractions = 0, abbreviations = 2 (pm and asap), non-verbal cues = 2 (two emoticons), interjections = 1 (oh).
CHV: Invitational rhetoric includes the willingness of the company to listen to the consumers concerns (“Let me know what you think”), asking for the consumers opinion (“How does that sound for you?”) and about further details (“Which store are you referring to?”). As for the other CHV components, the number of how often a component occurred was collected. For instance, if a company asked in its comment: “When did you order the coffee? Which McDonalds are you referring to?”, the variable asking for details was coded as 2. Intercoder reliability
Prior to coding, a pre-test was done to test the applicability of the codebook. After that, 10% of the sample was selected (N = 600, n = 60) for a second coder in order to test the extent to which independent coders come to consistent coding results (Tinsley & Weiss, 1975). The variables that were taken into consideration are Valence of consumer post, Response of the company and Complaint handling: Apology. The values for Krippendorff's alpha and the percentage agreement for the three variables are presented in Table 3. All the three chosen variables reported a satisfactory intercoder reliability of at least α = 0.80, which indicates a high agreement between the two independent coders. Therefore, the coding can be considered as showing high intercoder reliability.
Table 3
Values for Intercoder reliability test
Variable name KAlpha %agreement
Valence .97 98,3%
Response of the company 1.00 100%
Complaint handling: Apology .94 96,6%
Results
The first part of the results section provides some general descriptive findings that are relevant for answering the research question. Thereafter, an overview over the analysed hypotheses will be presented. For all statistical tests, an alpha level of .05 was applied. General results
Apart from the insights that the hypotheses deliver, some general results regarding the consumer posts and company’s reactions will be presented first. To begin with, of all the analysed consumer posts (n = 600), the highest percentage were negative in tone (68.3%), followed by neutral (18.7%), positive (10.8%) and unrelated posts (2.2%). Figure 2 illustrates the valence of the consumers posts in the sample. Further, of all the analysed posts from consumers on the company websites 76.7% (n = 460) were answered by the companies, resulting in 23.3% unanswered posts. When it comes to differences between service and goods companies in this matter, it can be said that service companies answered 80% (n = 240) of the consumer posts while goods-producing companies answered slightly less, 73.3% (n = 220) of them.
Figure 2. Valence of consumer posts for both company types in percentages.
Considering the reaction time, Table 4 provides an overview of the average time the different companies needed to respond to the consumer posts2. The values are listed by service and goods companies separately, starting with the lowest average reaction time. It can be seen that all the considered companies, except Nestlé, answered at least half of the posts that were published on their corporate Facebook page. Of the service companies, Barclays answered the most (94%) posts in the shortest time (M = 0.14). For the goods companies, Foot Locker scored best in the reaction time (M = 0.46), and Tesco answered almost every post (96%).
Table 4
Number of posts and average reaction time of each company, categorized by company type
Company name Number of answered posts % of answered posts Average reaction time in hh.mm SD Service companies Barclays 47 94% 0.14 0.13 Southwest Airlines 40 80% 2.15 3.07 McDonalds 26 52% 2.56 3.56 UPS 43 86% 5.18 15.09 EON 45 90% 6.05 13.06 Vodafone 39 78% 22.52 22.56 Cumulative 240 6.33 Goods companies Foot Locker 42 84% 0.46 1.28 Tesco 48 96% 2.37 3.41 General Motors 34 68% 13.58 10.26 GAP 34 68% 14.47 11.52 Coca Cola 38 76% 26.33 27.49 Nestlé 24 48% 34.52 44.45 Cumulative 220 13.33
Regarding the occurrence of the complaint handling strategies in the companies’ reaction to consumer posts, the most used complaint handling strategy of service companies was apology: 66.3% of the service companies’ reactions to negative posts included at least one apology element, followed by closeting with 58.9%, justification (17.1%), denial (2.9%) and compensation (1.7%).
For the goods companies closeting was the most popular complaint handling strategy (71.2%), followed by apology (50.6%), justification (21.2%), compensation (5.8%) and denial (1.3%). All in all, this means that apology and closeting were the most often-used complaint handling strategies by companies in reaction to negative consumer posts.
Furthermore, complaint handling strategies that were strongly correlated were: compensation and apology, r(332) = .16, p = .003, closeting and apology, r(332) = .11, p = .04, and closeting and justification, r(332) = -.42, p < .001). All other complaint handling strategies were not significantly correlated. The negative correlation between closeting and justification means that when companies provided a justification in reaction to a consumer post they were unlikely to invite the customer to continue the conversation outside of the public context of Facebook, whereas if they gave no justification they were more likely to invite the customer to continue the conversation outside of the public context of Facebook. That apology was positively correlated with both compensation and closeting, which means that if the degree of apology in the company reaction increased, companies were also more likely to offer compensation and invite the customer to continue the conversation outside of the public context of Facebook, and the other way around.
When it comes to the usage of the CHV elements, one variable for message
These CHV elements in turn consist of different components that could each take any value between 0 and the highest number of the respective CHV element. For example, if a comment did not include any addressing elements, the value for 'addressing', which is a component of message personalisation, was 0.
We found that the most used CHV component in companies’ reactions was message personalisation with a mean of M = 1.76 (SD = 0.78) for service companies and M = 2.11 (SD = 1.42) for goods companies. Furthermore, service companies made slightly more use of informal speech (M = 0.32, SD = 0.37) than goods companies (M = 0.26. SD = 0.34). The same holds true for invitational rhetoric (Service: M = 0.17, SD = 0.27, goods: M = 0.13, SD = 0.28). Figure 3 illustrates the different CHV elements and to which degree they were used by the two company types.
Testing Hypotheses: H1 Reaction time related to valence of consumer posts
Hypothesis 1 assumed that companies, independent from the company type, in general react faster on negative consumer posts than on positive ones. The fourth row of Table 5 presents the mean reaction time of both company types on positive, negative, neutral and unrelated consumer posts. If the company reacted on a negative post (n = 331), they did within an average time of M = 10.41 hours (SD = 20.50). The n = 43 answered positive posts received a reaction from the company within an average time of M = 4.18 hours (SD = 5.43). An independent samples t-test showed a significant difference between the reaction times to positive versus negative posts, t(372) = -1.99, p = .047. Contrary to Hypothesis 1, from this we must conclude that average reaction time was significantly faster in reaction to positive consumer posts than in reaction to negative consumer posts.
Table 5:
Reaction time depending on valence of consumer post
Valence of post Positive Negative Neutral Unrelated
Number of posts 65 410 112 13
Percentage 10.8% 68.3% 18.7% 2.2%
Number of answered posts 43 331 82 4
Reaction time M = 4.19 SD = 5.43 M = 10.41 SD = 20.50 M = 7.33 SD = 12.12 M = 53.58 SD = 37.28
A one-way ANOVA was conducted to compare the effect of all three levels of the valence of the consumer posts (positive, negative, neutral) on the reaction time. Only positive, negative and neutral posts were included in the test as the unrelated posts would not
There was no significant effect of the valence of the consumer posts on the reaction time (F(2, 453) = 2.79, p = .062).
However, as the assumption of equal variances was violated, post-hoc comparisons using the Games-Howell test were applied. This indicated that the mean reaction time for the positive posts (M = 4.19, SD = 5.43) was significantly different from the mean reaction time for the negative posts (M = 10.41, SD = 20.50), p < .001. The mean reaction time for the neutral posts (M = 7.33, SD = 12.12) did not significantly differ from the positive and negative posts conditions.
Summarized, the results of the ANOVA suggest that companies react significantly faster on positive posts than on negative ones, with the neutral posts in-between. Therefore, the first hypothesis must be rejected as the expected effect was not observed, but rather the opposite, meaning that companies react faster on positive posts than on negative ones.
Testing Hypotheses: H2 Valence of consumer posts in relation to company form
The second hypothesis stated that there are generally more negative Facebook posts that target service companies than negative posts that target goods producing companies. Table 6 contains the absolute numbers and percentages of positive, negative, neutral and unrelated posts for the service and for goods companies in our sample.
Table 6
Overview: Valence of consumer posts presented by company form
Services Goods
Absolute Percentage Absolute Percentage
Positive posts 31 10.3% 34 11.3%
Negative posts 216 72% 194 64.7%
Neutral posts 48 16% 64 21.3%
Unrelated posts 5 1.7% 8 2.7%
As documented in Table 6, 72% of the posts that were published on the Facebook page of the analysed service companies were negative posts, while the percentage of negative posts on the Facebook page of a goods producing companies was 64.7%. In order to examine the relationship between the company form and the valence of the consumer posts, a chi-square test of independence was performed. As the focus of the second hypothesis is on the
proportion of negative posts, a new binary variable was created that groups all posts that were negative versus all posts that were not negative.
The chi-square test revealed a marginally significant relationship between the company type and the valence of the consumer post, X2 (1, N = 190) = 3.73, p = .054. Therefore, the second hypothesis can be partly accepted, as there were indeed more negative posts targeting services companies than goods companies.
When it comes to the reaction time on negative posts, it was assumed that companies that offer services react faster on negative posts than companies that offer goods. In order to test the hypothesis, an independent t-test was conducted.
Table 7 shows the average reaction time to negative posts by company type. Table 7
Average reaction time on negative consumer posts by service versus goods companies
N Mean SD
Service 175 6.26 14.18
Goods 156 15.27 25.31
A t-test revealed a significant difference in the reaction times on negative posts for the service and the goods companies, t(329) = -4.02, p < .001, meaning that services companies indeed reacted faster to negative consumer posts than goods companies. Therefore, the third hypothesis can be accepted.
Testing Hypotheses: H4 Use of apology related to company form
The fourth hypothesis targets the use of apology in the company’s reactions to consumer posts. It was assumed that companies that offer services make more use of apology in their reactions than companies that offer goods. As mentioned in the general results, 66.3% of the reactions of service companies included at least one apology element. Regarding goods companies, slightly more than half of the reactions contained an apology (50.6%).3
In order to test the hypothesis, first a new variable was computed with 1 = no apology used and 2 = apology used, independent from the number of times the companies made apologies in their reaction. A chi-square test was conducted and the relation between company type and apology making was significant X2(1, N = 140) = 8.52, p = .004.
3
The considered number of posts for the service companies was N = 175 as the following conditions needed to be fulfilled: Negative valence and reaction of the company. N = 156 posts for goods companies were considered.
Therefore, it can be concluded that service companies made significantly more use of apology in their Facebook webcare responses than goods producing companies.
The previously presented result gave insight into the usage of apology per se, not into the extent to which apologies were made in the companies’ reactions. Therefore, also an independent samples t-test was conducted taking into account the number of used apology elements in the company reaction, with 0 = no apology elements, 1 = 1 apology element, and 2 = 2 apologies elements. Two apology elements in one company reaction was the highest amount of apologies in the dataset. The independent samples t-test showed that service companies (M = 0.70, SD = 0.54) made significantly more apologies per reaction than companies that offer goods (M = 0.53, SD = 0.56), t(332) = 2.78, p = .006.
Both results confirm that service companies made more use of apology in their reaction to consumer posts, why the fourth hypothesis can be accepted.
Testing Hypotheses: H5 Use of CHV related to company type
The last hypothesis proposed that companies that offer services apply a higher level of CHV in their reactions to consumer posts than companies that offer goods. In order to test this expectation, an independent samples t-test was conducted with a CHV variable that was composed by summing up the separate CHV elements as the dependent variable. As described above, CHV is composed of message personalisation, informal speech and invitational
rhetoric. For the present t-test one CHV variable was computed that includes all different CHV components. This scale could range from 0 (no CHV component present in a company reaction) to 19, which was the highest value of the number of CHV elements in the sample (Tesco for instance used 16 first person pronouns in one response to a consumer post).
The mean score for the service companies was M = 0.8 with SD = 0.41 and for the goods producing companies M = 0.9 with SD = 0.63.
The t-test showed a significant difference between the service and goods companies,
t(458) = -1.97, p = .05. As a higher mean implies applying a higher number of CHV elements, companies that offer goods made significantly more use of CHV elements in their reactions on consumer posts than companies that offer services. Therefore, the last hypothesis must be rejected as it is contrary to what was expected.
Conclusion & Discussion
The aim of the present study was to provide insights into the Facebook webcare strategies of service and goods companies. In the present conclusion and discussion section, the several hypotheses will be presented and discussed first. After that, implications for theory and practice, and limitations and ideas for future research are presented.
Critical reflections of the findings
Hypothesis 1: The first hypothesis assumed that companies react in general faster on negative online consumer posts than on positive ones. It was assumed that companies want to prevent fast-spreading of unfavourable news that can negatively impact their (online)
reputation (Hornik et al., 2015). However, the analysis revealed a contrary effect: Rather than reacting faster on negative consumer posts, companies, independent from the company type, replied faster to positive posts.
It could be the case that companies need to consider a reaction to a negative post more carefully than a reaction to a positive post. Due to the negativity bias (Skowronski & Carlston, 1989) people assign more attention to negative statements than to positive ones. A company likely is aware that the public observes its reaction to a negative consumer post very precisely and therefore plans its reaction accurately. In contrast, if a consumer publishes a positive post on the Facebook page of an organisation, the author already spreads a positive message towards the company.
Therefore, the reaction of the company does not need to be as thought-out because the risk for a reputation damage hardly exists. To summarize, when a consumer reports a negative
sentiment towards a company online, the company may want to prevent aggravating the situation by an inappropriate reaction which is why responding could take more time.
Hypothesis 2: Hypothesis 2 suggested that there are more negative posts about service companies than about goods companies. This assumption was partly accepted as there was indeed a higher percentage of negative posts for service companies observed, but with a marginally significant relationship between the company type and the valence of the posts. It was assumed that service companies are more vulnerable to negative feedback of consumers as the qualities of services are not tangible and rather hard to qualify (Willemsen et al., 2011).
The finding in this study that there are indeed more negative comments that target service companies than goods companies is in line with Hornikx and Hendriks’ (2015) earlier results. By analysing statements about brands with services and goods on Twitter, they also found that the percentage of negative tweets was higher for service companies than for goods companies.
However, the hypothesis was only marginally accepted, meaning that the difference was not very strong in the present study. A reason for that could be that consumers may think they are more likely to get a refund when they complain about a negative experience with a product than with a service. Therefore, they may use public platforms to pressure goods company more in the hope of receiving a compensation. This assumption is supported by the finding that 5.8% of the answered negative posts which targeted goods companies contained a compensation element whereas only 1.7% of the service companies responded with a possible compensation to dissatisfied consumers.
Hypothesis 3: In contrast to the first two hypotheses, Hypothesis 3 was clearly
accepted, confirming that the reaction time of service and goods companies was significantly different, in the sense that service companies reacted faster on negative consumer posts than companies that offer goods. The finding supported the assumption that service companies more than goods companies are aware of the potential loss of reputation and damage to trust coming from negative consumer posts and try their best to regain their trust by reacting fast.
Hypothesis 4: When it comes to the complaint handling strategies, it was predicted that service companies make more use of apology in their reaction to negative consumer posts than goods companies. The findings showed that the assumption can be accepted and that service companies indeed made more use of apology in their webcare responses. This confirms the need that service companies feel to apologize not only for the economical, but also for the psychological consumer losses.
Hypothesis 5: The assumption that companies that offer services will apply a higher level of conversational human voice (CHV) in their webcare reactions than companies that offer goods, was not confirmed in our study. Rather, a reverse effect was observed, meaning that companies that offer goods used more CHV in their reaction to consumer posts on Facebook.
An explanation could be that the level of CHV should be in line with the image of the company. If a company wants to be perceived as young and 'hip', they are more likely to apply a high level of CHV than a conservative company (Cummings, 2016). Therefore, it could be the case that the analysed goods versus service companies differ in the way they want to be perceived.
General discussion
Compared to previous studies, the present investigation found a relatively low number of unanswered consumer posts. Dekay (2012) found that only 17.6% of the organisations operating in the banking, software and service sector reacted to negative comments on Facebook. In contrast, all of the companies in the present sample answered at least around half of the posts that were published on their Facebook page. The contradicting results can be explained by different study designs: First, Dekay (2012) collected the information if the company had an official Facebook page, but did not exclude the company if this was not the case. Only 62.5% of the 40 examined corporations in Dekay’s study had an official Facebook page. In contrast, having an official Facebook page was a necessary condition for the
companies to be considered in the present study. The degree of professionalism and
integration in corporate structures is higher for officially organised social network pages than social networked pages organised by third parties (Van Noort & Willemsen, 2012), which could be an explanation for the different percentage of answered posts in our study versus the study by Dekay.
Second, the study of Dekay was conducted in 2012. Because social media is a fast moving medium and the relevance of webcare increased especially in recent years, it could be assumed that in the four years since Dekay’s study was published companies’ practices have changed and that companies more and more recognize the opportunities and challenges of social media webcare.
Additionally, it can be stated that the high percentage of negative posts (68.3%) for both company types means that Facebook is a popular channel for consumers to raise awareness of their dissatisfaction.
18.7% of neutral posts further lead to the conclusion that consumers use corporate Facebook pages also regularly to exchange information or ask questions without transporting an opinion towards the company.
These results contradict the earlier findings of Jansen et al. (2009) who found a higher amount of positive tweets than negative ones. Hornikx and Henriks (2015) found as many positive as negative tweets. A possible explanation is that both studies used Twitter in their investigation that is used more popularly for positive statements: Other than on Facebook, the maximum number of characters in a tweet are limited which implies that Twitter users cannot vent an elaborate complaint, for which they may turn to Facebook.
Finally, an interesting insight from our study is that ‘closeting’ (moving the discussion to private spheres, for instance by asking the consumer to send a private email to discuss the issue further) was the most popular complaint handling strategy for goods producing
companies, in contrast to service companies that used mostly apology as a complaint handling strategy. The tendency of goods companies to move the critical discussion to private spheres (‘closeting’) reveals that they do not consider the public dialogue with the consumers on Facebook an appropriate channel to communicate with them about their dissatisfaction. It could be that companies fear that a reaction to an unfavourable post could in turn initiate an expanding thread of negative statements (McCarthy, 2010). Service companies may seek the public dialogue with the consumers more as the dialogue could help other social network members to solve a similar problem. A negative statement about a product on the other hand only harms the corporate reputation publicly and does not provide any helpful information for other consumers.
Implications to practice and theory
As mentioned above, 23.3% of all the consumer posts in our sample remained without reaction from the company’s side, with more than half of the unanswered posts
(56.43%, n = 79) being negative posts. Although the main part of the posts were in fact answered by the company, it can be a risk to ignore a significant share of the consumer feedback on social media as a consumer conversation could develop publicly without control from the company over the content. As Van Noort and Willemsen (2012) demonstrated, consumers evaluate a brand more positively when being exposed to nWOM if the company responded to the consumer statement. Although the study was conducted in the context of an online forum, it can be assumed that the same effect occurs on other platforms such as Facebook.
Not only the question if companies react on a post, but also when plays an important role for organisations and their relationship with consumers: For instance, both Coca Cola and Nestlé needed on average more than 24 hours to answer the posts on their corporate Facebook page. Customers could feel not valued if a company needs days to react to their post. This is why it is often recommended that companies integrate webcare in their corporate structure as a department that monitors activities of consumers on their corporate social media pages on a constant basis (Van Noort et al., 2014). Together with analysing other channels such as
traditional mass media, in this fashion organisations can identify potential reputation risks and react accordingly in a timely fashion.
The present study analysed the differences between the reactions of service and goods companies on online consumer posts. By analysing the companies according to their company type, a new categorisation was applied.
Also, the fact that the present study studied Facebook as webcare platform, extends the existing body of research that mainly focussed on webcare strategies on Twitter so far
(Hornikx & Hendriks, 2015; Jansen, Zhang, Sobel & Chowdury, 2009; Rybalko & Seltzer, 2010). The study of Dekay (2012) who investigated the reaction of large companies of various sectors on negative Facebook comments was extended by also analysing the content of the company responses regarding their complaint handling strategies and the usage of CHV. Limitations
The first limitation of the present study regards the sample: During the coding phase, it became clear that some companies cannot be clearly seen as either a service or a goods
company. Tesco for instance, a company that is best-known as a grocery retailer and thereby a goods company, also offers a service to deliver food to customers’ homes. As a result, some consumer posts in the sample targeted services rather than goods that Tesco offers.
Furthermore, the present study did not consider any other company attributes apart from the offer of services or goods. In order to compare these two company forms, it would be ideal to select companies that are comparable in size, image and the timeframe when they were founded.
Additionally, the size and the role webcare plays within the company were not assessed. In order to properly understand differences in webcare strategies, it could also be analysed which role webcare plays in the respective companies. Is webcare integrated in the customer service or in the marketing department? Does the company have a separate webcare department? These questions can be of added value when it comes to making comparisons between the webcare strategies of different companies.
Future research
The limitations of the present study also build opportunities for future research. Future studies could integrate qualitative data in the analysis, e.g. by conducting in-depth interviews with companies in order to get a better idea of the significance and organisation of the
webcare department in the company. With the help of these additional data, more precise statements could be made about webcare strategies of goods and service companies.
Additionally, the present study focussed on consumer posts and the reaction to them on Facebook. However, it was not analysed how the discussion continued. It could be that consumers are getting even more dissatisfied if the company attempts to remove their complaint from the public domain though getting into private contact with the consumer.
To summarize, listening to stakeholders on social media and getting in touch with them is central for the reputation of the company. As the initial example of the frustrated customer Ruth Clemens and the extensive reactions show: There is relevance for future research to investigate the spheres of negative posts on social media to further improve companies’ knowledge of ways to deal with webcare challenges. Although H&M replied to Ruth's complaint and apologised for her experience, they did that more than 24 hours after the post, leaving the community enough time and space to discuss the incident publicly.
References
Allan, M. (2016, June 16). This Woman's Facebook Post Calling Out H&M for “Unrealistically Small” Sizing Going Viral. Cosmopolitan. Retrieved from http://www.cosmopolitan.com/style-beauty/fashion/news/a60019/hm-sizing-viral-facebook-post/
Bansal, G., & Zahedi, F. M. (2015). Trust violation and repair: The information privacy perspective. Decision Support Systems, 71, 62-77.
Cohen, D. (2011, November 15). 96% of Small Businesses Are On Facebook: Report. Adweek, SocialTimes. Retrieved from http://www.adweek.com/socialtimes/facebook-small-business-4/370213
Cummings, H. (2016). Finding Your Brand's Voice. Distilled. Retrieved from https://www.distilled.net/tone-of-voice/
Dekay, S. H. (2012). How large companies react to negative Facebook comments. Corporate Communications: An International Journal, 17(3), 289-299.
Facebook Advertiser Help Centre (2016). Visitor Posts. Retrieved from https://www.facebook.com/business/help/community/question/? id=10156712265320427
Facebook Help Centre. (2016a). Messaging. Retrieved from https://www.facebook.com/help/307375982614147
Facebook Help Centre. (2016b). Verified Page or profile. Retrieved from https://www.facebook.com/help/196050490547892
Fortune (2015). Fortune 500. Retrieved from http://fortune.com/global500/.
Gelbrich, K., & Roschk, H. (2010). A meta-analysis of organisational complaint handling and customer responses. Journal of Service Research, 14(1), 24-43.
Goldsmith, R. E., & Horowitz, D. (2006). Measuring motivations for online opinion seeking. Journal of Interactive Advertising, 6(2), 2-14.
Gruen, T. W., Osmonbekov, T., & Czaplewski, A. J. (2006). eWOM: The impact of customer-to-customer online know-how exchange on customer value and loyalty. Journal of Business Research, 59(4), 449-456.
H&M Facebook (2016). In Facebook [corporate page]. Retrieved June 17, 2016, from https://www.facebook.com/photo.php?fbid=10154209496612482&set=o.5263729 94152420&type=3&theater
Ha, A. (2015, April 29). Facebook says there are now 40M active small business pages. TechCrunch. Retrieved from https://techcrunch.com/2015/04/29/facebook-40-million/ Hennig-Thurau, T., Wiertz, C., & Feldhaus, F. (2015). Does Twitter matter? The impact of
microblogging word of mouth on consumers’ adoption of new movies. Journal of the Academy of Marketing Science, 43(3), 375-394.
Hennig‐ Thurau, T., Gwinner, K. P., Walsh, G., & Gremler, D. D. (2004). Electronic word‐ of‐ mouth via consumer‐ opinion platforms: What motivates consumers to articulate themselves on the Internet?. Journal of interactive marketing, 18(1), 38-52.
Hornik, J., Satchi, R. S., Cesareo, L., & Pastore, A. (2015). Information dissemination via electronic word-of-mouth: Good news travels fast, bad news travels
faster!. Computers in Human Behavior, 45, 273-280.
Hornikx, J. & Hendriks, B. (2015). Consumer tweets about brands: A content analysis of sentiment tweets about goods and services. Journal of Creative Communications, 10(2), 176-185.
Jansen, B. J., Zhang, M., Sobel, K., & Chowdury, A. (2009). Twitter power: Tweets as
electronic word of mouth. Journal of the American society for information science and technology, 60(11), 2169-2188.
Kahneman, D., & Tversky, A. (1979). Prospect theory: An analysis of decision under risk. Econometrica: Journal of the Econometric Society, 47(2), 263-292.
Kelleher, T. (2009). Conversational voice, communicated commitment, and public relations outcomes in interactive online communication. Journal of communication, 59(1), 172-188.
Kelleher, T., & Miller, B. M. (2006). organisational blogs and the human voice: Relational strategies and relational outcomes. Journal of Computer‐ Mediated
Communication,11(2), 395-414.
Lee, S., Hwang, T., & Lee, H. H. (2006). Corporate blogging strategies of the Fortune 500 companies. Management Decision, 44(3), 316-334.
Lillqvist, E., & Louhiala-Salminen, L. (2014). Facing Facebook Impression Management Strategies in Company–Consumer Interactions. Journal of Business and Technical Communication, 28(1), 3-30.
McCarthy, C. (2010, March 19). Nestle mess shows sticky side of Facebook pages. CNET. Retrieved from http://www.cnet.com/news/nestle-mess-shows-sticky-side-of-facebook-pages/
Pollach, I. (2005). Corporate self-presentation on the WWW: Strategies for enhancing usability, credibility and utility. Corporate Communications: An International Journal, 10(4), 285-301.
Rybalko, S., & Seltzer, T. (2010). Dialogic communication in 140 characters or less: How Fortune 500 companies engage stakeholders using Twitter. Public Relations Review, 36(4), 336-341.
Schamari, J., & Schaefers, T. (2015). Leaving the home turf: How brands can use webcare on consumer-generated platforms to increase positive consumer engagement. Journal of
Interactive Marketing, 30, 20-33.
Searls, D., & Weinberger, D. (2001). Markets are conversations. In R. Levine, C. Locke, D. Searls, & D. Weinberger (Eds.), The cluetrain manifesto: The end of business as usual (pp. 75-114). New York: Perseus.
Shively, K. (2014, September 30). New Study: 80% of Fortune 500 Companies Active on Facebook and Twitter. SimplyMeasured. Retrieved from
http://simplymeasured.com/blog/new-study-80-of-fortune-500-companies-active-on-facebook-and-twitter/#sm.0001hlxijl11j7euxzfut8gqg720b
Skowronski, J. J., & Carlston, D. E. (1989). Negativity and extremity biases in impression formation: A review of explanations. Psychological bulletin, 105(1), 131-142. Smith, C. (2016, April 30). By the Numbers: 90 Amazing Facebook Page Statistics. DMR.
Retrieved from http://expandedramblings.com/index.php/facebook-page-statistics/2/ Spary, S. (2016, June 15). This Student Says H&M Size-16 Jeans Are So Small They Are
“Laughable”. BuzzFeed News. Retrieved from
https://www.buzzfeed.com/saraspary/this-student-says-hm-size-16-is-laughable? utm_term=.jf0w73rZN#.dvglWMg8N
Sweetser, K. D., & Metzgar, E. (2007). Communicating during crisis: Use of blogs as a relationship management tool. Public Relations Review, 33(3), 340-342.
