Running head: TAKE CARE – SAY ‘SORRY’? 1

Take Care – Say ‘Sorry’?

The Effects of Webcare Strategies on and the Influence of Message

Personalization and Message Authenticity

Master’s Thesis

Selina Peukert, Student ID: 10697012

Master’s Programme Communication Science

Persuasive Communication

Supervisor: Dr. Guda van Noort

Date of completion: 30th January 2015

TAKE CARE – SAY ‘SORRY’? 2

Abstract

Organizations are increasingly experiencing customer complaints on social media, as the Web

2.0 has empowered consumers to voice their negative remarks about a company publicly with a huge audience witnessing. The practice of responding to these negative comments, also referred to as webcare, has gained importance as online complaints can negatively affect consumers’ brand responses and lead to organizational damage. This research examined how different webcare responses and their degree of personalization can influence the consumers’ brand re- sponses and whether the consumers’ perceived authenticity of these responses would mediate the effects. These aims were operationalized in a 3 (accommodative vs. defensive vs. no webcare) x

2 (personalized vs. corporate messages) experiment. The results show that an accommodative webcare response is the most suitable in evoking favourable brand responses. A direct effect of message personalization was found, as personalized webcare responses, regardless of them being accommodative or defensive, lead to more positive brand responses. However, the results demonstrated that the webcare response types did not interact overall with message personaliza- tion, as only for the consumers’ positive word-of-mouth (PWOM) intention message personali- zation could significantly enhance the effectiveness of a defensive response. The effectiveness of accommodative webcare responses could not be improved by personalizing them. In addition, this research uncovered that the consumers’ perceived authenticity mediates the effects of the different webcare responses on the consumers’ brand responses. The findings are discussed tak- ing into consideration the challenges organizations face to find the right approach to deal with negative word-of-mouth (NWOM) on social media.

Keywords: Negative word-of-mouth, webcare, message personalization, authenticity of electronic word-of-mouth, complaint handling

TAKE CARE – SAY ‘SORRY’? 3

“I am sick to my stomach after seeing the Kent State sweater. Absolutely disgusted with the UO [Urban Outfitters] company. Planning to never shop UO or its other companies, Anthro

& Free People again”. “Airberlin, your customer service is horrendous”. “Two hours just to can- cel a line. AT&T you guys are insulting to every customer who does business with you.” These statements are just a few examples of consumers who use negative electronic word-of-mouth

(hereafter: NWOM) on social media, to vent for instance their anger, dissatisfaction or negative experiences with companies. Social media depicts a double-edged sword for companies: Con- sumer empowering and interactive Web 2.0 applications do not only offer great new marketing potentials, but also allow consumers to use this two-way communication to distort and hijack brand messages, as well as to voice their complaints (Fournier & Avery, 2011; Hennig-Thurau et al., 2010; Xia, 2013). A study revealed that about 50% of the social media users voice their com- plaints or negative feedback regarding brands or services at least once a month (Nielsen, 2012).

NWOM information has to be taken seriously, as it can have detrimental effects on the consum- ers’ brand responses (e.g., Davidow, 2003; Fournier & Avery, 2011; Lee & Song, 2010).

The complaint nature has undergone a fundamental change by evolving from a private conversation between the complainer and the company to being a public matter. By posting complaints visibly on social network sites, blogs, forums or brand communities, other consumers are able to witness the complaint situation (Constantinides & Fountain, 2008; Hennig-Thurau et al., 2010). The Web 2.0 enables complaints to spread easily, rapidly and widely and complaints therefore also depict a threat to the relationship a company has with the observing audience and endangers the brand responses of this audience, who might too engage in NWOM and hereby increase the range of organizational damage (Breitsohl, Khammash, & Griffiths 2010; Fournier

& Avery, 2011). The negative impact of NWOM and its potential to be multiplied, demonstrate that companies need to engage in effective complaint handling, as their actions online can have harmful effects on multiple stakeholders. The necessity to find customer satisfying strategies to

TAKE CARE – SAY ‘SORRY’? 4 address NWOM is furthermore emphasized by the fact that about half of the complaining con- sumers are dissatisfied with how companies handle their complaints (Estelami, 2000).

Webcare, “the act of engaging in online interactions with (complaining) consumers, by actively searching the web to address consumer feedback (e.g. questions, concerns and com- plaints)” (Van Noort & Willemsen, 2011, p. 133), has been found as such an effective strategy to counteract NWOM. Previous research on complaint handling and service failure recovery demonstrated that effective corporate responses cannot only mitigate the effects of NWOM, but also lead to customer loyalty (Orsingher, Valentini, & de Angelis, 2010), positive word-of- mouth (hereafter: PWOM; e.g., Davidow, 2003; Maxham & Netemeyer, 2002), repurchase inten- tions (e.g., Davidow, 2003) and commitment and trust (Tax, Brown, & Chandrashekaran, 1998).

As these outcomes are desirable for companies, the question arises as to what extent different webcare responses can contribute to reaching these goals.

Companies can respond to NWOM in several ways by for example apologizing, offering compensation or promise corrective actions (e.g., Lee, 2004). These accommodative webcare responses are oppositional to defensive webcare responses, which include denying the responsi- bility, attacking the accuser or shifting the blame to others (e.g., Davidow, 2003; Kerkhof,

Beugels, Utz, & Beukeboom, 2011; Lee & Song, 2010). Although engaging in webcare depicts a company’s opportunity to demonstrate that they are open to address the consumers’ criticism and needs (Willemsen, 2013), not all webcare responses show equal effectiveness. Accommodative webcare responses have been found to result in positive brand evaluations (e.g., Lee & Song,

2010) and PWOM (Gelbrich & Roschk, 2011). Research revealed that defensive webcare re- sponses are not very well accepted by consumers, as they feel disappointed in the company’s complaint handling process and have the perception that the company is at fault (Gelbrich &

Roschk, 2011; Lee & Song, 2010). Companies might also ignore NWOM altogether, which gives consumers the feeling that the company does not care about them or does not take their complaint seriously (e.g., Fournier & Avery, 2011; Lee & Song, 2010). Deriving from above

TAKE CARE – SAY ‘SORRY’? 5 stated, the first research aim is to investigate which of these webcare strategies are suitable in evoking favourable marketing-related brand responses, such as brand attitude, purchase intention and PWOM intention. By concentrating on a marketing perspective, this study will extend previ- ous research, which mainly focussed on complaint handling from a PR perspective (e.g., Kerkhof et al., 2011), in connection with a crisis (e.g., Lee, 2004) or from a customer service perspective

(e.g., Gelbrich & Roschk, 2011, Holloway & Beatty, 2003). Through taking into consideration the importance of the observing audience (e.g., Breitsohl et al., 2010), this study will not only provide managerial implications on how to integrate webcare in the marketing mix, but also state guidelines on what to respond to NWOM to have a positive effect on multiple stakeholders.

Asides the webcare response type being crucial at influencing the consumers’ brand re- sponses, it is also important how the response is conveyed, thus what kind of tone of voice is used (e.g., Park & Lee, 2013; Van Noort & Willemsen, 2011). Conversational human voice, which is “an engaging and natural style of organizational communication as perceived by an organization’s publics based on interactions between individuals in the organization and the in- dividuals in publics” (Kelleher, 2009, p. 177) has been found to be a valid predictor of the effec- tiveness of the different webcare responses (Van Noort, Willemsen, Kerkhof, & Verhoeven, forthcoming). Whereas past research concentrated on conversational human voice as a whole, this study will extend previous findings by breaking down the construct and examining one indi- vidual component, message personalization. It is assumed that message personalization, opera- tionalized through the company identifying who is responding on behalf of them and addressing the consumer by name along with using a multitude of personal pronouns, can improve the effec- tiveness of a message and evoke favourable consumer responses (Kwon & Sung, 2011; Park &

Lee, 2013). The second aim of this thesis consists out of examining how message personalization can enhance the effectiveness of webcare responses and providing insights for marketers on how they can convey their webcare responses in a human and personal manner.

TAKE CARE – SAY ‘SORRY’? 6

Preceding research in the customer service context (Gruber, 2011) and the PR context

(Kerkhof et al., 2011) found that the tone of voice of a message and its content interact with each other: It has been postulated that the effectiveness of accommodative responses can be enhanced if the responses are also personalized (Gruber, 2011), and research demonstrated that message personalization can compensate defensive responses, as consumers get the feeling that the com- pany wants to build a relationship with them (Kerkhof et al., 2011). Until now no research has tested the interaction between the different webcare response types and message personalization from a marketing perspective, in which the attainment of marketing-related goals is a central requirement of a company’s webcare efforts. Thus, the third aim of this study is to provide prac- tical value, as it is expected that consumers will react differently to the different webcare re- sponses when they get the feeling that they are dealing with a real person instead of an anony- mous organization (Park & Lee, 2013).

Although companies can determine the content of their response and decide which tone of voice they apply to it, consumers might perceive the message differently than intended by the company (Van Noort et al., forthcoming). For instance, saying ‘sorry’ is an apology, but to what extent it really portrays the company’s sincere regret, is up to the consumers to decide (Van

Noort et al., forthcoming). Greyser (2009) postulated that corporate authenticity is central at all times, but especially in critical situations (e.g. consumer complaints), companies need to make authenticity an inalienable constituent of their communication. Authentic messages are success- ful at being accepted by the audience and lead to favourable brand responses (Huang, Cai,

Tsang, & Zhou, 2011; Liao & Yang, 2012). Research found accommodative strategies to be ef- fective in eliciting high authenticity perceptions, as apologies evoke the consumers’ feeling that the company responds in a credible and genuine manner (Gelbrich & Roschk, 2011; Gruber,

2011; Liao & Yang, 2012). Additionally, it has been demonstrated that the perceived authenticity of a response can be enhanced when the response is conveyed with a personal tone of voice, as this increases the consumers’ perception of genuineness and sincerity (Bell & Zemke, 1987).

TAKE CARE – SAY ‘SORRY’? 7

This points towards an interaction of webcare response type and message personalization on the consumers’ perceived authenticity, thus theoretical value can be found in uncovering the influ- ence of perceived authenticity as underlying mechanism to explain the varying effectiveness of the different webcare responses and their degree of personalization. As the above stated findings derived mainly from the service recovery context and crisis communication, the fourth research aim consists out of examining the role of authenticity from a marketing perspective in order to determine its importance for webcare.

Asides from providing theoretical value, this thesis will guide marketers in their use of webcare on social media by determining how effective complaint handling can be achieved through an adequate content and tone of voice of organizational messages. By doing so, this re- search will offer valuable insights into webcare’s potential to be a successful marketing tool.

Theoretical Framework

Online Complaint Handling as Marketing Practice

The characteristics of Web 2.0 applications have facilitated the creation and dissemina- tion of online complaints: Publicly accessible social media sites serve as popular platforms for consumers’ complaints, which can easily be shared with a multitude of other consumers without having financial or physical efforts attached to it (Fournier & Avery, 2010; Hennig-Thurau et al.,

2010). Due to various consumer empowering tools to rate or rank companies, consumers have increasingly voiced their criticism of companies (Fournier & Avery, 2011). Fournier and Avery

(2011) postulate that these developments led to an “age of criticism” (p. 200), in which effective complaint handling has become a full-time job. Consumers, who are networked together based on a positive ground, can be valuable allies, but if consumers unite based on their shared criti- cism of the brand, they can pose a great threat to companies (Fournier & Avery, 2011). That this is a serious matter, underlines the fact that 70% of social media users rely on the experiences and opinions of others when making decisions (Nielsen, 2012). Research on the acceptance of elec-

TAKE CARE – SAY ‘SORRY’? 8 tronic word-of-mouth (hereafter: WOM) information has revealed that consumers believe and accept NWOM more than PWOM (Huang et al., 2011). The above stated illustrate the im- portance of webcare being an impression management tool for the observing audience, as the witnessing consumers play the decisive role in multiplying negative brand experiences online

(Breitsohl et al., 2010).

Webcare can be studied from various theoretical perspectives, which differ in their con- sideration in which department of a company complaint handling is allocated and what organiza- tional goals should be reached with it (Van Noort et al., forthcoming). Whereas a vast amount of research examines complaint handling in connection to service failure recovery efforts and an- chors webcare in the customer care department of an organization (e.g. Holloway & Beatty,

2003; Maxham, 2001), or examines the role of complaint handling as crisis management tool

(e.g., Lee, 2004), there also exists a research stream which emphasizes the marketing and PR potential of effective complaint handling (e.g. Kerkhof et al., 2011; Tax et al., 1998; Van Noort et al., forthcoming; Willemsen, Neijens, & Bronner, 2013). This study will specifically concen- trate on webcare as part of marketing, as this perspective challenges companies to embrace com- plaints in a way that they are the consumers’ attempts to strengthen their relationship with a company (Kim, Kim, & Kim, 2009; Shields, 2006; Willemsen et al., 2013). Any interaction with customers, even if they are of negative nature, can be the first step in fortifying the relationship with the customers, as effective complaint handling evokes a positive feeling, which can turn into trust and satisfaction (Kim et al., 2009). This perspective promotes the consideration that webcare should exceed being an ad-hoc strategy to mitigate the negative impact of complaints.

The Effects of Different Webcare Response Strategies

Although research on the effects of different webcare strategies is still to some extent limited, previous studies have found important contributions for when to offer webcare (Van

Noort & Willemsen, 2011) or what to respond to customers (e.g., Gelbrich & Roschk, 2011; Lee

TAKE CARE – SAY ‘SORRY’? 9

& Song, 2010). According to Gelbrich and Roschk (2011), organizational responses to com- plaints can be divided into compensation, favourable employee behaviour and organizational procedures. Whereas these different response types emerged out of their meta-analysis based on customer service literature, it provides valuable insights in regards to the corrective actions of a company: Apologies, compensation and attempts to redress are the most commonly used ac- commodative response strategies (e.g., Kerkhof et al., 2011; Lee, 2004; Lee & Song, 2010). The counterpart of accommodative strategies are defensive strategies, in which companies shift the blame to others, attack the accuser or deny their responsibility for the incident (e.g. Kerkhof et al., 2011; Lee, 2004; Lee & Song, 2010). Whereas these strategies have their origin in customer care literature (Gelbrich & Roschk, 2011), crisis communication (e.g., Lee, 2004) or in face-to- face complaint situations (Bell & Zemke, 1987), the differentiation based on the varying content and approach of the messages also applies to the social media context used in this study.

In the context of marketing, companies use accommodative webcare strategies to show their efforts to restore the relationship with their customers. Given the positive approach of these responses and an organization’s intention to make it up to the customer, previous research found them to be more effective than no response to NWOM or defensive responses (e.g., Kerkhof et al., 2011; Lee & Song, 2010). Compensations lead to perceived distributive justice, meaning that customers are satisfied with the complaint outcome, which in turn leads to repurchase intentions and PWOM (Gelbrich & Roschk, 2011). Companies tend to avoid offering apologies in their responses, as they consider them to be an admission of guilt (Gelbrich & Roschk, 2011; Sparks

& Bradley, 2014). Kerkhof et al. (2011) underline that apologies are superior to their denial counterparts, as they, although evoking the consumers’ feeling that the company is responsible for the incident, lead to higher credibility and a positive attitude towards the response. Bell and

Zemke (1987) furthermore postulate that apologies, as “acknowledgement of error” (p. 33), are vital in terms of easing annoyed customers and Tax et al. (1998) explain that they are necessary to restore the emotional costs of consumers. A response, which contains a combination of an

TAKE CARE – SAY ‘SORRY’? 10 apology and compensation leads to greater purchase intention, than a response only incorporat- ing compensation (Wirtz & Mattila, 2004). The same applies for apologies, as they also not lead to positive reactions if presented by themselves as webcare response strategy (Davidow, 2003), thus a connection of apologies and compensation is recommended in order to reach favourable brand responses.

Companies that react defensively to NWOM neglect to consider complaints as an oppor- tunity to fortify customer relationships. Lee (2004) found that consumers in the denial-of-crisis responsibility condition, in which the corporate answer consisted out of shifting the blame to others, minimizing the severity or not commenting, reported on more negative impressions of the company, ascribed the responsibility for the crisis more severely to the organization, were less sympathetic towards the company and mistrusted the organization more than the participants in the accommodative condition. Defensive webcare responses are not only ineffective at leading to favourable brand responses (Lee & Song, 2010), but they also do not evoke the consumers’ sym- pathy with the company (Kerkhof et al., 2011).

The above stated research findings differ in their perspective on complaint handling from this study, as many have their origin in customer service literature (e.g., Gelbrich & Roschk,

2011) or crisis communication (e.g., Lee, 2004), which is also reflected in the examined com- plaint outcomes. Whereas previous research focussed on complaint web sites (e.g., Lee & Song,

2010) or face-to-face situations (e.g., Bell & Zemke, 1987) as complaint environment, this study will examine the effectiveness of webcare response strategies in the social media context. Social media is not only a central vehicle for a company’s marketing activities, but also depicts an es- sential tool for a company’s webcare efforts since a lot of consumers use social media to engage with organizations (Nielsen, 2012). Hence, the first hypothesis reads as follows:

H1. An accommodative webcare response leads to more positive brand responses than a defen- sive webcare response and no response to NWOM.

TAKE CARE – SAY ‘SORRY’? 11

Consumers resent being ignored when they take the effort to engage with companies by writing a complaint (Fournier & Avery, 2011), but there is still disagreement among researchers as to what extent reacting defensively can lead to higher brand responses than not providing a response to NWOM. Dekay (2012) found that not responding cannot count as acceptable webcare strategy as it does not promote conversation and depicts a company’s missed out oppor- tunity to turn negative comments into valuable opportunities. Lee and Song (2010) demonstrated that participants in the defensive response condition reported on worse brand responses and higher blame attributions than consumers in the no response condition. Lee (2004) included the no response condition in the defensive webcare cluster, which calls for caution when interpreting her results as these two conditions will be treated separately in the present study. Nevertheless, the study by Lee (2004) offers a crucial insight: When independently comparing the no response condition with the ‘shifting the blame’ condition it was found that participants who did not re- ceive a corporate response attributed less responsibility to the company but had a greater nega- tive impression of the company than participants who were in the ‘shifting the blame’ condition

(Lee, 2004). This is contrary to Lee and Song (2010) as they claim that defensive responses lead to greater blame attributions than not responding. Interestingly, Lee (2004) additionally demon- strated that participants who did not read a corporate response had more trust and sympathy to- wards the company than participants in the ‘shifting the blame’ condition. This is again in line with the study of Lee and Song (2010), in which participants in the no response condition rated the brand more positively than participants receiving a defensive response.

Lee (2004) investigated the effectiveness of the different response strategies in the con- text of crisis communication and Lee and Song (2010) looked at organizational responses on a complaint website, thus examining the effects of proactive webcare (Van Noort & Willemsen,

2011), but it is arguable that ignoring consumers or responding defensively is acceptable in terms of webcare being a tool to manage customer relationships and reach marketing-related goals.

Thus, the following research question was formulated:

TAKE CARE – SAY ‘SORRY’? 12

RQ1. To what extent does a defensive webcare response lead to more positive brand responses than no response to NWOM?

The Influence of Message Personalization

Applying a human presence to organizational communication on social media is vital as it serves as cue for consumers that they are communicating with real people (Rybalko & Seltzer,

2010). Although social media platforms were established to facilitate the interaction between users and not as a promotion tool for marketers, they enable organizations to get closer to their customers, interact and engage with them in a more informal way and manage customer relation- ships (Fournier & Avery, 2011; Hansson, Wrangmo, & Solberg Soilen, 2013; Kwon & Sung,

2011). In order to be successful on social media and not come across as intruders, companies have to adapt their communication to the particular social media platform (Becker, 2013;

Grundmann, 2015).

Message personalization can be applied to webcare by the means of two strategies. First- ly, the organization can be personalized (Van Noort et al., forthcoming): Brand anthropomor- phism, in which human attributes are connected to non-human entities (Kwon & Sung, 2011), can be used to attach a human feel to the organization. Using first-person pronouns (“I”, “we”) and identifying the individual communicating on behalf of the organization by name or photo are effective personalization strategies (Kwon & Sung, 2011; Park & Lee, 2013). Secondly, webcare responses can be personalized through tailoring the message to the individual customer, thus personalizing the receiver. Companies can address the customer by name and use second-person pronouns (“you”, “yours”, “yourself”) (Kwon & Sung, 2011). As this study was conducted among German speaking participants, German informal and formal second-person pronouns were used to differentiate between personalized and corporate messages: “Du” (being the infor- mal translation of “you”) personalizes a company’s approach to the consumer and reflects the

TAKE CARE – SAY ‘SORRY’? 13 informal communication style on social media (Grundmann, 2015). “Sie”, on the other hand, the formal translation of “you”, conveys a distance and is perceived as rigid (Grundmann, 2015).

A personal communication style is an opportunity to fortify customer relationships and allows companies to signalize closeness and their openness to communication (Grundmann,

2015; Kwon & Sung, 2011). Park and Lee (2013), who conducted a study on the effectiveness of attributing a human presence to a company’s communication, demonstrated personaliz- ing the company facilitates to minimize the consumers’ feeling that they are communicating with an anonymous organization. Message personalization evokes favourable brand responses and has a positive effect for relationship marketing, as engaging in a more human and personal commu- nication style promotes the consumers’ PWOM intentions and enhances their favourable percep- tions of the relationship with a company (Park & Lee, 2013). By using the informal personal- pronoun “du”, consumers get the feeling of a higher emotional bond with the company and a company can encourage communication (Grundmann, 2015). This is in according with Kwon and Sung (2011), who found that the use of second-person pronouns on Twitter stimulates con- versations with the company and is successful at decreasing the impersonality of online commu- nications. In addition, account personal pronouns and non-verbal cues, such as abbreviations and emoticons, positively for the consumers’ feeling that companies are actually listening to them

(Kwon & Sung, 2011).

The effects of message personalization can be explained with assumptions from infor- mation processing theories: The Elaboration-Likelihood-Model (ELM) suggests that individuals will be more motivated to process information centrally and elaborate, if the information is per- sonally relevant to them (Petty & Cacioppo, 1986). The ELM furthermore states that an in- creased elaboration leads to consumers devoting more effort on assessing the quality of the mes- sage characteristics (Petty & Cacioppo, 1986). Thus, consumers concentrate more on the argu- ments made in the message rather than relying on simple and easy-accessible heuristic cues, which consumers do when processing messages peripherally (Petty & Cacioppo, 1986). In the

TAKE CARE – SAY ‘SORRY’? 14 context of webcare, personalized webcare responses provoke the consumers’ feeling that the message is highly relevant for them, which will promote their thoughtful processing of a compa- ny’s response. It has been proven that health care material is more successful when it is personal- ized (Kreuter, Bull, Clark, & Oswald, 1999), but previous research also found that message per- sonalization increases the effectiveness of marketing messages, as personalized messages lead to greater attention and heightened elaboration, as well as better information recall and higher per- ceived usefulness (Tam & Ho, 2005, 2006).

Whereas previous studies investigated the effects of message personalization on Twitter

(e.g., Kwon & Sung, 2011; Park & Lee, 2013), being a platform with very specific communica- tion boundaries, this research will concentrate on the effects of personalized webcare messages on . This study will add to the scope of past research, which solely concentrated on message personalization operationalized through personalizing the company (e.g., Park & Lee,

2013) or giving a descriptive overview of how companies use brand anthropomorphism elements in their online communication (e.g., Kwon & Sung, 2011). In the present research, message per- sonalization will be applied using two dimensions, personalization of the company and the re- ceiver, and it will be examined how this can lead to favourable brand responses. Based on the above stated, the second hypothesis reads as follows:

H2. There is a direct effect of message personalization on consumers’ brand responses, such that personalized webcare responses lead to better brand responses than corporate webcare responses.

When in fact a personalized approach accounts positively for the consumers’ increased attention on the message, Kerkhof et al. (2011) argue that the tone of voice of a message can also interact with the content of a message: A personalized denial response leads to a greater percep- tion of the communicated relational commitment of a company than a corporate denial response

(Kerkhof et al., 2011). Gruber (2011) supports the assumption of an interaction effect, as he ex- plains that complaining customers wished for a personalized approach in connection with an

TAKE CARE – SAY ‘SORRY’? 15 apology. Based on the findings of information processing theories, it is therefore expected that message personalization will increase the consumers’ attention on the message characteristics, which in turn will enhance the positive effects of accommodative webcare responses, as con- sumers devote more thoughts on the company offering them an apology and signalizing their efforts to make it up to them. Nevertheless, it is also expected that defensive webcare responses will lead to more positive brand responses if they are personalized (vs. corporate), as consumers value the customer-centralized approach and feel taken seriously (Kwon & Sung, 2011). Addi- tionally, it can be assumed that the positive effect of message personalization will make up for the negative nature of defensive webcare responses. Following this line of argument, the third hypothesis reads as follows:

H3a-b. The effects of the different webcare responses on brand responses are moderated by mes- sage personalization, such that a) the positive effect of an accommodative webcare response on brand responses will be increased by message personalization, and b) message personalization will improve the effectiveness of a defensive webcare response in evoking more positive brand responses.

The Role of Perceived Authenticity of Webcare Responses

With the content of a webcare response and the tone of voice applied to that response, companies try to determine how their response is perceived by the consumers. Nevertheless, a well thought out and communicated response might still be off target, because consumers per- ceive it differently than intended (Van Noort et al., forthcoming). This discrepancy between what is being communicated and how consumers understand it can lead to a failure of a webcare re- sponse (Van Noort et al., forthcoming).

An explanation why webcare responses can be ineffective is found when considering the consumers’ perceived authenticity of these responses. Authenticity constitutes a central charac- teristic of WOM information and authentic information consists, according to Huang et al.

TAKE CARE – SAY ‘SORRY’? 16

(2011), of three concepts: Credibility, trustworthiness and reliability. Huang et al. (2011) adapted this understanding from Rieh’s (2002) explanation of information authority, which is important for the judgement of web information. The standard of authenticity, and especially its compo- nents credibility and trustworthiness, is used to evaluate the content and the tone of voice of a message (Huang et al., 2011). The notion of authenticity being a multi-dimensional construct is also shared by Greyser (2009) as he argues that corporate authenticity consists of four parts:

Companies have to talk authentically, be authentic (the core values and previous behaviour have to be in accordance), stay authentic (stand by core values), and defend their authenticity in times of crisis. The first constitute is in line with Mitra and Watts (2002), who point out that an authen- tic voice has to speak genuinely and ethically. For the context of this study authentic communi- cation is the key, as the consumers’ perception of a webcare response’s authenticity is merely based on the content of the response and how it is framed.

Authenticity is central for the effectiveness of a company’s webcare message, as per- ceived authenticity serves as a valid predictor of the consumers’ acceptance of WOM infor- mation (Huang et al., 2011; Liao & Yang, 2012). This is supported by Sparks and Bradley

(2014), who examined the effectiveness of organizational responses to negative online hotel re- views, as they explain that an authentic organizational response which is perceived as honest, thorough and adequate, leads to positive customer evaluations. These positive evaluations also translate to favourable behavioural intentions, as high message authenticity accounts for the con- sumers’ intention to resend that information (Huang et al., 2011). Research has furthermore demonstrated that authentic and credible WOM information can lead to purchase intention (Liao

& Yang, 2012). In the context of customer care, authentic employee behaviour depicts the most important requirement for customer satisfaction (Gruber, 2011), which again is central for posi- tive brand responses (e.g., Gelbrich & Roschk, 2011; Orsingher et al., 2009; Tax et al., 1998).

Whereas the above stated findings underline that authenticity can lead to favourable brand responses (e.g., Huang et al., 2011; Liao & Yang, 2012), it remains questionable how dif-

TAKE CARE – SAY ‘SORRY’? 17 ferent webcare responses and their degree of personalization can influence the consumers’ au- thenticity perceptions. Past research did not examine the authenticity of webcare responses in particular, but rather concentrated on authentic customer service employee behaviour (e.g., Bell

& Zemke, 1987; Gruber, 2011) or the role of authenticity in crisis communication (e.g., Greyser,

2009). Nevertheless, these results hint towards an interaction between the variables webcare re- sponse type, their degree of personalization and authenticity: An apology and a personal ap- proach are antecedents for the consumers’ feeling of distributive justice (Gelbrich & Roschk,

2011). The justice dimension is a requirement for the consumers’ feeling that the company will solve their problems. However, the expectation that the company will provide a solution is also closely linked to the consumers’ feeling of being taken seriously, which again can only originate if a company’s complaint handling is perceived as authentic (Gruber, 2011; Min, Lim, & Magni- ni, 2014). In order to be perceived as authentic, companies need to convey their messages with a personal tone of voice, as this promotes the consumers’ feeling that the company is sincere and credible (Bell & Zemke, 1987).

So far, no research has tested the role of authenticity as underlying mechanism of the ef- fectiveness of the different webcare responses, but above stated does suggest that there is an in- teraction ofthe different webcare responses and their degree of personalization on the perceived authenticity of these responses, which in turn influence the consumers’ brand responses. Since previous research points to a positive effect of accommodative webcare responses on the per- ceived authenticity of these responses (e.g., Gelbrich & Roschk, 2011; Gruber, 2011; Min et al.,

2014), and also points to a positive effect of personalized messages on the consumers’ perceived authenticity (e.g., Bell & Zemke, 1987), it is expected that a response that is both accommoda- tive and personalized will have the most positive effect on the consumers’ perceived authentici- ty, which will have the most positive effect on consumers’ brand responses. Nevertheless, it is also expected that a personalized defensive response, in comparison to a corporate defensive response, will lead to a higher authenticity perception, as will targeting the message to the indi-

TAKE CARE – SAY ‘SORRY’? 18 vidual customer and personalizing the organization account positively for the messages’ credibil- ity, being a necessary premise for authenticity. To explore above stated, the final hypothesis reads as follows:

H4a-b. There is a mediated interaction effect of webcare responses and message personalization on brand responses through perceived authenticity, such that a) the positive effect of an accom- modative webcare response (vs. a defensive response) on perceived authenticity is stronger when the message is also personalized (vs. corporate); and b) consequently this positive effect on per- ceived authenticity results in more positive brand responses.

All the expected relationships can be seen in figure 1.

Figure 1. Conceptual model

Method

Participants and Design

To test the hypotheses and research question, an online experiment with a 3 (Webcare re- sponse: accommodative vs. defensive vs. no response) x 2 (Message personalization: personal- ized vs. corporate) between-subjects design was conducted. In total 196 participants (67.9% fe- male) participated in this experiment (Mage = 30.82, SD = 12.36). 59.2% of the participants indi-

TAKE CARE – SAY ‘SORRY’? 19 cated that their highest level of education was a University degree (Bachelor or Master). Almost half of the participants were students (49%) and the second largest group (31.6%) indicated to be working.

Stimulus Materials

The experimental material consisted out of a fictitious complaint posted on Lufthansa’s

Facebook page and was designed in resemblance to real complaints on social networks in terms of content and tone of voice. Facebook was chosen as a platform for the complaint interaction as it is not only a very popular social network, but also has the most users across all social media platforms (Nielsen, 2012). Furthermore, 47% of social media users engage in social care, being customer care over a social network and consumers are most likely to do that on Facebook (Niel- sen, 2012).

The fictitious customer, Anna Langer, indicated to be very unhappy with Lufthansa as her flight to New York got cancelled short notice, which forced her to wait a great amount of time at the airport in London. She pointed out that she was promised some sort of compensation but did not receive anything until now. The customer declared her disappointment with Lufthansa, de- manded an immediate solution and doubted if she would ever fly with the airline again.

It was opted to use a real brand, as this contributes positively to the external validity of the study. Willemsen, Neijens, Bronner and De Ridder (2011) postulate that NWOM has an es- pecially big influence on experience products, which can only be evaluated after the purchase, as negative reviews are only perceived as useful when they are written for experience products (vs. search products). Flights are considered to be such products, as their quality cannot be assessed prior to the purchase. Moreover, it can be assumed that consumers are highly involved in

Lufthansa’s products, as flights are often quite costly and booked with a lot of consideration (in regards to different prices, dates, etc.).

TAKE CARE – SAY ‘SORRY’? 20

Depending on the webcare response condition, Lufthansa reacted with an accommodative response, a defensive response or no response at all. Additionally, the messages were manipulat- ed in regards to their tone of voice (personalized vs. corporate). Each participant saw only one of the five responses (see figures 3, 4, 5, 6 and 7 in Appendix A).

Webcare response. For the manipulation of the webcare response, the complaint was followed by either no response (the screenshot contained a statement informing the participants that Lufthansa did not respond to the complaint), or a response which was accommodative or defensive. In the accommodative webcare condition, Lufthansa apologized and offered compen- sation for the negative experience in form of a voucher. In the defensive response condition,

Lufthansa denied its responsibility and shifted the blame to the customer by stating that accord- ing to their general terms and conditions, short term cancellations are not applicable for the re- ceipt of equalization benefits. Lufthansa further refused to help the customer and indicated that the customer will have to contact customer care for further questions.

Message personalization. The manipulation of message personalization was two-fold in this study: In the personalized condition, the message to the receiver was altered by using many second-person pronouns, “you”, and directly addressing the customer with her name (“Dear An- na”). These alterations of the message are along the conceptualization of Kwon and Sung (2011).

As the questionnaire was translated to German, “you” was translated to “du”, the German infor- mal personal pronoun. To further emphasize the personal tone of the personalized responses,

“Du” was written with a capital letter, which reduces the distance between the speakers by sug- gesting that they are acquainted with each other. Secondly, the organization was personalized by using an individual Lufthansa employee, Max, who spoke using the first person for Lufthansa: In the beginning of the response Max introduced himself as a Lufthansa Team member and ex- plained that he will take care of the request. The personalized messages were signed with

“/Max”. The use of identification by name is adopted from Park and Lee (2013). In the corporate response condition, there was no information on who is responding on behalf of Lufthansa, as

TAKE CARE – SAY ‘SORRY’? 21 few personal pronouns as possible were used and if they were used, the formal and distant “Sie” was used. The customer was not addressed by name and the message was signed with “Kind regards, Lufthansa”.

Pretest

A pretest consisting of a 2 x 2 within subjects design was conducted among 22 partici- pants (72,7% female, Mage = 25.50, SD = 2.69) to check whether the manipulations of the webcare response types and message personalization were successful. Participants were random- ly assigned to one of the four conditions. The no response condition was excluded as participants only saw the complaint but no manipulated response. The participants were exposed to the

Lufthansa’s Facebook page, on which they saw the complaint and one of the four responses.

Subsequently, participants received questions concerning the tone of voice manipulation (per- sonalized vs. corporate) and the type of webcare manipulation (accommodative vs. defensive).

The first questions regarding the manipulation of webcare response type was ‘Lufthansa is apologizing and tries to make it up to the customer’. Participants had to indicate their opinion on a 7-point Likert scale (1= “Completely disagree”, 7= “Completely agree”). As intended, there was a significant difference between respondents who saw an accommodative response and respondents who saw a defensive response by Lufthansa (t(20) = 3,61, p < .01; Maccommodative =

5.45, SD = 1.44; Mdefensive = 2.73, SD = 2.05). The second statement, ‘Lufthansa refuses to help the customer and declares that the problem lies outside of its responsibility’, disclosed again a significant difference between respondents who saw the defensive response and respondents who saw the accommodative response (t(20) = -6.02, p < .001): The statement was agreed upon more by participants in the defensive condition (M = 5.73, SD = 1.62) than by participants in the ac- commodative condition (M = 2.00; SD = 1.27). Therefore it can be said that the manipulation of webcare response type into accommodative and defensive was successful and no adaptations needed to be made for the main study.

TAKE CARE – SAY ‘SORRY’? 22

To test the manipulation of message personalization participants were asked to rate four statements on 7-point Likert scales (1= “Completely disagree”, 7= “Completely agree”). The first statement, ‘Lufthansa responded in an impersonal manner’, revealed that there was a signif- icant difference between the personalized and corporate webcare response condition (t(20) = -

4.12; p < .01). However, the mean value of 4.73 of participants in the corporate condition (see table 1) suggests that the corporate messages were considered to be quite neutral in terms of their degree of personalization as the mean value manifested a great proximity to the midpoint of the scale. The second statement, ‘Lufthansa responded with a formal tone of voice’, disclosed anoth- er significant difference between the two groups (t(20) = -3.18, p = .01). Again, the mean value of the personalized condition indicated a small distance to the midpoint of the 7-point scale. In order to achieve a more distinctive differentiation, the corporate messages in the final stimulus material were slightly adjusted to emphasize the impersonal, distant and formal tone of voice.

The third statement, ‘Lufthansa identifies who is responding on behalf of them’, exposed again a significant difference between respondents in the corporate and the personalized condition (t(20)

= 5.62, p < .001). Finally, there was a significant difference between the two groups on

‘Lufthansa responded in a personalized and individualized manner’ (t(20) = 6.07, p < .001). All the mean scores regarding the manipulation of message personalization can be found in table 1.

Table 1 Mean scores regarding tone of voice manipulation (pretest) Condition Personalized Corporate Characteristic M SD M SD Impersonal manner 2.09 1.14 4.73 1.79 Formal tone 3.45 1.53 5.55 1.51 Personalization of the company 5.91 1.14 2.45 1.70 Personalization of the receiver 5.91 1.04 2.54 1.57

Note. N = 22

TAKE CARE – SAY ‘SORRY’? 23

Procedure

Participants living in Germany and German-speaking participants living in the Nether- lands were approached by email, through the social network Facebook and via online forums with the request to participate in a study on ‘consumer-related issues and complaints’. The re- quest contained the link to the online experiment. An online experiment was suitable in this con- text, as it allowed the stimulus materials to look as true to real Facebook brand pages as possible.

Firstly, participants read an introduction, which informed them that all their answers were treated anonymously and that they could cease their participation at any time. Participants were then told that they are about to see a corporate Facebook page and asked to look at it carefully, as ques- tions regarding that page would follow. In the next step, participants were randomly assigned to one of the five conditions. After the exposure to the stimulus materials, respondents were asked questions about the authenticity of Lufthansa’s reaction on the complaint and about their brand responses. Next, participants had to answer questions regarding several covariates, such as pre- vious experience with Lufthansa and previous experience with writing online complaints. Lastly, a manipulation check was performed to test whether the stimulus materials were effective and participants were asked to provide some demographic data before being debriefed and thanked.

Measures

Perceived authenticity. To assess the participants’ perception of the responses’ authen- ticity, the authenticity scale of Huang et al. (2011) was used. Participants were asked to indicate on a 7-point scale (1= “Completely disagree”; 7= “Completely agree”) whether they perceived the reaction of Lufthansa to be credible, trustworthy and reliable. All the items loaded on one factor (EV = 2.4; R² = .80), proved to constitute a reliable scale and were averaged to form a composite score (M = 3.53, SD = 1.49, α = .87).

TAKE CARE – SAY ‘SORRY’? 24

Brand responses. The measurement of the consumers’ brand responses was multi- dimensional and included the assessment of their brand attitude, their purchase intention and their PWOM intention. It was opted to measure brand responses in three dimensions, as purchase intention and brand attitude have proven to be reliable measurements to assess the participants’ overall brand responses (Raney, Arpen, Pashupati, & Brill, 2003). As this study concentrates on the marketing perspective, it is also valuable to measure the participants’ PWOM intention as it significantly contributes to the success of online marketing activities (Huang et al., 2011).

Brand attitude. As a real brand was used in this study, it was important to ensure that participants indicated their brand attitude in relation to how Lufthansa handled the complaint in this specific situation and not how they perceive Lufthansa in general. Based on Lufthansa’s re- action, participants were asked to indicate their brand attitude with the means of six items on a 7- point Likert-scale (1= “Completely disagree”; 7= “Completely agree”): The company is good, interesting, appealing, pleasant, of high quality and likable (Mitchell & Olson, 1981; Voorveld,

Neijens & Smit, 2011). The scale items loaded all on one factor (EV = 4.02; R² = .67) and as in previous research, the scale proved to be reliable and the items were averaged to form a compo- site score (M = 4.14, SD = 1.73, α = .86).

Purchase intention. The participants’ purchase intention was assessed with the means of three items on a 7-point Likert scale (1= “Completely disagree; 7= “Completely agree”). The first two items, which were adopted from the purchase intention scale from Coyle and Thorson

(2001), were supplemented by a third, self-created, item: ‘It’s very likely that I will fly Lufthansa,

‘I will use Lufthansa the next time I need to book a flight’ and ‘I’m very likely to choose

Lufthansa over a different airline company for my next flight’. The three items all loaded on one factor (EV = 2.00; R² = .67) and proved to form a reliable scale (M = 3.59, SD = 1.35, α = .74).

PWOM intention. The participants’ PWOM intention was assessed by the means of a single-item measure on a 7-point Likert-scale (1= “Very unlikely”; 7= “Very likely”): ‘How like- ly are you to recommend Lufthansa to friends or family?’ The item was taken from Boulding,

TAKE CARE – SAY ‘SORRY’? 25

Kalra, Staelin and Zeithaml (1993). Rossiter (2002) argues that single-item measures provide valid measurements if the construct is clear and unambiguous. He furthermore explains that a multiple item measure is only necessary if a construct contains many abstract attributes (Rossit- er, 2002), which is not the applicable for PWOM intention.

Control Variables

To rule out alternative explanations and test the impact of cofounding variables on per- ceived authenticity, brand attitude, purchase intention and PWOM intention, the following varia- bles were included: Previous experience with Lufthansa, the valence of this experience and the frequency and experience with writing online complaints as well as the participants’ satisfaction with previous complaint handling.

The participants’ previous experience with Lufthansa was measured by asking them if they ever flew with Lufthansa (dummy-coded, 0 = no, 1 = yes; 76.5% said ‘yes’). Participants, who indicated that they flew with Lufthansa before, were directed to a second question asking them how they perceive their experience with Lufthansa (assessed on a 7-point bipolar scale: 1=

“Predominantly negative”; 7= “Predominantly positive”; M = 5.91, SD = 1.15).

In order to assess the participants’ experience with writing an online complaint, they were asked if they had ever written an online complaint (dummy-coded: 0 = no, 1 = yes; 19.4% said

‘yes’) and participants who answered with ‘yes’ received a follow-up question about the fre- quency and how the companies resolved their complaint: ‘How many online complaints have you written in the past year’ (1-2 (84.2%), 3-4 (10.5%), 5-6 (2.6%), more than 6 (2.6%)) and ‘How did the companies resolve your complaint(s)?’ (Assessed on a 7-point bipolar scale: 1= “Pre- dominantly unsatisfactory”; 7= “Predominantly satisfactory”; M = 4.05, SD = 2.17).

TAKE CARE – SAY ‘SORRY’? 26

Results

Manipulation and Confound Checks

Manipulation checks similar to those applied in the pretest were used to ensure that the experimental materials had the intended effects. Respondents, who saw the accommodative re- sponse agreed significantly more (M = 5.63, SD = 1.36) with the statement postulating that

Lufthansa apologises and tries to make it up to the customer, than respondents who saw the de- fensive response (M = 2.00, SD = 1.30; t(160) = 17.34, p < .001). Likewise, respondents in the defensive response condition also perceived Lufthansa’s response to be more defensive than re- spondents in the accommodative condition (Mdefensive = 5.20, SD = 1.84; Maccommodative = 1.75, SD

= 1.35; t(160) = -13.69, p < .001).

Regarding the manipulation of message personalization and the question if Lufthansa re- sponded in an impersonal manner, there was a significant difference between respondents in the personalized response condition and respondents in the corporate response condition (t(160) = -

5.94, p < .001). Furthermore, respondents in the corporate condition indicated significantly more strongly that they perceived Lufthansa’s response to be formal than respondents in the personal- ized condition (t(160) = -7.57, p < .001). In regards to the personalization of the company, par- ticipants in the personalized condition agreed significantly more with the statement that

Lufthansa identified who is responding on behalf of them than participants in the corporate con- dition (t(160) = 8.72, p < .001). Lastly, the manipulation check revealed that there was again a significant difference between the two groups (t(160) = 7.04, p < .001) on the statement that

Lufthansa personalizes the approach to the receiver. All the mean scores for the manipulation of tone of voice can be seen in table 2.

TAKE CARE – SAY ‘SORRY’? 27

Table 2 Mean scores regarding tone of voice manipulation (main study) Condition Personalized Corporate Characteristic M SD M SD Impersonal manner 2.67 1.77 4.43 1.99 Formal tone 3.24 1.80 5.28 1.61 Personalization of the company 5.02 1.84 2.55 1.77 Personalization of the receiver 5.41 1.74 3.33 2.03

Note. N = 160

Confound checks revealed that the participants’ previous experience with Lufthansa (rs =

-.24, p < .001) and the valence of this experience (r = .34, p < .001) were significantly related to one variable, purchase intention. The participants’ previous experience with writing online com- plaints and its frequency were not related to perceived authenticity, purchase intention, PWOM intention or brand attitude and were therefore dropped as covariates. Confound checks addition- ally revealed that the participants’ satisfaction with how companies resolved their previous online complaints was also significantly related to purchase intention (r= .32, p = .05).1

To test the hypothesis and the research question, several statistical analyses were con- ducted. H1 and RQ1 were tested with a Multivariate Analysis of Variance (MANOVA), in which the webcare response type was the independent variable with three conditions (accommo- dative vs. defensive vs. no response), and brand attitude, purchase intention and PWOM inten- tion were used as dependent variables. Because H1 and RQ1 also investigated the effects of the no response condition, they were tested together in one MANOVA. A second MANOVA was conducted to test H2 and H3a-b, in which the independent variable, webcare response type only included the accommodative and defensive condition, as the no response condition was not per- sonalized. Next to the webcare response type, message personalization was added as a second independent variable in the MANOVA, whereas the dependent variables stayed the same. Lastly,

TAKE CARE – SAY ‘SORRY’? 28 to test H4a-b, three bootstrapping analyses were conducted (for each dependent variable) to test whether perceived authenticity was an underlying mechanism for the effects.

Effectiveness of Webcare Response Type

In H1 it was hypothesized that an accommodative webcare response leads to better brand responses than a defensive response and no response on the consumer’s complaint, whereas RQ1 dealt with the question whether a defensive webcare response leads to more positive brand re- sponses than no webcare response. To test H1 and answer RQ1, a MANOVA was performed with webcare type as the independent variable, and brand attitude, PWOM intention and pur- chase intention as dependent variables. Using the Wilks’ Lambda, there was a significant differ- ence in the effects of the different webcare response types on the consumers’ brand attitude, their purchase intention and their PWOM intention (λ = .65, F(6,382) = 15.21, p < .001, pη² = .19).

When the results for the dependent variables were considered separately, all three dependent variables reached statistical significance: Brand attitude (F(2,193) = 44.33, p < .001, pη² = .32), purchase intention (F(2,193) = 4.91, p = .01, pη² = .05) and PWOM intention (F(2,193) = 8.72, p

< .001, pη² = .08). Post hoc analysis was conducted in order to examine which of the three condi- tions significantly differed from each other: The Games-Howell test revealed that the accommo- dative condition significantly differed from the defensive condition and the no response condi- tion. An exception was found for the participants’ purchase intention, where the results only dis- closed a significant difference between the means of the accommodative condition and the de- fensive condition. However, regarding RQ1, the differences between the means of the partici- pants in the defensive condition and the participants in the no response condition did not reach statistical significance (see mean differences in table 3 in Appendix B).2 As predicted by H1, the participants in the accommodative condition reported on the most positive brand attitude (Mac- commodative = 4.79, SD = .97; Mdefensive = 3.34, SD = 1.04; Mno response = 3.12, SD = 1.57), the highest purchase intention (Maccommodative = 3.94, SD = 1.25; Mdefensive = 3.32, SD = 1.29; Mno response = 3.38,

TAKE CARE – SAY ‘SORRY’? 29

SD = 1.55) and highest PWOM intention (Maccommodative = 3.69, SD = 1.40; Mdefensive = 2.96, SD =

1.28; Mno response = 2.71, SD = 1.47). H1 could therefore be supported as accommodative webcare responses lead to the most positive brand responses. Regarding RQ1, the post hoc tests revealed that it cannot be concluded that one of the two strategies is superior to the other, as there was no significant difference between respondents in the defensive webcare condition and respondents in the no response condition.

Effect of Message Personalization on Brand Responses

H2 stated that personalized webcare responses would lead to more positive brand re- sponses than corporate webcare responses. To test the direct effect of the moderator, a second

MANOVA including message personalization (dummy-coded: 1 = personalized, 2 = corporate) and webcare response type (dummy-coded: 1 = accommodative, 2 = defensive) as independent variables and brand attitude, purchase intention and PWOM intention as dependent variables, was conducted. Yet, for this hypothesis was only the direct effect of interest, thus the interaction effect will be explained in the next section.

Although, the p value of the Wilks’ Lambda statistics of .07 showed that the difference between participants receiving personalized webcare responses and corporate webcare responses did not reach significance, the results demonstrated that there is a visible trend towards a differ- ence between these two groups. When considering the values for each dependent variable sepa- rately, significant effects could be reported: Message personalization had a significant effect on the participants’ brand attitude (F(1,158) = 4.23, p = .04, pη²= .03) and their PWOM intention

(F(1,158) = 5.66, p = .02, pη² = .04), but it did not significantly influence the participants’ pur- chase intention (F(1,158) = .51, p = .48, pη² = .00). Participants, who were exposed to a person- alized webcare response indicated to have a higher PWOM intention (Mpersonalized = 3.55, SD =

1.33; Mcorporate = 3.11, SD= 1.41) and a better brand attitude (Mpersonalized = 4.20, SD = 1.24; Mcor- porate = 3.96, SD = 1.24) than participants in the corporate condition. These results only partly

TAKE CARE – SAY ‘SORRY’? 30 supported H2, as personalized messages lead to a more positive brand attitude and a higher

PWOM intention, but the participants’ purchase intention was not significantly affected.

Interaction Effect between Webcare Response Type and Message Personalization

H3a-b postulated that the effects of the different webcare responses on brand responses are moderated by message personalization, such that a) the positive effect of an accommodative webcare response on brand responses will be increased by message personalization, and b) mes- sage personalization will improve the effectiveness of a defensive webcare response in evoking more positive brand responses. The MANOVA which was conducted for H2, was used to exam- ine how the two variables, webcare response type and message personalization, interact with each other on the consumers’ brand responses. The results revealed that there was no significant interaction effect of the different webcare responses and message personalization on the con- sumers’ brand responses (λ = .96, F(3,156) = 2.06, p = .11, pη² = .04). Follow-up analysis re- vealed that when considering the interaction of webcare response type and message personaliza- tion separately on the three dependent variables, there was no significant interaction effect on brand attitude (F(1,158) = 2.59, p =.11, pη² = .02) or purchase intention (F(1,158) = 1.68, p =

.20, pη² = .01), but the results revealed a significant interaction effect on the consumers’ PWOM intention (F(1,158) = 6.19, p = .01, pη² = .04).

However, contrary to the expectations, the results revealed that personalized accommoda- tive responses did not lead to higher PWOM intentions, as participants in the corporate accom- modative response condition reported on a slightly higher PWOM intention (Mpersonalized accommoda- tive = 3.68, SD = 1.44; Mcorporate accommodative = 3.70, SD = 1.37). H3a could therefore not be sup- ported, as there was no significant interaction effect on the participants’ brand attitude and pur- chase intention, and the significant interaction on PWOM intention was oppositional to what was expected for accommodative responses. Nevertheless, H3b can be partly supported as the hy- pothesized interaction effect of a defensive webcare response and message personalization was

TAKE CARE – SAY ‘SORRY’? 31 significant for the participants’ PWOM intention: The personalized defensive response (M =

3.43, SD = 1.21) scored significantly higher on the participants’ PWOM intention than the cor- porate defensive response (M = 2.43, SD = 1.43). The interaction effect of the webcare response type and message personalization on PWOM intention can be seen in figure 2.

Figure 2. Interactionplot for PWOM intention

Interaction Effect between Webcare Response Type and Message Personalization on Brand

Responses explained by Perceived Authenticity

In H4a-b it was hypothesized that there is a mediated moderation effect of webcare re- sponse type and message personalization on brand responses through perceived authenticity, such that a) the positive effect of an accommodative response (vs. a defensive response) on per- ceived authenticity is stronger when the message is also personalized (vs. corporate); and b) con- sequently this positive effect on perceived authenticity results in more positive brand responses.

To investigate the effects, Preacher and Hayes’ (2008) bootstrapping method was used. In total, three bootstrapping analyses were run in the Model 8 of Hayes’ PROCESS macro (2013), in which webcare response type was the independent variable, perceived authenticity the mediator, message personalization the moderator and either brand attitude, purchase intention or PWOM intention the dependent variable. With 1,000 bootstrapping samples (N = 160) to estimate the

TAKE CARE – SAY ‘SORRY’? 32 bias corrected and accelerated confidence intervals (BCACI), the results disclosed that the webcare response type has a direct effect on perceived authenticity (b = -.95, SE = .21, p < .001,

95% BCACI [-1.37, -.53]). Yet, the direct effect of message personalization on perceived au- thenticity (b = -.02, SE = .21, p = .94, 95% BCACI [-.44, .41]) and the interaction of webcare response type and message personalization on perceived authenticity (b = -.45, SE = .43, p = .29,

95% BCACI [-1.30, .39]) was not significant. The findings support that the consumers’ per- ceived authenticity influences their brand responses (H4b), as authenticity has a positive effect on the participants’ brand attitude (b = .45, SE = .06, p < .001, 95% BCACI [.34, .56]), their purchase intention (b = .38, SE = .08, p < .001, 95% BCACI [.22, .56]) and their PWOM inten- tion (b = .48, SE = .07, p < .001, 95% BCACI [.36, .61]). However, the findings reveal that the indirect effect of webcare response strategies moderated by message personalization on brand responses through perceived authenticity was not significantly different from zero, as the 95% bootstrap confidence interval included zero for all dependent variables (Brand attitude: indirect effect = -.10, SE = .10, 95% BCACI [-.31, .08]; purchase intention: indirect effect = -.17, SE =

.17, 95% BCACI [-.55, .11]; PWOM intention: indirect effect = -.22, SE = .21, 95% BCACI [-

.69, .14]). Thus mediated moderation did not take place and H4a-b has to be rejected.

Notwithstanding, the analysis disclosed some unexpected results. As established above, there was a significant effect of the webcare response type on the participants’ perceived authen- ticity, which significantly affected the participants’ brand responses. To test whether perceived authenticity indeed mediated the effects of the different webcare response types on the consum- ers’ brand responses, additional bootstrapping analyses (Model 4 in PROCESS; Hayes, 2013) for each dependent variable were conducted. The webcare response type was the independent varia- ble, perceived authenticity was used as mediator and brand attitude, purchase intention or

PWOM intention was used as dependent variables. The bootstrap procedures (1000 samples, N =

160) revealed that there was indeed a mediated path from webcare response type to the partici- pants’ brand attitude (b = -.43, SE = .09, p < .001, 95% BCACI [-.68, -.24]), their purchase inten-

TAKE CARE – SAY ‘SORRY’? 33 tion (b = -.37, SE = .12, p < .001, 95% BCACI [-.65, -.18]) and their PWOM intention (b = -.47,

SE = .18, p < .001, 95% BCACI [-.77, -.24]). The direct effects of the webcare response type on the participants’ purchase intention and their PWOM intention become insignificant after per- ceived authenticity has been controlled for, thus complete mediation takes place and c’ becomes zero (Purchase intention: b = -.25, p = .20, 95% BCACI [-.64, .13]; PWOM intention: b = -.26, p

= .17, 95% BCACI [-.64, .12]). The direct effects of the webcare response type on the partici- pants’ brand attitude stay significant when authenticity is added to the model, thus partial media- tion is existent (b = -1.01, p < .001, 95% BCACI [-1.27, -.74]).3

Conclusion and Discussion

Companies have come to realize that it is necessary for the means of successful impres- sion management and effective complaint handling, to find appropriate ways to address NWOM on social media (e.g., Fournier & Avery, 2011, Kerkhof et al., 2011). This study investigated the effectiveness of different webcare strategies in evoking favourable brand responses and exam- ined how these responses need to be conveyed in order to maximize their positive effects. In ad- dition, it was researched whether the consumers’ perceived authenticity mediates these effects.

By doing so, this thesis aimed at providing guidelines for marketers on how to use webcare as a marketing tool and deal successfully with NWOM on social media.

This was executed in an experimental study, in which respondents were randomly as- signed to one of the five experimental groups. The experimental groups differed in the type of

Lufthansa’s webcare response (accommodative, defensive or no response) and the degree of per- sonalization of that response (personalized or corporate). After being exposed to Lufthansa’s

Facebook page with the complaint interaction, participants had to indicate their perceived au- thenticity of Lufthansa’s reaction and their brand responses.

The results revealed four interesting findings. First, in support of H1, this study builds on the results of previous research (e.g., Kerkhof et al., 2011; Lee & Song, 2010) postulating that

TAKE CARE – SAY ‘SORRY’? 34 accommodative webcare responses lead to the most positive brand responses among the three different webcare response types in this study (accommodative, defensive and no response). This study extends the scope of previous studies, which concentrated on the effects of organizational responses on customer care outcomes (e.g. satisfaction; Gelbrich & Roschk, 2011) or crisis communication outcomes (e.g. trust, responsibility for crisis; Lee, 2004); by showing that ac- commodative webcare leads to the attainment of favourable marketing-related brand responses.

This study therefore sheds light on the notion of Fournier and Avery (2010), who postulate that brand managers are unsure about what to respond to complaining customers. Furthermore, a re- search question was formulated as previous studies (e.g., Lee, 2004; Lee & Song, 2010) reported mixed results regarding the effectiveness of defensive webcare responses versus no response of the company. Contrary to previous research (e.g., Lee & Song, 2010), the results disclosed that the means of the two conditions did not significantly differ from each other. These findings em- phasize that both response strategies are not effective in evoking favourable brand responses, thus defensive responses and no response to NWOM can therefore not be considered as webcare, since companies using these approaches pass on their opportunity to improve customer relation- ships and neglect to turn negative experiences into positive ones (Dekay, 2012). Whereas previ- ous research (Lee, 2004; Lee & Song, 2010) disagreed on defensive responses or no response leading to higher blame attributions, companies using webcare as marketing tool should have the goal to solve complaints regardless of whose responsibility the issue is. Thus the question “Who is to blame for the incident?” becomes irrelevant and blame attributions are not an insightful out- come variable when taking a marketing perspective on webcare.

Second, the findings from this research revealed that there was a significant effect of message personalization on the participants’ brand responses (H2), but only on their PWOM intention and brand attitude, and not on their purchase intention. It is assumed that there is no direct effect on the participants’ purchase intention, as the purchase of flights is connected to great financial investments and consumers are hesitant to form behavioural intentions about that.

TAKE CARE – SAY ‘SORRY’? 35

H2 could therefore be partially supported, as personalized messages led to higher PWOM inten- tions and a more positive brand attitude than corporate messages. These finding are in line with the Elaboration-Likelihood-Model (Petty & Cacioppo, 1986), which postulates that increasing the relevance of information will enhance its probability of being processed centrally, which in turn accounts for a thorough elaboration of the webcare response. Participants appreciate the personalized approach and perceive personalized webcare messages to be of higher value for them, which was reflected in their more positive attitude towards Lufthansa and their intention to recommend Lufthansa to friends and family. This is in line with previous research findings, as

Kerkhof et al. (2011) argue that a personal tone of voice conveys the feeling that “an organiza- tion cares about the ideas, opinions, and reactions of the public on the incident” (p. 19).

Third, the analysis for H3a-b revealed that the interaction of webcare response type and message personalization was not significant for the participants’ brand attitude and purchase intention, yet exposed again a significant effect for the participants’ PWOM intention. For defen- sive responses it was hypothesized (H3b) and confirmed that a personalized defensive response leads to a higher PWOM intention than a corporate defensive response. This result adds to exist- ing literature by emphasizing that message personalization can be highly beneficial for the effec- tiveness of defensive webcare responses, as it makes up for the nature of defensive responses.

However, contrary to what was expected in H3a, this did not apply for accommodative respons- es: Personalized accommodative responses led to very slightly lower PWOM intentions than corporate accommodative responses, but the difference in the means of the two conditions was so small (.02), that a meaningful interpretation is difficult. Although these results are conflicting with the assumptions of information processing theories, they find support in webcare research from a PR perspective: It has been postulated that accommodative responses, regardless of them being personalized or corporate, evoke the feeling of sympathy and openness as a consumer sympathizes with a company that apologizes (Kerkhof et al., 2011). It seems that there is a ceil- ing effect for brand responses evoked by accommodative strategies, such that the mere use of an

TAKE CARE – SAY ‘SORRY’? 36 accommodative webcare response is enough to evoke more positive brand responses (vs. re- sponding defensively or not responding) and that the consumers’ brand responses cannot be en- hanced by personalizing these accommodative responses.

Lastly, it was hypothesized that the interaction of webcare response type and message personalization on brand responses would be mediated by the consumers’ perceived authenticity of the responses (H4a-b). Although, the results disclosed a significant direct effect of the webcare response type on the participants’ perceived authenticity and a significant effect of au- thenticity on the participants’ brand responses, the interaction of webcare response type and message personalization on the participants’ perceived authenticity was not significant and the mediated moderation could not reach statistical significance, thus H4a-b had to be rejected.

Unexpected Findings

However, the results of H4a-b hinted towards perceived authenticity being a mediator for the effects of the webcare response types on the consumers’ brand responses. This was con- firmed through additional analysis: When authenticity was added to the model, it completely mediated the effects of the webcare response types on purchase intention and PWOM intention, whereas for the consumers’ brand attitude the effects were reduced, but still significant, thus a partial mediation through authenticity took place. In line with past research, these findings sug- gest that favourable brand responses are enhanced if webcare responses are perceived as authen- tic (e.g., Liao & Yang, 2012) and that the authenticity of webcare responses depends on the mes- sage elements and the tone of voice applied to them (e.g., Gruber, 2011). Accommodative webcare responses elicit a higher authenticity perception, due to them containing apologies and compensation, thus being perceived as credible and genuine (e.g., Bell & Zemke, 1987). Per- ceived authenticity has been disregarded in previous studies; hence these results provide an in- teresting insight into the role of authenticity as underlying mechanism explaining the effective- ness of the different webcare responses in evoking favourable brand responses.

TAKE CARE – SAY ‘SORRY’? 37

Limitations

Despite its merits, this research has several limitations. Across all analyses, participants reported on small means in the lower, negative area of the 7-point Likert scales. Whereas brand attitude as attitudinal measure consistently scored the highest, the assessment of the participants’ behavioural intentions, purchase intention and PWOM intention, resulted in lower means. The results have to be considered with caution, as the ‘highest brand attitude/ purchase intention/

PWOM intention’ does not necessarily imply positive values, but rather describe means which are located around the midpoint of the scale or slightly below. There are two possible explana- tions for this finding: Research widely agrees on attitudes influencing intentions (Ajzen, 1991;

Spears & Singh, 2004), which explains that if the brand attitude means are already not very posi- tive, the logical sequence is that the purchase intention means and PWOM intention means will also not be substantially higher. Additionally, it can be said that behavioural intentions depict a higher commitment to a brand than just indicating the attitude. This is critical when the partici- pants’ purchase intention is of interest, as the purchase of flights, being experience products, bears a risk and is mostly done with a lot of consideration due to its costly nature.

Secondly, the stimulus materials of the study resembled a complaint situation on the so- cial network Facebook. While Facebook is an important platform for webcare (Nielsen, 2012), it is not the only social network site on which companies have to deal adequately with NWOM.

The generalizability of the results of this study is limited, as different social media platforms have various specific communication boundaries. On Twitter only 140 characters can be used in every message, which limits companies in their ability to reply in a very thorough fashion. A vast amount of research already examined the effectiveness of message personalization on Twit- ter (e.g., Rybalko & Seltzer, 2010; Kwon & Sung, 2011; Park & Lee, 2013), which underlines that future research is required to determine the effectiveness of webcare in the context other social media platforms. It is assumed that users of the different social media platforms vary in their demographics, their motivations to use the particular platform, their willingness to engage

TAKE CARE – SAY ‘SORRY’? 38 with companies and many more (Burke, 2014). This challenges companies to adapt their webcare according to the specifications of each platform, as identical messages will not achieve the same effects across all platforms. Especially since has now more users than Twit- ter and reports on its users showing very high engagement rates with companies (O’Reilly,

2014), research is needed to assess the webcare challenges companies face on Instagram.

Lastly, it needs to be mentioned that Lufthansa’s pilots struck in Germany during the time of the experiment. This could have affected the experiment’s internal validity, as participants might have been biased because of their annoyance with these circumstances. It is expected that the participants’ disapproval of the strike is reflected in their brand responses, thus their answers were distorted in a negative direction. This also serves as an additional explanation for the small means of the participants’ behavioural intentions, as it is doubtful that participants would express the intention to fly with Lufthansa or be willing to recommend Lufthansa to their friends and family, if they perceived Lufthansa’s service to be unreliable at that time.

Theoretical Implications

Social media platforms facilitate interactions with companies, which led to an increased display of criticism among consumers and a decreased control over a company’s messages

(Hennig-Thurau et al., 2010; Macnamara, 2010). The potential of NWOM to cause immense organizational damage explains the greater need of academic and managerial webcare research.

The first theoretical implication which can be drawn from this research is the debatable effectiveness of message personalization: Whilst a vast amount of research supports the assump- tion that message personalization increases the response’s effectiveness (e.g., Park & Lee, 2013;

Kwon & Sung, 2011), the results of this research raise the question of the right amount of per- sonalization and are therefore in line with research postulating that personalizing messages might not coercively be beneficial (e.g., White, Zahay, Thorbjornsen, & Shavitt, 2008). Willemsen

(2014) argues that too personalized messages in combination with an informal tone of voice can

TAKE CARE – SAY ‘SORRY’? 39 backfire. In this study, message personalization was operationalized by personalizing the compa- ny and the receiver: The customer was addressed by her first name, a lot of informal personal pronouns (“du”) were used and the individual replying on behalf of Lufthansa was introduced by name and signed the response. The results revealed merely partial support of the hypotheses as- suming that message personalization would improve the effectiveness of webcare responses, hence it is demonstrated that a greater amount of personalization elements might not always be better. As the complaint in this research was not written with a very personal tone of voice, the mixed results might have been due to Lufthansa’s response not being perceived appropriate in terms of using the same tone of voice as the complaint did (Becker, 2013; Grundmann, 2015).

Future research is therefore required to uncover the effects of corporate webcare responses which adapt accordingly to the tone of voice used in the complaint. Furthermore, more research is re- quired to identify the “right amount” of message personalization for webcare, especially in refer- ence to webcare responses being conveyed on different social media platforms with varying ap- propriateness of personalized communication (Burke, 2014).

Secondly, this research offers valuable theoretical insights into the importance of authen- ticity of webcare responses. Certain message elements, such as an apology or compensation, can enhance the participants’ perceived authenticity of this response, which in turn leads to more positive brand responses. This is in line with previous research (e. g., Bell & Zemke, 1987;

Gruber, 2011; Huang et al., 2011) and adds to the notion that messages have to be authentic in order to lead to favourable marketing outcomes, such as purchase intention (Liao & Yang, 2012).

This research is the first to identify that perceived authenticity underlies the effects of the differ- ent webcare responses on the consumers’ brand responses, thus being a central variable in de- termining the effectiveness of webcare. Future research can build on these findings and should focus on investigating what message components, besides apologies and compensation, elicit high authenticity perceptions. It would be of further interest to discover the relationship of au- thenticity and the consumers’ justice perception of a webcare response, as the justice perception

TAKE CARE – SAY ‘SORRY’? 40 has a central influence on the consumers’ brand responses (Gelbrich & Roschk, 2011; Orsingher et al., 2009). Authenticity is not only judged upon by the consumer, but also by the observing audience, thus future research should investigate how the authenticity perceptions of these stake- holders change depending on a company’s previous response behaviour, as responding with a standardized messages is perceived as inauthentic (Van Noort et al., forthcoming).

Managerial Implications

Several practical implications can be derived from these findings, which can guide mar- keters in their use of webcare as marketing tool, particularly in regards to the questions about the return of investment of social media activities. A webcare strategy based on the results of this study facilitates companies to keep social media spendings at a minimum and communicate ef- fectively.

Companies should find a positive approach to address NWOM information, by opting for accommodative webcare strategies which incorporate an apology and compensation. Apologies need to be “real” and companies are advised to express sincere regret and display efforts to make it up to the consumers by offering financial or psychological compensation (e.g., a voucher; Gel- brich & Roschk, 2011; Sernovitz, 2012; Sparks & Bradley, 2014; Wirtz & Mattila, 2004). An example for such an accommodative webcare response could be “We are very sorry to hear that.

We want to make it up to you and therefore would like to offer you a voucher”. The importance of apologies should not be underestimated, because regardless of the incident being the compa- ny’s fault, once the complaint is posted on the corporate Facebook page, the company is held responsible for the inconveniences (Champoux, Durgee, & McGlynn, 2012). By conveying an accommodative response in an empathetic manner, companies can signalize their understanding of the consumer’s anger (Min et al., 2014). This research and management literature demonstrate that not responding to consumers on social media is not appropriate, as communication on social media is another form of a business relationship in which ignoring one party is not accepted

TAKE CARE – SAY ‘SORRY’? 41

(Stern, 2011). The above stated applies for complaining customers as well as for the observing audience, which underlines the superiority of webcare compared to traditional customer care, as webcare is an impression management tool for multiple stakeholders.

Companies need to make sure that the content of their webcare response is in accordance with the tone of voice used (Van Noort et al., forthcoming). Message personalization needs to be applied with caution: Companies should know their customers’ expectations, needs, and style of communication, as well as their chosen tone of voice when addressing the company (Becker,

2013; Grundmann, 2015). A mirrored form of address is recommended when conveying a webcare response, which means that organizations react with the same tone of voice as chosen by their customers (Becker, 2013). Although a personal tone of voice is common on social me- dia, management research revealed that this might be different for interactions between compa- nies and consumers: 44% of the consumers indicated that they want to be addressed with the formal “Sie” by companies on social media, whereas only 13% explicitly wished to be addressed with the informal and personal “Du” (Keylens, 2011). These results emphasize that companies need to regard the characteristics of their customers (e.g. age) and the appropriateness of person- alized communication in their business sector (Grundmann, 2015). This research underlines that it is not advisable to overemphasize the personal tone of voice, but companies should abstain from addressing consumers with a strictly corporate voice on social media and rather engage in dialogic communication in which they introduce themselves with their real names, take a friend- ly tone and tailor the message to the individual customer (e.g., Sernovitz, 2012).

This omits to the last practical implication: Companies need to create their responses in a way that they come across as authentic, as authenticity has a positive effect on the consumers’ brand responses. This, again, underlines the vital importance of apologies, as consumers’ value their sincerity and genuineness, which elicits high authenticity perceptions (e.g., Gruber, 2011).

Sincerity, credibility and genuineness should therefore be irreplaceable components of an organ- izational webcare response. Organizations would be wise to use a personalized style of commu-

TAKE CARE – SAY ‘SORRY’? 42 nication on social media, as this facilitates the feeling of openness and honesty, both being cen- tral for authenticity (Pleil & Rehn, 2010). The informal personal pronoun “du” can enhance the perceived authenticity of a company’s response, but only if the target group displays the willing- ness to communicate with a company on a more personal level (Grundmann, 2015). Another possibility to increase authenticity of a message is to paraphrase the customer’s issue to signalize that the complaint was read thoroughly, the customer is being taken seriously and the company is actively listening to the customer’s concerns (Min et al., 2014). This emphasizes that webcare is a highly customer-centralized marketing tool, which aims at providing valuable and satisfying content for each individual consumer.

Notes

1. Although confound checks revealed that there were three control variables (previous ex-

perience with Lufthansa, the valence of this experience and satisfaction with previous

complaint handling) significantly related to the participants’ purchase intention, the co-

variates were excluded from the analysis: Firstly, the participants’ satisfaction with pre-

vious complaint handling was a follow-up question and only 38 participants answered it.

Tabachnick and Fidell (2007) argue that in order to ensure adequate power, sample sizes

in each cell must be sufficient. This control variable did not manifest sufficient power as

there were only 5 respondents in the no response condition, 15 respondents in the ac-

commodative response condition and 17 respondents in the defensive response condition.

Thus, generalizability was not possible and this variable was dropped as covariate.

Secondly, for the other two covariates, separate ANVOVAs and t-tests with mes-

sage personalization or webcare response type as independent variables and each control

variable as dependent variable were conducted. They revealed that the participants’ pre-

vious experience with Lufthansa did not significantly differ between personalized and

corporate webcare responses (t(160) = .27, p = .79) or across the three webcare condi-

TAKE CARE – SAY ‘SORRY’? 43

tions (F(2,193) = 1.79, p = .17, pη² = .02). Regarding the participants’ perceived valence

of this experience, two additional ANOVAs disclosed that there was no significant dif-

ference in the mean scores of participants in the personalized and in the corporate condi-

tion (F(1,120) = 2.77, p = .10, pη² = .02), nor were there significant differences across the

different webcare conditions (F(2,147) = .29, p = .75, pη² = .02). This serves as an expla-

nation why the participants’ previous experience with Lufthansa and the valence of this

experience were also not included as covariates in the analysis.

2. See Field (2009) for discussion which post hoc test is appropriate if equal group sizes

cannot be assumed.

3. See Tabachnick and Fidell (2007) for an explanation of the effects in a complete and par-

tial mediation.

References

Ajzen, I. (1991). The Theory of Planned Behavior. Organizational Behaviour and Human Deci-

sion Processes, 50,179-211.

Becker, S. (2013, March 28). Dürfen wir „Du“ zu Ihnen sagen? [Weblog Marketing]. Retrieved

from http://www.weblogmarketing.de/durfen-wir-du-zu-ihnen-sagen/

Bell, C. R. & Zemke, R. E. (1987). Service breakdown: The road to recovery. Management Re-

view, 76(10), 32-35.

Boulding, W., Kalra, A., Staelin, R., & Zeithaml, V. A. (1993). A dynamic process model of

service quality: From expectations to behavioural intentions. Journal of Marketing Re-

search, 30, 7-27.

Breitsohl, J., Khammmash, M., & Griffiths, G. (2010). E-business complaint management: Per-

ceptions and perspectives of online credibility. Journal of Enterprise Information Man-

agement, 23(5), 653-660.

TAKE CARE – SAY ‘SORRY’? 44

Burke, S. (2014, July 15). Which social media platform is right for your business? [Spokal Web-

log] Retrieved from http://www.getspokal.com/which-social-media-platform-is-right-for-

your-business/

Champoux, V., Durgee, J., & McGlynn, L. (2012). Corporate Facebook pages: when “fans” at-

tack. Journal of Business Strategy, 33(2), 22-30.

Constantinides, E., & Fountain, S. J. (2008). Web 2.0: Conceptual foundations and marketing

issues. Journal of Direct, Data and Digital Marketing Practice, 9(3), 231-244.

Coyle, J. R. & Thorson, E. (2001). The effects of progressive levels of interactivity and vivid-

ness in web marketing sites. Journal of Advertising, 30(3), 65-77.

Davidow, M. (2003). Organizational responses to customer complaints: What works and

what does not. Journal of Service Research, 5(3), 225-250.

Dekay, S. H. (2012). How large companies react to negative Facebook comments. Corporate

Communications: An International Journal, 17(3), 289-299.

Estelami, H. (2000). Competitive and procedural determinants of delight and disappointment in

consumer complaint outcomes. Journal of Service Research, 2(3), 285-300.

Field, A. (2009). Discovering statistics using SPSS (3rd ed.). London: SAGE Publications.

Fournier, S. & Avery, J. (2011). The uninvited brand. Business Horizons, 54, 193 – 207.

Gelbrich, K., & Roschk, H. (2011). A meta-analysis of organizational complaint handling and

customer responses. Journal of Service Research, 14(1), 24-43.

Greyser, S. A. (2009). Corporate brand reputation and crisis management. Management Deci-

sion, 47(4), 590-602.

TAKE CARE – SAY ‘SORRY’? 45

Gruber, T. (2011). I want to believe they really care. Journal of Service Management, 22(1), 85-

110.

Grundmann, M. (2015, January 5). „Du“ oder „Sie“ – Tipps für die richtige Kundenansprache

in Social Media [SoMe Weblog]. Retrieved from http://www.hashtag-some.de/du-oder-sie-

tipps-fuer-die-richtige-kundenansprache-in-social-media/

Hansson, L., Wrangmo, A., & Solberg Soilen, K. (2013). Optimal ways for companies to use

Facebook as a marketing channel. Journal of Information, Communication and Ethics in

Society, 11(2), 112-126.

Hayes, A. F. (2013). Mediation, Moderation, and Conditional Process Analysis. New York: The

Guilford Press.

Hennig-Thurau, T., Malthouse, E. C., Friege, C., Gensler, S., Lobschat, L., Rangaswamy, A., &

Skiera, B. (2010). The impact of new media on customer relationships. Journal of Service

Research, 13(3), 311-330.

Holloway, B. B., & Beatty, S. E. (2003). Service failure in online retailing: A recovery oppor-

tunity. Journal of Service Research, 6(1), 92-105.

Huang, M., Cai, F., Tsang, A. S. L., & Zhou, N. (2011). Making your online voice loud: The

critical role of WOM information. European Journal of Marketing, 45(7/8), 1277-1297.

Kelleher, T. (2009). Conversational voice, communicated commitment, and public relations out-

comes in interactive online communication. Journal of Communication, 59, 172–188.

Kerkhof, P., Beugels, D., Utz, S., & Beukeboom, C. (2011). Crisis PR in social media. An exper-

imental study of the effects of organizational crisis responses on Facebook. Paper present-

ed at the 61st Annual ICA Conference, Boston (USA), 26-30 May 2011.

TAKE CARE – SAY ‘SORRY’? 46

Keylens (2011). Kundenerwartungen im Social Web. Retrieved from

http://www.keylens.com/fileadmin/web_data/related_links/KEYLENS-

201208_CMO_Studie-V5.pdf

Kim, T. T., Kim, W. G., & Kim, H.-B. (2009). The effects of perceived justice on recovery satis-

faction, trust, word-of-mouth, and revisit intention in upscale hotels. Tourism Manage-

ment, 30, 51-62.

Kreuter, M. W., Bull, F. C., Clark, E. M., & Oswald, D. L. (1999). Understanding how people

process health information: A comparison of tailored and nontailored weight-loss materi-

als. Health Psychology, 18(5), 487-494.

Kwon, E. S. & Sung, Y. (2011). Follow me! Global marketers' Twitter use. Journal of Interac-

tive Advertising, 12(1), 4-16.

Lee, B. K. (2004). Audience-oriented approach to crisis communication: A study of Hong Kong

consumers’ evaluation of organizational crisis. Communication Research, 31(5), 600-618.

Lee, Y. L., & Song, S. (2010). An empirical investigation of electronic word-of-mouth: Informa-

tional motive and corporate response strategy. Computers in Human Behavior, 26(5),

1073-1080.

Liao, J., & Yang, J. (2012). The influence of electronic word-of-mouth authenticity on custom-

ers’ behavior. Paper presented at the Second International Conference on Business Compu-

ting and Global Informatization. doi: 10.1109/BCGIN.2012.117

Macnamara, J. (2010). ‘Emergent’ media and public communication: Understanding the chang-

ing mediascape. Public Communication Review, 1(2), 3-17.

Maxham, J. G. (2001). Service recovery’s influence on consumer satisfaction, positive word-of-

mouth, and purchase intentions. Journal of Business Research, 54, 11-24.

TAKE CARE – SAY ‘SORRY’? 47

Maxham, J. G., & Netemeyer, R. G. (2002). Modeling customer perceptions of complaint han-

dling over time: the effects of perceived justice on satisfaction and intent. Journal of Re-

tailing, 78(4), 239-252.

Min, H., Lim, Y., & Magnini, V. P. (2014). Factors affecting customers satisfaction in responses

to negative online hotel reviews: The impact of empathy, paraphrasing, and speed. Cornell

Hospitality Quarterly, Advance online publication. doi: 10.1177/1938965514560014

Mitchell, A. A. & Olson, J. C. (1981). Are product attribute beliefs the only mediator of advertis-

ing effects on brand responses? Journal of Marketing Research, 18(3), 318- 332.

Mitra, A., & Watts, E. (2002). Theorizing cyberspace: the idea of voice applied to the internet

discourse. New media & society, 4(4), 479-498.

Nielsen (2012). State of the media: The social media report 2012. Retrieved from

http://www.nielsen.com/content/dam/corporate/us/en/reports-downloads/2012-

Reports/The-Social-Media-Report-2012.pdf

O’Reilly, L. (2014, December 17). Now Instagram is dominating Twitter in another hugely im-

portant way. Business Insider UK. Retrieved from http://uk.businessinsider.com/instagram-

engagement-rates-up-to-50-times-higher-than-twitter-socialbakers-finds-2014-12?r=US.

Orsingher, C., Valentini, S., & de Angelis, M. (2010). A meta-analysis of satisfaction with com-

plaint handling in services. Journal of the Academy of Marketing Science, 38(2), 169-186.

Park, H., & Lee, H. (2013). Show us you are real: The effect of human-versus-organizational

presence on online relationship building through social networking sites. Cyberpsycholo-

gy, Behavior, and Social Networking, 16(4), 265-271.

Petty, R. E., & Cacioppo, J. T. (1986). The elaboration likelihood model of persuasion. Advances

in experimental social psychology, 19, 123-205.

TAKE CARE – SAY ‘SORRY’? 48

Pleil, T., & Rehn, D. (2010). Authentizität im Social Web. Erwartungen der Community an die

PR. prmagazin, 2, 61-66.

Preacher, K. J., & Hayes, A. F. (2008). Asymptotic and resampling strategies for assessing and

comparing indirect effects in multiple mediator models. Behavior Research Methods,

40(3), 879-891.

Raney, A. A., Arpan, L. M., Pashupati, K., & Brill, D. A. (2003). At the movies, on the web: An

investigation of the effects of entertaining and interactive web content on site and brand re-

sponses. Journal of Interactive Marketing, 17(4), 38-53.

Rieh, S.Y. (2002). Judgment of information quality and cognitive authority in the web. Journal

of the American Society for Information Science and Technology, 53(2), 145-161.

Rossiter, J. R. (2002). The C-OAR-SE procedure for scale development in marketing. Interna-

tional Journal of Research in Marketing, 19(4), 305-335.

Rybalko, S. & Seltzer, T. (2010). Dialogic communication in 140 characters or less. How For-

tune 500 companies engage stakeholders using Twitter. Public Relations Review, 36(4),

336-341.

Sernovitz, A. (2012, February 22). 10 ways to deal with upset customers using social media.

Social Media Examiner. Retrieved from http://www.socialmediaexaminer.com/10-ways-

to-deal-with-upset-customers-using-social-media/.

Shields, P. O. (2006). Customer correspondence: Corporate responses and customer reactions.

The Marketing Management Journal, 16(2), 155-170.

Sparks, B. A., & Bradley, G. L. (2014). A “Triple A” typology of responding to negative con-

sumer-generated online reviews. Journal of Hospitality & Tourism Research. Advance

online publication. doi: 10.1177/1096348014538052.

TAKE CARE – SAY ‘SORRY’? 49

Spears, N., & Singh, S. N. (2004). Measuring attitude toward the brand and purchase intentions.

Journal of Current Issues and Research in Advertising, 26(2), 53-66.

Stern, G. M. (2011, November 22). Facebook complaints: Companies are asleep at the wheel.

Fortune. Retrieved from http://fortune.com/2011/11/22/facebook-complaints-companies-

are-asleep-at-the-wheel-2/.

Tabachnick, B. G., & Fidell, L. S. (2007). Using multivariate statistics (5th ed.). Boston, MA:

Pearson.

Tam, K. Y., & Ho, S. Y. (2005). Web personalization as persuasion strategy: An Elaboration

Likelihood Model Perspective. Information Systems Research, 16(3), 271-291.

Tam, K. Y, & Ho, S. Y. (2006). Understanding the impact of web personalization on user infor-

mation processing and decision outcomes. MIS Quarterly, 30(4), 865-890.

Tax, S. S., Brown, S. W., & Chandrashekaran, M. (1998). Customer evaluations of service com-

plaint experiences: Implications for relationship marketing. The Journal of Marketing,

62(2), 60-76.

Van Noort, G., & Willemsen, L. M. (2011). Online damage control: The effects of proactive

versus reactive webcare interventions in consumer-generated and brand-generated plat-

forms. Journal of Interactive Marketing, 26(3), 131-140.

Van Noort, G., Willemsen, L. M., Kerkhof, P., Verhoeven, J.W.M. (forthcoming). Webcare as

an integrative tool for customer care, reputation management, and online marketing: A lite-

rature review. In Philip J. Kitchen and Ebru Uzunoglu (Eds), Integrated Communications

in the Post-Modern Era.

Voorveld, H. A. M., Neijens, P. C. & Smit, E. G. (2011). Opening the black box. Understanding

cross-media effects. Journal of Marketing Communications, 17(2), 69-85.

TAKE CARE – SAY ‘SORRY’? 50

White, T. B., Zahay, D. L., Thorbjornsen, H., & Shavitt, S. (2008). Getting too personal: Reac-

tance to highly personalized email solicitations. Marketing Letters, 19(1), 39-50.

Willemsen, L. (2013). Electronic word of mouth. Challenges for consumers and companies (Un-

published doctoral dissertation). University of Amsterdam, Amsterdam.

Willemsen, L. (2014, September 3). Hoe persoonlijk moet je zijn in webcare? [SWOCC Web-

log]. Retrieved from http://www.swocc.nl/kennisbank-item/hoe-persoonlijk-moet-je-zijn-

in-webcare/

Willemsen, L. M., Neijens, P. C., Bronner, F. & De Ridder, J. A. (2011). “Highly recommend-

ed!” The content characteristics and perceived usefulness of online consumer reviews.

Journal of Computer-Mediated Communication, 17(1), 19-38.

Willemsen, L. M., Neijens, P. C., & Bronner, F. A. (2013). Webcare as customer relationship

and reputation management? Motives for negative electronic word of mouth and their ef-

fect on webcare receptiveness. In S. Rosengren et al. (Eds.), Advances in Advertising Re-

search (Vol. IV) (pp. 55-69). Wiesbaden: Springer. doi: 10.1007/978-3-658-02365-2_5

Wirtz, J., & Mattila, A. S. (2003). Consumer responses to compensation, speed of recovery and

apology after a service failure. International Journal of Service Industry Management,

15(2), 150-166.

Xia, J. (2013). Effects of companies’ responses to consumer criticism on social media. Interna-

tional Journal of Electronic Commerce, 17(4), 73-99.

TAKE CARE – SAY ‘SORRY’? 51

Appendix A

Research Stimuli

Figure 3. Personalized accommodative webcare

TAKE CARE – SAY ‘SORRY’? 52

Figure 4. Corporate accommodative webcare

Figure 5. Personalized defensive webcare

TAKE CARE – SAY ‘SORRY’? 53

Figure 6. Corporate defenisve webcare

Figure 7. No webcare response

TAKE CARE – SAY ‘SORRY’? 54

Appendix B

Table 3. Post Hoc tests for H1 and RQ1 per dependent variable

Games-Howell

Dependent variables Brand attitude Purchase intention PWOM intention

Webcare response type Mdifference 95% CI Mdifference 95% CI Mdifference 95% CI Accommodative Defensive 1.45* (.16) [1.08, 1.83] .62* (.20) [.14, 1.09] .72* (.21) [.23, 1.22]

No response 1.67* (.29) [.96, 2.37] .55 (.30) [-.17, 1.28] .98* (.30) [.27, 1.69]

Defensive Accommodative -1.45* (.16) [-1.83, -1.08] -.62* (.20) [-1.09, -.14] -.72* (.21) [-1.22, -.23]

No response .22 (.29) [-.49, .93] -.06 (.30) [-.79, .67] .26 (.29) [-.44, .95]

No response Accommodative -1.67* (.29) [-2.37, -.97] -.55 (.30) [-1.28, .17] -.98* (.30) [-1.69, -.27]

Defensive -.22 (29) [-.93, 49] .06 (.30) [-.67, .79] -.26 (.29) [-.95, .44]

Note. Standard errors are in parentheses. CI = confidence interval. *p < .05. Mdifference = Mean difference of the compared conditions