Managing e-NWOM: Webcare as Online Reputation Management
An Experimental Study Examining the Effects of the Communication Style used in Webcare, the Context of the Message and the Role of Engagement on the Perceived Reputation
Esmee Roetman 10533370 Master’s Thesis
Graduate School of Communication Master’s programme Corporate Communication
dr. S.C. (Suzanne) de Bakker 20/06/2017
2 Abstract
Nowadays, every stakeholder can voice its opinion on social media. Opinions and
issues voiced on social media are accessible for a large public and can therefore improve or
seriously harm the reputation of an organization. That is why many organizations have a
webcare team responding to the complaints of stakeholders. Webcare serves as a tool in
support of customer relationship, reputation and brand management. This research attempted
to fill the gap by examining to what extent the communication style used in webcare
(conversational human voice vs. corporate voice) and the context in which a webcare
response was posted (single comment vs. multiple comments), predicts the perceived
reputation. Besides, engagement was added as a possible moderator in the relationship
between the communication style and the perceived reputation. These relationships were
investigated by conducting an online experimental survey among 169 participants. In this
study, a fictional Facebook message from KLM – Royal Dutch Airlines was chosen. This
study only found a statistically significant effect of the communication style on the perceived
reputation. The results indicate that the perceived reputation is more positive when KLM uses
a conversational human voice in webcare than a corporate voice. There were no statistically
significant effects found for the context in which a webcare response was posted and
engagement did not moderate the relationship. Therefore, future research should examine
other organizations and organizations in different sectors. Besides that, given the dialogical
nature of a conversational human voice, a conversation does not end at one single webcare
3 Introduction
New technologies like social media make it possible for organizations to communicate
directly with their stakeholders (Valentini, 2015). Social media are an element of the
communication mix that organizations need to deal with and that have grown in importance in
the communication mix of European organizations (Verhoeven, Tench, Zerfass, Moreno &
Verčič, 2012). The introduction of it has dramatically impacted and transformed day-to-day activities of public relations practitioners (Moreno, Navarro, Tench & Zerfass, 2015).
There are numerous advantages and opportunities that social media have to offer. As
an example, they enable a more symmetrical and two-way communication between the
organization and their publics and these are seen as essential to building mutually beneficial
relationships (Valentini, 2015). However, there are also definitely some drawbacks attached
to it. Through social media, every stakeholder can voice its opinion. This offers some serious
threats for organizations, as it can even harm their reputation (Verhoeven et al., 2012).
Issues voiced on social media by different stakeholders are accessible for a large
public and can therefore cause serious damage to an organization. A well-known example is
that of a complaint by a famous Dutch comedian, Youp van ‘t Hek, about the negative
experiences his son had with the telecommunications provider T-Mobile. Youp van ‘t Hek
reported negative sentiments regarding T-Mobile’s customer service on his Twitter account
when his son got a problem with his contract. He was certainly not happy with the fact that
the company apologized via Twitter. As a consequence, Youp van ‘t Hek used his Twitter
account to report even more negative sentiments regarding T-Mobile and he invited other
consumers to do so as well. This created a vicious circle of electronic negative word of mouth
(e-NWOM) and eventually reputational damage for the telecommunications provider
T-Mobile (Van Noort & Willemsen, 2011). This example shows how relevant it is for
4 for participation which may pose a serious risk of reputation damage for organizations.
Social media, in comparison to traditional media, have a great potential to make the
public relations profession more strategic, two-way, interactive, symmetrical or dialogical
(Grunig, 2009). Increasingly, stakeholders want to engage with organizations on social media.
That is why many organizations now have a webcare team. Webcare can be seen as a tool for
online reputation management. It is an example of communications that organizations use to
respond to online complaints and other (negative) messages from stakeholders. Through
webcare, organizations try to protect the organization’s reputation and prevent reputational damage. Webcare can be defined as follows: “The act of engaging in online interactions with (complaining) consumers, by actively searching the web to address consumer feedback (e.g.,
questions, concerns and complaints). Webcare is performed by one or more company
representatives (i.e., webcare teams) and serves as a tool in support of customer relationship,
reputation and brand management” (Van Noort & Willemsen, 2011, p. 3).
Previous research (Van Noort, Willemsen, Kerkhof & Verhoeven, 2015) has already
shown that a webcare response provided by an organization is more effective than no webcare
response. More specifically, when corporate feedback to a complaint is provided, this
positively influences the relationship between the organization and the public (Van Noort et
al., 2015). Besides, consumers are more satisfied when an organization posted a response in
reply to a negative comment than when a response was lacking (Van Noort et al., 2015).
Thus, consumers appreciate it more when organizations respond to questions or comments
than when they do not respond (Schultz, Utz & Göritz, 2011).
Previous research has also shown that not only the webcare strategy (reactive vs.
proactive) used is important for an organization when they engage with their stakeholders,
also the used communication style is an essential part. Prior research (Van Noort &
5 strategy is reactive in response to NWOM instead of proactive. The effect appeared to be
mediated by a communication style, namely the use of a conversational human voice versus a
corporate voice. Such a voice reflects attributes as being open to dialog, treat others as human
and show empathy (Van Noort et al., 2015). A conversational human voice also appeared to
be a mediating factor in the study of Dijkmans, Kerkhof, Buyukcan-Tetik and Beukeboom
(2015). Based on survey research, the authors demonstrated a positive reputational effect of a
consumers’ exposure to a company’s social media activities and the mediating role of a conversational human voice in this relation. Furthermore, Schamari and Schaefers (2015)
revealed that webcare can be used to increase consumer engagement on consumer-generated
platforms and this effect is explained by consumers’ perceptions of a brand’s conversational communication style. Thus, there is some correlational as well as experimental evidence for
the positive outcomes of a conversational human voice: the more an organization is perceived
as using a conversational style communication, the more the organization is able to foster
trust, commitment, satisfaction and control mutuality (Kelleher & Miller, 2006).
Still, there is not much research yet on how different communication styles
(conversational human voice vs. corporate voice) affect the reputation of the organization and
to what extent the context in which a webcare response is posted (single comment vs. multiple
comments) affects the reputation. The studies discussed earlier usually examine the effects of
a webcare response in reply to a single negative comment of a customer. However, in reality,
it is possible that negative comments are sometimes posted adjacent to or in response to other
negative comments (Van Noort et al., 2015). It is highly relevant for organizations to know
whether there are differences in the effectiveness of webcare with regard to the amount of
e-NWOM messages.
Besides this, it is still unclear to what extent the organization’s history with a customer
6 consumers who do or do not have a strong relationship with the organization (Van Noort et
al., 2015). Being interested in an organization, having positive feelings about it and thus
feeling more engaged can strengthen feelings of injustice when organizations fail to solve the
problem in a satisfying way (Van Noort et al., 2015). That is why in this study, the level of
engagement will also be taken into account because it is important that an organization is
attentive to who it is dealing with.
Therefore, this study fills these gaps. This study examines and adds to the literature
other factors that have not been examined before and which might have an influence on and
determine the reputation of an organization. This research thus gives a broader and more
complete perspective of webcare in organizations and that is why it is an addition to scientific
literature. It is investigated to what extent the communication style used (conversational
human voice vs. corporate voice) influences the effectiveness of webcare looking at the
perceived reputation. This experimental study also tried to fill the gap by taking into account
the context in which a webcare response is posted (single negative comment vs. multiple
negative comments) and what role the level of engagement plays in the relationship between
the communication style used and the perceived reputation. This leads to the following two
research questions which will be answered in this study:
RQ 1: “To what extent does the communication style used in webcare and the context in which a webcare response is posted predict the perceived reputation of that organization?”
RQ 2: “What role does engagement play in the relationship between the communication style used on social media and the perceived reputation of that organization?”
Theoretical framework
In this section, the perceived reputation and webcare will be discussed as well as the
7 influence of the context in which a webcare response is posted (single negative comment vs.
multiple negative comments) and the concept of engagement. The relationships between these
concepts are shown in Figure 1 at the end of this chapter.
Perceived reputation and Webcare
Reputation is an intangible asset that provides organizations with sustainable
competitive advantage in the marketplace (Ponzi, Fombrun & Gardberg, 2011). It is a
valuable asset and one of the organization’s most important resources. Perceived reputation can be defined as a construct that describes the perceptions of multiple stakeholders about an
organization (Ponzi et al., 2011). These perceptions are formed based upon the organization’s
past, present and future activities and the way in which these activities are communicated
(Tucker & Melewar, 2005). People rely routinely on the reputation of an organization in
making investment choices, career decisions and product choices (Fombrun & Shanley,
1990). A bad reputation can translate into financial damage and can even threaten the
organization’s survival (Coombs & Holladay, 1996). That is why in this study, the main focus is on what factors in webcaredetermine the reputation of an organization.
Research has already shown that negative expressions from stakeholders have a
stronger impact on the perception of other stakeholders than positive expressions (Baumeister,
Bratslavsky, Finkenauer & Vohs, 2001). This is because negativity can be linked to change
and more explicit emotions than positive expressions. Thus, organizations do good to respond
adequately to these negative expressions to prevent reputation damage.
Communication style
The communication style used by webcare teams can have an impact on the reputation
of that organization (Kelleher & Miller, 2006). More specifically, the communication style
8 satisfaction and control mutuality (Kelleher & Miller, 2006). A conversational human voice is
such a communication style that can be used by organizations in webcare to communicate
with stakeholders. According to Kelleher (2009), a conversational human voice is an
engaging and natural style of organizational communication. It reflects attributes such as
being open to dialog and providing prompt feedback, but also attributes that are typically not
associated with corporate communications such as communicating with a sense of humor,
admitting mistakes and treating others as human (Van Noort et al., 2015). Using this
communication style, organizations “humanize” the corporate voice. A corporate voice
includes that organizations speak with one voice and one identity (Levine, Locke, Searls &
Weinberger, 2000). Also, organizations using a corporate voice speak more formally with the
use of a distant corporate tone and language in their communications (González-Herrero &
Smith, 2008).
A conversational human voice appears to be an effective communication style for
organizations to react on stakeholders. For example, it has a positive impact on brand
evaluations (Van Noort & Willemsen, 2011). Why it is such an effective communication style
can also be attributed to the Social Presence Theory, which implies that an online medium
with a high social presence will convey a social context and provide two-way communication
and interaction (Cui, Lockee & Meng, 2013). Within this framework, the concept of social
presence is defined as the degree to which a person is perceived as a “real person” (Park &
Cameron, 2014). Based on this theory, a conversational human voice appears to be an
effective communication style for organizations to respond to comments and questions on
social media because interpersonal communication can be established (Park & Cameron,
2014). Organizations active on for example Twitter attempt to bring human personality to
organizational communication by using human representatives, personal pronouns and
9 Yet organizations still use a corporate voice (also called professional voice) in their
communications on social media, because organizations want to convey one specific identity
and want to speak formally with a distant corporate tone (Levine et al., 2000;
González-Herrero & Smith, 2008). This however, is often perceived by stakeholders as persuasive and
profit-driven and would therefore not be beneficial for organizations when responding to
stakeholders’ complaints. Given the positive influence of the conversational human voice on relational aspects of corporate reputation, it does not seem desirable to respond to complaints
using a corporate voice. Therefore, expected is that a conversational human voice has a more
positive impact on the perceived reputation than a corporate voice. The following hypothesis
is formulated and will be tested in this study:
Hypothesis 1 (H1): The perceived reputation will be more positive when the communication
style used in webcare is a conversational human voice than when it is a corporate voice.
Context
Context is a very broad concept. In this study the surrounding sentiment is particularly
relevant, meaning that the context in which a webcare response is posted can be defined as the
amount of e-NWOM messages related to one specific stakeholder’s post. Previous studies
have usually examined the effects of a webcare response in reply to one single negative
comment (Van Noort & Willemsen, 2011). However, negative comments on social media are
often posted adjacent to or in response to other negative comments. This is because the
presence of someone exhibiting a behaviour of a positive or negative valence, increases the
probability that other observers show the behaviour of the same valence (Schaefers &
Schamari, 2016). Therefore, when one negative comment on social media is posted, others
might follow as well and exhibit the same behaviour.
10 are called “observers”. Those are the ones who only read other consumers’ negative posts. These observers search for negative information about a certain brand and evaluate the brand
based on these negative posts. An e-NWOM observer might be influenced by different posts
on social media because a high level of consistency can create a sense of a social norm (Kim
et al., 2016). After the observer reads e-NWOM in high levels of consistency, he will thus
conform to the social norm and form negative attitudes towards the organization, which can
negatively affect future purchase decisions (Kim et al., 2016).
Negative online interactions between consumers are already found to have negative
effects on the consumer’s decision-making process, including brand evaluations and brand choice (Van Noort & Willemsen, 2011). It is already known that consumers are more satisfied
when an organization posts a response in reply to such a negative comment than when a
response is lacking (Van Noort et al., 2015). However, based on the fact that a high level of
consistency in e-NWOM can create a sense of a social norm, it is expected that the perceived
reputation of an organization will be less positive when the webcare response is surrounded
by more negative comments. This suggests that the sentiment of all these negative comments
together, are likely to influence the perceived reputation. The following second hypothesis
can be formulated:
Hypothesis 2 (H2): The perceived reputation will be more positive when the webcare
response is posted in response to one single negative comment than when the webcare response is surrounded by more negative comments.
Moreover, a possible interaction effect will be investigated between the communication style
used and the context in which a webcare response is posted. It is already known that a
conversational human voice has a positive influence on relational aspects of corporate
reputation (Van Noort & Willemsen, 2011). Besides this, a study done by Purnawirawan
11 comment when it is surrounded by more negative comments than when it is surrounded by
positive comments. Putting more effort in such a comment is important in order to diminish
the negative effects, because a high level of consistency in e-NWOM can create a sense of a
social norm among observers. Based on this, the following hypothesis can be formulated:
Hypothesis 3 (H3): The perceived reputation will be more positive when the communication
style used in webcare is a conversational human voice than when it is a corporate voice, but this effect will be more pronounced when the webcare response is posted in response to one single negative comment than when it is surrounded by more negative comments.
Engagement
In this study, the level of engagement will also be taken into account. Engagement can
be defined in terms of a combination of cognitive aspects (e.g. being interested in the
activities of a company), emotional aspects (having positive feelings about the activities of a
company) and/or behavioural aspects (e.g. participation in the activities of a company)
(Dijkmans, Kerkhof & Beukeboom, 2015). Achieving a high level of engagement is viewed
as desirable, because it may enhance a company’s reputation and brand loyalty. Previous research has already found that the level of engagement in a company’s social media activities is positively related to the corporate reputation (Dijkmans et al., 2015). Furthermore,
Schamari and Schaefers (2015) have shown that webcare directed at positive consumer
engagement can reinforce engagement intentions. Reacting to comments and questions on
social media increases the perception among observing consumers that a brand exhibits a
natural and engaging style of communication (= conversational human voice), which
increases engagement intentions.
Most studies focused on the direct effects of webcare responses and reputation related
12 2011). Additionally, a study done by Yang, Kang and Johnson (2010) studied engagement as
a mediator. It appeared that in a crisis situation, the conversational human voice was
important in creating engagement, leading to a more positive perception and evaluation (Yang
et al., 2010). The results of this study found that openness to dialogic communication is
important in creating and enhancing engagement among people. This is due to the fact that
people intend to enjoy having a dialogue with a “real” person. When people feel more engaged, the message can create positive affective reactions, positive company attitudes and
supportive WOM intentions (Yang et al., 2010).
However, as stated before, the communication style used by webcare teams (a
conversational human voice) can also have a direct impact on the reputation of that
organization. More specifically, the communication style can influence relational aspects of
corporate reputation, like commitment, trust, satisfaction and control mutuality (Kelleher &
Miller, 2006). This study looks at whether this relationship between the communication style
used and the perceived reputation is strengthened by the level of engagement and thus
whether this is a moderator. Based on the positive relationship between the conversational
human voice and engagement and the level of engagement in relation to the corporate
reputation, expected is that the effectiveness of a webcare response (looking at the perceived
reputation) might actually differ for consumers who differ in their level of engagement. Being
interested in an organization, having positive feelings about it and thus feeling more engaged
can strengthen feelings of injustice when organizations fail to solve the problem in a
satisfying way (Van Noort et al., 2015). This can eventually negatively affect the
stakeholder’s perception of the organization. Thus, the level of engagement might play a moderating role in the relationship between the communication style used and the perceived
13
Hypothesis 4 (H4): The effect of an organization’s social media communication style on the
perceived reputation will be moderated by the level of engagement.
Figure 1. Conceptual model
Method
Design
To answer the research questions and test the hypotheses in this study, an online
experimental study has been conducted. The experimental design in this study is a 2x2 Communication style Conversational human voice Corporate voice Context Single comment Multiple comments Engagement Perceived reputation H1 H2 H3 H4
14 between-subjects factorial design: Communication style (conversational human voice vs.
corporate voice) X Context (single vs. multiple). The two independent variables in this study,
communication style and context were manipulated and four conditions are the result of that.
All participants were randomly assigned to one of the four conditions which means they were
not all exposed to the same stimulus material. They were either exposed to a webcare
response with a conversational human voice or a corporate voice and either to one single
negative comment or multiple negative comments. Table 1 shows the experimental design.
Table 1. Design Diagram
Communication style
Context Conversational human voice Corporate voice
Single
Multiple
Stimulus material
For this study, the stimulus material consisted of a complaint from a stakeholder about
a service. This is because according to Huibers and Verhoeven (2014), most of the negative
expressions on social media are related to services. Service companies like those in the
tourism and travel industry may be more vulnerable to risks of e-NWOM than other
companies (Litvin, Goldsmith & Pan, 2008). The reason for this is that service products are
intangible and they need to be consumed before they can be fully evaluated. That is why in
this current study, the organization KLM – Royal Dutch Airlines was chosen.
Participants were exposed to a fictional print screen of a Facebook page from the
organization KLM. For the selection of a complaint reason to create an e-NWOM setting, a
15 Facebook regarding his negative experience with flying KLM. This name has been chosen
because it is a gender-neutral name. The complaint from Alex reads as follows: “Dear KLM, a
month ago you lost my baggage over my holiday flight. It took you 7 days to deliver it to me, so I had to buy a lot of new stuff at my holiday destination. Now, a month later, I haven’t received a compensation whatsoever! Your customer service really sucks..”. The complaint was similar for every participant. However, the communication style used in the webcare
response from KLM and the context in which the webcare response was posted was
manipulated and thus differed.
Firstly, webcare responses from KLM were created in which the communication style
was either a conversational human voice or a corporate voice. The conversational human
voice consisted in this current study of the use of the following elements: organizational
communication by using human representatives, addressing the complainants personally,
using personal pronouns (like “I”) and using non-verbal cues (emoticons and repetition of punctuation). These aspects of a conversational human voice are based on earlier research by
Kwon and Sung (2011) and Kelleher (2009). The webcare response written in a
conversational human voice can be found in Appendix A.
Additionally, the webcare response written in a corporate voice did not consist of the
aspects mentioned earlier. This suggests no personal pronouns were used and non-verbal cues
for example. However, the response was more formal, businesslike and shows detached
behaviour, meaning that the response did not include elements such as being involved and
showing personal interest. These aspects are based on earlier research by Levine et al. (2000).
The manipulation can be found in Appendix A as well.
Secondly, besides the communication style which was manipulated, half of the
participants were exposed to either a single negative comment (which is the complaint
16 in addition to and in response to Alex. To create a context in which multiple negative
comments were visible, four comments from other stakeholders were added to the first
complaint. This manipulation and all the stimulus material for the four different conditions
can be found in Appendix A.
Manipulation check
To check whether the participants in this study have perceived the experimental factor
communication style as it was manipulated, a manipulation check was included in the survey.
To test whether the participants perceived the webcare response from KLM correctly as either
a conversational human voice or a corporate voice, six statements were given. Participants
needed to answer on a five-point scale ranging from (1) totally disagree to (5) totally agree.
An example of a statement for the conversational human voice was: “With this response,
KLM is open to dialog”. The items are based on earlier research about the conversational human voice (Van Noort & Willemsen, 2011; Kwon & Sung, 2011; Kelleher, 2009). An
example of a statement for the corporate voice was: “With this response, KLM tries to
respond formally”. These items are also based on earlier research about the corporate voice (Levine et al., 2000; González-Herrero & Smith, 2008). All statements used for the
manipulation check can be found in Appendix B.
Before the analyses for the manipulation check could be done, an exploratory factor
analysis with Varimax rotation over the first three items and the last three items indicated that
the scales were unidimensional, because one component was revealed with an Eigenvalue
above 1, namely 2.66 and 1.65, which explained 88.50% respectively 54.82% of the variance.
These 3-items scales also proved reliable with a Cronbach’s Alpha of .94 and .61. The total
score of the conversational human voice scale and corporate voice scale was computed by
using the mean across the first three items (M = 3.22, SD = 1.37) and the last three items (M =
17 conversational human voice scale and the last three statements were computed and formed
into the corporate voice scale.
An independent samples t-test was conducted because the independent variable
communication style was a dichotomous variable with the two levels conversational human
voice and corporate voice, and the dependent variable was measured at pseudometric level (1
= strongly disagree, 5 = strongly agree). It appeared that the score on the manipulation check
item for conversational human voice was significantly different in the conversational human
voice condition than in the corporate condition, t (167) = 14.57, p = <.05, CI = [1.76, 2.31].
Participants exposed to the conversational human voice communication style scored higher on
the manipulation item intended to measure the conversational human voice (M = 4.23, SD =
.76) than participants exposed to the corporate voice (M = 2.19, SD = 1.03). Besides this, it
also appeared that the score on the manipulation check item for corporate voice was
significantly different in the conversational human voice condition than in the corporate
condition, t (167) = -2.91, p = .004, CI = [-.67, -.13]. Participants exposed to the corporate
voice communication style scored higher on the manipulation item intended to measure
corporate voice (M = 3.46, SD = .95), than participants exposed to the conversational human
voice (M = 3.07, SD = .85).
Sample
The online experimental survey was distributed through the researcher’s own personal network (via e-mail and Whatsapp) and on the social media Facebook and LinkedIn. Two
important characteristics of the target sample was that all participants should be at least 18
years old and they must have a sufficient command of the English language to fill out the
survey. They must be at least 18 years old because at that age, people can rationally decide
how they feel about certain organizations and brands.
18 explained above, namely the use of a conversational human voice with one single negative
comment, the use of a conversational human voice with multiple negative comments, the use
of a corporate voice with one single negative comment and lastly, the use of a corporate voice
with multiple negative comments. A total of 194 persons began the questionnaire and 169 of
them finished the entire survey. This means 25 persons were removed from the data set
because they stopped participating early in the study which resulted in many missing values.
Analyses were thus conducted over the total sample of 169 participants. In total, the
convenience sample comprises 104 female (61.5%) and 65 male (38.5%) participants. The
participants’ ages ranged from 18 to 79 years (M = 33.64, SD = 13.87). Furthermore, 66.3%
of the participants answered they have obtained a degree in either HBO or WO, 23.1% of the
participants answered their highest educational level completed is secondary education and
only 10.7% answered they have completed the educational level MBO.
To check whether the participants’ age was comparable over the two communication
style conditions (conversational human voice vs. corporate voice), an independent samples
t-test was conducted. Communication style was the independent variable in this t-test and age the
dependent variable. The results show that the participants’ mean age in the conversational
human voice condition (M = 32.80, SD = 14.33) was not significantly different from the
participants’ mean age in the corporate voice condition (M = 34.50, SD = 13.42), t (167) = -.80, p = .427. An independent samples t-test was also performed with communication style as
independent variable and gender (0 = male, 1 = female) as dependent variable. The results
show that there was no significant difference between the two conditions with regards to
gender, t (167) = .22, p = .828. There were no significant differences in the amount of males
and females in the conversational human voice condition (M = .62, SD = .49) and corporate
voice condition (M = .61, SD = .49). We can conclude from this that participants were evenly
19 control for age and gender and thus, they were not added as covariates in the analyses for
testing the hypotheses.
Measurements
Perceived reputation
As stated earlier, the perceived reputation is conceptualized as a construct that
describes the perceptions of multiple stakeholders about an organization (Ponzi et al., 2011).
To measure this dependent variable, a short measure of corporate reputation consisting of four
items was used, measured on a seven-point Likert scale from (1) strongly disagree to (7)
strongly agree (Ponzi et al., 2011). An example of such a statement is: “KLM is an
organization that I admire and respect”. All four statements can be found in Appendix B. This variable needed to be computed before the hypotheses could be tested. An
exploratory factor analysis with Varimax rotation over all four items indicated that the scale
was unidimensional, because one component was revealed with an Eigenvalue above 1,
namely 3.15, which explained 78.78% of the variance. The 4-item scale also proved reliable
with a Cronbach’s Alpha of .91. The total score of ‘perceived reputation’ was computed by
using the mean across the four items (M = 5.20, SD = 1.12).
Engagement
Engagement is in this study defined in terms of a combination of cognitive aspects
(e.g. being interested in the activities of a company), emotional aspects (having positive
feelings about the activities of a company) and/or behavioural aspects (e.g. participation in the
activities of a company) (Dijkmans et al., 2015). To measure the level of engagement, ten
items adapted from the scale by Hollebeek, Glynn and Brodie (2014) were used. This scale
consisted of the three dimensions of Consumer Brand Engagement (CBE), namely cognitive
20 items were measured on a seven-point Likert scale ranging from (1) strongly disagree to (7)
strongly agree. An example of a statement measuring cognitive processing is: “The brand
KLM stimulates my interest to learn more about KLM”. An example of a statement measuring affection is: “I feel very positive about KLM”. An example of a statement measuring
activation is: “Whenever I fly, I usually choose KLM”. All ten items can be found in
Appendix B.
Also this variable needed to be computed before the last hypothesis could be tested.
An exploratory factor analysis with Varimax rotation over all ten items indicated that the
scale was not unidimensional, because three components were revealed with an Eigenvalue
above 1, namely 4.25, 2.04 and 1.50, which respectively explained 42.52%, 20.41% and
15.00% of the variance. It was expected that the scale of engagement would not be a
unidimensional one, because the scale is based on the three dimensions of CBE, namely
cognitive processing, affection (= emotional aspects) and activation (= behavioural aspects).
The first three items together form the first component, namely the ‘cognitive processing’ scale, the next four items together form the second component, the ‘emotional’ scale and the
last three items together form the third component, the ‘behavioural’ scale. The first scale proved reasonably reliable with a Cronbach’s Alpha of .69 (M = 3.77, SD = 1.14), the second
scale proved reliable with a Cronbach’s Alpha of .91 (M = 4.78, SD = 1.18) and the last scale also proved reliable with a Cronbach’s Alpha of .93 (M = 3.32, SD = 1.59). The scores of the three engagement components were each separately computed by using the mean of the items.
Demographic variables and covariates
Three demographic variables were measured in this study, namely gender
(male/female/other), the age of the participant and the highest completed educational level
(elementary education, secondary education (VMBO, HAVO, VWO, VAVO), MBO and
21 measured as well to check whether these variables could be a possible explanation for the
results in this study. Brand awareness is one of these variables. It was measured at nominal
level by asking the question: “Do you know the brand KLM – Royal Dutch Airlines?”.
Participants either answered yes or no. Also the attitude towards KLM was measured. To
measure this variable, a seven-point bipolar adjective scale was used with three items:
bad/good, unpleasant/pleasant and unfavourable/favourable (MacKenzie & Lutz, 1989) (α = .92) (M = 5.83, SD = .98). Participants needed to answer how they would assess the brand
KLM in general. Furthermore, the Facebook use of the participant was measured using a
five-point scale asking how often they use Facebook, where (1) = never and (5) = several times a
day. At last, participants were asked whether they have had any problems with a travel
company in the past two years, where (1) = yes, (2) = no and (3) = I don’t remember.
Procedure
Pretest
Before the actual experiment was conducted, an oral pre-test had been done to check
whether the participants perceived the experimental factor as it was manipulated. Eight
persons were asked to answer six questions regarding the webcare response from KLM. The
six statements mentioned above in the manipulation check section were used. All of the eight
participants totally agreed with the three statements regarding the conversational human voice
webcare response (M = 5) and all of the eight participants also (totally) agreed with the three
statements regarding the corporate voice webcare response (M = 4.67). However, a few of the
respondents exposed to the conversational human voice webcare response expressed that they
also find the response quite professional. At this time, such a voice is also seen as quite
professional, even though it is a very informal response. Still, this statement used to measure
22 the corporate voice. To conclude, all participants perceived the communication style as it was
intended, so the manipulation was successful and no changes to the stimulus material were
made.
Actual experiment
This research, between the 28th of April and 8th of May 2017, was conducted online
using the Qualtrics program. The participants in this study were approached online via
Whatsapp, e-mail, Facebook or LinkedIn. The survey consisted of several parts (a copy of the
survey can be found in Appendix C). First, participants were exposed to a short introduction
about the research. After that, participants were asked whether they are familiar with the
brand KLM and how they would assess KLM. Then the participants were all exposed to the
(fictional) Facebook page from KLM with the complaint from stakeholder Alex. However,
the participants were randomly assigned to this single negative comment or to multiple
negative comments and to the webcare response of KLM using a conversational human voice
or a corporate voice. Participants were asked to read the posts carefully. Subsequently, after
being exposed to the scenarios, the participants were asked to answer a few questions about
the conversational human voice or corporate voice (= manipulation check). After that,
participants must rate the brand (based on the conversation they were exposed to) in terms of
engagement as well as the perceived reputation. The last part of the survey consisted of
questions concerning for example the Facebook use of the participant and some demographic
variables.
Analyses
To test the first three hypotheses, a factorial ANOVA was conducted with two
independent dichotomous variables, namely: communication style (conversational human
23 measured at pseudometric level, namely: perceived reputation (1 = strongly disagree, 7 =
strongly agree). By conducting a factorial ANOVA, the first two hypotheses could be
answered by looking at the main effects and the third hypothesis could be answered by
looking at the interaction effect between the communication style and context.
To test the last and thus fourth hypothesis, a multiple regression analysis was
conducted with the dichotomous variable communication style (conversational human voice
vs. corporate voice) as independent variable, the three components of the moderator
engagement (1 = strongly disagree, 7 = strongly agree) measured at pseudometric level as
independent variable and the perceived reputation (1 = strongly disagree, 7 = strongly agree)
measured at pseudometric level as the dependent variable. Through this regression analysis,
the moderation effect of the level of engagement on the communication style used and the
perceived reputation could be tested.
Results
Testing hypotheses
In order to test the first hypothesis “The perceived reputation will be more positive
when the communication style used in webcare is a conversational human voice than when it is a corporate voice”, the main effect was analyzed (as shown in the factorial ANOVA) with communication style (conversational human voice vs. corporate voice) as the independent
variable and the perceived reputation (1 = minimum, 7 = maximum) as the dependent
variable. The results show that there was a significant main effect of communication style on
the perceived reputation, F (1, 165) = 25.57, p < .05, η2= .13. Thus, the first hypothesis can be
confirmed. This test result proves that after being exposed to the conversational human voice
(M = 5.61, SD = .93), participants scored higher on the perceived reputation scale than
24 In order to test the second hypothesis “The perceived reputation will be more positive
when the webcare response is posted in response to one single negative comment than when the webcare response is surrounded by more negative comments”, the main effect was analyzed (as shown in the factorial ANOVA) with context (single vs. multiple) as the
independent variable and the perceived reputation (1 = minimum, 7 = maximum) as the
dependent variable. The results show that there was no significant main effect of context on
the perceived reputation, F (1, 165) = 1.85, p = .176. Thus, the second hypothesis must be
rejected. This test result shows that after being exposed to a single negative comment (M =
5.30, SD = 1.03), participants did not score higher on the perceived reputation scale than
participants that were exposed to multiple negative comments (M = 5.11, SD = 1.21).
In order to test the third hypothesis “The perceived reputation will be more positive
when the communication style used in webcare is a conversational human voice than when it is a corporate voice, but this effect will be more pronounced when the webcare response is posted in response to one single negative comment than when it is surrounded by more negative comments”, the interaction effect was analyzed (as shown in the factorial ANOVA). Communication style (conversational human voice vs. corporate voice) and context (single vs.
multiple) as both independent variables and the perceived reputation (1 = minimum, 7 =
maximum) as the dependent variable. The results show that there was no significant
interaction effect between these variables on the dependent variable perceived reputation, F
(1, 165) = .64, p = .423. Thus, the third hypothesis must be rejected. The results show that
people who were exposed to the conversational human voice and saw one single negative
comment (M = 5.65, SD = .16) did not score higher on perceived reputation than when they
were exposed to multiple negative comments (M = 5.56, SD = .16). The results also show that
people who were exposed to the corporate voice and one single negative comment (M = 4.97,
25 multiple negative comments (M = 4.62, SD = .16).
In order to test the last and fourth hypothesis “The effect of an organization’s social
media communication style on the perceived reputation will be moderated by the level of engagement”, a multiple regression analysis was conducted. The independent variable communication style (conversational human voice vs. corporate voice) as well as the
moderator variables cognitive engagement, emotional engagement and behavioural
engagement (1 = strongly disagree, 7 = strongly agree) were saved as standardized variables
and were added as independent variables. Besides that, they were also multiplied and added in
the regression model which was essential to test hypothesis 4. The dependent variable in the
regression analysis was the perceived reputation (1 = strongly disagree, 7 = strongly agree).
The results show that the regression model was significant, F (7, 161) = 31.62, p < .05.
The regression model could therefore be used to predict the perceived reputation, the strength
of the prediction is quite strong: 58.00% of the variation in perceived reputation could be
predicted on the basis of the used communication style and the level of engagement (R2 =
.58). As already shown in the ANOVA, communication style significantly affects the
perceived reputation. Communication style, b* = -.23, t = -4.26, p < .05, 95% CI [-.38, -.14],
had a significant, moderately strong association with the perceived reputation. It also
appeared that emotional engagement as well as cognitive engagement had a significant and
strong, respectively weak association with the perceived reputation, b* = .68, t = 11.85, p <
.05, 95% CI [.64, .89] and b* = -.15, t = -2.70, p = .008, 95% CI [-.29, -.05]. However, it
appeared that the interaction effect (for all the three components of engagement) was not
significant. Thus, the last and fourth hypothesis must be rejected. The effect of an
organization’s social media communication style on the perceived reputation is not moderated by the level of engagement. For cognitive engagement, b* = -.03, t = -.56, p = .578, 95% CI
26 behavioural engagement, b* = .03, t = .47, p = .636, 95% CI [-.10, .16]. For all these effects
other independent variables are assumed to be held constant.
Conclusion
The study shows that the perceived reputation is more positive when KLM uses a
conversational human voice in webcare (on Facebook) than a corporate voice. Thus, the
results in this study confirm the first hypothesis. This corresponds to previous research, which
also stated that a conversational human voice appears to be an effective communication style
for organizations to react on stakeholders (Van Noort & Willemsen, 2011). It is more
appreciated by stakeholders that organizations bring human personality to organizational
communication instead of speaking more formally, using a distant corporate tone of voice
(Kwon & Sung, 2011; Levine et al., 2000; González-Herrero & Smith, 2008). This implies
that in this study, the Social Presence Theory does explain why using a conversational human
voice in webcare is more effective than using a corporate voice (Cui, Lockee & Meng, 2013).
However, the last three hypotheses cannot be confirmed because the research results
do refute them. It was not the case that the perceived reputation is more positive when the
webcare response is posted in response to one single negative comment than when the
response is surrounded by more negative comments. That is why the second hypothesis
needed to be rejected. This outcome is contradictory to previous research, which showed that
negative online interactions between consumers are found to have negative effects on the
consumers’ decision making process, including brand evaluations (Van Noort & Willemsen, 2011). In this research, it is not the case that after the observer reads e-NWOM in high levels
of consistency, he will conform to the social norm and form negative attitudes towards the
organization, which eventually negatively affect the perceived reputation (Kim et al., 2016). It
did not make a difference whether participants were exposed to one single negative comment
27 Besides, this study showed that the effect of the communication style used on the perceived
reputation is not more pronounced when the webcare response is posted in response to one
single negative comment than when it is surrounded by more negative comments.
Nonetheless, there was one interesting finding. It appeared that two of the three
components of engagement, namely cognitive engagement and emotional engagement,
directly influence the perceived reputation. As expected and as previous research has already
shown, engagement partly positively influences the reputation (Dijkmans et al., 2015).
However, unexpectedly and contradictory to previous research, this study showed that
cognitive engagement negatively influences the perceived reputation. This effect turned out to
be weak, so it might be coincidental with regards to the small sample used. The final
hypothesis needed to be rejected as well, because although engagement partly influences the
perceived reputation directly, it does not moderate the relationship between the
communication style used and the perceived reputation.
Thus, the two research questions in this study, namely: “To what extent does the
communication style used in webcare and the context in which a webcare response is posted predict the perceived reputation of that organization?” and “What role does engagement play in the relationship between the communication style used on social media and the perceived reputation of that organization?” are now answered. The perceived reputation is more positive when a conversational human voice is used in webcare than a corporate voice.
However, the context in which a webcare response was posted did not influence the perceived
reputation and engagement did not play a moderating role in the relationship between the
communication style used and the perceived reputation.
Still, nowadays it is important to use a favourable tone of voice in webcare and to
28 Discussion
Despite the fact that only one hypothesis was confirmed in this study, the results do
not suggest that no relationship at all exists between the context in which a webcare response
is posted and the perceived reputation and how engagement influences the perceived
reputation. There are several plausible explanations for the results in this study. In this
chapter, limitations will be given as well as suggestions for future research.
Firstly, the sample size (N = 169) was quite small and the sample in this study does not
fully represent the general population. This implies that it is difficult to generalize the results.
As an example, the majority of the sample is highly educated (either HBO or WO). An
important suggestion for future research is to take a more diverse sample into account to
increase external validity. Due to the fact that the survey was completely in English, it might
be that a smaller sample was reached. Future researchers should also translate the survey in
Dutch to make sure that only Dutch speaking people are not excluded from filling out the
survey.
In addition, this study only focused on one organization, namely KLM – Royal Dutch
Airlines. This means that the results are probably not generalizable to organizations in other
sectors. As already stated, service companies are more vulnerable to risks of e-NWOM than
other companies (Litvin et al., 2008) and that is why the focus in this study was on KLM.
However, it is also relevant for future researchers to examine organizations in a different
sector, like a telecommunications provider such as T-Mobile or KPN.
Besides, another important limitation related to the organization used, is the restriction
to one issue. For the selection of a complaint reason to create an e-NWOM setting, a lost
luggage case was chosen. It might be possible that there are issues stakeholders can have with
a service and that have a lower or higher impact on the observers’ evaluation of the severity of the Facebook scenario given. Future research should integrate multiple issues and for example
29 distinguish between the dimensions of the severity of the issue.
Moreover, as this study incorporates a good and sufficient level of realism, because an
existing organization has been used, we can to some extent generalize the experimental
findings to practice. Unfortunately using an existing organization also means that the
participants easily form an image of the organization. The risk remains that participants are
biased and that one webcare response does not change how they would assess the
organization. This research did show that the attitude towards KLM was quite positive.
Related to this is that in this study context was defined as the amount of e-NWOM messages
related to one specific stakeholder’s post. Investigating the amount of e-NWOM messages was an extension to the scientific literature and was not studied before (Van Noort &
Willemsen, 2011). However, it is possible that multiple negative stakeholder comments do not
change the already positive attitude that stakeholders have towards the organization KLM.
Reputation is often strong and stable and cannot be changed in only one experimental study.
The perceptions about KLM are formed based upon the organization’s past activities as well
(Tucker & Melewar, 2005), and they are not only formed based upon one specific moment in
time where stakeholders are exposed to e-NWOM and one webcare message.
Additionally, this research focused on the social medium Facebook. However,
organizations are also very active on Twitter. Twitter is also a medium which makes it
possible to reply and respond quickly to complaints. Future research can look at whether the
effects on the perceived reputation are different for Twitter.
Furthermore, an important limitation is related to the stimulus material used. Although
the manipulation check for communication style was successful and significant, the
differences in the means between the manipulated version of the conversational human voice
as well as the corporate voice are quite small. This relates to the fact that the conversational
30 only one manipulated Facebook message. This study is restricted to the effects of a
conversational human voice in only one webcare response to negative feedback on Facebook.
However, given the dialogical nature of a conversational human voice, it is certainly possible
that the conversation between the stakeholder and the organization does not end at one single
webcare response. This means that the organization might be asked for another reply. Thus,
future research should further examine such dialogues.
Besides, another limitation in this study is that no manipulation check is done to check
whether participants saw one single negative comment from a stakeholder or if they saw
multiple negative comments. This had not been done in this research because it seemed
obvious whether the participants were exposed to either one negative comment or multiple
negative comments. However, it is still possible that they were not consciously aware of the
amount of e-NWOM. That is why future research should include a manipulation check related
to context.
Important to notice as well is that in this study it was stated that when people are more
interested in an organization, have more positive feelings about it and thus feel more engaged,
it can strengthen feelings of injustice when organizations fail to solve the problem in a
satisfying way (Van Noort et al., 2015). However, it might also be the case that people who
feel more engaged with the organization, are more forgiving when the organization responds
inadequately to negative comments and questions. Future research should further examine in
different sectors and contexts if and how engagement plays a role in the relationship between
the communication style used in webcare and the perceived reputation.
To conclude, the goal of this current study was to gain more insights in how to
effectively handle e-NWOM. The most important implication for practice is that webcare
teams should be aware of how they respond to e-NWOM. For webcare teams, it seems
31 and distant tone in their communications. Social media will always be uncontrolled areas for
participation so future research should further examine what webcare teams should do when
stakeholders express their voices online.
References
Baumeister, R. F., Bratslavsky, E., Finkenauer, C., & Vohs, K. D. (2001). Bad is stronger
than good. Review of General Psychology, 5(4), 323-370.
doi:10.1037/1089-2680.5.4.323
Coombs, W. T., & Holladay, S. J. (1996). Communication and attributions in a crisis: An
experimental study in crisis communication. Journal of Public Relations Research,
8(4), 279-295. doi:10.1207/s1532754xjprr0804_04
Cui, G., Lockee, B., & Meng, C. (2013). Building modern online social presence: A review of
social presence theory and its instructional design implications for future
trends. Education and Information Technologies, 18(4), 661-685. doi:10.1007/s10639
012-9192-1
Dijkmans, C., Kerkhof, P., & Beukeboom, C. J. (2015). A stage to engage: Social media use
and corporate reputation. Tourism Management, 47, 58-67.
doi:10.1016/j.tourman.2014.09.005
Dijkmans, C., Kerkhof, P., Buyukcan‐ Tetik, A., & Beukeboom, C. J. (2015). Online conversation and corporate reputation: A two‐ wave longitudinal study on the effects of exposure to the social media activities of a highly interactive company. Journal of
Computer‐ Mediated Communication, 20(6), 632-648. doi:10.1111/jcc4.12132
Fombrun, C., Shanley, M. (1990). What’s in a name? Reputation building and corporate strategy. The Acadamy of Management Journal, 33(2), 233-258.
González-Herrero, A., & Smith, S. (2008). Crisis communications management on the web:
32 handle business crises. Journal of Contingencies and Crisis Management, 16(3), 143
153. doi:10.1111/j.1468-5973.2008.00543.x
Grunig, J. E. (2009). Paradigms of global public relations in an age of
digitalisation. PRism, 6(2), 1-19.
Hollebeek, L. D., Glynn, M. S., & Brodie, R. J. (2014). Consumer brand engagement in social
media: Conceptualization, scale development and validation. Journal of Interactive
Marketing, 28(2), 149-165. doi:10.1016/j.intmar.2013.12.002
Huibers, J. & Verhoeven, J. (2014). Webcare als online reputatiemanagement: het gebruik van
webcarestrategieën en conversational human voice in Nederland, en de effecten
hiervan op de corporate reputatie. Tijdschrift voor Communicatiewetenschap, 42(2),
165-189.
Kelleher, T. (2009). Conversational voice, communicated commitment, and public relations
outcomes in interactive online communication. Journal of Communication, 59(1), 172-
188. doi:10.1111/j.1460-2466.2008.01410.x
Kelleher, T., & Miller, B. M. (2006). Organizational blogs and the human voice: Relational
strategies and relational outcomes. Journal of Computer-Mediated Communication,
11(2), 395-414. doi:10.1111/j.1083-6101.2006.00019.x
Kim, S. J., Wang, R. J. H., Maslowska, E., & Malthouse, E. C. (2016). “Understanding a fury in your words”: The effects of posting and viewing electronic negative word-of-mouth on purchase behaviors. Computers in Human Behavior, 54, 511-521.
doi:10.1016/j.chb.2015.08.015
Kwon, E. S., & Sung, Y. (2011). Follow me! Global marketers’ Twitter use. Journal of
Interactive Advertising, 12(1), 4–16. doi:10.1080/15252019.2011.10722187
Levine, R., Locke, C., Weinberger D., & Searls, D. (2000). The cluetrain manifesto: The end
33 Litvin, S. W., Goldsmith, R. E., & Pan, B. (2008). Electronic word-of-mouth in hospitality
and tourism management. Tourism Management, 29(3), 458–468.
doi:10.1016/j.tourman.2007.05.011
MacKenzie, S. B., & Lutz, R. J. (1989). An empirical examination of the structural
antecedents of attitude toward the ad in an advertising pretesting context. Journal of
Marketing, 53(2), 48-65.
Moreno, A., Navarro, C., Tench, R., & Zerfass, A. (2015). Does social media usage matter?
An analysis of online practices and digital media perceptions of communication
practitioners in Europe. Public Relations Review, 41(2), 242-253.
doi:10.1016/j.pubrev.2014.12.006
Park, H., & Cameron, G. T. (2014). Keeping it real: Exploring the roles of conversational
human voice and source credibility in crisis communication via blogs. Journalism &
Mass Communication Quarterly, 91(3), 487-507. doi:10.1177/1077699014538827 Ponzi, L. J., Fombrun, C. J., & Gardberg, N. A. (2011). RepTrak™ pulse: Conceptualizing
and validating a short-form measure of corporate reputation. Corporate Reputation
Review, 14(1), 15-35. doi:10.1057/crr.2011.5
Purnawirawan, N. A. (2013). Consumer responses to positive and negative online reviews.
Unpublished doctoral dissertation. Antwerp, Belgium: University of Antwerp.
Schaefers, T., & Schamari, J. (2016). Service recovery via social media: The social influence
effects of virtual presence. Journal of Service Research, 19(2), 192-208.
doi:10.1177/1094670515606064
Schamari, J., & Schaefers, T. (2015). Leaving the home turf: how brands can use webcare on
consumer-generated platforms to increase positive consumer engagement. Journal of
34 Schultz, F., Utz, S., & Göritz, A. (2011). Is the medium the message? Perceptions of and
reactions to crisis communication via twitter, blogs and traditional media. Public
Relations Review, 37(1), 20-27. doi:10.1016/j.pubrev.2010.12.001
Tucker, L., & Melewar, T. C. (2005). Corporate reputation and crisis management: The threat
and manageability of anti-corporatism. Corporate Reputation Review, 7(4), 377–387.
doi:10.1057/palgrave.crr.1540233
Valentini, C. (2015). Is using social media “good” for the public relations profession? A
critical reflection. Public Relations Review, 41(2), 170-177.
doi:10.1016/j.pubrev.2014.11.009
Van Noort, G., & Willemsen, L. M. (2011). Online damage control: The effects of proactive
versus reactive webcare interventions in consumer-generated and brand-generated
platforms. Journal of Interactive Marketing, 26(3), 131-140.
Van Noort, G., Willemsen, L. M., Kerkhof, P., & Verhoeven, J. W. (2015). Webcare as an
integrative tool for customer care, reputation management, and online marketing: a
literature review. In Integrated communications in the postmodern era (pp. 77-99).
Palgrave Macmillan UK.
Verhoeven, P., Tench, R., Zerfass, A., Moreno, A., & Verčič, D. (2012). How European PR practitioners handle digital and social media. Public Relations Review, 38(1), 162-164.
doi:10.1016/j.pubrev.2011.08.015
Yang, S. U., Kang, M., & Johnson, P. (2010). Effects of narratives, openness to dialogic
communication, and credibility on engagement in crisis communication through
35 Appendix A: Manipulations
The webcare response from KLM using a conversational human voice reads as follows:
“Dear Alex,
I’m so sorry to hear you lost your baggage. I understand your first days at your holiday destination must have been somewhat inconvenient. If you send me your booking number, I will personally look into your case and I’ll get back to you with a solution Hope that’s ok!! Kind regards, Emma.”
The webcare response from KLM using a corporate voice reads as follows:
“Dear Alex,
Sorry to hear about that. For this issue, please contact our Customer Service Centre at: www.klm.com/travel/nl_nl/customer_support”
Context with multiple negative comments:
Jamie: “OMG yes Alex I feel you, this happened to me a year ago. A few MONTHS later I received a compensation.. Really, the customer service is pathetic!”
Jessie: “You did receive a compensation? Well I still haven’t got it yet. KLM lost my baggage over my holiday flight as well. One bag was delivered 3 days later but one bag is still missing! Seriously, I’m never going to fly with KLM again..”
