Abstract
The article discusses the use of conversational human voice (CHV) to address negative eWOM on brand-generated social media platforms. Using the case of a crisis-ridden Danish bank, the article investigates the use of CHV outside service failures and its effects on critical publics in the context of intentional crises. The data consist of posts and comments from the bank’s Facebook page, following allegations of money laundering. The analysis reveals that CHV is used extensively by the bank to counter criticism; however, the degree to which the strategy is standardized or tailored seems to depend on whether users appear once, have a regular presence, engage in dialog, and are known to the bank’s employees. These findings suggest that while CHV is intended for more personalized communication to make users more sympathetic to the organization, its use will have to be contextualized to be effective.
Introduction
It has been more than 10 years since the financial crisis hit the world in 2008. Since then, financial markets and national economies are said to have recovered from one of the worst economic setbacks since the Great Depression. This may be true in economic terms, but for some of its key culprits the crisis ushered in a long and painful period of image restoration. Thus, large parts of the financial sector have struggled to retain legitimacy both as a result of the crisis and the subsequent disclosure of related events that reveal questionable business conduct. This is especially true of some banks in the Danish financial sector.
However, the conduct of financial institutions is only partly contributory to the difficult process of image restoration—their communication with stakeholders is another. While corporate communication used to be a top-down, largely management-controlled endeavor, the omnipresence of social media now provides unprecedented possibilities for consumers to publicly criticize organizational behavior and hold the institutions accountable by sharing criticism online (Decock et al., 2021; Ghosh, 2017; Holmgreen, 2020). In doing so, consumers take advantage of the media’s ability to act as a vehicle of mass communication where views and sentiments can be shared instantly as negative electronic word of mouth (eWOM). Consequently, institutions have felt the need to implement online communication strategies to mitigate the effect of negative public coverage, hoping that this will assist in restoring the corporate image. One of these strategies has been coined webcare (Van Noort & Willemsen, 2012).
Webcare is a way for corporations to monitor and intervene in negative eWOM, either reactively or proactively so. When carrying out reactive webcare, the corporation will respond to consumers’ specific criticism or complaints, whereas proactive webcare is a more general response to negative word of mouth. Studies show that webcare is promising in addressing and mitigating negative eWOM, but that it may also backfire when used without proper consideration (Van Noort & Willemsen, 2012). One of the recommended strategies is therefore to engage actively with the properties of social media, which allow for an open, informative, and dialogic approach to users (Utz et al., 2013). This dialog may take the form of what has been coined a “conversational human voice” (CHV; Van Noort & Willemsen, 2012, 134).
While the literature in the field of webcare is concerned with customers’ reactions when they have experienced a service failure, that is, have had negative experiences with the handling of complaints or defective products (Ghosh, 2017; Van Noort et al., 2014; Weitzl et al., 2018; Willemsen et al., 2013), little attention seems to have been given to cases where webcare is provided in response to negative eWOM following continuous corporate malpractice, that is, organizational behavior outside the communication on a specific service failure (for an exception, see Catenaccio, 2021). The handling of negative eWOM following corporate malpractice has, however, been treated extensively in the crisis communication literature (see, for instance, Coombs, 2004, 2007; Coombs & Holladay, 2002; Utz et al., 2013), but primarily with a view to addressing how corporations can manage a crisis in its strategic stakeholder communication to avoid damage to its image and reputation, that is, from a one-to-many communications perspective, where corporate communication specialists act as gatekeepers of information. Thus, studies deal, among other things, with crisis responsibility and stakeholder reactions to this, suggesting that the more intentional a crisis is perceived to be (for instance, when a misdeed has been committed), the stronger the attributions of organizational crisis responsibility. This potentially leads to a marked increase in negative reactions, including anger, and a higher degree of willingness among consumers to take action, even when the crisis has subsided (Coombs, 2004; Lerner & Tiedens, 2006; Utz et al., 2013). In this scenario, organizations are encouraged to be vigilant and adopt communication strategies with the purpose of diminishing the effects of negative consumer reactions. However, with the spread of social media platforms in stakeholder communication, asymmetric, one-to-many strategies appear to be less effective, especially since these platforms encourage dialog and facilitate instant and many-to-many communication, where users may become co-producers of what is “relevant, newsworthy and credible” (Utz et al., 2013, 41). Thus, the adoption of appropriate strategies, such as, for instance, conversational human voice, must necessarily be tailored to the organizational and communicative context in which they are applied (cf. Barcelos et al., 2018).
These observations will form the background of the article, which will analyze the communication between employees in a crisis-ridden Danish bank and critical publics on their brand-generated platform on Facebook. The analysis will be concerned with how intentional crises influence conversations with stakeholders by focusing on matters that are not associated with these crises and the strategies adopted by the employees to mitigate criticism and negative outbursts, that is, negative eWOM. Thus, the article addresses apparent lacunas in both webcare and crisis communication research on how to manage critical publics when they appear outside service failures, continuously address intentional crises, and do so on social media platforms. Specifically, I will be analyzing and discussing the following research questions (cf. Van Noort & Willemsen, 2012),
RQ 1: To which extent is conversational human voice (CHV) used to interact with critical publics on the bank’s Facebook page?
RQ 2: Does the use of conversational human voice (CHV) lead to a positive conversational outcome, defined as the user expressing satisfaction with the response and a positive attitude toward the bank?
In this way, the analysis reflects observations made in recent studies on webcare that language use is central to the realization of successful interactions between corporate voices and consumers (Decock et al., 2021). Altogether, the article contributes insight into contextualized and real-life conversations and their effects in brand-related social media by investigating the outcomes of a part standardized and part tailored CHV strategy used to influence three different types of users (defined as the number of times users appear on the Facebook page and the level of their engagement).In continuation of this, the article offers an alternative to experimental studies frequent in recent crisis communication and webcare studies, demonstrating by means of a case study the possible effects of conversational and linguistic strategies on consumer attitudes in real-life conversations.
The article is structured in the following way: Section 2 provides an overview of research on crisis responsibility and conversational human voice, with a view to combining insights from the two. Section 3 presents the case of the bank from where the data originates, along with details on the empirical data and their retrieval. This is followed by Section 4 which presents the analysis of data, followed by the discussion in Section 5.
Crisis Responsibility and the Relevance of Conversational Human Voice
In a number of now classical contributions to crisis communication research, Coombs (2004, 2007) and Coombs and Holladay (2002) advance the claim that the level of crisis responsibility assigned to organizations will determine stakeholders’ reactions to the crisis. Thus, taking their cue from attribution theory (Weiner, 1985, 1995), they argue that if stakeholders are convinced that an organization could control a crisis, it will be assigned responsibility, which in turn will “shape feelings and behaviors toward the organization” (Coombs, 2004, 267). Following this, the stronger the responsibility is perceived to be, the feelings directed at the organization also tend to be stronger and more negative, not only in relation to the actual event but also in the more long-term relationship (Weiner, 1985). Thus, the development and handling of a crisis are of critical importance to the long-term image and reputation of the organization. Furthermore, it is argued that crisis responsibility is strongest when the organization has a history of crises or a negative relationship with stakeholders, when control by someone outside the organization is low, and the crisis originates from within the organization, cf. Weiner’s three independent dimensions of locus, controllability, and stability (Coombs, 2004; Weitzl et al., 2018).
To address these issues, it is suggested that crisis managers take stock of the situation, define the type of crisis according to a predefined set of cues and definitions, and adopt appropriate crisis response strategies to influence the attributions stakeholders assign to a crisis or the organization itself. These may include the three primary strategies of denying (responsibility for) the crisis, diminishing the crisis, or rebuilding the perception of the organization (Coombs, 2007). For crises that are seen as intentional, rebuilding is considered an important approach, as it involves the adoption of an accommodative strategy of either apology or compensation (Utz et al., 2013). Another proposed, but secondary, strategy is bolstering, which involves reminding stakeholders of the past good works by the organization or reminding them that the organization is also a victim (Coombs, 2007). However, as has been demonstrated in later empirical research, identifying something as a crisis is sometimes difficult and may happen too late, leading to unsuitable and even damaging response strategies. This suggests that crisis typologies can only function as general guidelines to the identification and handling of crises (Holmgreen, 2015). Part of the reason for this may be found in the fact that contextual factors play a decisive role in determining when and how something becomes a crisis according to stakeholders, just as the channels through which crisis communication takes place influence the attributions stakeholders assign to the organization (cf. Barcelos et al., 2018). Thus, with the advent of social media as important communication platforms, the speed with which communication takes place has intensified, and so have stakeholder reactions, just as new “rules” for engaging with stakeholders have emerged, which, among other things, involve a stronger orientation toward interaction and dialogic content (Du Plessis, 2018; Schultz et al., 2011).
Dialog and Conversational Human Voice (CHV)
According to recent studies, the orientation toward dialogic content on social media changes the perception of the stakeholder as someone whose attributions can be changed through framing the correct messages by using a “corporate voice,” into someone with whom relationships must be built and catered to (Du Plessis, 2018; Javornik et al., 2020). As a rule, this requires more open, transparent, and authentic communication. However, as studies (Liu et al., 2011; Van der Meer et al., 2014) also indicate, social media allow organizations to control content, but not the reactions of users. Thus, users may be inclined to react based on issues, sentiments, and personal interpretation instead of facts, which makes it inherently difficult to predict how messages should be framed and relationships built. In terms of events that may develop into crises, the use of social media platforms also means that information and updates on events can be shared in real time, prompting users to react while still being engaged in conversation with the organization (Du Plessis, 2018; Utz et al., 2013). In addition, users can engage with other users to share information, bond online, vent emotions, as well as demand that the organization takes steps to resolve the crisis (Liu et al., 2011). Altogether, these factors suggest the need for a crisis response strategy that addresses the conversational, interactional, and person-oriented nature of social media. One such strategy may involve the concept of conversational human voice from webcare research.
Attributed to Kelleher (2009), conversational human voice (CHV) refers to an engaging and natural style of organizational communication as perceived by an organization’s publics based on interactions between individuals in the organization and individuals in publics (Kelleher, 2009, 177).
Originating in public relations and communication studies, CHV is seen as a way for the organization to become associated with more human-like properties, which allow consumers to form relationships with the brand it represents, among other things because it allows for a more natural and emotional engagement as opposed to a more distant and formal one, typically associated with corporate voice (Barcelos et al., 2018; Javornik et al., 2020; Kelleher, 2009; Kelleher & Miller, 2006). This, in turn, is found to influence brand trust (Gretry et al., 2017), engagement intentions (Schamari & Schaefers, 2015), and crisis communication (Park & Cameron, 2014), among others. However, other studies find that CHV does not change perceptions when used in connection with multiple negative reactions to the brand online (Barcelos et al., 2018). The latter is of particular interest to this article. According to Barcelos et al. (2018), these differences in findings may be attributed to the differences in situational contexts, for instance, understood as the platform in which the dialog takes place and the organizational realities surrounding the conversation. This is a position shared by the author of this article, and which will be investigated further in the analysis below.
An apparent challenge with using CHV as the concept on which to base investigations into brand-consumer relations is its fuzziness, that is, the lack of clearly bounded categories to investigate it in discourse (Gretry et al., 2017). Part of the problem can be found in the fact that its realization is partly dependent on the context in which it is used; however, there seems to be consensus that at least three dimensions may constitute conversational human voice. These include message personalization, informal speech, and invitational rhetoric (Liebrecht et al., 2021; Van Noort et al., 2014). Message personalization involves treating both parties in the conversation as real individuals and can be realised through personal greetings, addressing the consumer by their name, the use of personal pronouns (both first and second person), personal signatures, and the sharing of personal information. Informal speech, on the other hand, refers to the use of a “casual, familiar, and common style of communication” (Liebrecht et al., 2021, 126) and includes non-verbal cues (for instance, the use of capitals, emoticons/emojis, or tactile elements), verbal cues, interjections, and informal vocabulary and phrases (for instance, adjectives, common verbs and expressions, active vs. passive voice, informal phrases, and verb omissions; Gretry et al., 2017). Finally, invitational rhetoric covers strategies that are used to invite consumers to the conversation, such as inviting the consumer to engage in dialog, acknowledging consumers’ thoughts, suggestions, comments, etc., apologizing, showing sympathy and empathy, or humor, and well-wishing (Liebrecht et al., 2021).
The above overview is the result of an extensive literature review of articles (a total of 158) from 2006 until 2020 that operationalize CHV linguistically (Liebrecht et al., 2021). In this respect, it represents one of the most comprehensive studies to be conducted on the study of CHV and is, as such, valuable to future empirical investigations. However, if we acknowledge that the linguistic realization of CHV is contingent on the social context in which it is used, the approach and categories for studying its use will necessarily be based on what the data reveal to the analyst. Thus, in this article select posts and comments from the dataset will be analyzed within the context they appear.
As indicated by Liebrecht et al. (2021), investigating the effects of linguistic choices in webcare interactions is not a new phenomenon. However, many of the existing studies have either combined discursive analysis of webcare interactions with experimental studies or they have been exclusively concerned with experimental setups, focusing less on rigorous linguistic analyses of webcare interactions and more on the effects of linguistic choice on brand trust and organizational reputation (Decock, 2022). This indicates the need for more studies that investigate real life interactions in which context is a key component for understanding language use and the realization of CHV categories. Furthermore, as observed by Creelman (2021), studies of language in service contexts have primarily involved airline, tourism, and hospitality industries, reflecting the concerns of these industries. Thus, taking the study of CHV into the financial sector, specifically retail banking, represents a relatively new but no less relevant area of study.
The above considerations will form the background of the analysis.
Data and Analytic Method
Investigating how banks use language to engage with customers in social media is interesting for a number of reasons. First, banks are key providers of financial services to industry, business, and private customers to ensure growth and value creation in society; second, they were, as mentioned, the key culprits in the development of the financial crisis in 2008; and third, as a consequence, they have become widely associated with greed and the short-term financial gains for investors, presumably violating the social contract that is perceived as a key component in their operating licence (cf. Holmgreen, 2012; Van Herck et al., 2020). Thus, whereas banks may previously have been considered as institutions with a high level of integrity, this is less so in the 2020s. This leaves the sector with a continuous need for managing its image and countering negative eWOM in, for instance, its communication with customers and users online.
The case for the study will be Danske Bank, Denmark’s largest bank. Of all Danish financial institutions, this bank is possibly the one to suffer the longest and severest history of crises. Previously considered the epitome of due diligence and prudent investment (and still the preferred bank for business and the public sector), since the financial crisis in 2008, it has been plagued with cases of bad judgment and business decisions which have led not only to criminal investigations, but also bad publicity and a significant decrease in customer numbers. The cases involve, inter alia, risky investments in foreign banks, poor service of private customers, golden handshakes to incompetent top managers, investments in weapons manufacturing, and ignoring money laundering regulations. As a result, today many consider the bank the epitome of greed and poor judgment.
The data were collected in the years that have passed since the latest case, that of money laundering in the bank’s Estonian branch. They were retrieved manually from the bank’s brand-related Facebook page by perusing all posts and comments between September 2019, a year after the bank presented its own investigations into the matter, and late 2021, rendering a corpus of 32 users in total, who would be present in the dataset one or several times (for further information, see below).
For the selection of posts and comments, the following criteria applied:
Posts made by the bank should be on a topic unrelated to any of the bank’s previous crises (cf. Introduction).
Comments should refer to the crises by using one or more of the following keywords, including inflections and related word classes (English equivalent term in brackets): ○ “kriminel” and “kriminalitet” (criminal/crime), ○ “grådighed” and “grådig” (greed/greedy), ○ “mistænkt” (suspected), ○ “hvidvask” (money launder/-ing), ○ “svindel” and “svindler” (fraud/fraudster), ○ “korrupt” (corrupt).
These keywords reflect wider themes in the criticism of Danske Bank. Thus, following the money laundering case in bank’s Estonian branch, Danish media would make frequent references to the bank’s perceived criminal behavior, just as the bank would generally be associated with greed and the lack of concern for private customers.
Acknowledging the dialogic aspect of social media, the selection of relevant posts and comments was not restricted to a simple statement-response sequence (Creelman, 2021; Goffman, 1981) but would include all exchanges between the bank’s employee and the critical user 1 on the Facebook page. This means that a sequence may involve one or more statement-response exchanges. Altogether, 42 sequences that included the abovementioned keywords were studied. As is typical of Facebook, comments to posts will often involve more users who will engage in the discussion and comment on other users’ comments. However, since my focus is on the bank’s engagement with individual users, whose criticism is triggered by the bank’s post, my analysis will exclude these additional comments.
A three-step approach was adopted for sorting and analyzing the data. First, based on the close reading of the corpus, three types of users were identified. Thus, it appeared that critical users would interact with the bank’s employees in three primary ways, that is, they would either make a single comment, make more comments/engage in dialog during one sequence, or engage more extensively in dialog within and across sequences following their regular presence. This led to the identification of 32 users altogether, distributed across the following three types:
The one-off user who makes one or only a few comments, but does not engage with the bank’s employees,
The engaged user who enters into dialog with the bank’s employees,
The user who has a regular presence on the Facebook page, and who engages extensively with the bank’s employees.
The one-off user is the user who only appears once in the dataset and only makes one comment in the sequence and never responds to employees’ comments. The engaged user, on the other hand, will be the user who will engage in a more extended dialog with the bank’s employee within the same sequence. The regular user is the user who will engage in a more extended dialog with the employee and who will do so on a regular basis within and across sequences. The way in which these types of users may invite different responses and CHV strategies will be detailed in the analysis.
Second, with the identification of users in place, I would engage in close linguistic analysis of employees’ comments to these users, investigating the use of CHV applying the three key categories suggested by Liebrecht et al. (2021). This means analyzing the use of CHV on the basis of the following:
1. Message personalization
• Personal greetings, addressing the consumer by name, using personal pronouns (first and second), personal signatures, and the sharing of personal information
2. Informal speech
• Non-verbal cues (capital letters, emoticons/emojis, and tactile elements), verbal cues, interjections, and informal vocabulary and phrases
3. Invitational rhetoric
• Invitation to engage in dialog, acknowledging users’ thoughts, suggestions, comments, etc., apologising, showing sympathy and empathy, humor, and well-wishing.
Third, based on steps 1 and 2, I would study users’ reactions to assess the impact of the conversational choices, cf. the aim of the article, which is to investigate whether CHV leads to a positive outcome.
Analysis
The One-Off User
The users who only make one comment and do not engage further in the sequence or on the Facebook page are relatively frequent in the dataset (18 out of 32). These users typically write a critical comment to the bank’s post on a new activity or event even if the event is not associated with any of the bank’s previous critical cases. However, the comment will be triggered by the choice of words in the post or the subject that it deals with, as in the example below.
Sequence 1
The first sequence of exchanges is initiated by the bank making a post on one of its initiatives to combat fraud, that is, when cyber criminals try to scam customers out of money by, for instance, getting access to log-in credentials. The post includes the video of a team who is dedicated to this task and is personalized through reference to one of the team members, Louise. The video is accompanied by the following text (English translation in brackets),
2
Post 1
Mød vores kollega, Louise. Louise og hendes team har den tvivlsomme ære at kontakte kunderne, når vi har mistanke om, at de er blevet udsat for svindel. De professionelle svindlere finder hele tiden på nye metoder, og i Danske Bank oplever vi fortsat en stigning i antallet af forskellige typer af svindelsager.
(Meet our colleague, Louise. Louise and her team have the dubious honour of contacting customers when we suspect they have been the victims of fraud. The professional fraudsters develop new methods all the time, and in Danske Bank we continue to see a rise in the number of different types of scams)
This receives the following comment, Comment 1 (User 1) At det netop er Danske Bank, der ligger ud med dette kan jo bare undre mig – de største professionelle ”svindlere” er jo netop Danske Bank! (That it is Danske Bank who starts doing this just puzzles me – the biggest professional “fraudsters” are, after all, Danske Bank)
The bank’s employee immediately responds, Response 1 (Employee 1)
Hej NN. Vi ved godt, at der kan være delte meninger om os, især ovenpå de alvorlige sager i medierne. Men jeg ved, at mange af mine kolleger, der til daglig servicerer og rådgiver kunderne, vil blive rigtig kede af at blive kaldt svindlere. Fortsat rigtig god dag til dig.
Medarbejder 1 (Hi NN. We know very well that opinions about us may differ, especially after the serious cases in the media. But I know that my many colleagues, who service and advise the customers daily, would be really sad to be called fraudsters. Continued nice day to you. Employee 1)
To understand the use of a conversational human voice in response to critical users, we must start by investigating the context in which the conversation takes place (cf. Barcelos et al., 2018). In the case of a brand-related Facebook page, this means, among other things, investigating its general purpose. In this regard, the bank provides little information, besides stating that the page is a public page in which a set of guidelines must be followed, which involves observing a decent tone, not sharing personal information, etc. However, when studying the many posts made by the bank, it also becomes clear that this is the place where they advertise and position themselves in terms of new initiatives and services to customers, following a rising trend in online branding and social commerce (Barcelos et al., 2018). Furthermore, they do so against the context of intentional crises discussed above. This means that when they communicate new initiatives (as in Post 1), this may be interpreted as a strategy of bolstering (Coombs, 2007) to mitigate the negative effects of the crises and possibly to construct a new narrative for themselves. To do this, the bank has chosen an informal approach over a formal one, which means that they generally adopt a forthcoming and personalized language style.
This is also clear from Post 1. Here, the strategy is to appear inviting and informal, but not personal. This reflects the fact that this is a general post not directed at a particular user. From a CHV perspective, this involves the employment of the three steps of message personalization, informal speech, and invitational rhetoric, but also more formal aspects resembling corporate voice. This is visible in the opening line where the bank is encouraging the reader to engage through the directive “Meet our colleague, Louise.” This is followed by indicators of message personalization such as “our” (l. 1), and “we” (ll. 1 and 2) as well as indicators of informal speech through the general use of active verbs. Together with the mentioning of the colleague’s name and the video, in which Louise talks about her work and what she experiences as a good day at work, this leaves the impression of an accessible and forthcoming organization. However, this is yet again made more formal through the reference to the bank’s name in l. 3 “in Danske Bank, we” (cf. Liebrecht et al., 2021).
The strategy of combining the informal with the formal is abandoned in the employee’s response to User 1 in favor of an informal approach. This may be interpreted as a deliberate strategy to engage with the user and make them feel embarrassed about their comment or simply as a strategy that follows the general organizational outline for engaging with users. However, it is important to note that the user’s comment is not made on the content of the post but is apparently triggered by its reference to fraudsters and therefore the comment may be considered an implicit reference to and criticism of the bank’s previous conduct. In response, the employee adopts a conversational human voice in which they make use of all three steps of CHV. In terms of message personalization, they open their comment with greeting the user using their first name, cf. “Hi NN” (l. 1). They use first-person pronouns, “I” (l. 2) and “we” (l. 1) to indicate engagement with the topic and the user, and they sign the comment with their first name. Furthermore, reflecting invitational rhetoric, they acknowledge the user’s critical feedback by stating that they “know that opinions about us may differ” (l. 1). Together with the informal and colloquial concluding remark “Continued nice day to you” (l. 4), which serves the function of well-wishing, these choices suggest a writer who is keen to personalize the bank in its communication with users (cf. Gretry et al., 2017). The personalization of the bank is underscored by the employee’s reference to their colleagues as people who would be “really sad” (l. 3) to be called fraudsters. In this way, they manage to put the blame on the user for their insensitivity to the “real people” that populate the bank, moving the attention away from what they were criticizing.
The Engaged User
The users who enter into dialog with the bank’s employees in one or more sequences are also relatively frequent in the dataset (10 out of 32). They are users whose comments are also triggered by the posts made by the bank, but who will engage in a more elaborate dialog typically with the aim of adding to the criticism voiced in their first comment. Thus, they will add one or more additional comments once the bank has replied to their initial comment. This is illustrated in the following example.
Sequence 2
The second sequence is initiated by a post where the bank warns against false webpages used by cyber criminals to trick people on Black Friday. The bank writes, Post 1 Pas på (Watch out 
Black Friday er en mærkedag for internettets kriminelle. Her er 6 gode råd til at spotte en falsk webshop. Black Friday is a red-letter day for Internet criminals. Here are 6 good pieces of advice on how to spot a fake web shop.
As in Sequence 1, the opening post is phrased in an informal tone of voice. In fact, this is taken a step further by including an emoji to stress the call to attention in “Watch out,” and the adding of the hashtag, #blackfridaydk, possibly to add to the informality of the text (Liebrecht et al., 2021). Furthermore, the text is accompanied by six images where the bank elaborates on the six pieces of advice. Altogether, this leaves the impression that the bank tries to communicate about a serious matter in the informal and somewhat upbeat fashion frequently associated with Facebook.
This triggers the following comments from a user, Comment 1 (User 2)
Amen det er da intet andet end fantastisk, man fra Danske bank’s side advarer om kriminelle, når man selv driver en virksomhed fuldstændig som var det hele et stort slaraffenlands tagselvbord, Danske bank er det tættes på mafiaens stormagtstid i 30’erne, bare gemt bag et hav af advokater og lovgivninger fortolkninger til deres fordel .. selv om i jo godt nok kun slår folk stille og roligt ihjel med latterlige stavnsbindinger på høje renter og gebyrer ..
(Well, it is no less than fantastic that Danske bank warns against criminals when they run a business completely as if it were all one big buffet from the land of milk and honey, Danske bank is the closest you get to the golden age of the mafia in the thirties, just hidden behind a multitude of lawyers and legislation interpreted to their advantage .. even if, admittedly, you only kill people slowly by tying them in through high interest rates and fees ..)
This receives the following response, Response 1 (Employee 2)
Hej NN. Vi er bestemt ikke stolte af de sager, som vi har været involveret i. Vi er kede af de fejl, som vi har begået, og vi arbejder på at blive en bedre bank for både vores kunder og samfundet. Vi har lavet dette opslag, så vores kunder er opmærksomme på, hvordan de undgår at blive snydt på internettet. // Medarbejder 2
(Hi NN. We are certainly not proud of the cases we have been involved in. We are sorry about the mistakes we have made, and we are striving to become a better bank both for our customers and society. We have made this post so that our customers are aware of how they may avoid being cheated on the Internet. //Employee 2)
In this exchange, the user’s comment is apparently triggered by the reference to criminals in the post and less by the contents, as was also the case in Sequence 1. This allows them to voice harsh criticism by comparing the bank and its activities to those of the American mafia in the 1930s.
Even if the criticism is harsh, the employee’s response remains subdued. We may identify the same CHV strategies as in the exchange in Sequence 1, that is, the response is initiated by greeting the user using their first name, followed by the extensive use of the first-person plural pronoun “we” (for instance, ll. 1, 2, and 3) and active verbs, and concluded with the signature of the employee. In addition, attitudinal and emotional adjectives (for instance, l1. 1 “not proud,” and 2 “sorry”) are used to emphasize the expressed regret and apology in ll. 1 and 2. Previous research in webcare indicates that this is a way of taking users’ comments seriously and inviting them to engage in conversation (Grégoire et al., 2015). However, it also echoes the recommendations in the crisis literature that to contain an intentional crisis, apology is an important means of rebuilding the perception of the organization (Coombs, 2007; Utz et al., 2013). This strategy is supported by the employee’s efforts to bolster the bank’s image and reputation through the mention of the good deeds they are doing, cf. l. 2 “we are striving to become a better bank” (Coombs, 2007).
However, this reply does not go down well with the user, who replies, Comment 2 (User 2)
Ja det er jo meget bekvemt Danske Bank, som i helst vil se frem af, men det ændrer dog intet på Danske Bank driver forretning som Mafiaen. Og kommer man først i Danske Banks klør, er internetsvindel, et Meget lille problem i forhold til hvad Danske Bank er klar til at udsætte deres kunder for i flere år. . .
(Well, it is, indeed, very convenient, Danske Bank, how you prefer to look ahead, but it does not change anything about the fact that Danske Bank runs its business like the Mafia. And once you are in Danske Bank’s clutches, internet fraud is a Very small problem compared to what Danske Bank is prepared to do to their customers for several years . . .)
As appears from the comment, the user insists on comparing the business strategy of the bank to that of the mafia despite the employee’s apologies. This leads to the following response from another of the bank’s employees, Response 2 (Employee 3)
Hej NN. Der er ingen tvivl om, at vi ønsker at se fremad – men vi glemmer absolut ikke de ærgerlige sager, som vi har været indblandet i. Du må endelig ikke tøve med at kontakte os, hvis du ønsker, at vi tager en snak om dine tanker. Send os gerne en mail på
(Hi NN. There is no doubt that we want to look ahead – but we do certainly not forget the unfortunate cases we have been involved in. Please, do not hesitate to contact us if you want us to have a talk about your thoughts. Please, send an email to 
Rigtig god dag. //Medarbejder 3
Really nice day. //Employee 3)
In many ways, this response resembles the previous two responses in its adoption of CHV strategies, that is, it personalizes the message by employing a personal greeting (l. 1), first-person plural and singular pronouns (ll. 1, 2, and 5), second-person singular pronouns (ll. 2 and 3), active verbs, a standard Danish closing remark, and a signature. The employee also acknowledges the user’s comments and confirms the previous statement of regret. However, with this being the second response to the same type of criticism, part of the response has been adapted to this situation. While the employee remains friendly and approachable (for instance, through the addition of a smiling emoji), they clearly try (and manage) to close the sequence by referring the user to the bank’s email for further correspondence, phrased as a polite invitation (cf. l. 4, “Please, send an email to
The Regular User
The final part of the analysis will concern exchanges between a regular user and the bank’s employees. The regular user makes a frequent appearance on the Facebook page, engages in more extensive dialog, and will continue to be critical of the bank. Thus, the regular user will react to more of the bank’s posts, add additional comments once the bank has replied to the initial comment and engage with other users, as well. As a type, the regular user is not as numerous as the other two user types as they only represent 4 out of the total of 32 users.
Sequence 3
Sequence 3 is also initiated by the post that initiated Sequence 2. This is where the bank warns its customers against cyber criminals on Black Friday. The sequence starts with the regular user stating the following, Comment 1 (User 3) Danmarkshistoriens mest kriminelle firma advare mod andre kriminelle . . ... man kan efterhånden ikke stole på nogen (The most criminal firm in Danish history warns against other criminals . . .. you can hardly trust anyone anymore 
)
This receives the following response from the bank, Response 1 (Employee 4)
Hej NN. Det er snart længe siden, vi har set noget til dig (Hi NN. It has been a long time since we last saw you Vi ved godt, at din holdning til os ikke er positiv – men vi ønsker blot at informere vores kunder om nogle af de ting, som de skal være opmærksomme på, når de handler på nettet. Vi håber, du har det godt, og at du får en god dag De bedste hilsner fra Medarbejder 4
We know that your attitude towards us is not positive – but we just want to inform our customers of some of the things they should be aware of when they shop on the Internet. We hope you are doing well, and that you will have a nice day Best wishes from Employee4)
As will appear from this initial exchange, the user’s criticism of the bank’s previous conduct is reflecting that of the previous two exchanges. The bank is, in other words, seen as a criminal, presumably because of its involvement in, for instance, money laundering. This comment is once again triggered by the choice of words in the post and less by its contents. However, as with the comment in Sequence 1, the user explicitly questions the bank’s moral principles when it warns against the criminal conduct of others and underscores their attitude by adding no less than five loudly laughing emojis.
Even if the comment is comparable to the previous ones, the employee’s response is different, leaving the impression that this is a conversation between two persons who know each other well. Thus, the realization of the categories of message personalization and invitational rhetoric is even more prominent than in the two previous sequences. This is reflected in several places in the comment, for instance, in l. 1, where the greeting of the user is extended into the first sentence. Besides using first- and second-person personal pronouns (“we” and “you”) as well as a smiling emoji to signal friendliness and personal engagement, the employee acknowledges the regular presence of the user on the page by stating “It has been a long time since we last saw you” (l. 1). This may, in fact, also be interpreted as a slightly ironical remark to suggest that this user is known to be not only regularly present but is also perceived as annoying. The acknowledgement of the user (as a representation of invitational rhetoric) is repeated in the closing of the comment with phrases of well-wishing (cf. l. 3, “We hope you are doing well, and that you will have a nice day”), a smiling emoji and an extended closing (l. 4, “Best wishes”). However, given that the user is already known to be continuously criticizing the bank, such remarks may appear insincere. Nonetheless, the user continues down this path in the following, Comment 2 (User 3)
Hej Medarbejder 4. Tak for det, og glædelig jul til dig og dine (Hi Employee 4. Thank you, and a Merry Christmas to you and yours 
PS, så håber jeg for dig, at du her i det nye år, finder et samfundsnyttigt og respektabelt job, så når du er ude blandt andre mennesker med stolthed kan sige hvor du arbejder
PS, then, my hopes for you are that in the new year, you will find a respectable job which is useful to the community, so that when you are together with other people, you can proudly tell them where you work )
This receives the following response, Response 2 (Employee 4)
Jeg er ellers utrolig glad for mit arbejde, og det er fortæller jeg gerne de mennesker, som jeg møder på min vej (I am usually very pleased with my job, and I like to tell this to the people I meet along the way Du må også have en rigtig god jul med din familie og dine nærmeste - Medarbejder 4
I also wish you a very Merry Christmas with your family and your nearest and dearest - Employee 4)
Despite the perceived insincerity of the employee’s well-wishing in the first response, this continues to be a significant aspect of the comment and response that follow, as illustrated in Comment 2, l. 1 “Thank you, and a Merry Christmas to you and yours,” and in Response 2, l. 2 “I also wish you a very Merry Christmas with your family and your nearest and dearest.” Both are underscored with the use of emojis. This suggests that even if well-wishing may be considered ironic, or at least representative of the standard exchanges of pleasantries in preparation of the holidays, it still appears to function as an appeasement of the user through the building of rapport (cf. Liebrecht et al., 2021). By taking up the employee’s “invitation” and adding winking emojis, the user shows they appreciate the gesture, but still maintain an ironic distance.
Discussion
Using CHV in its communication, the organization will strive to appear human, approachable, and engaged to improve its overall relationship with its publics (Kelleher, 2009; Kelleher & Miller, 2006). For an organization like Danske Bank, which has struggled with a bad image and reputation for many years following numerous crises, adopting a CHV approach would appear to be a perfect solution, especially since studies find that it increases brand trust, when the brand is familiar (Gretry et al., 2017). As Denmark’s largest bank, its brand is, indeed, familiar. The question remains, though, whether the positive outcomes observed in these studies can be transferred to other types of online interaction such as the one analyzed above, where reputation is at stake. Studies suggest that positive or negative outcomes of a humanized voice are highly dependent on the context, so that what may engender positive outcomes in one context may not do so in another (Barcelos et al., 2018; Crijns et al., 2017). Additionally, what may count as conversational human voice is highly dependent on context, even if agreement exists in the literature that its main categories or tactics are message personalization, informal speech, and invitational rhetoric.
The analysis is indicative of these challenges. In their responses to users’ criticism, employees realize CHV through a relatively restricted set of categories involving personal greetings and signatures, personal pronouns, active verbs, apology/regret, and well-wishing. In addition, the initial responses to one-off and engaged users (Sequences 1 and 2) seem to follow a predetermined set of moves where the employee will
greet the user,
acknowledge the criticism and/or express regret,
elaborate on the contents of the post,
conclude the response with a closing pleasantry and/or a signature.
This confirms the findings of other studies in the banking sector (Creelman, 2021), but it also suggests that part of the communication strategy is inspired by recommendations in the crisis communication literature. Thus, as mentioned in the section on crisis responsibility, a common recommendation for warding off a crisis would be to adopt an apologetic response strategy (Utz et al., 2013). This is the strategy realized in the second move above. Another recommended strategy is bolstering (Coombs, 2007), that is, reminding publics of the organization’s good deeds, which is clearly represented in Sequence 1 (Post 1) as well as Sequence 2 (Response 1). Altogether, for the communication with the two mentioned types of users, there seems to be an underlying organizational manual for how employees should engage with (critical) publics that is in part inspired by what is currently used in other organizations and in part informed by studies in the field.
Deviations from this relatively strict set-up emerge when the conversation moves beyond the initial exchange with users or when it is taking place between the employees and regular users. This suggests that once a relationship has been established, the employees will have more communication options at their disposal. However, the freedom to explore these options also appears to be restricted when communicating with engaged users. We see this in Sequence 2 (Response 2), where the employee will open the response with moves 1 to 3 and then refer the user to the organizational email for further handling. Although this is slightly different from what appears in the initial exchange, it resembles the strategies observed in webcare studies. In their study of (de)legitimization strategies in company-customer communication via emails and social media, Van Herck et al. (2020) found that in more than half of the examined company comments on social media, the move Future contact/Offer to help would be used, mostly with the aim of redirecting the consumer to a private channel. What is interesting in relation to the present study is that this strategy is often categorized as a no-response, resulting in a negative consumer experience, effectively cancelling any trust in the organization and its desire to appear apologetic. However, despite the observed restrictions on employees’ replies, the employees still find ways to appear friendly and approachable, no matter the tone and contents of the users’ comments. This is, for instance, the case in Sequence 1 (Response 1) with the employee imparting the perceived sensibilities of their colleagues, and in Sequence 2 (Response 2), where the employee strongly encourages the user to get in touch with the bank, underscored by a smiling emoji. Examples like these suggest that employees are at liberty to exploit the potential for personalized messages inherent in CHV, creating a more contextualized and relevant response than a possible organizational manual may allow for. In the communication with regular users (Sequence 3), the freedom to choose between options appears even more clear. The use of CHV tactics will be firmly centered on personal greetings, personal pronouns, active verbs, and a closing pleasantry/signature; however, the exploration of additional CHV categories, such as emojis and humor/irony, and additional moves establishing rapport, seems more prevalent, possibly because this user is in touch with the employees on many occasions. As with Sequences 1 and 2, the adaptation of CHV to a more personalized style is also evident here. This is clearly seen in the closing pleasantries, which are elaborated versions of standard Danish phrases supported by a smiling emoji. Thus, to hope that someone is doing well (as in Response 1) or to extend the season’s greetings to include family and close ones (as in Response 2) is not typically associated with business correspondence, that is, a corporate voice, but gives the employee the possibility of mitigating the criticism through the building of rapport.
Conclusion
The article set out to investigate the extent to which conversational human voice (CHV) is used in communication with critical publics on a brand-related Facebook page (cf. RQ 1), as illustrated by the case study of Danske Bank. Additionally, the aim was to investigate whether this leads to a positive conversational outcome, defined as the user expressing satisfaction with the response and a positive attitude (cf. RQ 2).
From the analysis, it appears that the answer to these questions is not straightforward. If we acknowledge that CHV can be operationalized as message personalization, informal speech, and invitational rhetoric and that these categories can be realized linguistically in many ways, then CHV is, indeed, being used. However, the way it is realized seems to be partly contingent on the context and partly inspired by literature and practices in crisis communication. This results in what appears to be standardized responses in the initial exchanges with users, and only when these users are increasingly known to the employees, do they depart from the script. In other words, the bank, as organization, does not seem to consider the possibility of adjusting CHV strategies to the individual user in any significant measure, nor do they appear to be wary of the fact that negative eWOM is deeply rooted in frames of criminal activity, which allows the users to construct the bank as fraudsters, the mafia, and the most criminal organization in Danish history. Only when employees adopt a more personalized and appreciative style, that is, a contextualized realization of CHV, there are indications of a positive outcome.
This begs the question of whether CHV leads to a positive outcome when used outside customer complaints over service or product failures. If we assume that lacking comments from users are indications of satisfaction, then the answer is yes. However, the fact that responses can easily be spotted as being standardized, also by users, would suggest otherwise. Thus, if the bank continuously repeats that it regrets its involvement in dubious business conduct, but nothing else is communicated as to what it intends to do about this, keeping it a thing of the past, then a negative outcome of the conversation is more likely than a positive one. This, in a sense, confirms the claim found in the crisis communication literature that a history of intentional crises contributes to the forming of negative reactions and anger among publics, even in the long term (Coombs, 2004; Lerner & Tiedens, 2006; Weiner, 1985). And this is, indeed, also visible in the data, when the negative framing of the bank is repeatedly voiced, even several years after the occurrence of the critical cases. As suggested above, a more positive outcome may also be achieved through employees’ appreciation of criticism and a personalized style. This will, however, undoubtedly require a more consistent and strategic effort to be successful, as also illustrated by the regular user’s recurring criticism. Furthermore, what remains to be investigated is the extent and degree to which criticism has been voiced in the Facebook page. Even if we can read from the bank’s policy that a decent tone must be observed, it is not clear what exactly this entails. In other words, does it only means removing offensive material, or does it also involve sorting comments that promote criticism, which has the potential of doing significant harm to the image or brand? In the latter case, this would potentially mean that criticism is even more prominent than indicated in this study. Even if this is a small-scale study, it illustrates the complexities of addressing critical publics, suggesting that even if CHV would mean a more personalized communication style that ideally would make users more sympathetic to the organization and its brand, its adoption would need to reflect the communicative as well as broader organizational context to be successful. To establish this more firmly, however, analysis will have to be conducted on larger datasets across organizational boundaries, in which attention continues to be paid to the particularities of the industry, as also suggested by this and previous studies (cf. Barcelos et al., 2018).
Footnotes
Declaration of Conflicting Interests
The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author received no financial support for the research, authorship, and/or publication of this article.