Aspiration, appearance, argumentation or attack? Reconstructing stakeholder engagement images of German industrial companies on social media

Companies face increasing pressure to react to societal challenges, such as growing inequalities or environmental crises like climate change (Mubushar et al., 2025; Gümüsay et al., 2022). The term corporate social responsibility (CSR) describes the responsibility of companies towards society. Companies use CSR management to guide how they operate within interconnected ecological, social and economic systems, seeking to balance these relationships in a way that generates positive outcomes, such as long-term environmental stability, improved societal well-being and sustained value for stakeholders (Rasche et al., 2023). Stakeholder engagement is a crucial part of dealing with these challenges. This is especially true in the digital age, where attention to ethical concerns has increased, and interconnected forms of communication allow for the rapid spread of news. At the same time, social media platforms offer the opportunity for organisations to engage with their stakeholders (Elving and Postma, 2017).

The focus of this study is on stakeholder engagement in the context of CSR activities, defined as “the practical mechanism through which social responsibility can be enacted in corporate decision-making as a counterweight to the primacy of shareholder value” (Hine and Preuss, 2009, p. 383). In this sense, stakeholder engagement includes a moral component and aims to (co-) develop the CSR management of companies as well as gain legitimacy for corporate actions in connection with environmental and social concerns (Manetti and Toccafondi, 2012; Kujala et al., 2022).

From a theoretical perspective, stakeholder engagement on social media can be first and foremost analysed as a communicative process or “communicative event” (Schultz et al., 2013). Morsing and Schultz (2006) distinguish three stakeholder communication strategies. While there is no stakeholder engagement in the information strategy, the response strategy offers stakeholders the opportunity to provide feedback in a manner controlled by the company (surveys, etc.). Only the involvement strategy can be regarded as stakeholder engagement in the narrower sense, as the creation and giving of meaning to CSR measures is co-constructed in an iterative process (Morsing and Schultz, 2006). Two-way symmetrical communication or dialogue is therefore considered central for facilitating and implementing stakeholder engagement processes in the narrower sense (Kent and Taylor, 1998; Manetti and Toccafondi, 2012; Golob and Podnar, 2014; Kujala et al., 2022; Morsing, 2006).

In its ideal form, this dialogue is characterised by an attitude of openness and respect of the participants in a communication (Kent and Taylor, 1998). In this sense, it is a form of communication that helps to soften power imbalances while prioritising interpersonal engagement, shared meaning-making, mutual understanding, the co-construction of reality and empathetic or sympathetic interaction (Taylor and Kent, 2014). This does not mean that participants have to agree on all issues to establish a dialogue – most of the time, interests are conflicting, and it is not the goal to find agreement, but to create and negotiate relationships among multiple stakeholders (Kent and Taylor, 1998; Taylor and Kent, 2014). Social media platforms such as Twitter [¹] offer great potential for stakeholder engagement, as they host a large number of active users from diverse backgrounds. Moreover, their accessibility to different stakeholders leads to a reconfiguration of power relations (Castelló et al., 2016) and enables direct feedback in real time and interactive communication between participants (Kent and Taylor, 2016). Social media, therefore, has the potential to promote dialogue and transparency and include multiple voices and viewpoints, but at the same time presents a challenge for companies to meet stakeholder expectations and legitimise their actions (Schultz et al., 2013; Elving and Postma, 2017; Kent and Taylor, 2016).

Nevertheless, the promises of social media for corporate stakeholder engagement processes have not been fulfilled. The early claim of communication scholars such as Kent and Taylor that the capacity of social media, e.g., for relationship-building between organisations and the public is still largely untapped, and what is often called dialogue on social media is not dialogue in the scientific sense (Kent and Taylor, 2016), has since been confirmed by different empirical results. Studies such as that by Illia et al. (2017) have found that most companies favour interactions they can control, with only a small number of companies willing to engage in open discussions on issues prioritised by stakeholders and civil society (Illia et al., 2017). In their study of the CSR communication of banks on Twitter, Gómez-Carrasco et al. (2021) arrived at a similar result. They found that CSR messages exchanged among external stakeholders are more often related to the core business activities, while firm insiders are more likely to release information about supplementary CSR activities, such as philanthropic events, particularly in times of crisis. This leads to a discrepancy between the concerns of the stakeholders and the communication of the companies (Gómez-Carrasco et al., 2021). Elving and Postma (2017) interviewed company representatives about their stakeholder engagement strategy on social media and found that social media was mainly used as a marketing tool, and the sampled companies were not taking advantage of the interactive opportunities (Elving and Postma, 2017). Other studies show that even when companies engage in more dialogic communication, these efforts are often met with scepticism by the targeted audience (Colleoni, 2013).

Research has addressed this problem of absent dialogue, for example, by testing communication strategies and content that are effective for creating dialogue (Abitbol and Lee, 2017; Araujo and Kollat, 2018; Capriotti et al., 2021; Viglia et al., 2018; DiStaso and Bortree, 2012; Pizzi et al., 2021), focusing on how dialogic spaces in the virtual realm should be designed (Illia et al., 2017) or on the mindsets of communication professionals who use social media (Kent, 2022). Moreover, Kent and Taylor (2021) call for a different architecture of social media, allowing for more meaningful conversations and social change (Kent and Taylor, 2021). In addition to more text-based communication, Bilderback (2026) shows that Snapchat's unique features, such as augmented reality (AR) lenses, enable companies to create interactive and engaging sustainability content. Given that alternative social media platforms that foster dialogue and stakeholder engagement are not widely implemented and used yet, the question arises whether commercial social media platforms play a role in stakeholder engagement.

For our empirical analysis, we adopt the view that “communication constitutes organisation (CCO)”, which allows us to broaden the perspective of social media use for stakeholder engagement. The main function of the CCO approach in this study is to provide a rationale for the formative role of communication in CSR, exploring its consequences and further advancing communicative approaches to legitimisation in the networked society.

Based on this, we propose an approach for analysing stakeholder engagement on social media that remains grounded in normative dialogue. While it continues to consider dialogic intention as important for stakeholder engagement, it also recognises the constitutive nature of communication as a form of engagement, going beyond the normative aim of dialogue. The basic assumption is that when companies use certain hashtags on Twitter (cf. e.g. Albu and Etter, 2016), they respond to underlying social expectations by creating or shaping a stakeholder engagement image (cf. Crane and Glozer, 2016). This, in turn, leads to stakeholders engaging with it (or not) by replying or retweeting, thus influencing and shaping the corporate engagement image by interacting with it. We thus develop a novel framework and thereby extend theory where communication in a networked society is seen as formative for organisational identity and images (Dobusch and Schoeneborn, 2015; Dawson, 2018; Schultz et al., 2013; Crane and Glozer, 2016) and stakeholder engagement can be characterised as the stabilisation and expansion of organisational boundaries through communication with third parties (Schoeneborn and Trittin, 2013). Explaining stakeholder engagement on Twitter as a constitutive process takes into account the changing communicative dynamic of a networked society and thereby contributes to questions of legitimacy creation on social media (Schultz et al., 2013).

In our study, we therefore address the following research questions:

1. How can the engagement between German companies and climate activists as their stakeholders on Twitter be qualitatively described and analysed when adopting a constitutive approach to communication?

2. What practical and theoretical implications can be drawn from this case regarding stakeholder engagement on social media platforms such as Twitter?

Based on Crane and Glozer's (2016) “4Is concept” for CSR communication research, we reconstruct corporate images (of stakeholder engagement) that are shaped through communication within a dynamic and discursive space, where numerous stakeholder voices interact and contest meanings (Crane and Glozer, 2016).

A stakeholder engagement image represents the sense-making and sense-giving processes associated with a hashtag. These images are then used to conceptualise and evaluate stakeholder engagement processes on social media. The presented framework provides an opportunity to analyse the potential of social media for stakeholder engagement in the absence of “real” dialogue. We identify four basic types of stakeholder engagement: “Aspiration”, “Appearance”, “Argumentation” and “Attack”. These can serve as a starting point for considering how stakeholder engagement on social media can be supported, from both theoretical and practical perspectives. In the following section, a review of the literature is provided to describe the shift in perspective on communication and stakeholder engagement introduced by the view that “communication constitutes organization (CCO)”. Next, the methodological framework and its basic assumptions are presented, followed by a detailed description of the approach to data collection and qualitative data analysis. In the results section, a description of the images and subgroups is given, which is then discussed in light of their theoretical and practical contribution. The final section discusses limitations and identifies possible starting points for further research.

The reconstruction of images in this study, which are created by companies when interacting with certain “socially charged” hashtags, corresponds to the “communication constitutes organisation (CCO)” perspective. Following Crane and Glozer (2016), a CSR image is defined as a manifestation of CSR communication aimed at external stakeholders. From a strategic communication perspective, a positive CSR image and strengthened reputation are outcomes that can be achieved through coordinated communication efforts (cf. Ajayi and Mmutle, 2021; Mochales and Blanch, 2022). The CCO view highlights the formative aspects of communication and emphasises that CSR images are co-created by multiple actors, meaning they may diverge from the sender's intended meaning (Putnam, 2022; Schoeneborn et al., 2025). This image is dynamic and discursive as it does not represent the “real” corporate identity but rather one of many representations that are reproduced inside and outside the company shaped and changed through communication (Crane and Glozer, 2016).

The view that communication is performative and creates organisations is rooted in the speech act theory, as developed by authors such as Searle (1969) and (Austin, 1975; Christensen et al., 2021). According to this line of thinking, all utterances are performative rather than merely a means of conveying “objective” information about reality (Ludwig and Ruyter, 2016). Speech act theory separates locutionary acts, the act of saying something, from illocutionary acts, which concern the speaker's intended function in saying it, and perlocutionary effects, which refer to the impact that the utterance has on its audience (Ludwig and Ruyter, 2016). All these elements can be found in companies' online images, which are shaped not only by the company's intended presentation but also by how these presented images are taken up by platform users, who may create new speech acts and thereby alter the company's image (cf. Conventional speech act by Sbisà (2009)). This fluidity and co-creation of images distinguish them from related communication concepts such as narratives and frames. Unlike narratives, which provide structured storylines, images are dynamic and situational. While frames are actor-specific interpretive lenses that structure communication, images represent the emergent, co-created perceptions produced through these communicative processes.

This perspective on images is particularly relevant in the context of social media, where companies lose control over the content they communicate to a certain extent, making it difficult to strategically manufacture information to engage certain stakeholders or create a specific, uncontested image (Kim, 2019; Dawson, 2018).

The images reconstructed in this study are stakeholder engagement images, which means that the focus is less on the negotiation of the issue and the contested field itself (here, climate change) but more on the willingness of companies to negotiate such issues with external stakeholders.

The central argument of the CCO view is a shift in perspective from organisations as “pre-existing realities” or “social facts” in which communication occurs as a phenomenon, to organisations that are constituted through communication (Cooren et al., 2011, p. 1150) or, in other words, “organisations come into existence, persist, and are transformed in and through interconnected communication practices” (Schoeneborn et al., 2019, p. 476). In this sense, CCO combines insights from organisation studies with communication theory, adopting a social constructivist perspective on how communication shapes organising and organisations (Trittin-Ulbrich, 2016). In the context of CSR, this means that CSR communication can be understood as a dialogical process through which an organisation is co-constructed (Cooren, 2020). In this sense, there is no clear distinction between CSR communication and activities, as the negotiation of meaning or sense-giving and sense-making processes influences the practice of CSR (Schoeneborn et al., 2020). This suggests that even aspirational talk can ultimately strengthen CSR management (Koep, 2017; Christensen et al., 2021). Moreover, authors such as Schultz et al. (2013) or Castelló et al. (2013) point out the shortcomings of an instrumental view of CSR and stakeholder engagement in a digital society, which regards social media as a means or medium for achieving legitimacy, but also of the normative view that regards social media as a socio-technical “space” for dialogue with stakeholders. This critique is based on three reasons:

Firstly, discourses on CSR arise on social media that are outside the sphere of influence and sometimes even the knowledge of companies and are not controlled by companies (Schultz et al., 2013). This changes the role of companies, which no longer initiate the discourse but instead encounter a variety of competing interpretations and negotiations of reality on social media (Castelló et al., 2013). Secondly, consensus as a basis and goal for legitimacy does not seem to correspond to the realities in social media and conflict and disagreement may play a more important role (Castelló and Lopez-Berzosa, 2021). This also challenges the role of dialogue on social media: “The assumption of the more the dialog, the more the communality seems to fail to portray the complexity of the communicational dynamics” (Colleoni, 2013, p. 228). Thirdly, while from the perspective of moral legitimacy, CSR communication is often branded as greenwashing, the CCO view broadens this perspective by acknowledging the role of aspirational talk as performative for changes in the company (Schoeneborn et al., 2020).

The CCO perspective encompasses three main assumptions that are of particular relevance for studying stakeholder engagement on social media. Firstly, from this perspective, communication is seen as formative, meaning that the act of communicating itself constitutes the organisation or here “virtual image” of organisations on social media (Crane and Glozer, 2016). As Dawson (2018) argues, organisational identity becomes more of a communicative phenomenon than a managerial one in the interactive social media setting, making the CCO framework particularly valuable for this constantly evolving, fluid and highly engaging environment. Secondly, in the CCO perspective, this organisational image is not understood as a Homogeneous representation of what the company wants to present, instead organisational (CSR) culture and values are fluid and continuously negotiated, shaped by how participants make sense of their environment (Carradini et al., 2025). Images on social media are thus created through interaction with the “social media society” involving multiple and also dissonant voices (Schultz et al., 2013). Thirdly, communication is seen as a vehicle for co-developing the CSR management, or as Schoeneborn and Trittin (2013) state, CSR communication has the potential to broaden organisational boundaries by engaging third parties (Schoeneborn and Trittin, 2013).

The research design is based on the qualitative analysis of tweets sent from companies listed in the German DAX index that contain one or more of the hashtags most frequently used by the climate activist group Fridays for Future in the sample period (see Figure 1).

The analytical framework is based on two assumptions:

1. Firstly, we include the normative concept of dialogue not as a result or prerequisite of stakeholder engagement, but by evaluating the dialogical intention of tweets. We assume that tweets from companies contain certain characteristics that reveal this dialogic intention, which we define as the willingness of companies to interact with stakeholders. It comprises elements and forms of social media communication that are known to favour dialogues and that have been deductively derived from the literature. These features are supplemented by our own inductively derived characterisations of tweets.

2. Secondly, applying the CCO view, stakeholder engagement processes can be seen as a negotiation (or rejection) of a (corporate) image in virtual space. This image can evolve in response to a “socially charged” hashtag that acts as a co-authored representation of what, in our case, a climate-friendly organisation means and what tackling climate change entails (Albu and Etter, 2016; Castelló et al., 2016). This involves processes of sense-making of the “socially charged” hashtag, as well as sense-giving and corresponds to the idea of stakeholder involvement characterised by Morsing (2006). On a principal level, companies can react in two ways: distance themselves from social expectations (distancing) or strive for consensus with stakeholders (confirming) (see also Dawson, 2018). From a CCO perspective, this can be described as defending or maintaining vs. extending the organisational boundaries (cf. Schoeneborn and Trittin, 2013).

The research questions were answered by analysing conversations on Twitter between DAX-listed companies, which have high relevance for reducing CO₂ emissions in Germany (Böhm and Haerle, 2024), and stakeholders on central topics of climate activism. To identify the relevant hashtags for these conversations and to define the dialogic intention of the tweets, we used the climate activist group “Fridays for Future” as a representative of a stakeholder group with a strong member base in Germany (and internationally) that, at the time when the data sample was extracted, was very active and thus influential with regard to climate protection (Fridays for Future, 2024). We consider this example appropriate for studying stakeholder engagement on social media for three reasons: Firstly, this activist group exerts legitimising pressure on companies, which increases their willingness to engage with them. Secondly, the activist group uses social media for maintaining and building its own networks, which makes social media a suitable venue for initiating stakeholder engagement. Thirdly, the (industrial) DAX-listed companies are among the largest emitters of greenhouse gases in Germany, making them an important target for Fridays for Future activists. Moreover, according to Saxton et al. (2021), the “connective power” of Fridays for Future Germany as part of a well-established virtual network, and their “normative power” based on prestige and esteem, makes it more likely that German industrial companies interact with these stakeholders on social media (Saxton et al., 2021).

According to Miles and Huberman (2008), the sampling strategy corresponds to the intensity type, which is characterised by its wealth of information and the increased likelihood with which the studied phenomenon (here, stakeholder communication) manifests itself.

Working with data obtained from social media raises ethical questions, including the distinction between private and public data, which is not fully addressed by the guidelines for ethical “analogue” research. We therefore sought legal advice and consulted with the internal data protection officer, developing a data management plan in accordance with Gold's (2020) guidelines (Gold, 2020). Recognising that tweets are secondary and dynamic data, which may change in the future, our data management was designed not to impair users' autonomy or process any personally identifying data, such as that contained in user-generated tweet texts, on our systems. After describing and submitting our research design, we were granted access to the “Twitter Research API”.

The group of relevant industry accounts was compiled by manual search on Twitter for user profiles of publicly listed DAX companies (as of April 2022). Twitter accounts that met the following criteria were included in the “industry” group:

1. “Based” in Germany (handle description) and

2. Clearly representing the main DAX-listed enterprise or its main brand, or

3. Sustainability-related, public relations or customer service-oriented accounts, or

4. Dedicated to the communication of a specific business unit (e.g. energy) that is not fully independent and still shares the corporate identity of the main company.

Due to our research focus on stakeholder communication in the context of companies that are particularly relevant for reducing CO₂ emissions, we additionally excluded all DAX-listed companies whose business models are not based on physical products (mainly banks and insurance companies). In total, the search identified 98 Twitter accounts representing 40 DAX-listed companies on Twitter (see Appendix 7.1 for the full list of identified accounts).

In order to identify the most relevant topics for stakeholder engagement related to climate change, we analysed the Twitter history of the Fridays for Future activist group. We normalised and aggregated the most frequently used #hashtags in tweets published or retweeted by the Germany-based @FridayForFuture account. Since the beginning of the protests and within the time range of the query (01.08.2018–10.10.2022), a total of 4,572 tweets were either originally posted (3,713) or retweeted (859). We decided to only include the tweets originally posted by the account, which we used for a hashtag frequency and content analysis. After normalisation, we identified 3,453 single usages of 749 unique hashtags (in the form of “#keyword”) by the user @FridaysForFuture. After excluding all hashtags that were only used once and those that we assumed unnecessarily widened the scope of our analysis, such as country names or names of politicians (see Appendix 7.3), we identified 212 hashtags to represent the topics the climate activist group is concerned with on Twitter (see Appendix 7.2 for the complete list).

We then identified all tweets from DAX-listed companies that addressed at least one of the 212 topics raised by the activists. Due to the 1024-character limit on the Twitter Search API, the relevant hashtags for the Fridays for Future movement were split into groups of six, as were the 98 industry accounts (see Appendix 7.1). To identify relevant conversations, the keyword “from:” was prefixed to each of the industry account handles for the query. We iterated through each of the six sets of relevant hashtags for each of the six industry group accounts, resulting in 36 unique queries to the Twitter Research API, 34 of which delivered a non-zero number of matching tweets (see Appendix 7.4 for more detail on the code and query). This was performed using the command-line version of the Search Tweets Python API client, as it supports the Twitter API authentication method in version 2.0 and has rate limiting and paging implemented ^[2]. The client was developed and published by Twitter itself.

The dataset we obtained from Twitter in this way consisted of 1,662 tweets posted by the identified corporate accounts. All tweets that strayed from the topic and did not address sustainability or climate-related content (377 tweets), as well as tweets that did not result in any interaction, were excluded from the sample (1,033 tweets). Each thread that contains at least one tweet from the company and one reply from a stakeholder is referred to as a conversation. Conversations with people who identified themselves as belonging to the respective company (e.g. moderators of a chat) that posted the tweet were excluded from the sample (6 tweets). In the course of several rounds of analysis, one account was deleted and dropped from the sample (19 tweets). The adjusted sample, therefore, consisted of 227 conversations.

Stakeholders look for signs that a company is approachable and willing to engage in dialogue. This means that, for a company, it is equally important to consider what is communicated and how it is conveyed (Saxton et al., 2019). In a preliminary study, a series of 12 semi-structured interviews was conducted with activists from the German branch of the Fridays for Future group and one member of another climate activist group (Niehoff, 2026). The aim was to shed light on the use of social media by activists and their willingness to interact with companies. The interviews delivered three relevant findings for defining dialogic intention:

1. Activists welcome the opportunity to voice criticism, e.g. when they have the chance to speak at shareholder meetings or discuss issues with company representatives.

2. In the perception of activists, companies show little interest in interacting with stakeholder groups on social media. The biggest motivation for activist groups to interact with corporate tweets is to provide clarity when, for example, activist group activities are portrayed in a misleading light or to expose potential corporate greenwashing activities.

3. Asked about their views on cooperation with companies, activists expressed their biggest concern as the risk of being misused for greenwashing purposes.

These results indicate that interaction is possible from the activists' point of view, but credibility and trust are crucial factors. In addition, the lack of interest on the corporate side is an obstacle, so factors that support interactivity are important. This is supported by the literature, which also suggests interactivity, credibility and trust as important success factors for online dialogue (Abitbol and Lee, 2017; Capriotti et al., 2021; Zarzycka et al., 2024).

Tweets can include additional resources such as graphic resources (e.g. pictures, emojis), audio-visual resources (e.g. animated GIFs, videos, podcasts) and/or hypertextual resources (e.g. links, hashtags) (Capriotti et al., 2021). Using such additional resources can support interactions with the tweet and indicate an intention for dialogue (Abitbol and Lee, 2017). In addition, messages that include questions or an invitation to act are considered more interactive than messages containing only a statement or announcement (Abitbol and Lee, 2017). Recurring formats such as regular chats or discussion formats on environmental topics (code recurring format) are one way to support the active presence of companies on social media and thus their interactivity (Capriotti et al., 2021). Another prerequisite for more interactive conversations is the responsiveness of companies (code reply in conversation) (Capriotti et al., 2021).

The company-cause-fit of a company's posts can play a significant role in shaping its credibility (Abitbol and Lee, 2017). From the original marketing perspective (see Varadarajan and Menon, 1988) company-cause-fit is defined as the “perceived link between a firm's product line or brand image with a cause it supports” (Abitbol and Lee, 2017, p. 798). From a sustainable development perspective, companies must link their CSR activities to their core business processes as a main priority to commit to actual change. Otherwise, their efforts are merely philanthropic and could be regarded as greenwashing (Carroll, 1991; Schaltegger, 2015). For CSR communication, companies that report on CSR activities related to their core business also appear more trustworthy (Greenwood and van Buren, 2010; Morsing, 2006).

The content of posts should also be related to the “core business” of activists (code activist-cause-fit, see Appendix 7.5). This is especially important when companies pursue the goal of establishing a dialogue and engaging with stakeholders to seek their advice on issues of organisational, public or community concern (Taylor and Kent, 2014). Another factor related to the content that could potentially facilitate dialogue is the disclosure of relevant information concerning the core topics of activists.

Transparency in its dimensions of disclosure, clarity and accuracy supports the perception of trustworthiness (Schnackenberg and Tomlinson, 2016). Disclosure of relevant information, therefore, indicates an openness to dialogue (Abitbol and Lee, 2017; DiStaso and Bortree, 2012). DiStaso and Bortree (2012) found that to improve transparency, companies should not only provide information but also explain (code explanation) what the company does and how the actions might affect others, as well as take accountability for their actions (DiStaso and Bortree, 2012).

The codes derived from the literature are complemented by the tone of tweets, which can also influence the perception of credibility of the message (advertising vs. fact-oriented). To ensure a consistent interpretation of the tone of a tweet, characteristics were agreed upon and documented in code memos when necessary. For example, a tweet with an advertising tone is characterised by “climate protection statements used for self-promotion, i.e. anything in which the company praises itself”. All these identified codes signal the willingness of companies to engage in dialogue without guaranteeing a certain outcome. They form the basis of a more structured analysis part, which then can be widened by applying the CCO view on engagement.

The qualitative data analysis comprised two steps, combining a deductive approach with an inductive one. Integrating deductive and inductive strategies enables evidence that supports, challenges, refines or extends the guiding theory or conceptual model, allowing the theory to be better aligned with a specific sample (Fife and Gossner, 2024). In this case, we draw on existing research into dialogic interaction and social media communication, combining these insights with the CCO perspective. The second inductive step ensures that emerging themes and patterns in this new context are also included in the analysis.

The initial analysis of tweets was thus based on Mayring's approach for qualitative content analysis, which is applied for drawing conclusions about certain aspects of communication (Mayring, 2015). The first tweet of a company in a conversation was determined as the unit of analysis. This includes a category-based text analysis that proceeds systematically, rule-based, as well as theory-based (Mayring, 2015).

For the first analysis, a typifying structuring approach with deductive category application was chosen (Mayring, 2015, pp. 104–106). In this approach, a typology is created by comparing distinguishable dimensions of a category (in this case, dialogic intention and distancing/confirming attitude).

The categories of the initial code-system were derived deductively (see Section 3.2.1), containing the codes for interactivity and trust/credibility as well as the categories distancing and confirming. In the first round of coding, the code system was detailed by adding different dimensions and topics associated with the codes (see Appendix 7.5). The assignment of a code and its different dimensions was performed by two coders, who defined the categories in memos, formulated assignment rules and discussed anchor examples. Appendix 7.6 shows how a tweet was coded as an example. The coding was performed with the software MAXQDA.

The steps of the initial analysis are illustrated in Table 1 using the example of the code “relevance of information”. We started with the main code “disclosure of relevant information” and assigned descriptive subcodes or dimensions derived from the tweets to this code. We then grouped these dimensions under top-level codes (no relevance, low relevance, high relevance of information disclosed) and then assigned these top-level codes to the summarised top-level codes (categories) of high, medium and low dialogic intention.

The last step of analysis involved the assignment of tweets to the typology:

1. distancing-high dialogic intention,

2. distancing-low dialogic intention,

3. confirming-high dialogic intention and

4. confirming-low dialogic intention.

For this purpose, visuals were created with MAXQDA (see Figure 2). The “document portrait” is a visualisation based on the frequency of codes in one document (here, a corporate tweet) and the assigned code colours, in our case green for high dialogic intention, yellow for medium dialogic intention and red for low dialogic intention. The decision for one of the four types was based on each MAXQDA visual, meaning that the category dominating in the respective conversation was selected, and tweets with confirming content were separated from tweets with distancing content.

The second part of the analysis followed the steps of qualitative data analysis according to Miles and Huberman (2008), which consists of reducing the data by structuring it along codes and subcodes, finding patterns by linking the coded data and drawing conclusions.

For this purpose, and in addition to the deductively derived top-level codes, other top-level codes were added inductively during the coding rounds. New codes were identified as positive or negative amplifiers concerning dialogic intention. Examples are the mentioning of credible persons such as scientists, direct speech or reference to important events (e.g. Earth Day), which can support the credibility of the tweet when combined with other dialogic elements. We also identified amplifying factors that might have a negative effect, e.g. when relevant information is not included in the tweet and additional effort is needed to find it, or the use of passive speech rather than speaking directly to the reader. Furthermore, the category “thread characteristics” was introduced for different tweets and answer combinations in conversations (e.g. one tweet- > many answers; one tweet- > answer stakeholder- > answer company, etc.). It was also coded if the content stayed on-topic during the conversation and if the conversation was particularly long (more than three interactions).

This approach was applied to the four groups of tweets identified in the initial analysis, leading to a qualitative validation of the groups by finding common and distinguishing elements for them and refining them into subgroups where possible. This follows the approach of “describe, compare and relate” advanced by Bazeley (2009). Table 2 illustrates the results after the initial round of coding, and Figure 3 illustrates the refinement of stakeholder engagement images after several coding rounds.

To ensure the trustworthiness and validity of the coding, we conducted several rounds of analysis of selected tweets by two different coders, refined the coding system several times and agreed on definitions of the codes.

The analysis of the sample of German DAX companies that posted tweets with frequent hashtags from the climate activist group Fridays for Future Germany has shown that the stakeholder engagement images of these companies can essentially be described by four basic types. The scheme displayed in Table 2 illustrates the four images of Argumentation, Aspiration, Appearance and Attack, which are described in more detail below.

The “Aspiration” image is characterised by tweets with a high degree of dialogic intention, mainly because of a strong company-cause-fit of the topic discussed in the tweet and, to a lesser degree, by including concrete statements or relevant facts as substantial information. Here, companies reaffirm the general goal of combating climate change and the responsibility they bear in doing so.

The tweets are aspirational in the sense that, while most messages have an advertising tone, they all contain detailed information about specific projects, milestones already achieved or plans, which in turn, offer new entry points for discussions (resp. Conversation or dialogue). The facts are in equal parts directly contained in the tweets or provided by incorporating references (links leading to additional details). An anchor example for this group is:

In the sample analysed, the “Aspiration” image is represented by 50 tweets, which can be further divided into the two main subgroups concerned with “Finding common ground” (32 tweets) and “Testing ideas (12 tweets), and a smaller group concerned with “Initiating environmental discourse” (6 tweets).

The tweets on “Finding common ground” present milestones and goals achieved by companies in their commitment to sustainable development, e.g. climate change mitigation on a general, mostly company-wide level. The details are often inserted directly into the tweet, with links and images providing additional information. Recurring topics are science-based targets and climate neutrality. In a large proportion of the tweets in this subgroup, the companies do not further engage with stakeholder answers. In the subgroup “Testing ideas”, the tweets usually include corporate plans or ideas to mitigate CO₂ emissions that are often smaller in scale and more palpable. Here, corresponding details are often not part of the tweets, but hyperlinks to more detailed descriptions of the respective projects on company websites are used.

A subgroup (“Initiating environmental discourse”) of six tweets under the “Aspiration” image stands out for combining high dialogic intention with content that is not closely related to the company's actual business or activities. In these tweets, stakeholders are invited to participate in discussions about politics and also environmental topics such as climate change. This seems to be a recurring format for initiating discussions on various topics of societal relevance. Amplifying elements such as emojis and more informal language are used in the tweets. Five of the tweets lead to longer threads.

With 107 tweets, the “Appearance” image is the largest in our sample and is characterised by tweets with an overall low dialogic intention, while at the same time confirming societal expectations such as to engage in the fight against climate change. The tweets of the “Appearance” image share some of their characteristics with the “Aspiration” image tweets. They usually have an advertising tone and contain information about projects, milestones and environmental facts. The main difference is that no relevant concrete information is provided, and the information that is provided is mostly only weakly related to business activities (weak company-cause-fit). The content of the tweets is therefore only suitable for making the company appear to be committed to societal goals. An anchor example for this group is:

Three subgroups of the “Appearance” image can be identified in the sample: “Environmentally conscious–we're on the same page” (34 tweets), “We're doing something” (55 tweets) and “Engaging others” (18 tweets).

The subgroup “Environmentally conscious–we're on the same page” contains tweets that are characterised by addressing environmental issues without linking the issues in any way to the company's activities. Many tweets in this subgroup provide climate-related information on topics such as the climate impact of milk production, although the respective company's business model is in no way connected to milk production. Another variant is to disseminate information about panel discussions or lectures by, e.g. climate activists or scientists who are not related to the company, acting as a kind of multiplier for the positions of others in the discourse. Another type of tweet involves people from within the company, such as CEOs, talking about sustainable development and climate change, and what companies should do, but on a very abstract and general level.

Closely related to the “Environmentally conscious” subgroup is a set of tweets that provides company-related information on projects, sustainability goals and achievements, and thus was named “We're doing something”. The messages do not contain enough specific information to be aspirational. In addition, the messages of these tweets are often that solutions have been found, leaving no room for further discussion or improvement (see anchor example above).

A third subgroup of tweets is characterised by the presentation of projects in which the company has motivated others, e.g. employees or customers, to participate in climate and environmental protection measures. These could be staff meetings at which climate protection is discussed, or participation in public campaigns, such as competing with other companies and public institutions to see which employees cycle the most kilometres in a given period. While these projects demonstrate goodwill, they do not address the more significant internal problems of reducing corporate carbon emissions. This subgroup is labelled “Engaging others”.

Three tweets under the “Appearance” image stand out by advocating for higher carbon prices for the private sector, which at first glance might seem to contradict the company's economic goals. As the companies concerned do not make any follow-up statements, the overall dialogic intention is low.

Most subgroups have equivalents on the confirming side (see Figure 3), with companies communicating an aspiring image. There are parallels in terms of content between the subgroup “Environmentally conscious” (Appearance image) and the “Finding common ground” (Aspiration image), but they differ primarily in the relevance of the information published. Another pair is formed by the subgroup “We're doing something” (Appearance image) and the subgroup “Testing ideas” (Aspiration image); the main difference here also lies in the relevance of the information and the company-cause-fit. In addition, the subgroup “Engaging others” (Appearance image) contains two tweets that attempt to initiate some kind of discourse on environmental issues, similar to the three tweets in the subgroup “Initiating environmental discourse” (Aspiration image). These “Appearance” image tweets work with incentives such as a lottery for winning a Tesla, which resulted in the longest threads of the whole sample. The main difference is that on the aspirational side, the thread is moderated and thus stays close to the topic of sustainable development, while on the appearance side, standardised responses are given by the company, leading to a fuzzy discussion and thus an overall low dialogic intention.

The 13 tweets under the “Argumentation” image are characterised by a predominantly high dialogical intention. The tweets offer a clarification or defence of a corporate point of view and thus distance themselves from the (assumed) societal expectation associated with the hashtag used. This means that the distancing element is not characterised by comparing the viewpoints of different actors, but by the way the company argues in its tweets. The following anchor example can illustrate this:

The tone of tweets in the sample is mostly fact-oriented or defensive; none of the tweets in the sample has an advertising tone. The relevance of information varies across the group and is not a unifying element; instead, most of the messages relate directly to the business models (company-cause-fit) and more than half of the tweets in the sample include explanations of their own processes. Similar to the high-dialogic-intention-image “Aspiration” on the confirming side of the spectrum, a relatively large proportion of tweets present the relevant information directly in the tweets, but here, on the distancing side, only a few tweets use additional resources such as links to point to further information, graphics or other media forms.

The 38 tweets under the “Attack” image show a low dialogic intention and create a distance towards societal expectations. All tweets under the “Attack” image contain information with low relevance. Moreover, the connection between the information presented in these tweets and the business models of the companies concerned is unclear, resulting in a predominantly weak company-cause-fit. Tweets are attacking in the sense that they make statements in an “end of discussion” manner that is illustrated by a defensive tone or the use of climate-related hashtags to lobby for a company's own goals. This group differs from all others in that a relatively large proportion of these tweets use climate-related hashtags but do not really deal with the core topics or positions of climate activists (no activist-cause-fit). One example of this is a company that warns of the potential dangers of demonstrations in coal mining areas but does not address the concerns raised by such demonstrations. This appropriation of hashtags frequently used by climate activists is an important factor in the constitution of the “Attack” image in the present sample. In addition, compared to all other groups of tweets in the sample, the “Attack” image tweets are also more likely to directly address individuals or groups by mentioning them (@). At the same time, tweets in this group contain fewer “we-messages” than all other tweet groups.

A representative example of this group is:

Two subgroups can be identified in the sample: The “We're the good guys” (18 tweets) and “Lobbying” (20 tweets) subgroups. The “We're the good guys” subgroup mainly comprises tweets that deal with a specific recent battle of meaning between environmentalists and a company; in other words, these tweets are concerned with the contestation of interpretation and authority. While the tweets in this sample are both highly topical and specific, it can be assumed that similar controversies have occurred in the past or will occur in the future. On a more general level, these tweets aim to convince the wider “Twitter society” of specific arguments, while also distancing the companies concerned from climate activist groups that use the same hashtags but attach a completely different meaning to them.

The same applies to the second subgroup – “Lobbying” – but here in a more subtle, less confrontational way. This group of tweets provides information in the form of general explanations or mere statements about what governments (although often not addressed directly) should do to support private sector climate action. A representative example for the “Lobbying” subgroup is:

The information provided in these tweets is often supplied by consulting firms. Credibility is also generated by mentioning the CEO or other high-ranking members of the company and, to a lesser extent, by directly mentioning politicians in a tweet.

Analyses of the positive or negative amplifying factors alone reveal some differences between the images with distancing and confirming attitudes. While some of the tweets of the confirming images are related to (environmental) events, e.g. “Earth Day” or “Climate conference of the parties”, this is very rarely the case in distancing images. Tweets of the distancing images are more often addressed directly to persons or groups, but usually in a formal way.

A certain number of tweets are outliers and do not align neatly with the images. As they are important to shed light on the limitations of the framework, they are described here. Further research could clarify if other groups should be added to the framework.

Tweets without a clear low or high dialogic intention only occur on the confirming side of stakeholder engagement images, with four tweets. This might indicate a certain “fluidity” between the “Appearance” and “Aspiration” images. All these tweets contain information of little or no relevance to a stakeholder engagement process related to climate change and are part of conversations in which the companies did not interact with the answers of stakeholders.

This group contains tweets that, at first glance, contain confirming content about tackling climate change. However, a closer reading reveals that this confirmation is conditional on other factors (e.g. energy security), indicating a reinterpretation and limitation of the scope of the measures to be taken. “Ambivalent in reaction” tweets appear with high (4 tweets) and low dialogic intention (9 tweets), as well as an unclear dialogic intention (2 tweets).

The final sample included all tweets of DAX-listed companies that resulted in an interaction with stakeholders. The most basic form of a conversation is a tweet from a company and an answer from a stakeholder. In this study, all these interactions are labelled as “conversations”, including the original tweet and all its associated threads on Twitter (see Section 3.1).

Under the “Aspiration” image, half the conversations consist of the company's original tweet and one or a few answers from stakeholders, but without an interaction from the company. The other half of these conversations includes threads where the companies at least give one reply, and answers are of medium or even high quality in their fit. Most of those conversations are on-topic, i.e. the conversation remains focused on the same topic throughout.

The vast majority of conversations in the sample under the “Appearance” image are ones that consist of one corporate original tweet and one or more answers from stakeholders, but without an interaction from the company. The share of conversations that are “off-topic” at one point in the conversation is higher under the “Appearance” image than under the other three images.

Conversations under the “Attack” image resembled those under the “Appearance” image in that responses from stakeholders were mostly critical but related to the topic, with no further interactions from the companies. The share of longer conversations is higher than under the “Appearance” image, indicating that controversial topics provoked more answers. When companies engaged in conversations, which happened in 12 out of 38 tweets, the fit of answers was mostly low.

The conversations under the “Argumentation” image have the highest share of conversations in which companies engaged with stakeholders' answers (only slightly higher than under the “Aspiration” image) and the highest share of conversations that stayed on-topic. Overall, the conversations under the “Argumentation” image share some characteristics with the conversations under the “Aspiration” image, with an equal share of longer conversations between companies and stakeholders and basic conversations with only one tweet by the company and one stakeholder response, without further engagement by the company. The re-replies of companies are of medium fit with regard to the questions or comments from the stakeholders.

This study is informed by the understanding that stakeholder engagement aims to co-develop the CSR management of companies, as well as increase legitimacy for corporate actions in connection with environmental and social concerns of society (Kujala et al., 2022; Manetti and Toccafondi, 2012). Under the CCO perspective, CSR co-evolves through communicative processes that unfold in dynamic interactions across networked societies (Schultz et al., 2013) and hashtags function as hypertextual mechanisms that at the same time “coproduce an organizational actor and act as a pastiche of the organization” (Albu and Etter, 2016, p. 5). Following this line of reasoning, we confirm that in addition to the direct interaction with stakeholders on social media, stakeholder engagement materialises also as an organisational image through interaction with hashtags (as proxies for social expectations). By reconstructing images of stakeholder engagement, the study has shown that organisations respond in certain ways to “socially charged” hashtags, so that even in the absence of dialogue in its idealised form, other forms of engagement take place. This provides a basis for further consideration of the performative effects of stakeholder engagement on social media.

The main contributions of this study are as follows:

1. The development and application of an analytical framework for stakeholder engagement on social media based on a CCO perspective;

2. The detailed description of four basic types of stakeholder engagement images (and subgroups thereof) and the first assumptions about their impacts on stakeholder engagement and

3. Providing starting points for further research and practical application of the four basic types, as well as the framework.

These contributions are discussed in detail below.

This study further develops the theory of the CCO perspective on stakeholder engagement in social media, building on theoretical work in which a constitutive perspective is used to explain how stakeholder engagement and issues of legitimacy emerge in a networked society (Schultz et al., 2013). We extend this theoretical knowledge by describing and analysing these processes in practice. This is done by developing and applying an analytical framework for reconstructing images of stakeholder engagement. Based on empirical findings in the literature, we described the term dialogic intention as a factor in shaping the stakeholder engagement image, as well as the confirming or distancing attitude of the image in relation to the hashtag used. The combination of an initial typifying structuring based on these two factors and, in a second step, the qualitative evaluation and refinement of the groups ensured that the framework was validated with regard to our sample. The four basic types derived in this way, together with their respective subgroups, offer new opportunities to examine the performativity of stakeholder engagement to promote CSR management by companies on social media.

Based on the analytical framework we developed, four different basic types of images were identified:

The “Aspiration” image is mainly about what could be. While such forms of communication are often suspected of “greenwashing”, from the CCO's perspective, corporate communication can have a performative and transformative effect even if aspirational talk is not fully translated into action (Christensen et al., 2013). Schoeneborn et al. (2020) distinguish between three variants of formative CSR communication, depending on whether the iterative process begins with a CSR practice (walking-to-talk) or a CSR communication (talking-to-walk). Communication under the “Aspiration” image is particularly similar to the third type of formative communication, in which an analytical separation of practice and communication is not possible, as both take place simultaneously (t(w)alking). This is also reflected in the subgroups “Finding common ground”, in which milestones achieved or smaller projects implemented are often presented, in contrast to the subgroup “Testing ideas”, in which plans are usually discussed. An interesting subgroup, from a communication constituting perspective, is the group “Initiating environmental discourse”. It provides a kind of indirect approach to promote the environmental topic without any reference to the company itself. Nevertheless, the initiation of such discourses could have performative effects, as companies run the risk of being attacked as hypocrites if they are unable to provide information about their actions and ambitions when required in such conversations.

Approximately half the conversations under this image in our sample involved further interactions from the companies, which addressed stakeholder concerns with medium- or high quality. This might indicate a higher level of (inter-)activity, which can potentially lead to an increased performativity of the image.

The “Appearance” image is primarily about “broadcasting” (Etter, 2014). The main statement in tweets under this image is that the company shares the same values (subgroup “Environmental conscious”) and is acting accordingly (subgroup “We're doing something” and “Engaging others”). Even though the term “broadcasting” is used from the perspective of strategic communication, it also has a meaning from the CCO perspective, as the image created seems to be geared towards fixation and leaves little room for interaction, negotiation and change. This impression is mainly created by the firms through the provision of information that is not concrete and/or only loosely related to the business activity, which offers few starting points for discussion. In addition, there is little interaction associated with this image; companies rarely respond and longer conversations do not stay on-topic, e.g. respondents may complain about a company's services even though the original tweet was about climate change. Based on our sample, it can be assumed that this image is less performative in response to social expectations than the other images and has little or no impact on the company, but further research would be needed to confirm this assumption.

The plurality of voices, which are (partly) contradictory and do not always strive for consensus, is a central aspect of the CCO perspective (Schoeneborn and Trittin, 2013). Moreover, dissent and conflict are more important for social change than consensus and agreement (Schultz et al., 2013).

This is reflected and (within the limits of our sample) confirmed with “distancing” images, where companies set the stage for more conflictive interactions. This makes these images potentially interesting from a stakeholder engagement point of view.

The “Argumentation” image contains the smallest number of tweets in the sample and has no subgroups. All of these tweets provide a clarification of the company's position or a defence for the company's actions. Interactivity under the “Argumentation” image is similar to the “Aspiration” image, with slightly more conversations in which the company responds to criticism, mostly with a medium quality and a high share of conversations that stay on-topic. This means for our sample that the “Argumentation” image has the potential for challenging, influencing and affecting the boundaries of the organisation.

The tweets under the “Attack” image contain one-sided conflicts that usually remain without interaction. In the conversations under the “Attack” image, the stakeholders identified as opponents are not invited to negotiate, but rather put in their place. Criticism is generally not responded to by companies in these conversations, and if it is, the responses are usually of low quality. The constructed image seems to be more about winning the debate and providing the “true” interpretation of certain hashtags in the eyes of the wider Twitter society. On the other hand, the tweets are much more concrete and topic-specific than the tweets under the “Aspiration” image, for example. This could be an indication that companies are engaging more deeply with what is expected of them. Further research could investigate whether this could lead to a form of negative aspiration that is performative through defensive actions.

As a summary, we present three main findings in the context of our sample:

1. As demonstrated by the interactions under the “Attack” and the “Argumentation” image, the distancing from societal expectations might be more fruitful for online engagement and legitimisation than the confirmation of societal expectations. This is in line with studies that emphasise dissent and conflict and agonistic deliberation as an alternative approach to stakeholder dialogue and consensus orientation (Castelló and Lopez-Berzosa, 2021; Brand et al., 2020).

2. Furthermore, the struggle for meaning itself in the form of seemingly unconstructive interactions, as seen under the “Attack” image in our sample, might reflect incremental legitimisation processes in situations where rapprochement between companies and stakeholders is not an option.

3. The third finding suggests that not all aspirational talk is formative, and instead, aspirational talk online can be used to avoid stakeholder engagement while trying to gain legitimacy. This is reflected in the image of “Appearance” identified in our sample.

The framework could be used by companies to assess their own communication from a novel perspective. It might be interesting to review whether specific CSR topics influence the occurrence and frequency of certain images, if some images in general occur more often than others, and how images are distributed and developed over a longer period of time or related to certain occasions.

Due to the qualitative nature of the study and its limitation to a specific sample, the developed framework refers to stakeholder engagement between DAX companies and stakeholders with climate change-related interests and cannot be generalised to other contexts or social media platforms without further research. The study operationalises dialogic intention, relying on a definition of “ideal” dialogue that cannot be fully realised in reality. Notably, communication scholars such as Hammond et al. (2003) offer a more nuanced definition of dialogue and its consequences by acknowledging power relations as a characterising factor, and by more explicitly differentiating between the procedural aspects and the content of online dialogue. They also highlight the tension between convergent and emergent dialogue outcomes. This would be worth exploring further when analysing online conversations between organisations and activists.

Research shows that the choice of social media platforms can influence the success of stakeholder engagement (Bilderback, 2026). Further research could adopt a broader conceptualisation and operationalisation of overarching core aspects of platforms, such as interactivity, credibility and trust. This would allow conclusions to be reached that are more platform-independent, moving beyond the specific context of Twitter.

Another limitation is the rapidly changing context of social media research. In our case, this is exacerbated by the change of ownership of Twitter and its transformation into X. Research has shown that the change of ownership created diverse anticipations and, consequently, behaviours, with some users leaving the platform due to concerns about the potential increase in misinformation (Joh et al., 2025). This seems to be particularly the case with environmentally engaged users. Chang et al. (2023) demonstrated this by showing that, following the sale, nearly half of the 380,000 users in their sample became inactive on X, which was much higher than in the control group. In the context of climate change discourse, Joh et al. (2025) revealed a disruption to communication norms, with less pronounced leadership from users, emphasising the urgency of climate change (Joh et al., 2025). This further limits the collection of data for scientific purposes, making it unlikely that a similar, information-rich sample of tweets could be created at a later point in time. Tweets that were deleted during the research process were removed from the sample in the final round of analysis. Although the CCO framework does not aim to strategically use communication to maximise impact and reach (measured on Twitter with metrics such as likes, re-tweets and replies), interactivity is again key to generating and, more importantly, contesting and modifying images. Due to the study design, it is not possible to establish a causal relationship between the image and the resulting interactions, but the occurrence of more interactions with a certain quality could indicate a greater potential for stakeholder engagement and serve as a starting point for further research.

Further research would also be needed to see if these images can be reconstructed in other contexts, e.g. in relation to social issues. Quantitative studies could focus on a more causal relationship between an image and its performativity in terms of stakeholder engagement. Especially the performativity of the “Attack” image might be interesting and empirically not researched enough. In our sample, the “Attack” image is ambivalent in that it contains a lot of interactivities, which are, on the other hand, not constructive from the perspective of stakeholder dialogue.

From an analytical point of view, further research could explore the link between image creation and micro-textual approaches. For example, the concept of addressivity, which was originally proposed by Bakhtin, can be used to analyse how posts engage with readers (e.g. through the use of hashtags) (cf. Liaw and English, 2017; Vladimirou et al., 2021). Meanwhile, Goffman's concept of “footing” focuses on analysing the role of the speaker (e.g. author, animator or principal) and identifying their responsibilities within a discourse (cf. He, 2025).

Longitudinal studies could shed more light on the stability or fluidity of images and how and why they change over time. In addition, the company-specific distribution of images could be an interesting reference point for the study of stakeholder engagement in different industries and concerning other aspects of corporate CSR management.

The supplementary material for this article can be found online.