Opposing brand activism: triggers and strategies of consumers’ antibrand actions

Abstract

Purpose

In brand activism, a brand promotes contested sociopolitical causes to highlight its values. Brand activism also alienates those consumers who disagree with the cause, who might, consequently, target the brand with critical, negative or even aggressive actions. This paper aims to study the triggers and strategies of consumers’ antibrand actions given in response to brand activism.

Design/methodology/approach

Qualitative content analysis and multiple correspondence analysis were used to study consumer responses directed at a chocolate brand’s campaign that advocated civilized online conversions and opposed hate speech, a politically heated topic. In total, 1,615 messages were collected from social media platforms.

Findings

Field infringement, political accusations and questioned impact of the campaign triggered consumers to turn against the campaign. Strategies to undermine it included boycotting, discrediting the brand and trapping. Trapping – creatively using technological affordances to create harm to the brand – was typically triggered by political associations.

Research limitations/implications

Findings relate to the critical responses regarding one campaign only.

Practical implications

By understanding the political discussion around the chosen cause, including the opponents’ typical triggers and strategies, brand activism can more credibly advocate for contested social causes and communicate brand values.

Originality/value

Political antibrand actions are distinct from the previously identified functional and ethical antibrand actions, and they are noninstrumental by nature. Practices that are native to social media are central to political antibrand actions, and social media platforms contribute to how such disappointment is articulated and acted upon.

Citation

Pöyry, E. and Laaksonen, S.-M. (2022), "Opposing brand activism: triggers and strategies of consumers’ antibrand actions", European Journal of Marketing, Vol. 56 No. 13, pp. 261-284. https://doi.org/10.1108/EJM-12-2020-0901

Introduction

Companies have increasingly started taking stances on substantial sociopolitical questions that go beyond their immediate economic interests. In marketing practice, such corporate political advocacy has been translated as brand activism (Hoffmann et al., 2020; Wettstein and Baur, 2016). Frequently used examples of brand activism are Ben and Jerry’s support for marriage equality (Wettstein and Baur, 2016) and Nike’s support for racial justice (Hoffmann et al., 2020). Brand activism differs from traditional corporate social responsibility (CSR) programs and cause-related marketing because it addresses contested sociopolitical causes to communicate brand values. Simultaneously, however, it alienates consumers who do not agree with the stance or do not like the cause (Ciszek and Logan, 2018; Vredenburg et al., 2020). Given its risky nature, the purpose of this paper is to provide insights into the possible negative consumer responses brand activism creates. More precisely, we study what triggers their opposition and what kinds of strategies they use.

While some argue that controversy and debate are the reason why brand activism is effective – the debate draws the attention of the consumers who demand “purpose” from brands and companies (Chatterji and Toffel, 2018; Minár, 2016; Sarkar and Kotler, 2018) – empirical research has not found strong supportive evidence for this claim. Both Jungblut and Johnen (2021) and Mukherjee and Althuizen (2020) found that the negative effects of brand activism (boycotting, deteriorated brand attitude, decreased brand choice) are more prevalent among those who disagree with the promoted cause than the positive effects among those who agree with it. The backlash can also be intense. Hoffmann et al. (2020), for example, found that most Twitter responses to Nike’s Dream Crazy campaign were negative and the most aggressive responses involved references to school shootings, terrorism and slavery. In other words, Nike was met with intense antibrand actions because of the campaign.

Antibrand actions are consumer behaviors that stem from one’s dissatisfaction with a brand, and they can range from sharing negative word-of-mouth (WOM) and organizing boycotts to spreading hatred and threatening the company (Japutra et al., 2018; Johnson et al., 2011; Kähr et al., 2016). However, most existing research on antibrand actions regards situations where there has been a product or service failure or an ethical or moral breach (Palazzo and Basu, 2007; Zeelenberg and Pieters, 2004). These are situations that the company may potentially fix. However, sociopolitical stances cannot be reconciled, which arguably changes the dynamics of antibrand actions. It is therefore important to understand the motivations of consumers who engage in antibrand actions when a brand campaigns for a contested sociopolitical cause, and how they try to undermine such a campaign. Understanding the opposing consumers’ triggers and strategies might also provide more explanations for the somewhat surprising negative effects existing research has found regarding brand activism (Jungblut and Johnen, 2021; Mukherjee and Althuizen, 2020).

To empirically study political antibrand actions, we chose to engage in a revelatory case study and analyzed consumer responses targeted at a marketing campaign of a Finnish chocolate brand. The campaign centered around an online bot that tried to mitigate hate speech on social media. Hate speech, discriminating speech targeting specific groups and minorities, which is particularly present in social media, has been widely identified as a pressing societal problem (Gagliardone et al., 2015; Hardaker and McGlashan, 2016; Laaksonen et al., 2020) but has also become a politicized topic in many western countries (Pohjonen, 2018; Pöyhtäri et al., 2021). Because of the nature of the campaign and cause, we collected data from social media. By doing qualitative content categorization of a filtered data set of 1,615 social media messages targeted to the campaign, this study builds a thematic analysis of the triggers and strategies of the critical consumer responses.

The study contributes to the literature by illustrating how political antibrand actions are distinct from functional and ethical antibrand actions (Kähr et al., 2016) and that they are mostly noninstrumental by nature; The opponents were ready to burn bridges to the brand and they wanted to deidentify with the brand that, after a brand activist campaign, was considered a political opponent. A strong and iconic brand image arguably acted as a catalyst for the negative responses (Vredenburg et al., 2020). The triggers of consumers’ antibrand actions were found to be field infringement, political accusations and questions regarding the impact of the campaign. The opponents used a variety of creative, technology-afforded strategies to oppose brand activism, formulated here as boycotting, discrediting the brand and trapping. We argue that the brand publics serve as a springboard for the opponents to spread their own political ideologies and reach new audiences (Arvidsson and Caliandro, 2016; Johnson et al., 2019). The findings advance our understanding of the consequences of brand activism, which has become both a mainstream marketing tactic and an established dimension of corporate political advocacy.

Theoretical background

Corporate social responsibility, cause-related marketing and brand activism

Firms have been traditionally considered to act in the economic sphere of society, but recent developments have shown a tendency of firms to also act outside their economic activities to produce social or public good. Such extra-economic activities have been discussed from the perspectives of CSR and corporate citizenship. On the one hand, research on CSR discusses the roles and responsibilities corporations have in society and the various public campaigns they engage in to promote their views and legitimacy (Garriga and Melé, 2004). On the other hand, the growing political role of corporations has been discussed in relation to corporate citizenship (Crane et al., 2008; Pies et al., 2014; Scherer et al., 2014) and corporate political advocacy (Hoffmann et al., 2020; Wettstein and Baur, 2016). According to Scherer et al. (2014), corporations become political actors, as they launch CSR projects or otherwise contribute to the production of public goods. In what Matten and Crane (2005) call the limited view of corporate citizenship, corporations are considered to act philanthropically for strategic reasons to gain social and reputational capital to support their economic activities.

Particularly, these instrumental theories link companies’ CSR activities to the marketing and communication functions of a firm (Garriga and Melé, 2004). One way of using CSR in marketing is cause-related marketing, which is characterized by contributing usually monetary support to a designated nonprofit effort or cause (Brønn and Vrioni, 2001; Sheikh and Beise-Zee, 2011). According to this view, cause-related marketing can be used as a “signifier” of CSR and thus it represents the cause-specificity of CSR (Sheikh and Beise-Zee, 2011, p. 27). A key task in cause-related marketing is thus to choose the signifying cause the brand supports and communicates outwards. There are plenty of studies that inspect the company-cause fit. Generally, the better the perceived fit between the company and the cause, as well as the customer and the cause, the better the consumer response is (Barone et al., 2007; Gupta and Pirsch, 2006; Robinson et al., 2012; Yucel-Aybat and Hsieh, 2021). While most consumers react usually positively to CSR activities, an opposite group can also be identified. These people are skeptical about the implementation of supporting the cause or cynical toward the company's motives or both (Webb and Mohr, 1998).

Notably, research usually assumes that brands take on causes that are universally “good” (most charitable causes, for example), and the focus has been on the fit between the cause, contribution type, brand and target customers (Hildebrand et al., 2017; Gupta and Pirsch, 2006; Robinson et al., 2012; Torelli et al., 2012). What has been significantly less often considered is that some consumers might fully oppose the cause a brand chooses to support. This is likely when the cause is politically divisive. With these kinds of causes in mind, Sarkar and Kotler (2018) have introduced a concept called brand activism to refer to brand efforts that promote social, political, economic or environmental reforms which appeal to “progressive” (vs “regressive”) consumers and is an evolution of CSR and cause-related marketing.

As opposed to Sarkar and Kotler’s (2018) thinking, Vredenburg et al. (2020) argue that brand activism is distinct from CSR and cause-related marketing because it emphasizes the inherent company values the addressed issue reveals, rather than the consequences of the actions taken. Vredenburg et al. (2020) also define brand activism through the controversial or contested nature of the issue, but, unlike Sarkar and Kotler (2018), they claim that the issue can be either progressive or conservative in nature. Also, they claim that brand activism contributes toward the issue through both messaging (e.g. advertising) and practice (e.g. organizational practices, policies, donations and partnerships) that are expected to be long term and embedded. The latter factor most often differentiates authentic brand activism from inauthentic (“woke washing”) brands may be active in messaging about their stance, but their corporate practices and values do not align with the message (Mirzaei et al., 2022; Vredenburg et al., 2020).

Research on the effects of brand activism is still scarce. Notably, Mukherjee and Althuizen (2020) found through a series of experimental studies that consumer response to brand activism is not as tremendous as some have suggested (Sarkar and Kotler, 2018). In their studies, brand attitude and choice of those who agreed with the stance (immigration, abortion rights) were not affected, but the brand attitude and choice of those who disagreed deteriorated. A small increase in the brand attitude of the proponents of the stance (freedom of speech) was found when the brand faced a public backlash. It was theorized that the positive effect was because of the tendency to sustain in-group coherency when group values are challenged from the outside (Mukherjee and Althuizen, 2020). Moreover, Jungblut and Johnen (2021) found that when a brand takes a stance on a controversial political issue, the opponents of the issue are more likely to engage in boycotting than are the proponents likely to engage in buycotting. The finding was attributed to the collective nature of boycotting (vs the individual nature of buycotting) and to negativity bias, which makes consumers pay more attention to negative issues than to positive ones.

Antibrand actions

When a consumer is dissatisfied with a company, be it for a poor service experience or displeasing corporate practice, they can engage in antibrand actions, which are critical, negative, hostile or even aggressive behaviors targeted at a brand (Japutra et al., 2018; Johnson et al., 2011; Kähr et al., 2016). The range of such actions and their intensity is wide, from negative WOM or brand avoidance to spreading hatred or threatening a company, and they can take the form of communal behavior, for example by creating antibrand communities (Dessart et al., 2020). According to research, the more meaningful a brand was to a consumer before a dissatisfactory event, the more likely they are to engage in antibrand actions (Japutra et al., 2018; Johnson et al., 2011). In some cases, consumers engage in antibrand actions with the aim of getting the brand to correct its mistakes, restoring equity or reestablishing the relationship. In such a situation, the actions are instrumental by nature (Kähr et al., 2016). Sometimes, however, consumers are motivated by the desire to harm the brand without the wish to reestablish their relationship. Kähr et al. (2016, p. 26) call this kind of behavior consumer brand sabotage, which has the objective of causing harm to a brand by impairing its brand associations.

Perhaps most typically, antibrand actions arise from a functional problem a consumer has experienced and, as a result, negative WOM is shared with others. Motivations can be many: venting feelings, taking revenge, warning others, entertaining, strengthening social bonds, seeking comfort or advice or managing others’ impression of oneself (Wetzer et al., 2007; Zhang et al., 2013). Antibrand actions can also stem from a wider evaluation of the ethics of a company, and the aim might be an ideological and cultural change in the consumer society (Kozinets and Handelman, 2004; Palazzo and Basu, 2007; Thompson et al., 2006). This is often called consumer activism or antibrand activism. Besides negative WOM, a typical tool is boycotting, wherein consumers refrain from buying a specific brand and spread the word to encourage others to do so as well, or its opposite, buycotting (Hawkins, 2010; Sen et al., 2001). Other means of antibrand activists identified in the literature include culture jamming, adbusting, creating doppelgänger brand images, hashtag hijacking, hacking corporate websites and social media activism (Palazzo and Basu, 2007; Romani et al., 2015; Thompson et al., 2006; Veil et al., 2015). These forms of activism usually lend images and discourses from the popular culture and brands themselves to ridicule the brands and counter the consumption-centricity of the western world (Carducci, 2006).

The literature however rarely considers the fact that consumers’ antibrand actions can stem from the political statements of the brand or firm. Based on this notion as well as previous literature, we identify three categories of antibrand actions: functional, ethical and political. The first refers to actions that follow a dissatisfactory customer experience (Wetzer et al., 2007; Zeelenberg and Pieters, 2004; Zhang et al., 2013), the second refers to actions that follow an ethical or moral breach of a company (Kozinets and Handelman, 2004; Palazzo and Basu, 2007) and the third refers to actions that follow a politically unpleasing stance of a company. While the core reason behind functional and ethical dissatisfaction may be corrected, political dissatisfaction cannot be corrected unless the company withdraws its stance. Thus, divisive political statements by companies may spark criticism toward the company by consumers who oppose the stance. In such a case, consumers may engage in political antibrand actions to fight back on the issue and the company’s involvement in a cause that decouples the consumer from the company based on their political or moral values (Mukherjee and Althuizen, 2020). Political antibrand actions have not been systematically studied before, which provides an opportunity to create a research contribution.

Method and data

Case description

The study concerns consumer responses targeted at a marketing campaign of a Finnish candy producer Fazer called Lovebot Blue. Fazer’s most well-known brand is milk chocolate Fazer Blue. For several years, both Fazer and Fazer Blue have been ranked as the most or one of the most reputable brands in Finland (Kauppalehti, 2018). The company has often used “love” as a key marketing theme in its campaigns.

The campaign was organized around a bot called Lovebot Blue that operated on Twitter, YouTube, Instagram and various Finnish discussion forums. The aim of the bot was said to be to find as much hate speech online as possible (Fazer, 2019). Fazer says that for this purpose it has developed “a learning artificial intelligence that searches and interrupts hate speech in public social media discussions” and that there is “a moderator helping Lovebot Blue checking and directing the [bot’s] comments according to each community’s or platform’s rules” (Fazer, 2019). Further details of the ways the bot functions have not been disclosed. The hashtag the campaign used was #smallpieceoflove (translated from Finnish). Hate speech is a concept that lacks a clear definition and no legislation regarding hate speech exists in Finland. Therefore, it is not a policy issue but rather a contested sociopolitical issue that has raised a lot of concern during the past few decades (Laaksonen et al., 2020; Pohjonen, 2018; Pöyhtäri et al., 2021). On its campaign website, Fazer defined hate speech as “words, expressions or pictures that spread or agitate hate towards an individual or a group of people” (Fazer, 2019). The definition was based on the Finnish National Institute for Health and Welfare’s definition of hate speech (THL, 2017).

Below are examples of the kinds of messages the bot posted as a reply when it had identified a message containing hate speech (translated from Finnish):

Hi, you don’t always have to agree with others but you can still choose your words so that you won’t hurt them. Right? #smallpieceoflove

Now stop. I’m not actually involved in this discussion but I have to say that this kind of discussion style goes way over the line. Each and everyone needs to be treated with respect. #smallpieceoflove

The bot never further continued the discussion to which it intervened nor did it respond to questions directed to it. However, social media accounts of the parent company were used to respond to a few claims and questions regarding the campaign or the CSR practices of the company. Apart from a few campaign-related posts on Fazer’s social media sites, the main content of the campaign was delivered via the interventions in people’s social media discussions. At the end of the campaign, Fazer stated on Facebook that the bot had identified over 20,000 messages it categorized as hate speech, and it intervened in over 500 of them (Fazer, 2018).

By choosing a single case, critical consumer responses to a brand activism campaign, we follow the logic of purposive sampling (Mills et al., 2009). The aim is not to verify hypotheses, but rather to provide detailed, closeup perspectives to a case that is unique yet likely to appear again, thus providing a revelatory case previously inaccessible to investigate (Mills et al., 2009; Mason, 2017). Such cases suggest possibilities and help form new questions (Donmoyer, 2000). The campaign is expected to create rich data on consumers’ antibrand actions for three reasons. First, it deals with an issue that is extremely controversial and political in nature; hate speech has been widely discussed and debated in Europe, particularly in connection to the so-called European refugee crisis (Ojala et al. 2019, Pöyhtäri et al., 2021). Second, the campaign strategy builds on a logic of intervention or even intrusion by commenting on ongoing discussions by citizens. Such an approach is unusual for a marketing campaign by a consumer brand and is likely to spark a lot of consumer reactions. Finally, the campaign promotes automation and technology as a solution to a wicked societal problem, which imbues the campaign with an additional aura of technological awe. All these reasons make it a revelatory case to study consumer responses of contemporary, technology-aided brand activism.

Data collection

Our data was collected using a commercial social media data provider Futusome. The company extensively follows Finnish-language content across various social media platforms at the approximate rate of 500,000 new messages per day and is known to provide the most comprehensive access to Finnish social media data. Sources include publicly available textual content from Facebook pages, Twitter, Instagram, YouTube, blogs, discussion forums and news article comments (Pöyry et al. 2018). We queried the Futusome database with two keywords: the username of the bot (@lovebotblue) and the campaign hashtag (#smallpieceoflove) for messages published between the campaign launch day October 22, 2018, and the end day December 31, 2018. Querying for messages that included more common keywords, for example, messages that contained both words “Fazer” and “hate speech” would have significantly increased the data set, but we chose to focus on messages that were sent as a direct response to the campaign. The messages in the sample include direct replies to the bot as well as messages that participate in the discussion about the campaign by using the official hashtag or the name of the bot.

The resulting data set contains 1,615 unique messages posted by 655 unique usernames. Of these messages, the majority were from the platforms on which the bot was active: 1,021 tweets from Twitter, 69 Instagram posts or comments and 367 messages from the Suomi24 forum, which is the largest online forum in Finland. The remaining 158 messages were comments from other online forums. All the messages include a timestamp, the username of the author, message content, message platform and type (e.g. twitter_tweet or twitter_retweet), the original URL and various fields extracted from the message content such as hashtags or links posted.

Data analysis

To chart out the message sentiment, as well as the triggers and strategies of the critical messages, we followed the procedure of qualitative, inductive content analysis (Berelson, 1952; Krippendorff, 2004; Kuckartz, 2014), while also being inspired by frameworks of qualitative iterative coding (Gioia et al., 2012; Tracy and Herberger, 2018). The unit of analysis was one message, which could represent one or more theoretical categories.

Overall, our analysis had a qualitative focus: we were interested to map the accusations the online commenters targeted the brand with (triggers) and the various means they used to undermine the campaign (strategies). These “why” and “how” factors are missing in the studies that concern consumer responses to brand activism (Jungblut and Johnen, 2021; Mukherjee and Althuizen, 2020) and provide important knowledge on how political antibrand actions are played out. Therefore, while we discuss the prevalence of the categories in the data, our main objective is not to map their frequencies but to discuss the types of triggers and strategies of political antibrand actions and to understand their relationship with each other.

Before engaging in the coding, both authors browsed through the data using both an interface provided by the Futusome company and a spreadsheet program after the data was downloaded. During the first stage of the coding, each message was categorized according to its overall sentiment: negative or not negative (positive or neutral). Because the purpose of the study was to study political antibrand actions, positive or neutral messages or messages that did not take a stance on the issue were excluded from further analysis. A message was considered negative if it contained expressions of negative feelings or actions toward the campaign, brand, company or the bot or expressions of criticism, critical requests for more details, direct abuse, disconnecting with the brand or demands to correct its actions (Dessart et al., 2020; Kähr et al., 2016). Following the theoretical premises of consumer antibrand actions, negative brand behavior and consumer brand sabotage, as well as our initial screening of the data, we also labeled messages containing hashtag hijacking or ironic imitation of the campaign as negative. An intercoder reliability test was conducted with a random set of 165 messages (10% of the full data set; messages by the campaign bot or Fazer company were excluded before taking the sample) that were labeled by both authors. We calculated the level of agreement using Krippendorff's alpha (ɑ = 0.708, see Krippendorff, 2004), which exceeds the lowest acceptable limit of α ≥ 0.667.

In the second phase of the analysis, we conducted a qualitative, iterative coding focused on the negative messages only to identify the types of responses. We used a random set of 600 messages to generate the classification scheme together with both authors. This number of messages was chosen because the categories started to reach saturation after 500 messages. During this stage, we held frequent peer briefing sessions with both authors. Then, the first author coded the remaining messages using the developed classification. During this stage, the classes could overlap. If the content of the message was unclear, the message context was consulted using the corresponding URL.

After the classification, the first-order codes were synthesized into second-order top-level categories of field infringement, political accusation, campaign impact, boycotting, discrediting the brand and trapping. Finally, we placed these categories under the two broad themes of triggers and strategies of political antibrand actions (Table 1). To visually explore the relationships between the top-level categories, we used multiple correspondence analysis (Figure 4). Correspondence analysis extracts underlying dimensions of categorical data using a statistical calculus, and the data points are visualized along these dimensions (Clausen, 1998). The analysis was calculated using the R:FactoMineR package (Lê et al., 2008). In the final model (Table 2), we included three dimensions that together account for 64.26% of inertia in the model, which is considered an appropriate goodness-of-fit (Hair et al., 2014).

Finally, as a note on research ethics, social media data is personal data that always includes some identifiable information about the person who wrote the message. During the analysis, all usernames were present, but for reporting the results, we have taken extra precautions not to reveal the identity of the discussants or to cause any potential harm to them, as the content in our data is rather provocative (Markham and Buchanan, 2012). Therefore, we do not display the names of private persons’ social media accounts. Further, all the quotations presented here as examples were translated from Finnish to English, which, to some extent, works as fabrication (Markham, 2012).

Results

The share of the campaign-related messages was notably high right after the launch of the campaign, with almost 70% of all messages sent during the first three weeks (Figure 1). When categorizing the messages, approximately 63% (N = 1,011) of the total amount were critical of the campaign and 37% (N = 604) were not. While there were plenty of different kinds of noncritical messages, they were less diverse in style and content than the critical messages. The positive messages for example retweeted the messages of the campaign or news regarding it and said supportive words (e.g. “Such a good initiative!”). As this study focuses on the critical messages, we next describe the identified triggers and strategies identified in the messages classified as negative and then explore their relationships using multiple correspondence analysis.

Triggers of political antibrand actions

Our first analytical interest was to identify what triggered the consumer to react negatively to the campaign, that is, what they offered as explanations for their negative stance. Based on the analysis, three categories of triggers of political antibrand actions were identified: field infringement, political accusations and the impact of the campaign.

Field infringement. The most common criticism of the campaign related to accusing the company of limiting people’s freedom of speech and being engaged in censorship. This association emerged on the day the campaign was launched, the first negative tweet referring to the problems with freedom of speech in China (Figure 2). Numerous referrals were also made to Russia, the former Soviet Union, Nazism, Fascism and the Orwellian dystopia. The political nature of hate speech and freedom of speech was evident in the material: the campaign was immediately connected to politics. For example, the following tweet was shared multiple times online:

#LovebotBlue project, not only #Fazer’s political statement, but also a statement regarding freedom of speech. Bear this in mind next time you go to a supermarket. #fazerboybott

Often related to the criticism of limiting freedom of speech, the discussants also criticized the company’s interference in issues that were not its core business and that it should stick with candy production instead of taking part in politics. These comments were raised particularly as direct responses to the bot’s interventions. As an example, after having vilified Muslims and being intervened by the bot, a person responded in the following manner: “@LovebotBlue F*** off, you troublemaker chocolate bot. No one asked you anything.”

Doing interventions in people’s social media discussions was also associated with acting as police or as though the company possessed law enforcement authority. For example, one discussant argued that “the bot has taken over power that belongs to the court of law” and another that “acting as hate speech police is not a duty of a candy factory.” The arguments were frequently and often humoristically associated with the Finnish police, which has gained a presence on social media. Some also argued that the campaign appeared as the company was “the moral police” or “the thought police”.

Many critics also described their feelings regarding the campaign as a disappointment, such as in the picture posted on Twitter, depicted in Figure 3. Many also said that they had preferred Fazer’s products before but expressed how, unfortunately, they could not do so anymore. This sentiment related particularly to the “Finnishness” of Fazer Blue; through its connection to political questions, the campaign contradicted the commenters’ patriotic values and the values they thought the brand represented. In sum, the critics seemed to disapprove of the campaign because it had stepped outside its legitimate field.

Political accusations. It is evident in the material that hate speech and freedom of speech are part of a larger, divisive political debate as many critics associated the prevention of hate speech with other political issues. The most common accusation was that the company was supporting too liberal immigration policies or multiculturalism. The following tweet illustrates the association:

I’ll f****** laugh my brains out when @FazerSuomi explains the plunge in the stock market to the investors. ‘For the immigrants needed to be saved and think about the #hatespeech online.’ idiots

In addition to immigration-related questions, the critics associated the campaign with liberal, green and/or leftist parties in general and accused the company of being politically engaged with them. As an example, this accusation is visible in the following tweet, along with references to other political issues such as geopolitics and Russian influence:

I’m increasingly convinced that LovebotBlue’s ‘artificial intelligence’ is based on the fact that with Soros’ money Fazer has hired some red-green trolls in their chocolate. no, in their troll factory.

The critics also claimed that the campaign had materialized because of the “social justice warriors” active in the company and its marketing department. The Finnish version of the term has been widely adopted among the supporters of antiimmigration movements in Finland to refer to the political supporters of the liberal left-wing and green parties and people concerned about the treatment of immigrants and racism.

Some also associated the campaign with nongovernmental organizations that operate with immigration or have supported asylum seekers’ human rights. A widely shared tweet, accompanied with a picture of the campaign mascot and the face of the then Minister of the Interior and the Amnesty International logo, for example, made a parody of the campaign by saying: “@LovebotBlue Hi, I’m Lovebot Amnesty. I hate borders. No one is illegal. Becoming a less developed country is a matter close to our hearts.”

Impact of the campaign. Some critics questioned the campaign’s ability to prevent hate speech or claimed that it increased hate speech by agitating those who were prone to use insensitive language. These messages were not necessarily against the cause itself but were critical of its true impact and consequences. For example, in one discussion thread, someone said: “I’m not sure if people’s rage should be fed with these kinds of bots.”

More typical, however, was to criticize the campaign for how it defines hate speech or to argue that hate speech could not be defined. Many for example replied to the bot’s interventions with a question on how it defined hate speech or claimed that what they said was not hate speech because it was “true.” Concerning these arguments, many complained that emotions were not allowed to be shown anymore or that people get upset from too little, such as the following message:

Nowadays any angry talk can be defined as hate speech. You don’t even need to be angry. Ambiguous word monster, created because people hurt their feelings from hearing the truth. Sensitive people […] #smallpieceoflove #nicespeech #hatespeech #censorship

Strategies of political antibrand actions

The consumers in our data did not only express their disappointment or hostility toward the campaign but also made calls and declarations for action, as discussed in antibrand action and negative brand behaviors literature (Dassart et al., 2020; Romani et al., 2015). Through our analysis, we identified three categories of strategies for political antibrand actions: boycotting, discrediting the brand and trapping.

Boycotting

Boycotts form a traditional way of opposing a company or its actions (Hawkins, 2010). Also, in the studied case, many people suggested that they started boycotting the company because of the campaign. The critics claimed that they refused to buy the company’s products from now on and encouraged others to boycott the company as well. Some also said they hoped or predicted that the company would get into financial trouble because of the campaign. Alternatively, they claimed that they started to favor competitors’ products because of the campaign. For example, the following message was a response to the bot’s intervention:

@LovebotBlue It’s lovely to enjoy Finnish chocolate by Brunberg, a company that doesn’t interfere with politics but produces excellent products for us Finns.

Boycotting also gave rise to typical forms of online campaigning and social media activism: the discussants started to use hashtags “fazer boycott” or “avoid fazer products” to spread the idea of the boycott.

In addition, many associated the company with other companies that have engaged in similar brand activism campaigns. Several messages reflected the idea that any company supporting liberal or progressive values should be boycotted. A person for example tweeted:

I must say that finlayson’s campaign [a textile company supporting sexual minorities] created such a strong counter-reaction in me that I haven’t visited their stores ever since, even though I like their products. The same happened with fazer.

Discrediting the brand

There were many ways the critics tried to bring awareness to the company’s own responsibility problems. The underlying thought was that these problems would discredit the brand so that there was no basis for the campaign. Many accused the company, for example, of problems with sourcing its cocoa, the environmental impact of the company or tax avoidance. In a conversation about the ethicality of the company’s sourcing of cocoa, one person for example stated: “A certification system for preventing the use of child labor exists but Fazer is not committed to using it. You should buy chocolate from someone who is engaged in ethical principles.” Another critic highlighted that the company should take upon other causes than the prevention of hate speech:

I can’t believe a candy factory is involved in something like this. – If you want to give back to society, pay your taxes and give up on your holding companies.

Another strategy to discredit the brand was to remind about the company’s own politically incorrect brands, both current and past. This strategy involved mostly humorous and ironic ways of speaking, for example by reporting “hate pictures” to the bot and by attaching a picture of an old Fazer product that would today be considered insensitive.

Along the same lines, some reminded of the health impact of chocolate and suggested that the company is not eligible to promote a social cause because its products have “such a harmful impact on public health,” as one discussant put it. Some ironically argued that the campaign and the resulting boycott were good because now they might lose some weight or complained that the quality of the company’s products had been deteriorating. The ethicality problems were summed together by some to argue that the company was not entitled to take upon the campaign:

@LovebotBlue Everyone at Fazer, screw you. You exploit developing countries with your shitty chocolate and destroy kids’ teeth. You assholes can’t afford to be ethical.

Finally, many suggested the implementation of the bot was poor in a way that ridiculed the company and belittled any possible benefits it may have. They, for example, suggested that the bot did not work based on artificial intelligence at all or that the way it worked was amateurish. Some also tried to make the company look foolish by suggesting that the entire campaign was merely a way for an advertising agency to profit from a marketing trend.

Trapping

Besides discrediting the brand in general, many tried to make the campaign turn against itself. This strategy resembles setting a trap for the company – in all cases aided by some feature of technology, either platform affordances or by aiming to game the bot algorithm. One of the popular traps was testing the bot by trying to make it behave ridiculously, assuming that it was operating automatically without human control. This was mainly done by writing dirty or politically incorrect words one after another. This strategy was usually related to the notion that the bot was poorly implemented or that it was silly in the first place. Some also explicitly stated this mentality:

Hi #LovebotBlue – I wonder if you could be provoked to become a homophobic Nazi just like the Microsoft bot?

Another way of testing or fooling around with the bot was to report one’s political opponents to the bot, for example when a liberal politician posted something insensitive or something fully normal. This way the opponents signaled disapproval of the politicians but also made fun of the campaign. Many also reported and encouraged others to report the bot to Twitter, or another platform, for spamming, for example. These messages were often retweeted by others to get the message through to the platform officials. Similarly, the bot was reported to the parent company for example by saying:

Hey @FazerSuomi, why is @LovebotBlue breaking community rules, harassing and trying to silence people on Twitter?

Some also reported the campaign to various news outlets for inappropriate pictures that included the campaign hashtag and were thus automatically displayed on the campaign website: “News tip: Fazer won’t take a vulgar photo away from its website for children to see.” Finally, some argued that the campaign had created an unlawful person register and suggested the company had violated the law.

As another trap, there were multiple instances where the critics had created social media accounts that resembled the bot but either posted humorous messages in the bot’s name or merely posted the same messages the bot had posted before but in contexts where the meaning of the text changed. For example, in message threads about crimes, an account pretending to be the bot posted messages saying, “the world would be a better place if we adopted a little gentler attitude towards each other.” Making a parody of the bot was also popular, some highlighted the political motivation of the campaign, such as an account called “LovebotRed,” while others made fun of the campaign, such as an account called “BlueLovebot”:

Now stop! I must come to say that the discussion has gone over the line. Does anyone believe that with this kind of discussion style you can bring about any good? At least you should’ve eaten tasty Fazer Blue at the same time. This kind of toilet humor is not acceptable. #smallpieceoflove

Finally, hashtag hijacking took place as well, particularly in discussions related to immigration and without a clear intention to address the campaign – the hashtag was used in contexts where no one spoke about the campaign or the company as such. The campaign hashtag “small piece of love” was often used together with hashtags “richness” and “dream” when talking about crimes committed by immigrants to ridicule the arguments of those who support more liberal immigration policies. This behavior increased toward the end of the campaign when a series of alleged rapes by asylum seekers occurred in Finland. Consequently, as also shown in Figure 1, the last week of the analysis period was almost fully dominated by messages categorized as negative ones and particularly by those that used the campaign hashtag to refer to the rapes.

Relationships between the triggers and strategies

In the final stage of our analysis, we used correspondence analysis to explore the relationships between the identified triggers and strategies. Figure 4 displays the correspondence analysis and its two main dimensions. Together, these two dimensions explain 47.64% of the model variance. The full model, containing three dimensions, is reported in Table 2. The explorative visualization shows how different triggers and strategies of political antibrand actions are associated. Discrediting the brand and boycotting initiatives are close to one another in the top-left quadrant of the plot, and on the right side of the figure, political accusation and trapping are positioned at a close distance. The visualization can be interpreted so that two dimensions distinguish political consumer antibrand actions: Dimension 1 represents a continuum from the more consumership-oriented responses to the politically inclined ones: political accusations and the creative strategies of trapping are closely related. Dimension 2 depicts a distinction from campaign-related criticism and comments (i.e. discussions on the campaign impacts and doubts if the bot works) to explicit consumer sabotage directed toward the company. Dimension 3 is characterized by accusations of the company exceeding its accepted field of action and comments that question the effectiveness of the chosen approach.

Discussion

The purpose of this study was to identify triggers and strategies of political antibrand actions aimed at a marketing campaign that supported a politically divisive cause and used an algorithmic intervention approach. In the following section, we will first summarize the results, and then discuss the theoretical and practical implications of the study.

Three top-level categories of triggers were identified: field infringement, political accusations and questions regarding the impact of the campaign. The first trigger shows how consumers evaluate the legitimacy of the brand to take upon the cause and regard the adopted role of a political actor as noncompatible with the brand image. The second trigger relates to the political associations of the cause: the company was accused of supporting too liberal immigration policies or multiculturalism. As the word hate speech has been used in political discussions (Gagliardone et al., 2015; Pohjonen, 2018), it is assumed that some degree of disapproval was inevitable, regardless of how the campaign had been implemented. Based on the analysis, people drew inferences on a much larger set of issues than probably expected by the brand. The parent company stated that the campaign was not politically motivated, but the discussants still made multiple political associations. A similar array of associations was also observed in how consumers criticized Nike’s Dream Crazy campaign (Hoffmann et al., 2020). The third trigger relates to the way the campaign was implemented. Some thought the campaign was ineffective in impeding hate speech or that it would only increase it. Many also criticized the way the campaign defined hate speech. These comments, however, were mostly intertwined with the first two categories, as they pinpoint the divisive and politicized concept of the cause.

Cause-related marketing literature has widely discussed the cause-company fit and mainly argues that the two entities should “make sense together” (Robinson et al., 2012, p. 129) and that the brand should be perceived to be sincere in its motives to support the cause (Barone et al., 2007; Koschate-Fischer et al., 2012). Even though research has identified a cynical group of consumers who are usually not persuaded by cause-related marketing (Webb and Mohr, 1998), the triggers identified in the study only partly aligned with concerns about whether the campaign can make a real difference or with cynicism toward the motives of the company. More often, the critics were triggered by the fact that, by making this seemingly political move, the brand was stepping out of its institutionalized and legitimate field of action (Scherer et al., 2014; Suchman, 1995).

Our data also demonstrates how chains of political antibrand actions can be complex: while it could be argued that the proposition to speak kindlier would associate with chocolate and the general marketing theme of love, the use of a politicized term triggered some consumers. Critics did not approve of the cause the company had taken upon because they opposed the values the cause was deemed to represent and had not thought the company would publicly support them. What is particularly stressing here is the generally appreciated and somewhat patriotic reputation of the studied brand: the brand is strongly associated with its country of origin, which is why the often-associated political theme, immigration and multiculturalism, could be argued to contradict the image of the brand. This supports previous literature that has shown that brands are particularly susceptible to antibrand activism if they use emotional branding strategies, are positioned as “iconic” or have had a close relationship with consumers (Japutra et al., 2018; Johnson et al., 2011; Thompson et al., 2006; Vredenburg et al., 2020).

The main strategies of political antibrand actions were identified to be boycotting, discrediting the brand and trapping. Boycotting was exhibited in a rather traditional sense of the concept: consumer power has traditionally been thought to base on their collective purchasing power (Sen et al., 2001), and refusing to buy the products of the opposed brand was one of the key strategies also in the studied case. Declaring a boycott seemed however more like a signifier of disapproval of the company’s political statements-regardless of whether boycotting takes place or not, the statement itself functions as a signal of disappointment. According to Johnson et al. (2019), using boycott-related hashtags in social media may also help connect with like-minded consumers, which show how declarations of boycott serve multiple purposes.

The second strategy was about discrediting the brand, which meant throwing light on the company’s own alleged responsibility problems and the unhealthy nature of its products. These strategies align with previous literature that has discussed the problem of virtue signaling and the increased accountability demands for companies that actively promote their CSR (Palazzo and Basu, 2007). Further, some made remarks about the company’s past and present brands that would today be considered politically incorrect. This strategy related particularly to the fact that the campaign itself raised the question of what can be said and how, which urged the discussants to turn the tables. This is also a sign of increased accountability of companies that engage in brand activism.

The third strategy was perhaps most closely related to the locale of the campaign-social media and the internet. While boycotting and discrediting the brand can be considered rather typical forms of consumer activism (Kozinets and Handelman, 2004; Palazzo and Basu, 2007; Romani et al., 2015), the strategy of trapping was more specifically related to social media. This strategy is a form of antibrand action afforded by social media platforms: not only publishing opinions and aiming to influence others through them but also hijacking hashtags, trying to fool the algorithm, generating hyperlinks to other people and actors and creating fake copycat accounts. These traps aimed to turn the campaign against itself by teasing or tweaking it. Thus, the strategy was to create harm to the company with the means of consumer brand sabotage (Kähr et al., 2016) and culture jamming (Carducci, 2006), albeit most of these means have not been discussed in the marketing literature before.

The correspondence analysis revealed associations between the triggers and strategies of political antibrand actions. It was found that the politically oriented antibrand strategies differentiate from the more traditional means of antibrand actions such as boycotting and discrediting the brand, which resembles negative WOM (Zeelenberg and Pieters, 2004; Zhang et al., 2013). Political accusations were closely associated with the strategy of trapping, which used the possibilities of social media to generate harm to the company. This implies that consumers who are politically active, at least when it comes to freedom of speech and immigration-related issues, are also creative to play out their antibrand actions and ready to invest time and effort to try to harm the brand.

Theoretical implications

Brand activism is an increasingly popular marketing practice (Moorman, 2020), which refers to a brand publicly taking a stance on a contested sociopolitical issue (Mirzaei et al., 2022; Sarkar and Kotler, 2018; Vredenburg et al., 2020). This paper studied consumers’ antibrand actions – consumers’ negative or even hostile behaviors targeted at a brand – that can follow when a brand launches a campaign that promotes a contested cause. Building on previous literature, we identified three types of antibrand actions: functional, ethical and political. The two former types have been widely studied before but the latter type significantly less. While functional and ethical antibrand actions are usually instrumental (consumers might hope that their actions make the firm correct its mistakes or wrongdoings), political antibrand actions in our study were mostly noninstrumental (Kähr et al., 2016); the critics did not ask the brand to withdraw its stance. Instead, they mostly deemed the company as their political opponent that deserves to be ridiculed. The critics also attacked the brand on the pretext of functional and ethical problems, but the primary reason for the allegations was the campaign. Thus, political antibrand actions highlight the intertwining of consumption, brands and politics.

Research has found that brand activism is likely to affect and activate those who oppose the cause the brand promotes, rather than those who agree with it (Jungblut and Johnen, 2021; Mukherjee and Althuizen, 2020). Our research material was in line with these findings – the critical or even hostile comments outnumbered the positive or neutral ones, and the opponents used a variety of strategies to undermine the campaign. It has sometimes been suggested that supporting challenging sociopolitical causes and engaging in CSR activities that go beyond the immediate boundaries of the firm would, through identification, positively affect those who share the same values (Palazzo and Basu, 2007; Sarkar and Kotler, 2018). It however seems that brand activism is particularly noticed among opponents of the cause who, then, want to publicly deidentify with the brand. Based on Vredenburg et al. (2020), we believe that the strong reputation and iconic image of the studied brand as well as the lack of previous similar campaigns contributed to the backlash. Some for example commented that they were disappointed that, of all brands, this brand launched such a campaign and that they were sorry that they now had to boycott it. We argue that, in these cases, consumer–brand identification was unexpectedly challenged, which led consumers to react.

With regards to the reactions, the platform of antibrand actions plays a notable role. As political discussions widely take place on social media (Wright et al., 2015), it is natural that social media is also a principal space where consumers discuss sociopolitical statements of brands – even in the case of campaigns that are not as social media oriented as the one described in this study. Our analysis shows how political discussions spread outside formal political arenas, themes and actors and how they occur in connection with consumer brands. By engaging in a politically contested topic and by discursively framing the campaign in a political manner, a brand can act as a trigger for an online political debate – arguably one strategy for a firm to become a political actor (Scherer et al., 2014). Such activity could be considered one form of meta-level participation in the formation of the public discourse (Pies et al., 2014). On the other hand, consumers also use the brand publics to spread their political ideologies outside their usual networks and discussion arenas (Arvidsson and Caliandro, 2016; Johnson et al., 2019).

We therefore suggest that the practices that are native to social media are central to political antibrand actions. Our analysis shows how consumers, who are interested in politics and oppose a sociopolitical statement of a brand, creatively use the affordances of social media platforms to express their disagreement and to pressure the brand. While engaging in political antibrand actions has existed before social media, social media platforms distinctively contribute to the ways in which such disappointment can be articulated and acted upon. Particularly, the strategy of trapping shows how platform features such as hashtags, flagging and reporting and generating fake accounts are used to oppose the political actorhood of the brand. Some of these resistance strategies borrow from cultural jamming by altering and parodying the original message (Carducci, 2006; Klein, 1999; Harold, 2004; Madden et al., 2018; Thompson et al., 2006). These practices also reflect the varying nature of networked publics and online consumer cultures on different platforms (Arvidsson and Caliandro, 2016; Van Dijck, 2013); the most hostile voices were found on anonymous online forums, where subcultures appropriated the campaign language and hashtags for their own purposes and the campaign gained an afterlife. If queried six months after the end of the campaign, the hashtag “small piece of love” was still frequently used on certain anonymous online forums in immigration-related discussions.

Practical implications

For brand managers, the study offers several implications. First, it is crucial to know that when engaging in brand activism, the hostile voice of politically motivated consumers can be very loud; both in the studied case as well as in others (Hoffmann et al., 2020; Trott, 2020), the critical comments outnumbered the positive or neutral comments. These comments were also more varied in style and content as the critics found several ways to undermine the campaign and the brand. However, the fact that a politically contested topic will create critical comments and hostile reactions should not be the reason to avoid brand activism. By understanding what triggers the opponents and what are their opposing strategies, brands can design campaigns that are less susceptible for consumers’ antibrand actions, for example by avoiding certain trigger words, proactively addressing most typical criticism and ensuring alignment in other communications. Preparedness helps create campaigns that more credibly address the chosen cause because the discussion will focus more on the core questions.

Political associations of the cause form a major trigger to attack the brand. Thus, when taking upon a politically controversial cause, one needs to understand the corresponding political debate and the strategies the critics will most likely use. This can be done by reading news stories and social media discussions on the cause and the surrounding societal debate from a variety of sources – both mainstream and niche, high- and low-esteemed sources. In the studied case, the campaign tried to impede hate speech online and the critics first blamed the brand for limiting freedom of speech and, later, for supporting liberal immigration policies. Because of the latter association, that campaign hashtag was hijacked and used in discussions on alleged crimes committed by asylum seekers. While the lifecycle of the discussions is of course impossible to fully predict, companies should familiarize themselves with the debate they are about to enter and understand the typical arguments and associations.

Nevertheless, a backlash might still occur. According to Mukherjee and Althuizen (2020), a public backlash might be desirable, as it makes the proponents of the cause notice the campaign and start supporting the brand. This might have been the effect that Nike’s Dream Crazy campaign created – despite receiving negative comments on social media, Nike’s sales and the stock price soared following the campaign (Eyada, 2020). As a possible explanation, the campaign and the following backlash created so much media coverage that similar visibility would have been impossible to reach otherwise. However, as opponents can find some very creative ways of opposing brand activism, brands need to make sure the discussion about the cause and the brand is brought to wider arenas and is something the general public can also understand and follow. Thus, supporting brand activism with paid advertising and other communications and creating an integrated brand message becomes critical (Batra and Keller, 2016).

Finally, the measurement of the success of a political brand activism campaign requires a multitude of indicators and measures. If the success is measured only by social media hits or dictionary-based sentiment analysis, the results are likely to be skewed because of the numerous negative comments. As positive comments were found to be more restrained, the measurement of social media likes, shares and other “easy” reactions might better indicate how those who do not oppose the cause respond to the campaign. Accounting for the volume and content of news stories in traditional media is important particularly when the goal is to create a wider societal discussion around the cause and the brand. Moreover, market research on alternative campaign messages or test marketing are also advised to learn in advance how different consumer groups would respond to the campaign.

Limitations and future research agenda

Certain limitations should be kept in mind when interpreting the results of this study. The share of negative social media messages does not equal the share of people who oppose the cause or the brand because of two reasons: First, negative content is more likely shared online than positive content (Koh et al., 2010). Second, as the studied topic is susceptible to trolling behavior, there might have been fewer distinct persons behind the messages than the number of aliases or usernames might suggest. Moreover, when studying online conversations, or any textual content separated from its context, it is difficult to determine the motives and meanings behind a single tweet or post (Krippendorff, 2004). Drawing inferences is particularly difficult in online communication where sarcastic and ironic messages are common (Laaksonen et al., 2020; Gal, 2019). Furthermore, the fact that we chose to limit ourselves to messages that contained the hashtag or the username of the campaign creates another limitation as a great deal of other relevant messages were left out (McKelvey et al., 2014). This way, however, we could ensure that our data related to the campaign was intended for the brand and other people to see.

As the analyzed data concerned consumer responses toward a marketing campaign of a Finnish brand, the results reflect the political climate in Finland at the time of the campaign. Immigration was a hot political topic in Finland during and after the so-called refugee crisis in Europe in 2015 (Ojala et al. 2019), which is suspected to play a role in the volume and nature of the consumer reactions. However, other sociopolitical causes are controversial in other ways, and they have different connections to party politics and to the political agenda at a given time. The paper also portrays a marketing campaign that uses a surprising tactic of technology-afforded intervention, a rare feature in any marketing communications. Future research should therefore inspect other sociopolitical causes, campaign choices, brand types and cultural contexts to produce knowledge on the contextuality of consumers’ antibrand actions.

As brands have increasingly started to take a stance on politically divisive issues, there is a lot to learn about the economic and social impacts of such initiatives. We suggest that future studies should investigate the effects of brand activism over time using a variety of data and methods. Modeling the sales effect as well as the formation of public opinion because of brand activism is particularly interesting and valuable, and researchers could use public polling data on political and social values, marketing investment data and sales data for this purpose. Another approach is to analyze consumer response to brand activism and the following public debate using experimental methods, for example. This requires paying attention to representative sampling of participants – it is expected that consumers who are skeptical about brand activism are more difficult to get to participate in academic research than others (Tellis and Chandrasekaran, 2010). Finally, we call for future research to critically investigate the preparedness of companies to respond to consumers’ antibrand actions and explore possible response strategies.

Figures

Figure 1.

Volume of messages over time

Figure 2.

Meme criticizing the campaign

Figure 3.

Screenshot posted on Twitter of the campaign’s service that allowed customers to design their own Fazer Blue wraps

Figure 4.

MCA visualization of the associations between trigger (T) and strategy (S) categories

Table 1.

Classification scheme: top level categories, their classes and their frequency and share in the critical messages and their description

Top level category	Class	N	(%)	Explanation
Triggers of political consumer hostility
Field infringement	Limiting freedom of speech	96	9.5	Brand limits people’s freedom of speech or tries to censor social media content
	Interfering in others’ business	21	2.1	Brand interferes in things that are not its business or that it should remain in candy production
	Acting as a police	77	7.6	Brand acts like a police or a moral police
	Disappointment	7	0.7	Disappointment in the brand’s engagement in the cause
Political accusations	Immigration or multiculturalism	224	22.2	Brand supports liberal immigration policies, multiculturalism, Islam and/or Judaism
	Political motivation	114	11.3	Brand is politically motivated by liberal, green and/or leftist parties or actors
Impact of the campaign	Inefficient or contradictory effects	17	1.7	Can the campaign actually impede hate speech or does it only agitate more of it
	Definitions	37	3.7	How is hate speech defined or something cannot be hate speech if it is true
Strategies of political consumer hostility
Boycotting	Refusing to buy the company’s products	72	7.1	Will not buy the company’s products anymore and encourages others to do as well or hopes the company will get into financial trouble
	Favoring competitors	20	2.0	Prefers competitors’ products from now on and encourages others to do so as well
	Association with other boycotted companies	10	1.0	Linking the brand to other brands that have engaged in campaigns with similar political values
Discrediting the brand	Company’s own responsibility problems	38	3.8	Blaming the company for their own social responsibility problems and does not afford to promote the cause
	Company’s own politically incorrect brands	21	2.1	Reminding about the company's own brands that can be deemed politically incorrect
	Unhealthy products, bad quality	19	1.9	Reminding that chocolate is bad for one’s health or that the products are of bad quality or overpriced
	Poor implementation of the campaign	24	2.4	Suggesting that the campaign is poorly implemented or the campaign is only a way for an advertising agency to cash in
Trapping	Testing the bot	19	1.9	Listing politically incorrect words or curse words to get the bot to react or to appear foolish
	Reporting one’s opponents to the bot	26	2.6	Reporting one’s political opponent to the bot to signal disapproval of their ideologies
	Reporting the bot to the online platform	17	1.7	Reporting the bot to the online platform in question for violating its rules
	Reporting the bot to the parent company or press	25	2.5	Reporting the bot to the parent company for violating the rules of an online platform or censorship or reporting the bot to the press or to the general public for breaking the law or social norms
	Pretending to be the bot, bot parody	244	24.1	Creating accounts that appear to be the bot and copy-pasting the bot’s comments to unsuitable contexts or creating parody accounts
	Hashtag hijack	145	14.3	Using the campaign hashtag in political discussions to ridicule one's political opponents

Table 2.

Correspondence analysis model details: dimension contribution by variable, variance and cumulative variance

Category	Dimension
	Dim 1	Dim 2	Dim 3
T_Field_	0.590	1.136	3.434
T_Field_X	4.676	9.004	27.204
T_Impact_	0.021	0.247	2.088
T_Impact_X	0.632	7.437	62.755
T_Political_	6.394	0.239	0.205
T_Political_X	39.705	1.484	1.276
S_Boycott_	0.175	2.505	0.006
S_Boycott_X	2.769	39.642	0.091
S_Discret_	0.095	2.008	0.012
S_Discret_X	1.483	31.422	0.187
S_Trap_	6.109	0.685	0.385
S_Trap_X	37.352	4.191	2.356
Variance	0.281	0.193	0.168
% of variance	28.138	19.326	16.791
Cum. % of variance	28.138	47.464	64.255

Notes:

T = triggers; S = strategies