Why so toxic? A framework for exploring customer toxicity

Customers are increasingly empowered in their interactions with firms. Sometimes they help firms but, unfortunately, they can also become “toxic” and hurt them. Customers are toxic when they engage in deliberate and potentially harmful acts towards a firm driven either by a reparatory or damaging mental state following a transgression. Whilst the existing literature has studied customers’ negative actions against organizations, critical questions remain as to how and why customers become toxic. We structure a fragmented field of research on customer toxicity and explore customers’ mental state before they decide to do nothing (non-complainers), avoid the brand, act against firms with either a reparatory mental state—and, thus, often constructive in nature (e.g., to initiate change)—or with a toxic mental state and destructive objectives (e.g., to harm and punish a firm). We highlight that the impact of these actions on a firm can still be “toxic” even without intention of harming and punishing. Furthermore, we outline the conceptual domain of customer toxicity and shift the focus from negative behavior to customers’ mental state, by integrating the marketing, aggression, and psychology literatures. We discuss the theoretical implications of our study and explore how future work may further examine organizations’ interactions with toxic customers. Finally, we provide managerial recovery techniques depending on customers’ mental state at a particular time.

amount of harm you cause the firm and the type of response the firm would need to initiate to deal with it.
Prior work shows that when things go wrong, some customers stay silent (Ro & Mattila, 2015), some avoid the brand, some share their experiences to initiate organizational change, while others decide to voice their concerns to vent, seek compensation, or actively harm the company (Kähr et al., 2016). Publicly sharing negative experiences, irrespective of whether customers aim to help or harm the firm, can represent a real and significant threat to any firm, particularly in a marketplace where the Internet has enabled fast and far-reaching sharing of negative information by customers. It can impact sales and purchase propensity, brand beliefs, and even stock prices (e.g., Monga & John, 2008;Luo, 2007;2009;Berger et al., 2010). Therefore, it is important for academic researchers and managers to understand this phenomenon, the ways in which it may be mitigated, and possible coping strategies.
While existing research has conceptualized different toxic behaviors of customers, such as negative word-ofmouth (NWOM), exaggerated NWOM, brand sabotage, etc. (Romani et al., 2013;Kähr et al., 2016;Rotman et al., Picture yourself after a terrible customer service experience, or an incident where a firm has severely let you down. How would you react? Would you do nothing? Or would you complain directly to the firm, and consider making a post on their social media pages, so as not to let them get away with it? Or would you voice your anger on third-party social media sites, attack the firm publicly, and perhaps even exaggerate your negative experience to harm the firm as much as possible? Or maybe you may just like to vent your frustration on a discussion board, take a deep breath, and simply promise yourself never to buy from that firm again? The type of actions you take, and your mental state, can determine the 2017), several key questions remain unanswered. Specifically, while previous studies have made a strong case for individual customer behaviors and consequences, the literature remains largely fragmented and disjointed. These negative customer behaviors can be arranged on a continuum, ranging from non-complaint behaviors (Voorhees & Brady, 2005), brand avoidance (Bechwati & Morrin, 2003), and different types of constructive and destructive punitive actions (Romani et al., 2013). Needed is an integration of that body of research, to develop a comprehensive framework that focuses on the entire process from the perspective of customers' mental state, ranging from customers who do nothing (non-complainers), to customers who avoid the brand, warn others, or seek revenge (see Khamitov et al., 2020;Lages et al., 2023). Thus, in this paper, we establish that toxicity has numerous outcomes, some of which are intended to be helpful (e.g., constructive feedback or non-complaint), whereas some result in loss of the customer base (e.g., brand avoidance), or harm the brand (e.g., revenge). In this paper, we focus on customers' reactions to quality failure of a product or service and to brand transgressions and integrate the literature on negative behaviors that intend to harm and damage, and negative behaviors that are more constructive in nature, yet can still cause damage to the brand. This will help us understand what drives the extent of aggression customers exert against firms, shed light on when and why customers turn to different negative behaviors, identify pathways for future research, and develop managerial tools to prevent a transgression escalating and causing financial harm (see Fig. 1).
In this research, we employ theory synthesis to summarize and integrate the fragmented literature on the abovedescribed behaviors (Jaakola, 2020). More specifically, we integrate concepts across the fields of marketing, aggression, and psychology (rumination) to help us structure a fragmented body of research on customer toxicity. In doing so, we outline the conceptual domain of customer toxicity, which we define as deliberate and potentially harmful acts of customers following a transgression, driven either by a reparatory mental state (e.g., constructive in nature, such as to initiate change and improvement)-or a damaging mental state (e.g., with destructive objectives, such as to harm and punish the firm). Although the underlying mental state of customers may be different, the effect of these toxic behaviors tends to be the same: they can damage the firm's image, reputation, and associations, and even hurt its financial performance (Grégoire & Fisher, 2008;Monga & John, 2008;Park et al., 2013;Fornell et al., 2016;Jain & Sharma, 2019). We posit that whether the customer does nothing, abandons the firm or turns to toxic behavior-and if so, which toxic behavior-is contingent on rumination. Rumination is defined as a set of conscious thoughts and repetitive evaluation of negative and damaging features of a situation (Martin & Tesser, 1996).
Our study, therefore, makes three key contributions to the literature on negative customer-firm relationships. First, we develop a conceptual framework from the initial transgression through to toxic behavior, integrating the literature on non-action, avoidance, and constructive and destructive behavior, and exploring the consequences of toxic behavior. Second, we shift focus merely from negative behaviors to an investigation of what drives them, i.e., the mental state of customers when they engage in toxic behavior. In so doing, we introduce the construct of customer toxicity. Third, we differentiate this toxic mental state from other related constructs through the lens of rumination, and consequently propose a set of important future research directions.
In addition to making important theoretical contributions, we also identify key managerial implications. Our study examines and highlights the different options available to managers to intervene, and the kind of recovery actions that can mitigate the circumstances. We posit that there are multiple instances, from the initial transgression to toxic customer behavior, when the firm can still engage in recovery actions and provide redress. An understanding of customers' mental state is pivotal from a managerial perspective because a customers' mental state is a key driver behind the nature of their toxic behavior. In the next section, we offer a discussion of the growing literature on toxic customer behavior. We then present our conceptual framework and research propositions. We conclude by discussing the implications for future research and practice.

Customer toxicity
The term "toxicity" originates from the Latin word toxicum-'poison'-and is defined as something that is "extremely harsh, malicious, or harmful" ((Merriam-Webster, 2022a, b). In the medical literature, toxicity has been defined as the "degree to which a substance (a toxin or poison) can harm humans or animals" (Shield, 2022). In the leadership literature, toxic leaders are those that are "exploitative, abusive, destructive … corrupt and poisonous" (Walton, 2007: 20). Similarly, customers can be toxic and harm firms. Table W1 (in Web Appendix) reviews prior literature on toxic customer behaviors. In the leadership literature, leaders have been shown to become toxic to maintain personal status (Stein, 2007). In the context of employee-customer toxicity, employees turn toxic towards customers when customers behave aggressively and place unreasonable pressure on them (Kern & Grandey, 2009). In the context of customer toxicity, the pathway towards toxic behaviors begins when something goes wrong in an interaction between the customer and the firm. In this study we are concerned with customer toxicity triggered by perceptions of a firm's wrongdoing (i.e., a transgression), rather than because of a customer's own negative feeling (e.g., toxicity in the front-line employee context). Also, we focus on customer behavior directed towards a firm, rather than the mistreatment of individual employees.
The spark that ignites a toxic customer relationship is usually a transgression, or negative stimuli originating from the firm, that fall below a customer's expectations and/or violate customers' values. Kähr et al. (2016) define these negative stimuli as performance-and value-based stimulus violations (Weun et al., 2004;Chan & Wan, 2008;Brigden & Häubl, 2020;Mukherjee & Althuizen, 2020). Performance-based stimulus violation denotes a quality failure of a product or service, whereas a value-based stimulus violation arises when the firm's behavior (such as an antisocial or unethical act) is in conflict with the customer's values (Kähr et al., 2016). Following a transgression, customers can engage in multiple behaviors to voice their concerns, facilitate change, vent their feelings, be heard, or hurt the firm. These behaviors can range from non-complaint and  Berkowitz (1989) denotes as an aversive event. These events are transgressions, either quality failures of a product or service (e.g., a damaged baggage at the airport), or brand transgressions where a firm's behavior, for example, conflicts with its customer's values (Kähr et al., 2016). The aversive event gives rise to negative affect, which triggers expressive-motor reactions, such as feelings or thoughts (Berkowitz, 1965).
If the aversive event creates a discrepancy between the customers' current state (e.g., no compensation for the damaged baggage at the airport), and desired end state (e.g., an undamaged baggage), and the customer continues to dwell on the discrepancy, rumination will ensue (Martin & Tesser, 1996). There are different perspectives and definitions of rumination (Nolen-Hoeksema et al., 1997;Martin & Tesser, 1996). We follow Martin & Tesser's (1996) seminal work on rumination, as it concerns discrepancies between current and desired outcomes, including problems or failures. Martin and Tesser (1996: 1) define rumination as "a class of conscious thoughts that revolve around a common instrumental theme and that recur in the absence of immediate environmental demands requiring the thoughts." The literature identifies three styles of rumination. The first one is task-irrelevant; in this case, the individual distracts oneself by thinking about an unrelated aspect to the aversive event (Ciarocco et al., 2010;Denson et al., 2006). Second, the individual can turn inwards, and focus on the self, thereby engaging in self-focused rumination. During self-focused rumination, an individual compares their current state with their personal standard, and if there is a discrepancy, they experience negative affect. This is followed by one's need to discuss the event with others and engage in emotional activities to relieve anger . The third rumination style is provocationfocused rumination. One engages in provocation-focused rumination when one replays the anger-provoking incident and failed goal attainment in one's mind, and this may result in revenge and more aggressive feelings (Bushman et al., 2005;Denson et al., 2006).
Rumination does not only have different styles, but also different intensities. There is generally no agreement on how many recurrent thoughts constitute intense rumination (e.g., Martin & Tesser, 1996). However, research has established that repeating the same negative response, looking for alternative solutions, or renegotiating the desired goal can result in greater rumination intensity (Martin & Tesser, 1996). Rumination style and intensity can trigger different negative behaviors in individuals (Beckman & Kellman, 2004), such as anger relief, and vengeful and aggressive behavior. Drawing on the above arguments, we propose that rumination will drive whether the customer forms a desire for reparation and turns to constructive action, or a desire for brand avoidance to deviant and dysfunctional customer behaviors (Reynolds & Harris, 2006, 2009. Kähr and colleagues (2016) differentiate between hostile and instrumental behaviors, while Grégoire and Fisher (2008) define these as retaliatory behaviors. Table 1 defines the various behaviors further. A common theme emerging from this body of research is that customers can act against firms with either 'constructive' aims, (e.g., to induce change, or warn others), or with 'destructive' aims (e.g., to harm, cause impairment and punish). The literature defines both forms of actions as deliberate customer acts, with an inherent underlying goal to achieve something (either a change, a self-serving action, or harm; Grégoire & Fisher, 2008). The impact of these actions on a firm can be negative even if they were not initiated by a customer with the intention of harming and punishing, but merely to facilitate change. Therefore, we collectively refer to these deliberate and potentially harmful acts of customers as toxic customer behaviors, driven by different aims and objectives. For instance, a company called Robinhood (a financial services company that facilitates commission-free trades of stocks) was recently the target of toxic customer behavior when it revealed that it would impose restrictions on trading of some stocks. This led to an online firestorm and PR crisis on Twitter and TikTok, which resulted in a brand image crisis, and 82.87% of Robinhood's social media mentions turned negative on January 28th, 2021 (Czarnecki, 2021). Even though customers did not necessarily intend to harm the firm when they tweeted or posted on TikTok (some of them merely wanted to initiate change and discuss the events with a constructive and reparatory, rather than destructive and toxic mental state), their actions still harmed the firm and prompted it to reverse its decision.

Aggression theory and the role of rumination
When something goes wrong in an interaction between the customer and the firm, customers can feel frustrated. Because aggression plays a central role in the toxicity framework, it is useful to look at aggression theories from social psychology for guidance, such as the frustration-aggression hypothesis (Berkowitz, 1989). The frustration-aggression hypothesis states that frustration evokes negative affect, after which expressive-motor reactions initiate a flight or fight response, which may lead to aggressive inclinations (intention to fight reaction). Not every aggressive inclination leads to aggressive behavior, and the pathway is moderated by individual and environmental variables (Berkowitz, 1989;Dill & Anderson, 1995: 71) argues that frustrations arise from aversive events that "generate aggressive inclinations only to the extent that they produce negative affect". In the current context, we argue that frustration can arise from

A process model of customer toxicity
Our framework (see Fig. 2) identifies the stages customers go through mentally before deciding on the action to take after a negative incident. The framework establishes the role of rumination in driving whether the customer engages in a fight or flight behavior. The next component distinguishes between these two tendencies and explores customers' mental state before deciding. A central premise of our framework is that customers may exhibit different toxic behaviors towards the firm at different points in time, driven by their underlying desires and mental state. The final component of the framework explores the consequences of these toxic behaviors on the firm and its performance.
To devise our framework, we rely on theory synthesis, which provides a "conceptual integration across multiple theoretical perspectives" (Jaakola, 2020: 22) and can therefore help structure a fragmented field by analyzing it through a specific theoretical lens. Because the topic of toxic customer behavior is fragmented across different literatures, theory synthesis can help us identify and underscore commonalities that build coherence (Jaakola, 2020). Our theory synthesis study thus offers a new view of toxic customer behavior through linking previously unconnected pieces (marketing, aggression, and psychology literatures) in a novel way (through the lens of customers' mental state). We systematically review the literature on toxic customer behaviors as outlined in Table W1, along with the literature on rumination and aggression theory. We also include studies in the marketing and management literatures that look at managerial practices that address customer toxicity, along with consequences of customer toxicity. We refrain from reviewing the literature on other forms of transgression (e.g., product-harm crises, data breaches). This review was planned, conducted, and reported in adherence to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement (Page et al. 2021). Further details can be found in Table W4 and Figure W1.
With these qualifications in mind, we searched 16 leading marketing and management journals and reviewed all issues from 1970 onward (published and online first articles) in the following peer reviewed English journals:  (2) We introduce a new construct of customer toxicity that differentiates between the mental state of customers.
(3) Through integrating the literature on marketing, aggression, and psychology, we identify opportunities where management can intervene and stop a customer turning to toxic behavior.  Table W3 summarizes the motives behind toxic behaviors. However, the literature that identifies underlying emotions and cognitions driving toxic behavior does not explicitly answer the question of how customers decide to turn to either reparation or retaliation following the transgression. To investigate the mental process of customers driving this decision, we turn to rumination. Rumination occurs when an individual repetitively focuses on the negatives and the damaging features of a situation, such as the provocation incident or its impact on the self . Rumination regulates emotions that arise in response to stress (Beckman & Kellman 2004), and different rumination styles can evoke negative behaviors towards the firm. For example, Denson, Pedersen and Miller (2006) found that provocation-focused rumination increases aggressive behavior, while Strizhakova, Tsarenko and Ruth (2012) found that rumination acts as a mediator of anger on negative WOM behavior. Building on the arguments of Beckman and Kellman (2004), Denson, Pedersen and Miller (2006), and Strizhakova, Tsarenko and Ruth (2012), we posit that rumination style will be a key determinant behind the type of behavior the customer engages in:

Fig. 2 Conceptual model and propositions
Proposition 2 An unresolved incident can induce more adverse customer response because of rumination intensity.

Part 2: Desire to fight or flight
Customers do not always act against firms to harm them. Often customers do not react to transgressions at all. They may continue using the brand, or leave the firm and find a substitute product, without initiating a fight with the firm (Ro & Mattila, 2015). And even in cases where customers react to a transgression, they may do so merely because they would like reparation (Joireman et al., 2013), rather than with the explicit aim of damaging the firm (Kähr et al., 2016). In fact, most toxic behaviors are often motivated by more constructive, rather than destructive, retaliatory aims and active problem-solving (Kähr et al., 2016). We now turn to a discussion of the literature on customers' tendency to fight or flight (Day & Landon, 1977;Fornell & Wernerfelt, 1987;Blodgett et al., 1997), which is important to understand customer toxicity.
Research has identified several different types of reactions to transgression, and different customer types: the noncomplainers, the satisfied, and the dissatisfied complainers (Voorhees & Brady, 2005)-in other words, those who fight and those who flight. After a transgression, an individual can either not complain and continue buying from the firm, not complain but abandon the firm, complain and stay with the firm, or complain and exit (Fornell & Wernerfelt, 1987;Blodgett et al., 1997). The literature has identified important moderators driving whether the customer fights the firm, or flights. First, firm-specific factors include the extent to which the firm facilitates customer complaints, which can lead to fight behavior, but consequently can also increase customer loyalty towards the brand (Umashankar et al., 2017). The rapport between the firm and the customer may also govern whether the individual decides to complain: higher rapport decreases complaint intentions and leads to higher postfailure satisfaction (DeWitt & Brady, 2003). Whether the brand is perceived to be sincere or exciting can also impact customers' reaction to brand transgressions: customer-brand relationships suffer when a sincere brand violates a stimulus, but the relationship with an exciting brand can be reinvigorated after a transgression (Aaker et al., 2004).
Second, customer-specific factors, such as a customer's attention to social comparison information may drive flight behaviors. Customers with higher attention to social comparison information experience higher uncertainty in terms of how others would react to their actions, hence may avoid complaint behavior (Kim et al., 2016). Furthermore, when customers' dissatisfaction after a transgression is minor, the Proposition 1 Rumination style influences the type of toxic behavior the customer turns to following an aversive event (transgression).
We posit that even if a customer has formed cognitive and emotive responses following a transgression, firms can still avoid an adverse customer response and inhibit rumination. Management can mitigate the situation before the customer decides to fight. These mitigation strategies are outlined in Fig. 1 and are further discussed in our managerial implications section. Lack of such initiatives can result in detrimental consequences. For example, if customers share their dissatisfaction online, negative emotions may be amplified, leading to more adverse actions (Lopez et al., 2018). If the company cannot resolve the situation, they can face an even angrier customer. Customers can get emotional twice: once following the transgression, and once following the failed recovery. Consequently, customers can respond with more intense negative emotions and damage the customer-firm relationship (Valentini et al., 2020). Any unresolved recovery attempt can prompt more intense emotions, such as rage, which can trigger more adverse customer behavior (Surachartkumtonkun et al., 2015).
We theorize that the driving mechanism behind such adverse customer behavior following a failed recovery lies not only in rumination, but also in its intensity. Existing research has established that repeating the same negative response, looking for alternative solutions, or renegotiating the desired goal results in greater rumination intensity (Martin & Tesser, 1996). Failed recovery or amplified negative emotions can push customers to re-evaluate and ponder on the negative event multiple times, therefore increasing the intensity of their rumination. Research that measured rumination over a period of 25 min to 8 h found that ruminating about a provocation increases the likelihood that even a minor trigger, such as annoyance at a situation, will turn to aggression (Bushman et al., 2005).
Although existing research has not yet established the role of rumination intensity on aggression (Martin & Tesser, 1996), we posit that the more one repeats the same negative responses and provoking incident in their mind, the more intense their rumination will become, triggering a more aggressive reaction. Therefore, we propose that the more intense customers' rumination is, the more likely the customer is to turn to more toxic behavior. Thus, a failed recovery from the company, or amplified negative emotion can trigger more adverse reactions not just because of rumination, but more specifically, because of its intensity. Management should intervene and try to rectify the situation on time, to ensure rumination does not intensify. In accordance with the preceding discussion, we propose that: We now explore these four different pathways considering the role of rumination, starting with the noncomplainers.
Desire to flight Customers can, after the transgression, decide not to complain or engage in inactive or passive problem-solving. As rumination governs customer responses , we posit that rumination and the firm-specific, customer-specific, and eventspecific factors will guide whether the customer decides to flight. These non-complainers often do not reach out to the firm, are less negative than customers who fight, and may even stay with the firm over time, without ever engaging in a 'fight' behavior (Fornell & Wernerfelt, 1987;Blodgett et al., 1997;Voorhees & Brady, 2005). In contrast, some customers may decide to exit and leave without a fight; the literature defines these customers as avoiders; they avoid the firm because of a severe transgression. Such customers are unlikely to forgive and forget, choose to disassociate themselves from the firm altogether, avoid purchasing from the firm, and simply exit the relationship (McCullough et al., 1998;Bechwati & Morrin, 2003;Grégoire & Fisher, 2008;Fetscherin & Sampedro, 2019).
We propose that whether one turns to brand avoidance or non-complaint is contingent on one's rumination style and its intensity. We argue that an individual who turns to a flight response does so as a result of task-irrelevant rumination: the individual distracts oneself by thinking about an unrelated aspect of the adverse event (Ciarocco, Vohs & Baumeister Denson et al., 2006). Distraction can result in the customer forgetting about the event, and thus resulting in non-complaint. Alternatively, distraction can also increase the cognitive accessibility of the transgression and negative thoughts and trigger a more intense reaction (Wegner et al., 1987). We thus argue that the rumination style alone does not explain flight behavior, but one needs to look at the number of recurrent thoughts too (referred to as rumination intensity by Martin & Tesser 1996). Low rumination intensity explains the reason why noncomplainers are usually less negative and do not want to fight. Non-complaint is preceded by less negative emotions and cognitions, thereby making it easier to distract oneself and think about unrelated aspects to the aversive event. Thus, non-complainers may be less inclined to reevaluate the situation and ruminate less about the aversive event. On the other hand, brand avoidance is frequently preceded by a severe transgression (McCullough et al., 1998;Bechwati & Morrin, 2003;Grégoire & Fisher, 2008;Fetscherin & Sampedro, 2019). As a severe transgression elicits more intense negative emotions and cognitions, it becomes harder to think about unrelated aspects to the aversive event (Kähr et al., 2016). As multiple rounds of task-irrelevant rumination can increase customer may decide to flight, whereas one is more prone to fight when the incident induces major dissatisfaction (Richins, 1983). Customers' involvement with the firm also influences their satisfaction levels; for instance, customers who are highly involved and experience a situation that gives rise to dissatisfaction with the core elements of the experience and product, report lower satisfaction levels overall than customers with low involvement (Goodman et al., 1995). Selfesteem may also govern customers' reaction; high self-esteem individuals are more prone to abandon the brand and switch; whereas low self-esteem customers stay with the brand quietly and exhibit a flight tendency (Consiglio & van Osselaer, 2019). Customers with low emotional intelligence react more negatively to a transgression, because they attribute the negative emotions and associations directly to the brand (Ahn et al., 2016). An individual's commitment towards the brand can also have an impact on switching or complaint behavior. When the transgression is serious, affective commitment will induce switching intentions. In contrast, a mild transgression experienced by a committed customer will not trigger a response (Ganesan et al., 2010). Moreover, financial constraint can reduce a customer's desire to fight and share his or her experiences with others (Paley et al., 2019).
Third, event-specific factors, such as a firm's recovery actions, may also drive fight or flight behavior. Customers who receive a dissatisfactory recovery are more negative, spread NWOM, experience purchase regret, and have lower level of repurchase intentions, than noncomplainers or complainers who receive a satisfactory recovery (Voorhees & Brady, 2005). Whether the brand failure is experienced as part of a group or alone also tends to drive customers' post-failure intentions: customers experience more intense negative emotions and complaint intentions if they experience the negative incident in a group, rather than alone (Albrecht et al., 2017). Transgression severity, the switching costs associated with the product or service, and the availability of alternatives as contextual variables can also influence an individual's response to a transgression (Beverland et al., 2010). In addition, monetary compensation can reduce customers' negative emotions; and post-failure intentions (Valentini et al., 2020;Reynolds & Harris, 2009) highlight the importance of psychological obstructionism and personality traits driving the severity of dysfunctional customer behavior (Reynolds & Harris, 2009). To sum up, whether the customer fights or flights is influenced by firm, customer, and event-specific factors. In addition to these, we also posit that rumination plays an important role. For example, the aggression literature has found that rumination can lead to a flight response and abandoning the goal (e.g., Denson et al., 2006), as well as anger-relieving actions and revenge behavior (e.g., Bushman et al., 2005;Denson et al., 2006). initiate change (e.g., by providing feedback and suggestions; Romani et al., 2013;Whiting et al., 2019).
In contrast, customers can harm the firm through destructive punitive actions (Romani et al., 2013), which the literature also refers to as customer revenge behavior (Obeidat et al., 2017;Grégoire et al., 2010;Zourrig et al., 2009), hostile aggression (Kähr et al., 2016) or retaliation (Grégoire & Fisher, 2008). The drive here is usually a desire to communicate negative feelings (to express the self), to harm the firm and damage its image, or to inflict financial hardship on the firm (vengeance). In addition, customers may also share negative reviews because of self-serving functions, i.e., to communicate information about themselves, for example through impression management techniques, emotion regulation, information acquisition, social bonding, and persuasion (Berger, 2014).
The drivers behind individuals' choice of reparation or revenge include a desire to make a difference, the perceived impact of their actions, their past experience with the firm, their levels of public self-consciousness and self-threat, their self-esteem, self-enhancement, and social norms (Sen et al., 2001;Klein et al., 2004;Harris, 2008;Johnson et al., 2011;Dunn & Dahl, 2012;Philp et al., 2018). The perceived level of interest to the self and society, the costs associated with stepping up, and the perception of success can also drive one's propensity to engage in toxic behavior (Sen et al., 2001;Klein et al., 2004;Ahn et al., 2016). In summary, existing studies have presented similar concepts for retaliatory or reparatory customer behavior. The literature has widely explored the question of why one engages in a toxic action against firms, and what kind of toxic behavior one engages in depending on their intent. However, there is scope to explore the mental state and thinking process of customers who engage in reparatory behavior or revenge, to understand what drives the extent of aggression customers exert against firms. More specifically, exploring customers' mental state can shed light on how customers decide to turn to a more (or less) aggressive behavior, and consequently, help develop managerial tools to prevent a transgression escalating and causing financial harm (see Fig. 1).

Part 3: Demand for reparation or retaliation
Demand for reparation If the individual engages in reparatory behavior, the individual usually chooses an active path to problem solving-for example, reaching out to the organization online, on social media, through complaint procedures or legal action (Aquino et al., 2006;Grégoire & Fisher, 2008). In these cases, the individual tries to solve a problem in an active, and not retaliatory manner. We propose that demand for reparation is preceded by self-focused the cognitive accessibility of the transgression and remind customers of their negative emotions and cognitions, the customer is more likely to deliberately choose to reject the brand and leave the firm, rather than to stay with the firm over time (Lee et al., 2009). Therefore, we posit that flight behavior is preceded by task-irrelevant rumination, and while non-complainers do not ruminate about the aversive event intensely, brand avoiders tend to engage in intensive rumination: Proposition 3 The customer decides to turn to flight behavior after task-irrelevant rumination. Low intensity of task-irrelevant rumination is more likely to lead to non-complaint; high intensity of task-irrelevant rumination is more likely to result in avoidance behavior.
As highlighted earlier, the aggression literature has found that rumination can also result in anger-relieving actions and revenge behavior Bushman et al., 2005;Denson et al., 2006). If the customer decides to fight, we need to explore what makes a customer turn to either anger relieving actions to initiate change, or to revenge the firm. We explore this dilemma now.
Desire to flight Prior literature has distinguished between customers who actively complain and leave the firm, and those who complain, but stay with the firm over time (Fornell & Wernerfelt, 1987;Blodgett et al., 1997;Voorhees & Brady, 2005). Customers who have a desire to fight the firm after a transgression can do so driven by a dominant desire for reparation or retaliation (Grégoire & Fisher, 2006, 2008Grégoire et al., 2009;Beverland, Lates, Lindgreen and Chung 2010;Grégoire et al., 2010;Joireman et al., 2013;Kähr et al., 2016;Grégoire et al., 2018). Research has widely studied the question of why customers harm a firm following a transgression (e.g., Sundaram et al., 1998;Hennig-Thurau et al., 2004;Romani et al., 2013). A common theme emerging from the literature is that customers engage in toxic behaviors driven by various objectives. For example, customers can engage in constructive punitive actions (Romani et al., 2013), also referred to as instrumental aggression by Kähr et al. (2016) or desire for reparation and reconciliation by Grégoire and Fisher (2008) and Funches, Markley and Davis (2009). Customers typically engage in these punitive actions to inform other customers (altruism), to obtain an issue resolution (Lovelock & Wright, 2002;Hennig-Thurau et al., 2004), an apology or compensation, to be understood by the organization, or to et al., 2018). Furthermore, firms may adjust their defense and recovery efforts depending on the purchase context. For instance, in case of a utilitarian purchase context, an accommodative recovery response works better than a defensive one, while in a hedonic context, a defensive response from the firm is more beneficial (Johnen). In addition, firms may stress their positive motives (e.g., through apologies and compensation) to work towards more positive recovery efforts (Joireman et al., 2013).
Customer-based factors that determine the success of a recovery strategy include downward social comparison, whereby customers compare their experiences to others with a worse experience, hence viewing their own in a better light (Bonifield & Cole, 2008;Antonetti et al., 2018). A customer's interaction expectation (e.g., the ability to talk with an employee post-recovery) can also determine whether a firm's apology efforts are effective (Min et al., 2020). If the customer participates in the recovery efforts, their post-recovery satisfaction increases (Van Vaerenbergh et al., 2018). Also, customers tend to be satisfied with complaint handling if it is perceived to be just (Orsingher et al., 2010), and if the response to the failure is proportional to its magnitude (Smith & Bolton, 2002).
Event-based factors can influence whether a firm's recovery efforts are successful, or if they result in recovery disconfirmation. For instance, the stability of cause influences the type of firm reaction the customer expects; if the cause is due to a stable cause (constant, not a temporary shortcoming or fault), customers tend to prefer a refund (Folkes, 1984). Recovery efforts also need to be tailored to the market (Borah et al., 2020). The framing of the conflicting event also influences recovery effectiveness: the customer is more prone to be satisfied with the apology or economic restitution if the firm focuses on the events that lead to the failure (task-based framing of conflicts). Customers are less likely to be satisfied with an economic compensation when the focus is on the person at fault, rather than the events (personal-based framing style; Beverland et al., 2010).
If the customer demands reparation and engages in toxic behavior, but the firm's subsequent actions (e.g., apology or compensation) leads to recovery that exceeds expectations, and thus leaves the customer more satisfied than if the transgression had not occurred at all, the customer will experience a recovery paradox. However, it is important to note that customers tend to experience a recovery paradox only when they encounter a single case of failure. If customers are faced with multiple instances of failure from the same company, they will not experience a recovery paradox, even if the firm's recovery efforts are deemed extremely satisfactory by the customer (Maxham & Netemeyer, 2002).
A transgression followed by an unsuccessful recovery leads to double deviation (Joireman et al., 2013;Grégoire rumination: an individual compares their current state with their personal standard or expectations, and if there is a large discrepancy, they experience negative affect. The negative affect, which is a pivotal driver of aggressive behavior (Anderson & Bushman, 2002;Berkowitz, 1993), leads to aggressive priming of the individual. Aggressively primed individuals who ruminate are more likely to react in an aggressive manner (e.g., Vasquez et al., 2013). Self-focused rumination results in one's need to discuss the event with others and engage in emotional activities to relieve anger . Such behaviors can include, for instance, sharing their views and experiences on social media, or trying to initiate change by engaging others into the conversation: Proposition 4 Customers are more likely to form a desire for reparation if they engage in self-focused rumination.
When a customer decides to demand reparation, the consequences may be three-fold: first, the recovery efforts are successful and exceed customers' expectations, potentially leading to a recovery paradox; second, the recovery efforts are successful and reset satisfaction, but do not exceed expectations; or, third, they are unsuccessful, resulting in recovery disconfirmation. A dissatisfied customer experiences a recovery paradox when their perception of the recovery performance is so great that they are more satisfied than if the transgression had not occurred in the first place. This heightened satisfaction state relative to pre-failure levels can increase loyalty and strengthen the social ties between the customer and the firm (Blodgett & Anderson, 2000;Umashankar et al., 2017). Successful recovery may be the result of compensation or apology. For instance, monetary compensation can mitigate the effects of negative emotions, while communicating clearly with the customer and setting clear expectations can induce more positive emotions (You et al., 2020;Orsingher et al., 2010). Successful recovery is also driven by firm-based factors, such as a customer's prior experience with the firm (Tax et al., 1998;Vázquez-Casielles et al., 2007). The recovery strategy of a firm is also important. For example, shifting the focus away from blaming the firm and highlighting customers' contributions can increase customers' self-esteem and consequently their post-recovery satisfaction (You et al., 2020). Frontline employees also play a key role; if frontline employees over-display their positive affect and overdo relational work during the recovery efforts, customers' beliefs about the effectiveness of problem-solving decreases (Marinova et al., 2018). Customers also experience higher post-failure satisfaction when they know that the employee behind the failure was reprimanded by management (Pugh they deem appropriate from the firm), abandon the goal (flight), or alternatively, turn to aggressive behavior (Pedersen et al., 2011). Proposition 5a Rumination style will determine whether the customer turns to reparatory or retaliatory behavior or decides to 'flight.' Demand for retaliation If the customer starts to ruminate over the anger-provoking incident, the individual may turn to vengeful behavior with the explicit intention to harm and retaliate. Customers are also more likely to turn to revenge after a severe transgression or when they feel betrayed by the firm (Joireman et al., 2013). Aggression theory outlines that provocation-focused rumination results in more aggressive behavior (Pedersen et al., 2011). In such cases, we argue that customers form a toxic mental state with the intention to harm and take revenge on the firm. Revenge behavior can be twofold: direct revenge behavior directed at the firm, and indirect revenge behavior, performed behind the firm's back (Grégoire et al., 2018). In practical terms, direct revenge behavior includes customer complaints, stealing and sabotage, whereas customers who share NWOM, exaggerated NWOM or boycott engage in indirect revenge behavior. Drawing on aggression theory and the frustrationaggression hypothesis, when one chooses to fight, aggressive inclination emerges; an intention to be the one who gets back at the firm. Berkowitz (1989) argues that aggressive inclinations are cognitive and affective responses to negative affect, that may lead to exhibiting aggressive behaviors. Accordingly, we put forward the following proposition: Proposition 5b Customers are more likely to become toxic with the explicit intent to harm the firm following provocation-focused rumination.
As previously argued, we denote all deliberate acts by customers directed at firms with constructive or destructive aims as toxic behaviors. Nevertheless, there needs to be a clear distinction in customers' mental state when they choose to engage in constructive or destructive actions. Customers with a reparatory mental state decide to engage in a toxic behavior when they have formed a desire for reparation following self-focused rumination. But what about a customer's mental state that drives a desire for revenge? We theorize that customers' mental state when they wish for retaliation is a toxic mental state. Our definition of customer toxicity comprises four key elements. First, a "state of mind" which is one's mental state at a particular time (Collins, 2022), and a "person's mood and the effect that mood has on a person's thinking and behavior" (Cambridge & Dictionary, 2022). The definition of state of mind implies et al., 2018). Such double deviation, or multiple cases of failure from the same firm, initiate a service recovery disconfirmation (Smith, Bolton & Wanger 1999;Smith & Bolton 2002;Chih et al., 2012) and further levels of rumination. A customer can continue to think obsessively about the discrepancy between their current state and their personal standards, thereby engaging in another round of selffocused rumination . Rumination can initiate further negative behaviors toward the firm (Worthington, 2006;Pronk et al., 2010). Customers may start to think not just about the discrepancy between their current and desired state, but also about the transgression, its causes, and consequences (Nolen-Hoeksema, 1991). Customers may start thinking about the angerprovoking incident, and how it made them feel, thereby triggering provocation-focused rumination. More specifically, provocation-focused rumination directs one's attention to the anger-provoking incident, increases self-reported anger, and can lead to planning aggressive acts of retaliation (Bushman, 2002;Caprara, 1986;Sukhodolsky et al., 2001). Consequently, provocation-focused rumination triggers revenge behavior and more aggressive emotions (Bushman et al., 2005;Denson et al., 2006). Alternatively, customers may, following rumination, decide to abandon their goal. Martin and Tesser (1996) established that rumination can only cease after goal attainment or abandonment. Following an unsuccessful recovery, when customers realize that they cannot attain their goal (e.g., free compensation, a new product, or an apology), they may decide to engage in taskirrelevant rumination and subsequently, abandon said goal. Therefore, following an unsuccessful recovery, the individual will re-evaluate the situation and engage in another round of rumination. We posit that rumination style will prompt customers to decide whether to flight, or to fight. Following self-focused rumination, customers may decide to restore their well-being, and engage in reparatory behavior again, for example by further raising the issue with the firm, or by writing another review to initiate change (Cohen & Areni, 1991;McCollough et al., 2000;Maxham & Netemeyer, 2002;Gaab et al., 2005;Sembada et al., 2016;Kähr et al., 2016;Valentini et al., 2020). Customers may also decide to abandon their goal, and hence engage in a flight behavior following task-irrelevant rumination (Martin & Tesser, 1996). If customers start to dwell on the anger-provoking incident and engage in provocation-focused rumination, more aggressive feelings, and consequently, a desire for retaliation may follow (Bushman et al., 2005;Denson et al., 2006). As rumination can only be stopped by attaining a goal or abandoning the goal (Martin & Tesser, 1996), we posit that the customer will engage in this spiral of rumination until they either attain their goal (i.e., obtain a recovery of the mental process of toxic customers, the firm may be able to inhibit the toxic behavior, by facilitating and helping in equity restoration, relying on well-targeted recovery efforts, or trying to stop rumination. It is important to note, however, that a customer can have a toxic mental state even if the customer does not engage in toxic behavior. Based on these arguments, we posit that: Proposition 5c A customer is more likely to turn to toxic behaviors when, following provocationfocused rumination, the customer forms a desire for retaliation, followed by an aggressive inclination to act on that desire.

Consequences of customer toxicity
Customer toxicity can be very damaging. For example, it can affect the financial performance of a firm, including sales and propensity to buy (Gopinath et al., 2014). It can weaken the stock price, stock returns, and cash flow, and increase stock volatility (Luo, 2007(Luo, , 2009. It can also influence market share (Tang, 2017;Lin & Kalwani, 2018). Customer complaints have been shown to have a larger effect on stock and shareholder value than customer satisfaction (Gruca & Rego, 2005;Luo & Homburg, 2008;Anderson & Mansi, 2009;Fornell et al., 2016). Customer performance (e.g., customer satisfaction, customer-brand attachment, loyalty, and brand associations) can also suffer because of toxic customer behavior. Negative publicity can negatively affect perceptions of a brand (Monga & John, 2008) and dilute brand image (Monga & Hsu, 2018). The "love becomes hate" effect is also a prevalent aftermath of a serious transgression: brand love and strong brand attachment can turn into more intense negative emotions, such as hate, following a transgression (Grégoire & Fisher, 2008;Jain & Sharma, 2019). In other words, the customers who are most strongly connected with a brand are those that are most likely to turn against it violently when they are let down. Brand loyalty and customer satisfaction levels can also be impaired following a transgression. If the firm does not offer adequate recovery, customer loyalty towards a firm can suffer. Dissatisfaction following such incidents can lead to customer exit (Buttle & Burton, 2002).

Contributions to research
The main objectives of this study were to explore and enhance our understanding of customers' reactions to transgressions (and, thus, aid managerial responses to them), that customer toxicity is an internal state at a particular point in time, and the effect that this internal state has on a customer's behavioral state is "toxic", i.e., potentially harmful towards firms. Second, the word "toxicity" implies that a customer's action needs to be destructive or harmful. Toxic customers choose activities that they believe will get back at the firm. After all, a "poison" is "something destructive or harmful," therefore the toxic behavior a customer engages in, at the end of the day, needs to be able to harm the firm in some way ( (Merriam-Webster, 2022a, b). Therefore, customers who avoid the firm, do not complain, or engage in a toxic act driven by a reparatory mental state will not have a toxic mental state, even if the behavior they perform may be toxic. Third, customer toxicity is triggered by rumination arising from a transgression. Drawing on the frustrationaggression hypothesis, the presence of negative emotions following a transgression will not necessarily lead to toxic behaviors (Berkowitz, 1989). These emotions may arise in a poor encounter with a firm, but do not necessarily induce aggressive actions from a customer. We argue that the presence of negative affect is pivotal for toxicity, but one must also ruminate on the anger-provoking incident (provocation-focused rumination) and consequently form an aggressive inclination to become toxic.
Thus, aggressive inclination is a key element of the construct of customer toxicity, which differentiates it from responses such as inaction, avoidance, and a reparatory mental state. The frustration-aggression hypothesis posits that a negative affect is followed by an aggressive desire before aggressive inclination is formed. Intention and desire are different constructs (Malle & Knobe, 2001); an action begins with a desire and is followed by an intention before the action itself is executed. Whereas desire is not accompanied by a commitment to act, intention arises when one is committed to sticking with the choice they made (Bratman, 1978;Mele, 2000). Therefore, even if customers have a desire to harm, they do not have a toxic mental state if the aggressive intention is not present. For instance, one may be too apathetic to engage in toxic behavior or not committed enough to act. Alternatively, one may not have the required time and effort to engage in toxic behavior and intentionally harm the firm (Voorhees, 2006). Prior research has shown that whether the customer perceives to have power over the firm (Gelbrich, 2010) and the firm's responsiveness to harmful messages (Voorhees & Brady, 2005) can govern whether one decides to act on a dominant desire to harm the firm. However, drawing on aggression theory, we posit that it is not important whether one causes harm and engages in toxic behavior for a toxic mental state to be present; it is the presence of the aggressive inclination that is pivotal for customer toxicity (Richardson, 2014). With a clearer sense Third, we differentiate customer toxicity from other related constructs outlined in Table 1. By introducing the concept of customer toxicity, we look beyond one type of harmful act (such as NWOM), and a single emotion or cognition driving that, to an overarching framework that helps understand the mental process customers go through when deciding whether to fight or flight (see Table W2). We distinguish customer toxicity from similar concepts, such as customer incivility, deviant customer behavior or instrumental and hostile aggression (see Table 1 for further elaboration). We have shown that customer toxicity is a broader concept that (1) integrates the disjointed literature on different customer behaviors as a response to a transgression in a public manner, (2) combines the literature on brand avoidance, non-action, reparation and retaliation, and (3) introduces the mental processing of customers through the lens of rumination. Discussion of future research questions is outlined in the 'Limitation and Future Research Directions' section and Table 2 below.

Managerial implications
Our study has several managerial implications. First, we highlight how firms may employ different preemptive and defensive strategies from the initial transgression through to toxic behavior. We show the importance of understanding customers' mental state at a particular time and identify the appropriate recovery effort to employ. For instance, management may offer a compensation or apology after the incident, or even once the customer has complained and engaged in reparatory behavior. Monetary compensation can mitigate the effects of negative emotions, and communicating clearly with the customer and setting clear expectations can induce more positive emotions (Schoefer & Diamantopoulos, 2008;Valentini et al., 2020). Besides reactive strategies, firms can also use preemptive approaches to minimize the potential incidence of toxicity, even in advance of an actual transgression. This may be achieved for example through the blemishing effect (Ein-Gar et al., 2011), expectation management (Kopalle & Lehmann, 2001;Diehl & Poymor, 2010), inoculation (Compton & Pfau, 2009;Ivanov & Parker, 2011;Mikolon et al., 2015) and customer immunization (Merlo, Esingerich & Hoyer, 2023). The blemishing effect, for instance, involves providing weak negative information about the product, firm, or service encounter prior to a transgression (Ein-Gar et al., 2011). Similarly, inoculation theory states that by confronting customers with a weaker form of negative argument, customers' cognitive defense strategy is stimulated. This way, customers can develop a defense mechanism against future attacks, such as a service failure (Mikolon et al., 2015), which could to understand customers' mental state when engaging in a vengeful action (and in doing so introduce the construct of customer toxicity), and to differentiate this toxic mental state from other related constructs, such as avoidance, constructive punitive actions, and non-actions. In line with these objectives, we have developed a conceptual framework of customer toxicity, and we contribute to the marketing literature at least in three key ways.
Firstly, we bring attention to processes through which customers can harm companies (Dietz et al., 2004;Liao & Chuang, 2004). The existing literature has conceptualized the different negative behaviors, their antecedents and consequences, and the underlying emotions and cognitions in a disjointed, and surprisingly unconnected manner. We have offered a construct that brings clarity and integration to the existing body of work. In doing so, we have argued that whilst the aftermath of customers 'toxic' behavior may be the same, differences can lie in the role of rumination and customers' mental state. We have introduced the construct of customer toxicity, which denotes a deliberate and potentially harmful act of customers following a transgression, driven either by a reparatory or damaging mental state towards a firm. This new construct allows academics and practitioners to differentiate between customers' mental states before engaging in toxic actions and appreciate the difference between a reparatory mental state driven by constructive aims, and a toxic mental state, driven by retaliatory aims. With the rise of social media, and the emergence of fast-growing platforms, such as TikTok, understanding the mental state of customers will become increasingly important. Introducing the concept of customer toxicity highlights not only the need for a better and less fragmented understanding of customers mental state, but also invites future research to ascertain the role of rumination further (see Table 2).
Second, we posit that customers' intention to harm a firm is driven by rumination style and rumination intensity. Our model sheds light on customers' mental state before engaging in such behaviors, hence providing a novel insight to the marketing literature, and a fruitful area for future research (see Table 2 for future research directions). By highlighting this mental state, our framework provides a better understanding of when and why customers choose to exert toxic actions towards firms. We posit that customers stepping up against firms and harming them can be a result of taskirrelevant, self-focused or provocation-focused rumination that will drive the extent of aggression customers exert. By acknowledging that customers turn to destructive punitive actions because of their rumination style, we present an overarching framework that can help firms understand, and address toxic behavior in the future. when the service is slow or inefficient (e.g., waiters may provide an extra free appetizer, or an iPad for the kids). A toxic customer may decide not to engage in toxic behavior after all in case of successful recovery (Grégoire & Fisher, 2008;Funches et al., 2009).
Second, Proposition 1 suggests that rumination will influence the type of toxic behavior the customer turns to following a transgression. The literature has argued that a value-based violation (i.e., a transgression that goes against the moral stance of the customer) leads to more aggressive customer action and more harmful repercussions than those violations that are merely service-based, such as a service failure (Kähr et al., 2016). Looking at this finding through inhibit rumination. Expectation management can also act as a preemptive strategy; lowering customers' service-related expectations can lead to higher customer satisfaction postfailure (Diehl & Poynor, 2010;Kopalle & Lehmann, 2001). Once an individual has formed an intention to harm a firm, the individual can also engage in toxic behavior. Opportunity may be impeded by distractions, time constraints, and complexity and amount of information (Hoyer & MacInnis, 2010). For instance, the customer may be too distracted to engage in toxic behavior or is not committed enough to the cause itself to engage in action (Berkowitz, 1989). For instance, frontline employees at the Ritz Carlton often use a simple practice of "resetting the clock" to distract customers

Propositions
Suggested new research directions Proposition 1: Rumination style influences the type of toxic behavior the customer turns to following an aversive event (transgression).
• Does a performance-based stimulus violation evoke the same intensity of rumination as a value-based stimulus violation (and vice versa)?
Proposition 2: An unresolved incident can induce more adverse customer response because of rumination intensity.
• Which emotions and cognitions evoke a more (vs. less) intense rumination/self vs. provocation-focused rumination? • Do some emotions and cognitions evoke a more intense rumination than others? • At what point does an emotion and cognition turn to rumination? • How many consecutive rounds of rumination constitute an intense rumination? Proposition 3: The customer decides to turn to flight behavior after task-irrelevant rumination. Low intensity of taskirrelevant rumination is more likely to lead to non-complaint; high intensity of task-irrelevant rumination is more likely to result in avoidance behavior.
• Which firm-, customer-, and event-specific factors trigger higher (vs. lower) rumination intensity? • Can the firm reduce customer' rumination intensity, and thereby turn a potential brand avoider to a non-complainer?
Proposition 4: Customers are more likely to form a desire for reparation if they engage in self-focused rumination.
• Can the company stop self-focused rumination? If so, how? • How intensely does one need to ruminate to turn to reparatory behavior?
• Is a customer more inclined to turn to self (vs. provocationfocused rumination)? Is self (vs. provocation-focused) rumination linked to customer personality traits and characteristics? Proposition 5a: Rumination style will determine whether the customer turns to reparatory or retaliatory behavior or decides to 'flight'.
• How many times does the customer try to recover the situation through self-focused rumination and reparation? • When does the customer turn to provocation-focused rumination from self-focused rumination? • How can management encourage customers to abandon their goal and flight following rounds of rumination? Proposition 5b: Customers are more likely to become toxic with the explicit intent to harm the firm following provocation-focused rumination.
• How many rounds of self-focused rumination result in provocation-focused rumination? • When drives the customer to engage in provocation-focused rumination immediately after transgression? • What drives the intensity of rumination in governing whether one turns to retaliation or reparation? • Does more intense rumination push one towards a provocation-focused rumination faster? Proposition 5c: A customer is more likely to turn to toxic behaviors when, following provocation-focused rumination, the customer forms a desire for retaliation, followed by an aggressive inclination to act on that desire.
• How many rumination rounds are necessary for the formation of retaliatory desire? • Can the company inhibit the formation of a desire for retaliation, after the customer has started ruminating? Table 2 Future research directions effective depending on customers' mental state. For instance, management should not disregard customers who flight and engage in task-irrelevant rumination. Even if customers are trying to distract themselves from the transgression, multiple rounds of rumination can push non-complainers into brand avoiders. Understanding how many rounds of rumination the customer has engaged in can ensure management acts on time to retain their customers. Management could consider reaching out to those customers who have not been in touch after a severe transgression, provide an explanation and apologize. By turning attention to the customers who stay silent, management can ensure that non-complainers do not turn into brand avoiders over time.
If the customer has a toxic mental state, then an apology or compensation may not always be effective. For example, when United Airlines violently removed a passenger from their airplane due to overbooking, an apology and offering to reimburse every passenger's ticket on the flight was not enough to contain the online firestorm the incident generated. The recovery efforts should have been more substantial, considering customers' mental state at the time. Airlines such as United Airlines receive daily negative tweets from unsatisfied customers, to which they usually respond with a pro-forma tweet encouraging customers to contact the airline directly to solve the issue. Management should utilize online sentiment analysis tools to detect negative sentiment across platforms and respond to customers quickly and adequately. If the customer is complaining about the discrepancy between their current state and personal standard, they may be engaging in self-focused rumination and have a reparatory mental state. If the customer is reiterating the anger-provoking incident multiple times, they are likely to be engaging in provocation-focused rumination and have a toxic mental state.
Consequently, and in line with Propositions 4 and 6, it is important for firms to carefully assess customers' mental state. For example, knowing that a customer is driven by a need for reparation, managers can ensure that the problem is adequately resolved in a timely fashion, and try to exceed expectations to trigger a service recovery paradox (Maxham & Netemeyer, 2002). When customers share their opinion online to initiate change, signaling to customers that management takes their concern seriously, and is working towards a change, is important. Similarly, by understanding that in such cases, the customer is engaged in selffocused rumination, management can recover the situation by reducing the discrepancy between the customer's current state and personal standard. When the customer turns toxic, remedial actions also need to be adjusted. At this point, the customer is ruminating over the anger-provoking incident, so management should emphasize how it is addressing the the lens of Proposition 1-2, customers may respond more aggressively to value-based violations because of more intense rumination. Managers should therefore understand the role of rumination intensity and adjust recovery efforts accordingly. For instance, if we take the example of Starbucks, the firm experiences service-based failures frequently (such as a bad service experience). The company may address these basic complaints by replying to a tweet or responding directly to the negative review on different platforms. However, when Starbucks banned its employees from wearing attire advocating political, religious, and personal causes, it found itself in the middle of a firestorm and boycott. It was not enough to merely apologize to customers who spoke up against the firm. Starbucks had to issue a public apology, donate to an organization advocating the cause at hand, and issue T-shirts to employees with graphics that advocate the movement that prompted its ban on the attire in the first place (Aratani, 2020). Therefore, management should respond to a value-based violation differently than to a performance-based transgression and ascertain customers rumination intensity in the first place. When it comes to value-based violations, customers may be ruminating more intensely, so recovery efforts that reduce rumination intensity need to be implemented. We propose that rumination intensity can be reduced by offering multiple recovery options to customers. Like Starbucks' response, we propose management can offer multiple recovery options to reduce rumination intensity (e.g., provide an apology, along with compensation, and communicate the specific ways in which the firm will address the incident). This way, management can forego another round of rumination that would be triggered by an inappropriate recovery.
Third, Proposition 2 suggests that not resolving a customer incident can induce more adverse customer responses because of rumination. Companies can use social media tracking tools to identify unresolved incidents as early as possible. A study by Herhausen, Ludwig, Grewal, Wulf and Schoegel (2019) found that if management reacts to online toxic behavior timely, they may prevent it from escalating. If a firm responds to online tweets and comments before it gains adequate traction by fellow customers, it can offer an apology and/or compensation on time and avoid serious repercussions. To put this into perspective, Herhausen and colleagues (2019) showed that if a firm reacts to a negative review or comment after it has reached 6,600 likes, it is unable to contain the 'online fire'; however, if the firm manages to respond after the initial post only received 7 likes, this 'online fire' is put out quickly, before it has a chance to escalate. Fourth, Propositions 1, 3 and 5 suggests that rumination style will determine the type of behavior customers engage in. Therefore, there is no silver bullet when it comes to dealing with customer toxicity: different solutions are a goal, customers may start ruminating more intensely. This suggests that developing effective channels for customer participation and engagement may help minimize toxic behavior (Eisingerich et al., 2014).
Rumination may stop when there is goal abandonment. However, as people ruminate about goals they perceive to be central to their well-being (e.g., a compensation from a firm, an apology, or a satisfactory service resolution), giving up said goal can result in a sense of losing one's identity (Wicklund & Gollwitzer, 1982) and may evoke negative affect, such as frustration or even aggression (Klinger, 1975). Koole and colleagues (1999) found that self-affirmation through substitution can also stop rumination. By affirming an important aspect of the self, one can reduce their failure-related cognitions. Companies often educate their sales-team to use positive scripting and affirm customers with sentences such as "I realize that this situation is difficult, but rest-assured we will find a solution for you" or "I would feel the same way if this happened to me. We will sort this out." By providing affirmation to customers, management can halt rumination.
If management cannot ensure that the customer attains (or abandons) their goal after the initial round of complaints, they could also try and "distract" customers. Martin and Tesser (1996) argued that distraction can relieve individuals from ruminative thoughts. However, the effect is only temporary, as the transgression itself is not addressed. Nevertheless, a temporary distraction by management could reduce the intensity of rumination (Martin & Tesser, 1996). That is, temporarily blocking customers' negative thoughts may provide a distraction, and help reduce the intensity of rumination, thereby minimizing the likelihood of a customer turning to toxic behavior. For example, restaurants often provide customers with free drinks or appetizers, and luxury brands often provide customers with a free glass of champagne or coffee while waiting. The customer service community clearly believes in the distraction technique. By directing customers attention to something physical and concrete, such as a computer screen, a brochure or file, customers' anger can be reduced (Bacal, 2018). Figure 1 summarizes the managerial implications through the process model.

Limitations and future research directions
Our conceptual framework is, by necessity, a simplification of a set of complex phenomena. At the same time, these complexities represent interesting avenues for future research. These avenues concern for example the process of rumination, the drivers behind a tendency to fight or flight, and the formation of aggressive inclination. First, the tendency to engage in root cause of the problem to reframe the anger-provoking incident in customers' minds.
Fifth, this study highlights the importance of the rumination process that drives customers' desire for reparation or retaliation. As highlighted in the propositions, understanding customers' mental state behind their actions can help managers respond appropriately (Silva et al., 2017;Cohen & Areni, 1991). More specifically, it is important to understand that a failed recovery effort towards a customer who demands reparation may not always and immediately result in toxic behavior driven by a demand for retaliation; in some instances, customers decide to re-evaluate the situation, and try a different form of active problem-solving behavior (Cohen & Areni, 1991). For instance, after a transgression, the customer may turn to complaint behavior with a desire for reparation; nevertheless, if the complaint behavior does not bring about the desired outcome, another round of self-focused rumination may prompt the customer to raise the issue through a more public medium, such as Twitter or a review site. To put this into perspective, many reviews published by customers indicate that they have tried raising the same issue with the firm through their live chat or customer support already, without any luck, hence they turn to a more public medium. However, management needs to make sure they can grasp whether the customer is engaging in their first round of rumination or has started to ruminate more intensely. For example, management should look out for tweets or reviews that mention that the customer has raised the issue multiple times with the company. The customers who have engaged in more rounds of rumination are more likely to develop a toxic mental state. Following up with these customers through online review sites and offering a remedy directly can make sure the customer does not develop a toxic mental state and wish to take revenge on the brand.
Last, the question of how to stop ruminative thinking has been explored in the psychology literature (Beckmann, 1994;Pyszczynski & Greenberg, 1987;Wegner, 1994). Rumination can cease through goal attainment, goal abandonment, and distraction (Martin & Tesser, 1996). Goal attainment may be the most effective way of stopping rumination, yet in the context of transgression, a goal attainment is often outside of the customer's control, and in the hands of the firm at fault. Firms can increase the likelihood of goal attainment if they provide multiple ways for customers to complain. For example, having a chatbot support, a live chat option or a clearly signposted customer care number can ensure customers raise the issue with the firm directly, and hence ruminate less. Many companies still make the mistake of not disclosing an email address or phone number for the public to use for complaints. By making it harder to find the most appropriate way to contact the firm and attain

Declarations
Conflict of interest The authors declare that they have no conflict of interest.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons. org/licenses/by/4.0/. a fight or flight response is a key factor in our model. However, further research could examine in more detail other relevant drivers behind the tendency to develop a fight or flight response and observe the boundary conditions and contingency factors behind these. For example, more work is needed to assess the underlying mechanism behind rumination; more specifically, how many failed rumination rounds trigger provocation-focused rumination following a failed recovery. Furthermore, the literature distinguishes between rumination style and rumination intensity, but we invite future research to look at both style and intensity driving different forms of toxic customer behavior. We hope that our framework can also inform future research on leadership and management style and encourage research to explore how management can structure organizations to address customers at the rumination stage before a fight tendency is developed.
Second, our process model shows that there are multiple pathways towards a toxic mental state. An individual may first want to initiate active constructive actions and engage in a toxic act with the underlying desire for reparation. However, one may still end up engaging in a toxic behavior with the underlying intention to harm and revenge if they start to ponder on the anger-provoking incident. Therefore, the process of rumination is a key construct of the process model that needs to be studied further. For instance, the literature does acknowledge the presence of rumination as a key driver behind the reappraisal process, but does not directly address its importance in driving toxic behaviors (Kähr et al., 2016). Hopefully, future research will be directed at examining the mechanism behind rumination and observing rumination intensity and style in driving toxic behavior.
Finally, a third interesting line of work pertains to combining the literature on aggression theory with the marketing literature. This is a particularly relevant issue for the process model and the understanding of customer mental state before engaging in toxic behavior. Past research has suggested that an individual first forms a desire to fight, followed by an aggressive inclination (Berkowitz, 1989). Nevertheless, Berkowitz (1989) claims that an aggressive inclination does not necessarily result in aggressive action. Just because a customer has developed a need for revenge, it does not automatically translate into toxic and aggressive behavior. Current research, however, does not explore this distinction in sufficient detail, and does not establish the factors that may influence whether one engages in toxic behavior, or just forms an aggressive inclination and does not engage in any further act. Clearly, this is an area that merits further attention. Table 2 outlines our proposed future research pathways in more depth.