Mind your tone: how swear words in online complaints influence bystander perceptions of airlines and hotels

Social media platforms, such as Facebook, have become central channels for consumer complaints and brand responses (Adil et al., 2022; Herhausen et al., 2023; Kim et al., 2025; Zhang and Zhang, 2025). Consumers are increasingly empowered to publicly share positive and negative service experiences (Al-Badawi et al., 2025; Golmohammadi et al., 2021; Huang and Ha, 2024; Li et al., 2023; Nazifi et al., 2021; Nguyen et al., 2025). These platforms facilitate rapid complaint resolution, yet they also transform public complaint responses into public and highly visible interactions. In the United States, 33% of consumers report observing anonymous social media complaints, and 84% receive responses from brands (Kowalewicz, 2024; Yaqub, 2024). Complaint management in digital environments is therefore no longer purely dyadic; it is audience-facing.

A defining characteristic of contemporary online complaints is the variation in communicative tone. Complaints range from civil (polite and respectful) to uncivil (rude, aggressive, or profane). Prior research has documented the rise of online incivility (Bacile et al., 2018; Bhutto et al., 2022; Wolter et al., 2023). In the event of a service failure, consumers may use harsh or offensive language in public (Bacile, 2024; Wolter et al., 2023), which can negatively affect a brand’s reputation and escalate the negative impact of the situation. Several swear words are commonly used in uncivil online complaints, including “f*ing,” “jack shit,” and “bastard” (Béal et al., 2023; Dineva and Daunt, 2023; Lafreniere et al., 2022). Such uncivil complaints often contain emotionally charged language reflecting frustration or anger, which can intensify interactions and shape how they are perceived by others (Bhutto et al., 2026).

Due to the public visibility of social media complaints, other users observe how brands respond. These users, often referred to as bystanders or silent observers, form judgments about the fairness and professionalism of the brand based on what they witness (Huang and Ha, 2024; Johnen and Schnittka, 2019). Uncivil complaints escalate negative sentiment (Bacile et al., 2025; Bacile, 2020; Fujii, 2025; Huang and Ha, 2024) and increase reputational risks when poorly managed (Marlow, 2024). Poor complaint handling carries significant financial consequences, with estimates suggesting annual losses of up to $494 billion (Customer Care Measurement and Consulting, 2020). Accommodative responses, a common strategy in the service sector, involve accepting responsibility, apologizing, taking corrective action, expressing regret, and seeking additional information (Esmark Jones et al., 2018; Lopes et al., 2023).

Prior research on online complaints has expanded rapidly, and a number of existing studies have examined response strategies, complaint tones, and bystander evaluations in different ways. Several studies (e.g. Dens et al., 2015; Chang et al., 2015; Weitzl and Hutzinger, 2017; Piehler et al., 2019; Herhausen et al., 2023; Kim et al., 2025; Zhang et al., 2024) have focused on brand response strategies, such as apology, compensation, explanation, accommodative response, defensive response, and active listening. These studies indicate that accommodative brand responses can improve consumer outcomes, including fairness perceptions, brand attitudes, word-of-mouth intentions, trust, and purchase intentions. While they establish that response strategy matters, the aforementioned studies do not clarify whether accommodative responses remain equally effective when the complaint on social media (e.g. Facebook) is expressed in a civil versus an uncivil tone.

A number of researchers have explored online incivility, harsh language, trolling, and consumer-to-consumer (C2C) conflict (Bacile et al., 2018; Batista et al., 2022; Dineva and Daunt, 2023; Wolter et al., 2023; Riedel et al., 2025; Jeseo and Tatara, 2025), and have provided important insights into harsh reviews, C2C incivility, uncivil comments, trolling, and illegitimate reviews. However, these studies focus on related but distinct forms of incivility and do not directly examine whether a civil versus an uncivil complaint changes the effectiveness of an accommodative response.

Bystander-focused studies have also shown that observers evaluate brands after witnessing online complaint interactions (Weitzl and Hutzinger, 2017; Hutzinger and Weitzl, 2021; Ku et al., 2021; Liao et al., 2025; Floriano, 2025). These studies demonstrate that observing audiences form brand attitudes, trust, satisfaction, and purchase intentions after witnessing complaint handling. However, these studies do not fully explain how bystanders interpret an accommodative response when the original complaint differs in tone. This is important because bystanders do not evaluate the brand response in isolation; they observe the complaint and the response as a single interaction.

The present study addresses these limitations through an integrated 2 × 2 experimental design by (1) focusing on accommodative response versus no response, (2) examining complaint tone as civil versus uncivil, and (3) adopting a bystander perspective to test how observed complaint handling influences brand attitude and purchase intention in the airline and hotel sectors. By doing so, this study provides a clearer understanding of how publicly visible online complaint interactions are evaluated when bystanders observe both the complainant’s tone and the brand’s response.

Specifically, the following two research questions (RQs) guided this study:

Drawing on findings from two experimental studies, this paper contributes to the online complaint and brand response literature by showing that bystanders form brand attitudes and purchase intentions by interpreting the complaint and the accommodative brand response as part of the same interaction context. In so doing, the study adds to social learning theory (Bandura, 1977) by showing that social learning in the context of online complaints is shaped by both the complainant’s tone and the brand’s response; bystanders learn not only from how the brand behaves but also from the tone of the consumer behavior that precedes the response.

Online complaints on social media have become a primary channel through which consumers express dissatisfaction, seek remedies, and publicly evaluate brands (Kim et al., 2025; Kim and Kim, 2019; Kowalewicz, 2024; Sharma et al., 2023; Zhang and Zhang, 2025). Unlike traditional complaint channels, these interactions are visible to a wide audience, transforming complaint handling into a public process that extends beyond the complainant–firm relationship.

Complaints are behavioral expressions of dissatisfaction that may serve multiple purposes, including emotional release, problem resolution, and interpersonal communication (Fornell and Westbrook, 1984; Hogreve et al., 2019; Kowalski, 1996). In digital environments, the tone in which complaints are expressed has become increasingly important. Prior research distinguishes between civil complaints, which are polite and respectful, and uncivil complaints, which involve rude, aggressive, or profane language (Anderson et al., 2014; Porath et al., 2011; Wolter et al., 2023). Synthesizing these insights, recent work has defined consumer incivility as rude, disrespectful, and emotionally charged customer behavior directed at service employees in online or offline settings, often characterized by ambiguous intent and broader negative consequences for service environments (Bhutto et al., 2026). Such uncivil communication has been associated with increased anxiety and negative observer reactions (Porath et al., 2011), polarization of attitudes in online settings (Anderson et al., 2014), and social discomfort or reduced engagement among audiences exposed to hostile interactions (Bacile et al., 2018; Wolter et al., 2023).

Consequently, online complaints, particularly those with an uncivil tone, can negatively affect consumers’ brand perceptions (Wolter et al., 2023). A study of uncivil words used by consumers provides evidence that aggressive language expressions, such as “f*ing,” “you bastard,” and “shit” (Béal et al., 2023; Dineva and Daunt, 2023; Lafreniere et al., 2022), as well as emotional slang expressions, such as “i h8 ur brand” (Bacile, 2024), can be used to analyze consumer behavior. In service contexts, this lack of civility not only disrupts the relationship between complainants and brands but also influences how observing users evaluate brand performance (Béal et al., 2023; Huang and Ha, 2024).

Due to the high public visibility of these interactions, their effects extend beyond the complainant. Evidence suggests that uncivil complaints reduce engagement, polarize audience responses, and negatively influence bystander evaluations of brands (Wolter et al., 2023; Huang and Ha, 2024). Consequently, brands may face heightened reputational risks when complaints are expressed in an uncivil tone (Golmohammadi et al., 2021; Hutzinger and Weitzl, 2021). In summary, complaint tone plays an important role in shaping how brands are evaluated in public digital environments.

Accommodative responses are widely used by firms to address customer complaints. These responses typically involve acknowledging the issue, apologizing, explaining the situation, and offering corrective actions (Esmark Jones et al., 2018; Lopes et al., 2023). As summarized in Table 1, a substantial body of research demonstrates that accommodative responses improve consumer evaluations. Prior studies have shown that responses characterized by apology, empathy, and corrective action enhance satisfaction, brand attitudes, and behavioral outcomes such as word-of-mouth and purchase intentions (Dens et al., 2015; Chang et al., 2015; Weitzl and Hutzinger, 2017; Piehler et al., 2019). These positive effects extend beyond complaining customers to include observing audiences. Bystanders who witness accommodative responses are more likely to form favorable impressions of the brand and update their evaluations accordingly (Hutzinger and Weitzl, 2021; Ku et al., 2021).

Table 1 also highlights that the effectiveness of brand responses varies across contexts. Defensive responses may outperform accommodative responses depending on communication style and context (Johnen and Schnittka, 2019), and combining an explanation with compensation may produce stronger behavioral outcomes than single strategies (Piehler et al., 2019). In addition, recent work indicates that consumers may react less positively to AI-generated responses than to human-crafted replies due to concerns about authenticity and empathy (Tan et al., 2025). Other studies have shown that response effectiveness may depend on the nature of the complaint or interaction context, such as trolling or illegitimate complaints (Jeseo and Tatara, 2025; Mulcahy et al., 2023; Riedel et al., 2025). These findings suggest that accommodative responses are generally effective, but their impact depends on the context in which they are delivered.

Due to the public visibility of social media complaint interactions, they are frequently observed by bystanders who are not directly involved in the interaction but may still form evaluations of the brand (Batista et al., 2022; Singh and Mishra, 2023; Wolter et al., 2023; Huang and Ha, 2024). These observers play an important role in shaping brand reputation, as their attitudes and behavioral intentions are influenced by what they witness.

Prior research has shown that specific characteristics of brand responses influence bystander evaluations. Empathetic and active listening responses reduce negative reactions (Herhausen et al., 2023), relational commitment enhances trust (Kim et al., 2025), personalized or timely responses improve satisfaction and purchase intentions (Floriano, 2025), and public responses, compared to private ones, strengthen positive brand evaluations (Liao et al., 2025). These findings indicate that bystanders respond not only to whether a brand responds but also to how the response is delivered.

However, bystanders do not interpret brand responses separately from the complaints that trigger them. In publicly visible online complaint interactions, the complaint and the response are observed together. Prior studies have shown that accommodative responses can improve consumer evaluations (Dens et al., 2015; Hutzinger and Weitzl, 2021; Liao et al., 2025), while uncivil communication can negatively affect brand-related evaluations and audience perceptions (Bacile et al., 2018; Wolter et al., 2023; Huang and Ha, 2024). What remains unclear is whether bystanders interpret the same accommodative response differently when the original complaint is civil rather than uncivil.

This study addresses this issue by examining how complaint tone (civil versus uncivil) shapes the effect of accommodative responses on bystanders’ brand attitudes and purchase intentions.

A conceptual model was developed to examine how bystanders interpret brand responses to online complaints (Figure 1). Complaint handling in social media environments is publicly visible, allowing observers to assess both the complainant’s behavior and the brand’s response to it.

This study builds on social learning theory (Bandura, 1977), which explains how individuals learn attitudes and behaviors by observing others’ behaviors and their consequences. During online complaint interactions, bystanders observe two cues: the tone of the complaint and the response provided by the brand. A civil complaint signals respectful communication, whereas an uncivil complaint indicates rude or norm-violating behavior. Accommodative responses indicate responsiveness, accountability, and concern for the customer.

Bystanders do not interpret brand responses in isolation. They observe both the complaint and the response as parts of the same public interaction. Thus, the tone of a complaint may influence the interpretation of an accommodative response. When the complaint is civil, an accommodative response may appear appropriate and strengthen brand evaluations. The tone of the complainant may undermine the positive effect of an accommodative response when the complaint is uncivil.

This study examined whether accommodative responses influence bystanders’ purchase intentions through brand attitudes and whether the process differs for civil and uncivil complaints.

In social media environments, complaint handling is publicly visible, allowing bystanders to observe how brands respond to dissatisfied customers. These observations serve as informational cues that influence how individuals learn by evaluating the brand and form behavioral intentions (Hutzinger and Weitzl, 2021; Ku et al., 2021).

Accommodative responses, such as apologies, explanations, corrective actions, and expressions of concern, signal that the brand is responsive, accountable, and customer-oriented (Gelbrich and Roschk, 2011; Lopes et al., 2023). When bystanders observe such responses, they may infer that the brand takes customer concerns seriously and is willing to address service problems. Prior research has shown that accommodative responses improve consumer evaluations and behavioral outcomes, including brand attitudes, word-of-mouth intentions, and purchase intentions (Dens et al., 2015; Weitzl and Hutzinger, 2017; Piehler et al., 2019). More recent studies have demonstrated that observers update their trust and behavioral intentions based on publicly visible complaint handling (Liao et al., 2025; Kim et al., 2025; Zhang and Zhang, 2025). Thus, when bystanders observe an accommodative response rather than no response, they are more likely to evaluate the brand favorably and develop stronger purchase intentions. In light of this reasoning, the following hypothesis is proposed:

Although accommodative responses may influence behavioral intentions, this effect is likely to operate through bystanders’ evaluations of the brand. Brand attitude reflects an individual’s overall evaluation of a brand and is a key predictor of behavioral outcomes (Hagtvedt, 2011). From a social learning theory perspective, bystanders observe how a brand behaves in response to a complaint and use this observed behavior to form an evaluative judgment about the brand.

When brands respond in an accommodative manner, bystanders may interpret the response as a sign of professionalism, responsibility, empathy, and customer orientation. These observed cues can improve bystanders’ attitudes toward the brand because the brand appears willing to listen, acknowledge concerns, and take corrective action. Prior studies have shown that accommodative and empathetic responses improve brand attitudes among consumers and observers (Dens et al., 2015; Esmark Jones et al., 2018; Hutzinger and Weitzl, 2021). Related research also indicates that response characteristics such as empathy, responsiveness, personalization, and relational commitment strengthen brand evaluations and trust (Chang et al., 2015; Lee and Song, 2010; Herhausen et al., 2023; Kim et al., 2025; Floriano, 2025).

Positive brand attitudes, in turn, are expected to increase purchase intentions. If bystanders form favorable attitudes after observing an accommodative response, they are more likely to consider purchasing from the brand. Conversely, when a brand does not respond, bystanders may lack positive cues about the brand’s responsiveness, which can weaken brand attitudes and reduce purchase intentions. Thus, accommodative responses are expected to influence purchase intentions indirectly through brand attitude. Based on this reasoning, the following hypothesis is proposed:

Bystanders observe not only the brand’s accommodative response but also the tone of the complaint that triggered the response. Drawing on social learning theory (Bandura, 1977), this study argues that bystanders learn from the full public interaction by observing two connected behaviors: the complainant’s tone and the brand’s response. Complaint tone provides contextual information about the interaction, while the brand response provides information about the brand’s behavior. When the complaint is civil, it is expressed in a respectful and socially acceptable manner. In such cases, bystanders may perceive the complaint as legitimate and the brand’s accommodative response as appropriate. The response may therefore strengthen brand attitude because the brand appears responsive to a reasonable customer concern. Prior research has shown that accommodative responses can improve brand evaluations and behavioral intentions when observers perceive the brand as responsive and customer-oriented (Dens et al., 2015; Weitzl and Hutzinger, 2017; Zhang and Zhang, 2025).

In contrast, when the complaint is uncivil, using rude, hostile, or aggressive language, bystanders may perceive the complainant as violating social norms (Béal et al., 2023; Dineva and Daunt, 2023; Wolter et al., 2023). In such cases, the uncivil tone may change how bystanders interpret the brand’s accommodative response. Although an accommodative response still signals responsiveness, its positive effect may be weaker because bystanders evaluate the response within a more negative and norm-violating interaction context. A rude tone may draw attention away from the brand’s positive behavior, reduce the perceived appropriateness of the accommodative response, or make the overall interaction less favorable. Therefore, the indirect effect of accommodative responses on purchase intention through brand attitude is expected to be stronger when complaints are civil and weaker when complaints are uncivil. This reasoning leads to the following hypothesis:

This paper details two experimental studies conducted at different times. Study 1, focusing on the airline context, was conducted first, in May 2024. Study 2 was conducted in the hotel context in July 2025, after the data for Study 1 had been collected and analyzed. Study 2 was designed as a conceptual replication to assess whether the hypothesized relationships hold across distinct service settings rather than being context-specific.

The airline and hotel sectors were selected because they are prominent high-contact service settings in which service failures, customer complaints, and brand responses are frequently visible to other consumers on social media. Both sectors involve experiential and relatively high-involvement purchase decisions, meaning that prospective customers may rely on publicly observable complaint interactions when evaluating service providers. At the same time, the sectors differ in the nature of the service experience: airline services involve transportation, connections, and baggage handling, whereas hotel services involve accommodation, cleanliness, and facility quality. Testing the proposed relationships across these distinct but comparable settings therefore provides an appropriate conceptual replication and helps assess whether the findings extend beyond a single service context.

Both studies followed a between-subjects 2 × 2 experimental design testing complaint tone (civil vs. uncivil) and brand response (accommodative vs. no response). The respondents were randomly assigned to one of four conditions. The scenarios were developed to resemble consumer–brand interactions on Facebook and designed to simulate realistic social media experiences.

Purposive sampling was used to recruit participants through the Qualtrics panel. Because the experimental stimuli simulated Facebook-based consumer–brand interactions, both studies screened for US adults who reported Facebook use at least monthly to ensure that the participants were familiar with the platform and could meaningfully interpret the scenarios. Almost 280 million Americans use Facebook, making it an appropriate context for studying online complaint behavior (NapoleonCat, 2025). Along with screening questions, Qualtrics offers skip logic, which allows infrequent users to be automatically excluded from participation if they do not meet the predetermined criteria. The questionnaire included two instructional attention checks at the beginning and one at the end to ensure data quality (Oppenheimer et al., 2009). In practice, the participants were instructed to select a specific response (e.g. “Somewhat disagree”) for a specific question unrelated to both studies to verify their attentiveness. During the analysis, those who failed attention checks were removed.

Study 1 was conducted in the airline context, and data were collected via a Qualtrics panel. Participant recruitment and compensation were administered by Qualtrics. Of the 873 responses, 679 (77.8% as a response rate) were included in the final analysis after 140 straight-line responses, 34 failed the attention test, and 20 incomplete responses were removed. The participants completed the experiment, on average, in five minutes.

Study 2 was administered using questionnaires on Qualtrics, and the participants were recruited via the Connect CloudResearch data platform by the researchers themselves. Study 2 replicated the 2 × 2 design in the hotel context to test the robustness and generalizability of Study 1’s findings. Out of 637 initial responses, 553 (87% response rate) indicating usable participants were retained for testing purposes. A total of 26 respondents failed the Facebook usage criteria, 39 failed the attention test, and 19 provided incomplete responses. The participants in this study received a small compensation for participation, and the average completion time was approximately seven minutes.

The participants were instructed to imagine planning a summer vacation using Blue Horizon Airlines’ (Study 1) or Majestic Stay Hotels’ (Study 2) Facebook pages. Each participant read a brief contextual scenario describing this situation before viewing the experimental stimuli. For example, in Study 1, the participants read the following:

In Study 2, the participants were presented with an analogous hotel scenario:

In both studies, the message thread included a consumer complaint (civil versus uncivil) followed by either an accommodative brand response or no response. Each message thread contained one focal complaint, to which the brand either provided an accommodative response or did not respond, and two additional contextual complaints posted by other consumers. Complaint tone was manipulated at the message-thread level rather than only within the focal complaint. Accordingly, both the focal complaint and the two contextual complaints used either civil or uncivil language. This design choice was intended to represent publicly visible social media environments in which bystanders encounter an overall conversational stream rather than an isolated consumer–brand exchange. However, the contextual comments in the uncivil conditions included stronger language than the focal complaint. Therefore, participants’ perceptions may have reflected the overall tone of the message thread, and not solely the tone of the focal complaint. This design feature is acknowledged when interpreting the findings and is discussed further in the limitations section.

Table 2 summarizes the four experimental conditions across both contexts, and the stimulus materials corresponding to each experimental condition are provided in Appendix A.

Both studies used four variables in the experiment. Two variables were manipulated, and two were adapted from established scales. The first manipulated variable, online complaint tone, was coded as civil tone (0) or uncivil tone (1), while brand response was coded as no response (1) or an accommodative response (2). A civil complaint was expressed in socially accepted language, such as “some excuse,” “difficult,” or “crazy.” Uncivil complaints used swear words or insults, such as “Stupid excuses,” “Jack shit airline,” “Why the f**k is it so goddamn,” “Filthy,” “Lame excuse,” or “F*ing joke.” In addition, in both the airline and hotel contexts, established scales adapted from previous research were used. As a measure of brand attitude (α = 0.94; adapted from Hagtvedt, 2011), three items were used on a Likert scale ranging from one (“strongly disagree”) to seven (“strongly agree”) in the airline context: “I have a favorable attitude towards the airline brand,” “I have a positive attitude towards the airline brand,” or “I have a good attitude towards the airline brand.” A measure of purchase intention (α = 0.95; adapted from Grewal et al., 1998) was adapted with the following three items: “I would make purchases from this airline brand,” “I would consider purchasing from this airline brand,” and “The probability of purchasing from this airline brand is high.”

In the hotel context, brand attitude (α = 0.98; adapted from Hagtvedt, 2011) was measured with the following items: “I have a favorable attitude towards the hotel brand,” “I have a positive attitude towards the hotel brand,” and “I have a good attitude towards the hotel brand,” using the same seven-point Likert scale. Purchase intention (α = 0.96; adapted from Grewal et al., 1998) included the following: “I would book a stay with this hotel brand,” “I would consider booking a stay with this hotel brand,” and “The probability that I would consider booking a stay with this hotel brand is high.” According to the reliability coefficients, both scales demonstrated excellent internal consistency, indicating that the adapted measures were appropriate for both service contexts.

In addition to the primary variables, both studies considered control variables such as age, gender, employment status, education, Facebook content interaction, reading online reviews, and social ethics. Social ethical sensitivity, represented by the single-item statement “I have respect for others (their culture, privacy, personality, etc.),” was adopted from Toti and Moulins (2017). This item was chosen from three statements because it is linked to the perspective of civil and uncivil complaints.

The final datasets comprised 679 valid airline responses and 553 valid hotel responses. The participants were geographically spread across the US, and the majority were aged 25–45 years. In both contexts, more than half of the sample reported using Facebook at least weekly. In the airline sample, 49.6% of the respondents were female, and in the hotel sample, 49.9% were female. Table 3 summarizes the demographic distribution.

Pre-tests were carried out to ensure that the online survey functioned as designed, to validate manipulations before the main data collection, and to confirm that the scenarios were perceived as sufficiently realistic. In terms of validating manipulations, two identical manipulation checks were used in both Study 1 and Study 2. Manipulation check 1 (MC1) tested whether the respondents perceived the complaint tone as intended. This was operationalized via the following question: “To what extent did the customer-generated comments include uncivil language [rude or inappropriate language]?” Manipulation check 2 (MC2) tested whether the respondents considered the brand’s response to be accommodative. This was operationalized via the question, “How accommodative was the response from the brand?” Scenario realism was operationalized through the question, “How would you rate the realism of the presented content?” In the manipulation and realism checks, the response scale ranged from 1 (“not at all”) to 7 (“very much”) and 1 (“very unrealistic”) to 7 (“very realistic”), respectively. Both the realism check and the manipulation checks confirmed that the manipulations worked as intended.

In Study 1 (the airline context, N = 30), uncivil complaints were rated significantly higher (p < 0.001) in incivility (mean [M] = 5.75, standard deviation [SD] = 1.52, n = 16) than civil complaints (M = 3.21, SD = 1.76, n = 14). The accommodative response was perceived as significantly more accommodative (M = 4.73, SD = 1.033, n = 15) than the no response (M = 1.87, SD = 0.915, n = 15) (p < 0.001). Scenario realism was above the midpoint (M = 5.10, SD = 1.24, N = 30) (p < 0.001).

In Study 2 (the hotel context, N = 60), uncivil complaints were perceived as more uncivil (M = 5.50, SD = 1.17, n = 30) than civil complaints (M = 1.40, SD = 0.89, n = 30) (p < 0.001). Furthermore, the accommodative response was viewed as more accommodative (M = 5.06, SD = 1.93, n = 34) than no response (M = 1.42, SD = 0.504, n = 26) (p < 0.001). Again, the scenario realism was above the mid-point (M = 5.42, SD = 1.49, N = 60).

The manipulation and realism checks were operationalized similarly to the pre-tests. In Study 1 (N = 679), the MC1 results showed that uncivil complaints were rated significantly higher in incivility (M = 5.49, SD = 1.70, n = 325) compared to civil complaints (M = 2.44, SD = 1.86, n = 354) (p < 0.001). MC2 showed that the brand response was perceived as more accommodative (M = 4.50, SD = 1.54, n = 450) compared to no response (M = 1.62, SD = 0.78, n = 415) (p < 0.001), signifying the effectiveness of the manipulation. The participants rated the realism of the scenarios significantly above the midpoint (M = 5.30, SD = 1.328, N = 679).

In Study 2 (N = 553), the MC1 results suggested that uncivil complaints were rated significantly higher in incivility (M = 5.53, SD = 1.65, n = 277) than civil complaints (M = 2.10, SD = 1.76, n = 276) (p < 0.001). MC2 indicated that the brand response was perceived to be more accommodative (M = 5.66, SD = 1.46, n = 294) compared to no response (M = 2.01, SD = 1.68, n = 259) (p < 0.001), validating the manipulation. The participants rated scenario realism above the midpoint (M = 5.61, SD = 1.27, N = 553).

To test H1, a PROCESS macro [Model 1] (Hayes, 2013) was used to examine the effects of an accommodative response to online complaints on bystanders’ purchase intentions compared to no response. The results supported H1 (Airline: β = 0.93, standard error [SE] = 0.16, t = 5.88, p < 0.001, 95% CI [0.62, 1.24]; Hotel: β = 1.26, SE = 0.17, t = 7.34, p < 0.001, 95% CI [0.92, 1.60]), revealing that social media users who observed an accommodative brand response reported significantly higher purchase intentions than those who observed no response.

Using a PROCESS macro [Model 4] to evaluate H2, both studies found that the accommodative response had a significant indirect influence on purchase intention as a function of brand attitudes in both contexts (airline: β = 0.51, 95% CI [0.34, 0.69]); hotel: β = 1.20, 95% CI [1.01, 1.39]). Direct effects were not significant, indicating full mediation.

To test H3, a PROCESS macro [Model 7] was utilized in both studies to examine whether complaint tone moderated the indirect effect of accommodative brand responses on bystanders’ purchase intentions via their perceptions of brand attitude. Both the hotel and airline contexts had significant indexes of moderated mediation (hotel: Index = −0.37, SE = 0.16, 95% CI [–0.69, −0.05]; airline: Index = −0.39, SE = 0.18, 95% CI [–0.74, −0.04]), indicating that the indirect effects were stronger when civil complaints were involved.

As shown in Figure 2a (the airline context) and b (the hotel context), bystanders’ brand attitudes were significantly affected by complaint tone and accommodative response across both studies. Although accommodative responses generally improved brand attitudes, this effect was significantly stronger for civil complaints than for uncivil complaints, indicating a moderating role of complaint tone. Specifically, the positive influence of accommodative responses was attenuated when complaints were expressed using uncivil language.

A significant indirect effect was found for civil complaints in the hotel industry (Effect = 1.40, SE = 0.13, 95% CI [1.15, 1.66]) compared to uncivil complaints (Effect = 1.03, SE = 0.12, 95% CI [0.79, 1.28]). A similar indirect effect was observed in the airline sector for civil complaint tone (Effect = 0.70, SE = 0.12, 95% CI [0.46, 0.96]) compared with uncivil complaints (Effect = 0.32, SE = 0.13, 95% CI [0.06, 0.58]). In both contexts, significant interactions were found between response type and complaint tone (hotel: β = −0.49, p = 0.026; airline: β = −0.45, p = 0.032).

In both the airline and hotel contexts, the direct effects of an accommodative response on purchase intention were not significant when brand attitude was taken into account (hotel: 0.13, p = 0.203; airline: 0.12, p = 0.108), indicating full mediation. These results support H3, namely that accommodative responses increase purchase intentions, especially in cases of civil complaints, by enhancing brand attitudes.

Among the control variables, gender, Facebook interaction, and employment status significantly influenced brand attitudes across both studies. Purchase intentions were also affected by age and education. Specifically, females and those with higher Facebook interactions demonstrated more positive brand attitudes, while younger and more educated bystanders showed higher purchase intentions.

Drawing on two experimental studies, this research examined how bystanders evaluate accommodative brand responses to online complaints. Addressing the previously identified gap regarding the limited integration of response strategies, complaint tone, and observing bystanders’ evaluations, the findings show that accommodative responses positively influence bystanders’ purchase intentions through brand attitude. However, this effect is weaker when complaints are expressed in an uncivil tone. These results demonstrate that the effectiveness of public complaint handling depends not only on how brands respond but also on how the complaint is expressed in publicly visible interactions.

Consistent with H1, accommodative responses increased bystanders’ purchase intentions across both service contexts. This aligns with prior research showing that apologies, explanations, and corrective actions improve consumer outcomes (Dens et al., 2015; Hutzinger and Weitzl, 2021; Piehler et al., 2019). Extending this work, the present findings show that such responses also influence observing audiences, who form behavioral intentions based on publicly visible complaint handling.

Supporting H2, brand attitude fully mediates this relationship. This finding clarifies the mechanism underlying bystander responses: observers first evaluate whether the brand behaves fairly and professionally, and these evaluations then shape purchase intentions. This is consistent with interactional justice theory, which emphasizes respectful and transparent interpersonal treatment in shaping fairness perceptions (Tax et al., 1998; Gelbrich and Roschk, 2011). The results suggest that fairness evaluations are formed through observation rather than direct interaction.

The results also support H3, showing that complaint tone moderates the indirect effect of accommodative responses. When complaints are civil, accommodative responses generate stronger positive effects on brand attitudes and purchase intentions, whereas these effects are weaker for uncivil complaints. While prior research has examined the negative consequences of uncivil communication (Bacile et al., 2018; Wolter et al., 2023), the present findings show that complaint tone influences how brand responses are interpreted, highlighting the importance of considering both elements jointly in public complaint settings.

From a social learning theory perspective, the results suggest that bystanders form evaluations by observing interactions between complainants and brands (Bandura, 1977). Extending prior applications of the theory (Hutzinger and Weitzl, 2021; Ku et al., 2021; Weitzl and Hutzinger, 2017), the findings indicate that bystanders consider not only observed behaviors but also the context in which these behaviors occur. In particular, complaint tone influences how brand responses are interpreted, suggesting that observational learning in digital complaint environments is shaped by both brand behavior and interaction context. The consistency of findings across airline and hotel contexts further suggests that these effects are not limited to a single service setting. Although effect sizes vary, the overall pattern remains stable, indicating that publicly visible complaint handling influences bystander evaluations across different service environments.

This research contributes to the online complaint and brand response literature in three ways. First, this study contributes by examining the combined role of accommodative brand responses, complaint tone, and observing bystanders in publicly visible online complaint interactions. Prior studies have examined accommodative or related brand responses (Dens et al., 2015; Weitzl and Hutzinger, 2017; Piehler et al., 2019), online incivility or complaint tone (Bacile et al., 2018; Wolter et al., 2023; Huang and Ha, 2024), and bystander evaluations (Hutzinger and Weitzl, 2021; Ku et al., 2021; Liao et al., 2025). However, they have not clearly shown how bystanders evaluate accommodative responses when the triggering complaint is civil versus uncivil. By testing these elements together, this study shows that bystanders form brand attitudes and purchase intentions by interpreting the complaint and the brand response as part of the same public interaction.

Second, this study contributes by showing that complaint tone changes the effectiveness of accommodative responses. Across the airline and hotel contexts, accommodative responses improved bystanders’ brand attitudes and purchase intentions, but these positive effects were weaker when complaints were uncivil rather than civil. This finding refines prior research showing that accommodative responses generally improve consumer outcomes (Dens et al., 2015; Piehler et al., 2019; Weitzl and Hutzinger, 2017) and demonstrates that the value of accommodation depends not only on the brand’s response but also on how the original complaint is expressed. Thus, the study advances the understanding of online complaint handling by showing that complaint tone influences how bystanders interpret accommodative responses.

Third, this study contributes to social learning theory (Bandura, 1977) in digital service contexts. The theoretical development proposed that bystanders learn from publicly visible complaint interactions by observing both the complainant’s tone and the brand’s response. The findings support this theorization by showing that bystanders’ evaluations are shaped by both observed behaviors. Specifically, complaint tone changes how accommodative responses affect brand attitude and, in turn, purchase intention. This extends social learning theory by demonstrating that observational learning in online complaint settings is based not only on brand behavior but also on the consumer behavior that precedes the brand response.

The findings provide actionable implications for three distinct stakeholder groups involved in managing public digital complaint environments: brand managers, social media platform managers, and policymakers. Each group plays a different role in shaping how complaint interactions unfold and are interpreted by observing audiences.

For brand managers (e.g. airline and hotel brands), the results highlight that accommodative responses remain an effective strategy for improving bystanders’ brand attitudes and purchase intentions. However, their effectiveness is not unconditional. When complaints are expressed in uncivil tones, the positive impact of such responses is significantly weakened. This suggests that managers should move beyond generic response approaches and adopt tone-sensitive communication strategies.

In practice, this means that brand managers should develop structured response protocols that account for the tone and context of complaints. While maintaining professionalism and empathy is essential, immediate accommodative responses to highly aggressive or abusive complaints may not always yield the intended positive outcomes. Instead, managers may consider delayed, reframed, or moderated responses that maintain brand dignity while avoiding the reinforcement of norm-violating behavior. Training frontline social media teams to recognize complaint tone and adjust response strategies accordingly is therefore crucial. In general, managers should recognize that the handling of complaints on social media is not limited to resolving individual issues but is a public performance that is observed by a wider audience. A small improvement in the way responses are framed can have a significant impact on the perception of a brand and on observers’ purchase intentions.

For social media platform managers (i.e. those responsible for designing and governing platforms such as Facebook), the findings highlight their central role in shaping the communication environment in which brand–consumer interactions occur. Since uncivil communication weakens the effectiveness of otherwise appropriate brand responses, platforms directly influence whether complaint interactions remain constructive or become harmful.

Platform managers should therefore invest in proactive moderation systems, including automated tone detection, sentiment monitoring, and early intervention mechanisms, to identify and manage uncivil or abusive content. Such tools can help prevent escalation and maintain communication environments that support constructive engagement between brands and users. In addition, clearer community standards and enforcement policies are needed to discourage aggressive language and promote respectful interaction norms.

Importantly, platform design choices, such as the visibility of responses, comment ranking algorithms, and reporting mechanisms, can shape how interactions are interpreted by bystanders. By prioritizing respectful and informative exchanges, platforms can enhance the overall quality of interactions, strengthen user trust, and support more effective brand response outcomes for firms operating within their ecosystems.

For policymakers and regulators, the findings suggest that online incivility is not only a social concern but also a market-relevant issue that affects how effectively firms can respond to consumer complaints. Because uncivil complaint environments reduce the positive impact of accommodative responses, improving the quality of digital communication can enhance both consumer protection and market efficiency.

Policymakers should consider developing guidelines and regulatory frameworks that promote responsible platform governance, including transparency in content moderation practices and accountability for managing harmful communication. Emerging research shows that firm responses to social media interactions must be strategically managed, as engagement can sometimes amplify negative dynamics depending on the context (Cheng et al., 2024). This highlights the need for policies that support not just responsiveness but also appropriate and context-sensitive interaction management. In addition, regulators should encourage the adoption of AI-assisted moderation tools, standardized reporting systems, and clear behavioral guidelines that help distinguish legitimate consumer complaints from abusive or norm-violating content (Bhutto et al., 2026). Such measures could contribute to safer and more constructive digital environments, improve trust in online marketplaces, and support more effective complaint-handling practices across industries.

This paper has several limitations that should be acknowledged and addressed in future research. First, the experimental design tested only accommodative versus no-response strategies. The literature indicates that accommodative responses build bystander trust better than defensive or minimal actions do (Chang et al., 2015; Johnen and Schnittka, 2019). Future research should expand the experimental design by applying multiple response styles, such as the use of humor or sarcasm, which may yield different insights.

Second, complaint tone was manipulated at the message-thread level: both the focal complaint and the two contextual complaints were written in either civil or uncivil language. This approach reflected the way bystanders encounter complete conversational streams on social media. However, the uncivil contextual comments contained stronger language than the focal complaint, which may have amplified perceived incivility and influenced the manipulation checks. Consequently, the findings identify the effect of the overall interaction tone rather than isolating the causal effect of the focal complaint alone. Future studies should manipulate only the focal complaint while holding contextual comments constant.

Third, the use of scenario-based experiments may have limited external validity, as the participants evaluated hypothetical situations rather than real consumption decisions. Although such designs are common in service research, future studies could complement these findings using field data, archival social media interactions, or behavioral experiments involving actual consumers to enhance ecological validity.

Fourth, the empirical context was limited to the airline and hotel sectors, which represent high-contact service environments. While this improved internal consistency, it may constrain generalizability. Future research should examine whether the observed effects hold across other contexts, such as healthcare, public services, or e-commerce, where expectations, emotional intensity, and interaction norms may differ.

Finally, this study focused on situational factors (i.e. complaint tone and response strategy) but did not account for individual differences among bystanders, such as prior brand attitudes, empathy, or digital literacy. These factors may influence how observers interpret complaint interactions and evaluate brand responses. Incorporating such variables could provide a more nuanced understanding of heterogeneity in bystander reactions.

This paper concludes with the importance of context-sensitive accommodative responses in digital service environments, where complaint interactions are publicly observed by large audiences. While such responses generally improve bystander perceptions, their effectiveness depends on the complaint’s tone, with uncivil communication weakening their impact. These findings underscore the need for strategically managed responses that consider both brand actions and consumer behavior. Future research should build on these insights to better understand how firms manage public complaints and strengthen customer trust.

Maqsood Hussain Bhutto is currently a Visiting Researcher at Birmingham Business School, University of Birmingham, UK. The authors gratefully acknowledge the Foundation for Economic Education (Liikesivistysrahasto), Finland, for supporting this research.

The supplementary material for this article can be found online.