From brand ambassadors to brand adversaries: antecedents of luxury employee brand hate

Physical retail is a dominant distribution channel for the global personal goods luxury market. According to Bain and Company (2025), monobrand stores account for 36% of total sales, compared to 20% for online. Moreover, it has been observed that the presence of brick-and-mortar locations has a direct impact on online sales (CBRE, 2025). Luxury brands continue to invest in expanding or upgrading their physical retail footprint. While physical store attractiveness and the overall retail experience are important, sales associates play a crucial role in the success of a luxury retail strategy (Dion and Arnould, 2011). Client advisors and other frontline employees are instrumental in conveying the brand’s ideology. They also form strong bonds with clients, particularly in service-intensive retail contexts, where frontline employee–brand relationships contribute directly to customer loyalty and brand equity (Jung et al., 2021). The prominence of sales associates and retail-level employees highlights the need for companies to attract, train and retain frontline talent (Solca, 2024). Consequently, luxury companies have implemented various employee-focused marketing and engagement initiatives. For example, LVMH launched an employer brand campaign, “Where Dreams Become Careers”. This is particularly relevant because luxury brands are facing a retail labour crisis (Danziger, 2024). According to a consultancy study, 51% of retail employees plan to leave their current jobs (CXG, 2024).

This rising dissatisfaction is increasingly being expressed publicly through digital platforms such as Glassdoor and Indeed, where current and former employees can share negative perceptions of their employer, potentially affecting the firm’s employer brand. Despite the potentially significant impact of employees’ Negative Word of Mouth (NWOM), research remains largely focused on consumers, with limited attention paid to the antecedents of employee NWOM and employee brand hate (Lee et al., 2022). This is significant as Bryson et al. (2021) highlighted NWOM as a significant predictor of luxury brand hate, suggesting that both current and prospective consumers are responsive to negative messages disseminated by others.

It is therefore essential to examine the reasons why employees may develop feelings of hatred toward their luxury employers. To address this research gap, the current study explores how intense employee dissatisfaction in the luxury goods sector leads to negative brand-oriented behaviours. The objective is to uncover the key drivers of employee brand hate, in both general and luxury retail-specific settings.

Research on emotional brand relationships in luxury retail remains limited, particularly from the perspective of frontline employees (Hyun et al., 2021). Previous research has examined brand hate toward luxury brands in the context of specific luxury fashion brands (Atwal et al., 2021), in situations such as gaming (Guzzetti et al., 2024), or across the luxury industry as a whole (Bryson et al., 2013, 2021).

A review of the literature indicates that retaining or losing high performing employees impacts competitive advantage, operational efficiency, and work effectiveness (Lam et al., 2022). A stimulating and meaningful workplace is associated with improved employee performance and retention (Scroggins, 2008).

As the primary interface between organizations and customers, frontline luxury employees play a central role in brand management by conveying brand values and cultivating strong client relationships (Wallace et al., 2013; Merk and Michel, 2019). To attract and retain customers, personal selling functions as a key differentiator from competitors. It also represents a primary source of value in distinguishing luxury from non-luxury goods (Kapferer and Bastien, 2012).

The highly customized and emotionally demanding nature of luxury service work exposes frontline employees to greater emotional stress than employees in non-luxury service settings (Brant, 2016; Sherman, 2007). These negative emotional states, including frustration, anger, and anxiety, may increase the likelihood of counterproductive workplace behaviours such as decreased morale and productivity, retaliation, service sabotage, and workplace deviance (Merk and Michel, 2019; Phillips et al., 2006; Harris and Ogbonna, 2012; Kumar and Pansari, 2016; Malik et al., 2019).

Research on brand relationships has primarily focused on consumers (Fetscherin and Heinrich, 2015; Fournier et al., 2012) and positive brand relationships (Hegner et al., 2017). However, recent studies have presented a taxonomy of negative emotions that consumers may direct toward brands. Among these, the concept of brand hate (Zarantonello et al., 2016; Hegner et al., 2017; Kucuk, 2016a, b; Bryson et al., 2013) is recognized as the most intense and extreme form of negative consumer sentiment defined as “an intense negative emotional affect towards the brand” (Bryson et al., 2013, p. 395). Building on this literature, we define employee brand hate, based on Kucuk’s (2016a, b) concept of brand hate, as a psychological state whereby employees form intense negative emotions and detachment toward brands.

Such a negative emotional state can also be understood through organizational research showing that employees may react particularly strongly when workplace experiences are perceived as unfair or as breaches of relational expectations. Research on psychological contract violation and organizational justice suggests that employees may develop strong feelings of anger, betrayal, and retaliatory intent when workplace experiences are perceived as unfair (Morrison and Robinson, 1997; Skarlicki and Folger, 1997).

Sternberg (2003) proposes that hate is typically expressed through emotions such as disgust, contempt, and anger which aligns with the components of brand hate identified by Fetscherin (2019). Zarantonello et al. (2016) differentiate between active and passive brand hate, further classifying responses into avoidance-like, approach-like, and attack-like behaviours. However, brand meanings are not only applicable to organizations' external relationships with customers but also impact employee attitudes and behaviour in contemporary organizations (Brannan et al., 2015). Raggiotto et al. (2023) demonstrate that an employee’s negative emotional bond with the retailer affects negative dysfunctional brand-oriented behaviours. In retail settings, dysfunctional behaviours refer to customers and employees that “intentionally or unintentionally, overtly or covertly, act in a manner that, in some way, disrupts otherwise functional service encounters” (Harris and Reynolds, 2003, p. 145). Brand hate has significant implications, as it can lead to “the purposeful and deliberate intention to avoid or reject a brand, or even to act out behaviors that demonstrate this rejection” (Bryson et al., 2013, p. 395). In the case of employees, this could mean ultimately the decision to leave the organization.

From an internal branding perspective, employees are not only organizational members but also key carriers of brand meaning (Punjaisri and Wilson, 2011; Burmann and Zeplin, 2005). Research on employee-based brand equity further emphasizes that employees play a central role in delivering brand value through their behaviours and interactions, effectively acting as brand ambassadors (King and Grace, 2010; Morhart et al., 2009). Employer branding literature highlights that employees’ perceptions of their work environment influence their identification with the organization and their willingness to advocate for or against it (Backhaus and Tikoo, 2004).

WOM is a form of online or offline communication that occurs between non-commercial sources and receivers regarding brands, products, and services (Keller, 2007). WOM is widely recognized as one of the most influential forces shaping consumer behaviour (Brown et al., 2005). Depending on the message's valence, WOM can be categorized as either positive word-of-mouth (PWOM) or negative word-of-mouth (NWOM). Consumers typically engage in PWOM after positive consumption experiences, whereas they resort to NWOM following negative experiences (Hollebeek and Chen, 2014). Previous research has demonstrated that negative information exerts a stronger influence on receiver responses compared to positive information (Papathanassis and Knolle, 2011). This is attributed to the greater analytical depth and perceived utility of negative information in impression formation (Papathanassis and Knolle, 2011).

Discretionary brand-oriented behaviours have traditionally been associated with consumers but are increasingly recognized as relevant among employees, who play a vital role in organizational brand management (Lee and Suh, 2020). Employee brand love, expressed through PWOM, has been shown to enhance affective commitment and reduce turnover intention (Mittal et al., 2022; Kim and Legendre, 2021). Conversely, negative brand-oriented behaviours conveyed through NWOM in external settings can damage a brand's reputation (Lee et al., 2022; Garas et al., 2018; Harris and Ogbonna, 2012).

Employees may engage in negative brand-oriented behaviours, such as disseminating NWOM, defined as “unflattering, uncomplimentary, or critical comments that employees make about the staff, customers, efficiency, or competency of their employing organization” (Harris, 2013, p. 467).

This phenomenon can be attributed to employees who serve as key sources of insider information regarding company practices (Cable and Turban, 2001). As organizational insiders, employees are expected to be well-informed about their company. As a result, their NWOM is often more impactful on external audiences than consumer-generated NWOM (Harris and Ogbonna, 2012). Information on internal practices such as working conditions often originates from employees who have direct experience with these aspects (Keeling et al., 2013). Concerns about workplace climate, job fit, and professional development further enhance the credibility and value of employee-generated insights (Parker et al., 2003). Customers expect employees to have extensive knowledge of their company’s services and products. For this reason, employee-generated NWOM is often perceived as more credible than peer-generated NWOM (Bansal and Voyer, 2000). As a result, employee NWOM can undermine the brand’s credibility (Garas et al., 2018) and have severe repercussions on brand image and reputation, deterring consumers from purchasing and discouraging job seekers from applying (Keeling et al., 2013; Van Hoye and Lievens, 2009).

Traditionally, employees' opportunities to communicate positive and negative workplace experiences have been limited to internal communication channels sanctioned by the organization. However, in the digital era, employees can instantly disseminate messages to large audiences with minimal effort (Ravazzani and Mazzei, 2018). Employee voice which refers to the constructive expression of ideas, suggestions, or concerns intended to improve organizational functioning (Liang and Yeh, 2020) poses a significant risk for reputation-sensitive organizations due to its potential negative consequences (Opitz et al., 2018). As a result, employees who act against their organization’s interests may function as brand saboteurs (Wallace et al., 2013).

Despite this emerging phenomenon, empirical research on the impact of employee communication on corporate reputation remains scarce (Miles and Mangold, 2014). Although scholars emphasize the importance of presenting a consistent, unique, and attractive organizational image to external stakeholders (Van Riel and Fombrun, 2007), the negative effects of employee NWOM on brand reputation remain underexplored (Van Hoye and Lievens, 2009; Harris, 2013; Lee and Suh, 2020). Limited research has linked employee NWOM to emotional exhaustion, underscoring the need for further investigation into its determinants and consequences (Lee and Suh, 2020).

This perspective aligns with research on employee voice, which highlights the role of communication structures in shaping how and where employees express concerns (Morrison, 2014). In particular, when internal voice mechanisms are perceived as ineffective, employees may resort to external channels to communicate dissatisfaction.

A key factor influencing such behaviour is psychological safety, defined as the shared belief that individuals can express themselves without fear of negative consequences (Edmondson, 1999). Work environments with psychological safety have been associated with reduced internal voice and increased reliance on alternative forms of expression. This is significant because, in the broader brand hate literature, NWOM is typically conceptualized as a consequence of negative experiences rather than as a potential antecedent of reputational damage.

This study examined employee reviews published on Link to the website, an online platform where current and former employees anonymously review companies without fear of retaliation from their employers. The website provides valuable insights into employee perceptions of their jobs and employers, including aspects such as corporate culture (Corritore et al., 2020). As Alvarez Castillo et al. (2022, p. 296) note, “Such employee-generated content from these digital platforms offers less risk of obtaining biased views of how the employees feel or felt being part of their organization”. This is significant as negative attitudes are more likely to be disseminated through social networks (Allsop et al., 2007).

The dataset was based on self-selected, anonymous online reviews, meaning that employees’ internal emotional states were not directly observable. The analysis therefore centred on textual expressions of dissatisfaction, hostility, and negative evaluations. Within this framework, employee brand hate was operationalized as the discursive manifestation of intense negative brand-oriented sentiment.

To identify the antecedents of NWOM, we employed a semi-supervised natural language processing (NLP) technique for topic modelling. Opinion mining was used to process the data, identify patterns, and systematically categorize the content using established categorization rules (Mayring, 2004). A subset of manually categorized data was subsequently used to generate classifier models through machine learning methods, applying semi-supervised active learning (Shams, 2014). Active learning involves training a model on labelled data and using it to classify unlabelled instances. The model's classifications were reviewed by a human expert in a process referred to as “judgment” and the correctly labelled instances were added to the pool of labelled data. Semi-supervised methods were deemed suitable for this research, as they require minimal labelled data for training models (Chapelle et al., 2010).

The study followed a three-stage approach. The first step was data collection. To acquire a large number of comments, an automated web crawling technique was employed (Moon and Kamakura, 2017). Specifically, Octoparse, a web scraping software, was used to extract data from Link to the website. The focus was on reviews from the nine most valuable luxury brands according to Kantar (2023): Louis Vuitton, Hermès, Chanel, Gucci, Dior, Cartier, Rolex, Saint Laurent, and Prada. The software collected all comments written in English from January 2008 to August 2024, removing duplicate or irrelevant entries. For each job review, we obtained data including company name, rating (on a 1-to-5 scale), tenure, job title, location, review title, and stated pros and cons of working at the company. The final sample consisted of 2,721 comments (Table 1). We then coded each comment by the employee’s department, allowing for comparisons across job functions.

Table 1 provides an overview of how employees evaluate their work environment. Louis Vuitton stands out with the lowest average rating of 2.92, accompanied by the highest standard deviation of 1.31. This suggests both relatively low overall satisfaction and widely varying opinions among respondents. Prada shows a similar weak evaluation, with a mean rating of 3.16 and a level of dispersion close to that of Louis Vuitton. The two brands not only register the lowest overall ratings but also have the highest proportion of feedback from retail employees—83.4 and 76.5% respectively. Overall, with a mean rating of 3.41 and a standard deviation of 1.26, the data signal underlying workplace concerns.

The second step involved the coding of comments. In line with the research objective, we focused exclusively on negative comments, as valenced evaluations are strong predictors of overall attitudes toward a given subject (Brinol et al., 2004). To capture employee dissatisfaction, 'negative reviews' were operationalized as those with an overall rating below 3.4 on Glassdoor’s 5-point scale. This threshold was chosen because it corresponds to the sample mean and falls within Glassdoor’s reported retail-sector average range of 3.3–3.5. Scores below this benchmark can therefore be interpreted as signalling employee misalignment and dissatisfaction.

To ensure the reliability of the semi-supervised machine learning process, a content analysis on a subsample was created through manual annotation, following an iterative interpretative and comparative process of emergent coding (Stemler, 2015). Two researchers performed open coding and categorization ensuring internal validity. In cases of disagreement, discussions were conducted to reach a consensus-based evaluation. Specifically, 100 reviews for each analysed brand were randomly selected and manually labelled, resulting in a manually coded subsample of 900 reviews. This set was used to train the algorithm and validate the topic assignment. The consistency of manual coding was verified using Cohen’s Kappa, achieving a score of 0.82, indicating substantial agreement (Cohen, 1960). This manually coded subset was used to train and validate the classification model using an 80/20 train-test split. To assess the model’s reliability, we calculated standard performance metrics including Precision, Recall and the F1-score across the pre-identified categories (Table 2) (Powers, 2011). The analysis yielded a weighted average F1-score of 0.82, with Precision and Recall consistently above 0.75 across all categories, confirming the model's ability to accurately distinguish between complex employee-related topics. The confusion matrix (available upon request) further validated the classification, showing minimal overlap between drivers. The few observed misclassifications primarily occurred between the labels 'Management' and 'Company Culture', likely due to the inherent semantic overlap in how employees describe organizational leadership.

Given the moderate dataset size, a supervised machine learning approach using Python’s Scikit-learn library was adopted due to its suitability for small-to medium-sized textual datasets. Reviews were pre-processed using the Natural Language Toolkit (NLTK) and vectorized through Term Frequency-Inverse Document Frequency (TF-IDF) to prepare the data for thematic classification. A semi-supervised topic modelling approach combining Latent Dirichlet Allocation (LDA) and supervised classification was then employed to identify key themes within the reviews.

Unlike traditional unsupervised topic modelling, where themes emerge without predefined labels, this study adopted a semi-supervised learning framework to balance exploratory discovery with analytical rigor. LDA was used to identify initial latent themes within the review data, which were subsequently refined and validated to establish the final ten drivers of employee brand hate and reduce post-hoc interpretation bias. Specifically, two researchers independently examined the 'top words' and a sample of original reviews with the highest probability scores for each cluster. Only clusters that demonstrated high semantic consistency and were independently validated by both researchers were refined into the ten final drivers. This approach ensured that the categories were grounded in the actual narratives of the employees rather than being arbitrarily imposed. To facilitate the scaling of this classification, a manually pre-labelled subset of 900 comments was used to train the model and map the LDA-generated topics to these validated themes. To ensure reliability, a confidence score (CS) was monitored based on the probability distribution output by the classifier. The training phase continued until an optimal confidence level was reached (CS = 0.78), following the approach of Zhang et al. (2010).

This methodology enabled efficient scaling of thematic categorization while maintaining interpretability. The combination of unsupervised topic modelling and supervised classification ensured a structured and reliable approach to sentiment and thematic analysis. The analysis resulted in ten topic categories related to employee brand hate.

The third and final step involved conducting a regression analysis to examine the relationships between a set of indicator variables (coded themes), a set of covariates (employee seniority, department, region, word count), and a distal outcome variable (overall company rating).

As shown in Table 3, the opinion mining analysis identified ten categories reflecting the most prevalent issues related to employees’ brand hate in the luxury goods industry. These categories were subsequently segmented according to three first-order dimensions of brand hate following the framework of Odoom et al. (2019): (1) experiential avoidance, (2) deficit–value avoidance, and (3) moral avoidance.

While the categories are analytically distinct, some themes are conceptually related. Where relevant, these interconnections are acknowledged without duplicating explanations.

Experiential avoidance: Experiential avoidance refers to employees' negative emotional responses stemming from day-to-day experiences. These can be distinguished into static experiences, such as public humiliation by a manager, and dynamic experiences, such as a consistent lack of recognition (Kranzbühler et al., 2018). Research consistently identifies a direct correlation between negative past experiences and brand hate (Hegner et al., 2017; Kavaliauskė and Simanavičiūtė, 2015; Platania et al., 2017; Zarantonello et al., 2016).

1. Workflow: This category reflects the operational inefficiencies that hinder employees’ ability to perform their roles effectively. Key issues are linked to the use of outdated software systems, unclear processes, resource constraints, and slow decision-making. Inadequate inventory management systems are a particular source of frustration, leading to stock shortages and difficulties in fulfilling customer demands. A lack of communication between corporate offices and store-level teams is another issue. Employees also report that corporate decision-making appears disconnected from store-level operations, particularly when they feel that headquarters imposes policies without fully understanding operational challenges.

2. Hierarchical structure: This category refers to the company’s rigid organizational structure and slow decision-making processes. Employees complain about a traditional, hierarchical approach, with limited flexibility, slow adaptation to change, and corporate bureaucracy that slows decision-making. The work environment in luxury companies is often described as hierarchical, rigid, and heavily influenced by the cultural norms of the country where the company originated. Employees frequently mention the lack of openness, with personal opinions and creativity being suppressed in favour of strict adherence to top-down instructions. A recurring theme is that headquarters are perceived as overly controlling, leaving limited local flexibility and growth opportunities for the company.

3. Work-life balance: This category relates specifically to structural aspects of work scheduling and time demands. Employees reported irregular working hours, mandatory overtime, work during holidays and weekends, and a lack of control over vacation scheduling. The lack of predictability in schedules is a recurring issue, with employees often informed about their shifts at the last minute, making it difficult to balance personal commitments. Many employees report a lack of opportunities for flexible working arrangements, such as remote work or part-time hours, with particular challenges faced by those returning from maternity leave.

Deficit value avoidance: Deficit-value avoidance occur when the consumer “does not perceive the brand to provide adequate value” (Berndt et al., 2019, p. 182). In the case of employee brand hate, this reflected a perception that the employer fails to deliver on functional benefits such as remuneration, fringe benefits, working hours, and training.

1. Compensation: A significant number of reviews pointed to dissatisfaction with compensation. Employees frequently mention low salaries, limited fringe benefits and a lack of transparency in the payment of bonuses. There was also frustration over unequal pay between full-time and part-time employees. A frequently reported concern the disparity between the compensation of managers and non-managerial employees, with the latter often feeling undervalued and overworked.

2. Workload and performance: This category focuses on performance expectations. The strong emphasis on performance metrics, such as KPIs, coupled with competition among staff, creates considerable pressure. Employees often feel that they are in constant race to meet targets. This leads to frustration, particularly when sales quotas are perceived as unattainable and the process of achieving them becomes exhausting.

3. Internal growth: This category reflected limited opportunities for career progression and professional development. Reviews mentioned frustration with the HR department, such as long recruitment periods and insufficient training. Internal growth opportunities are perceived as being scarce and highly competitive. Inexperienced management and the absence of clear communication from HR further impede career progression, making it difficult for employees to see a clear path to advancement. Internal growth is constrained by slow promotion cycles, favouritism, and ineffective management, while high turnover reflects the dissatisfaction of employees who feel undervalued and professionally stagnant.

Moral avoidance: Kucuk et al. (2016a, b, p. 43) define moral avoidance as “an ideological mismatch between a consumer’s ideological beliefs and the values represented by the brand.” Bryson et al. (2013) report that consumers' perceptions of corporate social performance (CSP) do not seem to be a strong source of brand hate, whereas Hegner et al. (2017) identifies ideological incompatibility as its strongest driver. In the context of employee brand hate, this phenomenon may be less closely related to CSP and instead reflect perceived moral injury, which occurs when employees witness or participate in actions that violate their deeply held moral beliefs (Litz et al., 2009). This is particularly critical, as DeTienne et al. (2012) found that moral stress in service workers is significantly associated with lower levels of job satisfaction.

1. Atmosphere and wellbeing: This category captured the psychological and social climate of the workplace, including interpersonal tensions, stress, and experiences of toxic behaviour and bullying. The reviews also highlighted a culture of fear and manipulation, contributing to a sense of emotional exhaustion and diminished morale among employees. Reports of a toxic culture are sometimes further aggravated by microaggressions, bullying, favouritism, and discrimination.

2. Rules and etiquette: This category relates to formal expectations regarding employee appearance and behaviour. Employees are required to adhere to rigid dress codes, often dealing with uncomfortable uniforms, specific grooming standards, and inconsistent enforcement of these rules across the team. Employees report being scrutinized on both their appearance and their behaviour, creating a stressful and unpleasant work environment.

3. Micromanagement and nepotism: This category focused specifically on managerial practices. Employees expressed dissatisfaction with direct management practices, including excessive supervision, lack of autonomy, and favouritism in task allocation and promotions. Micromanagement is pervasive, with overbearing supervision from managers who demand constant reporting, scrutiny, and control over every aspect of the work, often with limited autonomy for employees to make decisions or take ownership of their tasks.

4. Organizational Culture: This category reflected broader organizational norms and values, including systemic issues related to inclusion, diversity, and informal power structures. Comments indicate significant issues related to xenophobia, sexism, and racism within certain organizations. Employees of colour frequently describe disparities in career opportunities, with many feeling excluded from leadership roles compared to colleagues of other ethnicities. Sexism is also a pervasive issue, especially regarding career progression, with women disproportionately affected by gender-biased company cultures. Several employees mentioned that promotions and rewards are often given to men or employees who are part of a certain 'boys club'. These findings can be interpreted through the lens of organizational justice theory, which emphasizes the importance of fairness in outcomes, processes, and interpersonal treatment (Colquitt, 2001). The results are consistent with research suggesting that authentic and transparent diversity and inclusion (DEI) communication enhances employer attractiveness and fosters stronger employee engagement (Confetto et al., 2023).

To identify the issues most frequently associated with employees’ brand hate across the ten categories, statistical analyses were performed. A preliminary examination revealed that the issue of “Internal growth” was mentioned most often, followed by “Micromanagement and nepotism” and “Work-life balance”. In contrast, the “Hierarchical structure” category was reported only 71 times (Table 4). Further analysis of co-occurrences indicated that “Internal growth” is most commonly paired with “Micromanagement and nepotism”, and subsequently with “Compensation” and “Atmosphere and well-being”. These findings suggest that employees tend to link “Internal growth” practices with compensation-related concerns and the overall work environment when discussing their experiences.

The analysis examined whether employees’ dissatisfaction with internal growth opportunities persisted across seniority levels and departments. Results revealed that employee seniority (junior, mid-level, senior) does not significantly affect negative reviews, which consistently centre around “Internal growth”. However, the “Management and nepotism” category also appeared among the top three issues in comments for all levels of seniority. Specifically, for mid-level employees, concerns about “Management and nepotism” are nearly as prominent as those related to “Internal growth”. This suggests that mid-level employees have a relatively balanced focus on both professional development and management-related issues. “Internal growth” remains the main issue also for employees working in manufacturing and corporate departments.

As the majority of comments originated from retail employees, we focused our analysis on this segment of the industry. Employees in the retail sector, who make up 70% of our sample, revealed that dissatisfaction is most pronounced with regard to “Management and nepotism”, followed by concerns about “Internal growth” and “Work-Life balance” (Table 5). In addition, when focusing specifically on the corporate levels in the retail sector, differences emerge in the issues most frequently reported by junior, mid-level and senior employees. Unlike senior retail employees, who follow the overall pattern of dissatisfaction with “Internal growth”, junior retail employees express greater dissatisfaction with “Atmosphere and well-being,” while mid-level employees are more critical of “Management and nepotism”. This suggests that different levels of seniority within the retail sector highlight distinct areas of concern.

We examined whether retail employees’ concern about their work experience was primarily reflected in their ratings or their narrative comments. To facilitate this analysis, the comments were transformed into a quantitative variable that accounts for the number of categories mentioned in each comment. The initial phase of the correlation analysis involved a hypothesis test. The null hypothesis (H0) stated that there was no statistically significant correlation between the two variables. The alternative hypothesis (H1) proposed that such a correlation exists. Given the extremely low p-value (***p < 0.001), we can reject H₀, thereby concluding that the correlation between the two variables is statistically significant.

We consequently investigated whether the correlation was strong or weak and whether it was positive or negative. Spearman's rank correlation coefficient, a nonparametric measure, was used to determine the correlation between the two variables, as the data did not follow a normal distribution (Kumar and Abirami, 2018). The analysis yielded a coefficient of ρ = − 0.35. Based on Cohen’s conventions, where coefficients of approximately 0.10, 0.30, and 0.50 correspond to small, medium, and large effect sizes, respectively (Cohen, 1988), the correlation between the two variables can be classified as a medium negative relationship. This result suggests that as the rating increases the number of issues reported in the comments by retail employees tends to decrease. In practical terms, higher ratings are only partially associated with fewer categories mentioned, and considerable variability remains, indicating that the association is not sufficiently robust to support a consistent or predictive trend between the two variables.

A broad analysis of the comments revealed considerable variation in their lengths, measured in terms of word count. To examine the relationship between comment length, the number of categories mentioned, and the ratings, we introduced a variable termed word count that captures the total number of words in each comment. We then employed a multiple linear regression model, estimated using the ordinary least squares (OLS) method, where the retail employees’ ratings serve as the dependent variable (Y), and word count (X₁) along with the number of categories mentioned (X₂) function as independent variables. The linear regression model Y = β₀​ + β₁​*X₁ + β₂​*X₂ + ε is specified as follows:

In the regression analysis conducted, it was assumed that Y may be potentially explained by X₁ and X₂ through some coefficients that show how different variables will contribute differently. It is also assumed that the error (ε) present in the model is random, and that may be caused by the following four reasons: some variables may not be considered in the model, some measurement issues may arise, the relationship between the variables may not be perfectly linear, and random effects may influence the relationship. In addition, the results are considered valid, as four hypotheses were specified prior to running the regression model. The errors have a mean of zero—indicating that the size of the error does not depend on the magnitude of the measurement—and are normally distributed, uncorrelated, and exhibit constant variance (Poole and O’Farrell, 1971).

The linear regression model is statistically significant, and all coefficient estimates are significant at conventional levels (Table 6), indicating a meaningful association between the predictors and the rating outcome. The results reveal that both an increase in the word count and the number of categories mentioned in a comment are associated with lower rating scores. Holding all other variables constant, the model indicates that for each additional word written in a comment, the rating decreases by 0.007 points on the 5-point scale. Assuming that a 10-word comment is associated with a predicted rating of 4.00, results suggest that a 50-word comment would be expected to receive a lower rating of 3.72. While the per-word coefficient is small, this comparison demonstrates that moving from a concise to a lengthier comment reduce the overall retail employee rating. Furthermore, for each additional category included in a comment, the rating decreases by 0.35 points. This means that a reviewer mentioning six categories would be expected to provide a rating approximately 1.05 points lower than a reviewer mentioning three only, ceteris paribus. These findings demonstrate a negative relationship between the rating and both predictors.

The regression model accounts for 17.5% of the variance in rating scores, which is considered a meaningful effect size in behavioural and social science research (Cohen, 1988). Nevertheless, the remaining unexplained variance suggests that word count and the number of categories mentioned capture only part of the full complexity of retail employees’ sentiment. This highlights opportunities for future research to identify additional factors that influence ratings.

Additionally, we investigated the correlation between retail employees’ country of origin and their concern with organizational culture and diversity. The dataset included 151 comments referencing these themes, with North America and Europe accounting for the highest number of comments—a reflection of the larger number of respondents from these regions. To assess the correlation, we created a dummy variable coded as 1 for comments that mention organizational culture and 0 for those that do not. However, the resulting p-value (p = 0.317) indicates that there is no statistically significant relationship between country of origin and the presence of comments related to organizational culture.

The final aspect examined is the temporal evolution of retail employee reviews. Although the data spans from 2008 to 2024, we concentrated our analysis on the years 2021, 2022, and 2023, which correspond to the periods with the highest number of reviews (Figure 1).

The analysis indicated that, on average, rating scores have decreased over time by 3.44%. Table 7 shows that over the three-year period analysed, the median rating remained consistent at 4 in both 2021 and 2023, falling to 3 in 2022. Despite this fluctuation, the variability and underlying distributions exhibited a comparable overall pattern throughout the years. During the entire period considered, ratings fluctuated between 2 and 5, predominantly clustering between 3 and 4 and presenting some outliers with a value of 1, which represents an extreme low rating. Overall, these findings highlighted that, although the general trend remained consistent, the majority of ratings in 2022 were lower compared to the other two years. This decline may be attributed to the fact that 2022 saw the highest volume of reviews, which may have influenced the overall trend.

The three brands with the highest number of reviews by retail employees in the period taken into consideration are Louis Vuitton, Gucci, and Chanel. The analysis shows that there was a gradual decline over time, highlighting an overall negative trend in brand ratings – with a 13.5% decrease in the average brand rating for Louis Vuitton, 5.3% for Gucci, and 2.3% for Chanel over the three-year period.

Our study extends the literature on brand hate by shifting the focus from consumers to employees. The findings suggest that luxury brands are increasingly subject to employee brand hate. This multifaceted phenomenon emerges across experiential, moral, and value-based dimensions, which reflect the evolving dynamics of employee-brand relationships. Negative responses often arise from overlapping causes. For example, work-life balance issues may reflect experiential, deficit-value, and moral avoidance simultaneously.

While prior research has largely examined consumer-driven NWOM and its impact on brand hate (Bryson et al., 2013; Fetscherin, 2019), we provide empirical evidence that employees also engage in NWOM, significantly affecting organizational reputation. The findings confirm that negative employee experiences are key drivers of NWOM, consistent with previous research on workplace dissatisfaction and brand-oriented dysfunctional behaviours (Lee et al., 2022; Raggiotto et al., 2023).

The findings identify limited career progression as the most prominent source of employee dissatisfaction, followed by management-related concerns and poor work–life balance. Career stagnation frequently coincided with dissatisfaction regarding compensation, while hierarchical structures and perceived favouritism further intensified negative employee sentiment and NWOM within the luxury retail sector.

The regression analysis further supports this result, showing a significant but modest negative correlation between employee ratings and the number of workplace issues reported. Each additional concern mentioned in a review corresponds to a lower overall company rating. Longitudinal analysis shows that, although overall ratings have remained relatively stable, a decline occurred from 2021 to 2023. This suggests a worsening employee perception of workplace conditions in recent years, particularly in brands with a high volume of reviews. This trend might underscore a growing dissatisfaction among employees and highlight the urgency for luxury brands to address workplace dissatisfaction before it further impacts employer branding and talent acquisition.

Moreover, our study builds on organizational behaviour theories by demonstrating how hierarchical structures, micromanagement, and nepotism contribute to emotional exhaustion and subsequent brand hate. These results align with previous findings on workplace dissatisfaction and turnover intention (Scroggins, 2008; Lam et al., 2022). Additionally, we highlight the critical role of frontline employees in luxury retail, where high emotional labour and customer interactions intensify the impact of workplace dissatisfaction on brand reputation (Merk and Michel, 2019; Kapferer and Bastien, 2012).

These findings can also be interpreted through the lens of psychological safety, as environments characterized by micromanagement and fear may suppress internal voice while encouraging externalized forms such as NWOM (Edmondson, 1999). Similarly, organizational justice theory explains why perceived inequities in career progression and management practices trigger strong negative emotional responses and brand-directed hostility (Colquitt, 2001).

This study also contributes to the literature on employee engagement and employer branding by illustrating how internal brand dissonance affects both workforce retention and consumer perceptions. The intersection of workplace culture, employee sentiment, and corporate reputation remains an area requiring further academic inquiry, particularly in high-pressure sectors like luxury retail. The statistical analysis also highlights that dissatisfaction with management practices is particularly high among mid-level employees, where the concern over nepotism is almost as prominent as career stagnation.

The integration of sentiment analysis and topic modelling demonstrates the efficacy of computational techniques in extracting meaningful insights from large-scale employee-generated content. Unlike traditional surveys, which may be subject to social desirability bias, online reviews provide a more candid reflection of employee sentiment, making this approach particularly useful for studying sensitive workplace issues.

The findings reveal that employee brand hate is not a monolithic construct but is rooted in specific, identifiable triggers. In particular, limited career progression and management-related concerns appear to erode employer brand equity by transforming employees from brand ambassadors into adversaries engaging in NWOM. The visibility of these narratives on platforms such as Glassdoor may also increase reputational and recruitment-related costs for luxury firms.

The current luxury brand literature typically focuses on positive consumer-brand relationships. Our findings have important managerial implications for luxury brands aiming to reduce employee brand hate and mitigate its negative effects on corporate reputation. Specifically, the findings provide an overview of the main drivers of brand hate, offer insights into how these issues are experienced by employees, and identify key areas for managerial intervention.

This study also supports the argument that wellbeing in the workplace can be considered as a multidimensional construct (Grant et al., 2007). This is consistent with Biétry and Creusier (2015), who identified four key dimensions that shape employees' perceptions of well-being: managerial support, social interactions, workplace environment and work-life balance.

In the context of luxury employee brand hate, managers should focus on improving both functional and emotional benefits to strengthen employee satisfaction and organizational commitment. Prior research highlights the importance of employees feeling respected, valued, and supported in shaping workplace satisfaction (Eisenberger et al., 1986; Halbesleben, 2006). These insights suggest that luxury brands should adopt a more integrated approach to employee wellbeing, focusing especially on the following four key strategic priorities.

Employee Wellbeing and Work Environment: Luxury brands should prioritize employee wellbeing through improved workload management, more predictable scheduling, a more supportive work culture, and healthier day-to-day working conditions. Such measures may improve employee satisfaction, retention, and long-term employer brand equity.

Career Development and Advancement: Organizations should strengthen transparent career pathways, merit-based promotion systems, and internal mobility opportunities to address concerns about career stagnation and favouritism. Clearer progression frameworks may help reduce frustration and strengthen employee commitment to the organization.

Leadership and Management Practices: Reducing micromanagement and strengthening inclusive leadership practices are essential to improving employee autonomy, trust, and organizational commitment. Employees frequently reported frustration with excessive control, limited autonomy, and inconsistent managerial behaviour. Luxury firms should therefore invest in leadership development initiatives that promote fair management practices, emotional intelligence, and supportive supervisory relationships across different organizational levels.

Internal Communication and Employee Voice: Luxury firms should implement stronger internal feedback mechanisms and employee engagement systems to address dissatisfaction before it escalates into external NWOM. The prevalence of negative employee narratives across different seniority levels suggests that dissatisfaction is often systemic rather than isolated. Structured communication channels, employee feedback systems, and internal reporting mechanisms may help organizations identify workplace concerns earlier and strengthen overall organizational trust and engagement.

Despite its contributions, this study has several limitations. First, the data are drawn exclusively from reviews on Glassdoor which may introduce self-selection bias, as employees who choose to post reviews often hold particularly strong opinions—whether positive or negative—compared to the broader workforce. To enhance the robustness of our findings, future research should incorporate additional data sources, such as internal employee feedback surveys. Moreover, online reviews capture expressed narratives rather than directly measured emotional states and may therefore reflect a discursive amplification of negative experiences.

Further investigation is indeed needed into the psychological and emotional factors driving employee dissatisfaction and brand hate, particularly through qualitative research methods such as in-depth interviews. In addition, employees should not be treated as a homogeneous group. According to Lovich et al. (2025), “nearly all organizational attempts at employee value delivery treat employees like a monolithic group, rarely recognizing segments with differentiated characteristics and needs.” Our research supports the importance of segmenting employees, for example by functional role, to identify the reasons for employee hate. Comparative studies across different retail sectors could help determine whether the identified antecedents of brand hate are unique to luxury retail or more widely applicable. Given the distinctive characteristics of luxury companies, the generalizability of these findings to other sectors may be limited.

This study also opens several avenues for future research. From a statistical perspective, the OLS model explains 17.5% of the variance using only two independent variables, suggesting that a substantial proportion of variability remains unaccounted for. Future research could enhance predictive accuracy by incorporating additional variables or employing alternative analytical approaches.

Finally, future research should integrate qualitative employee sentiment with internal organizational metrics to measure the impact of employee’ NWOM on financial outcomes. Specifically, researchers could collaborate with luxury brands to map drivers of Brand Hate directly against measurable economic outcomes, such as precise employee turnover rates, recruitment costs per hire, and longitudinal changes in employer brand equity. Such studies would provide a more granular understanding of the economic value of employee dissatisfaction and offer a financial justification for investments in internal brand management.