Artificial intelligence (AI) is being increasingly used in retail to automate service delivery and improve personalization, and one of the goals of using AI is to strengthen customer loyalty. In this paper, we report findings from a small exploratory study in the United Arab Emirates (UAE) that examines how retail consumers relate AI-enabled experiences to their loyalty intentions. The study proposes and pilots a structured survey instrument capturing AI awareness, perceived personalization, perceived value, service convenience, trust, perceived intrusiveness, and loyalty in UAE retail settings. Survey data were collected from 25 respondents and used to assess directional associations between these constructs. Contrary to widely held expectations, the results show that more positive evaluations of AI-enabled retail services were not associated with higher loyalty; in several cases, the observed relationships were weak, non-significant, or negative. Trust levels were also generally modest, suggesting that functional improvements enabled by AI may not translate into stronger relational outcomes in consumer-retailer relationships. While the small sample size limits generalization and the study should be interpreted as an early pilot, the findings raise important questions about the assumed loyalty benefits of AI adoption in retail. The paper contributes early evidence from a UAE context and highlights trust and human-centered service design as central considerations for future large-scale research.
Keywords: Artificial intelligence in retail; Customer loyalty; Trust and privacy; Service automation; Human-centric AI.
Introduction
Artificial intelligence (AI) is now part of the competitive strategies adopted by most retailers. It supports personalization, automates parts of customer service, and enables data-driven decision making at scale. AI recommendation systems, chatbots, and predictive analytics are widely used to improve efficiency and customer experience, and retailers often assume that these improvements will naturally lead to stronger customer loyalty (Shankar, 2018; Davenport et al., 2020). In other words, AI is increasingly treated not only as a tool for operations, but also as a way to strengthen long-term customer relationships. In addition to personalization and service automation, AI is increasingly used for customer sentiment and insight generation from digital interactions, further strengthening its role in data-driven customer understanding (Bibi et al., 2025).
Yet the link between AI adoption and loyalty is not as clear as industry narratives suggest. AI-driven personalization and automation can improve convenience and perceived value, but loyalty is not purely functional. It is also shaped by relational and emotional factors that technology may not fully replicate (Huang & Rust, 2018; Prentice et al., 2020). Consumers may appreciate speed and personalization, while still feeling uneasy about how their data are used, how transparent AI decisions are, or whether the shopping experience is becoming too impersonal. Studies increasingly highlight the role of trust, privacy concerns, and the perceived loss of human interaction in shaping consumer responses to AI-enabled services (Van der Aa et al., 2021; Puntoni et al., 2021). This creates a broader tension in AI-driven retail, often framed as a personalizationâprivacy and automationâhumanity trade-off (Rosado-Pulido et al., 2023).
Despite growing research in this area, two issues remain underexplored. First, much of the literature focuses on adoption, satisfaction, or short-term service outcomes, while the assumed translation of AI improvements into customer loyalty is examined less consistently. Second, trust-related concerns such as perceived intrusiveness are frequently discussed as important drivers of acceptance, but they are not always measured alongside loyalty outcomes in empirical studies. These gaps matter because AI may raise service performance while failing to strengthen the relational foundations that loyalty depends on.
These questions are especially relevant in rapidly digitalizing retail environments such as the United Arab Emirates (UAE). The UAE retail sector has high levels of technological investment and strong competitive pressure, alongside a culturally diverse consumer base. At the same time, there is increasing public and regulatory attention to data protection and responsible AI use. While AI adoption in the region continues to accelerate, there remains limited consumer-centered evidence on how AI-enabled retail experiences relate to loyalty in practice.
Against this background, the present study reports an exploratory pilot investigation of AI-driven retail experiences and customer loyalty in the UAE. The study proposes and pilots a structured survey instrument capturing AI awareness, perceived personalization, perceived value, service convenience, trust, perceived intrusiveness, and loyalty. Using descriptive and exploratory analysis, the study examines whether the commonly assumed positive relationship between AI-enabled retail services and loyalty is reflected in customer perceptions, and whether trust-related tensions emerge as a potential explanation. Given the pilot nature of the study and its small sample size, the results are not intended to support causal claims. Instead, the aim is to provide early empirical signals that can inform future large-scale studies and help refine how loyalty is conceptualized in AI-enabled retail.
The contribution of this conference paper is threefold. First, it provides early empirical insight into AIâloyalty dynamics within an underexplored regional context. Second, it places trust and perceived intrusiveness at the center of the analysis of consumer responses to AI-enabled retail services. Third, by reporting counterintuitive pilot findings, the study contributes to ongoing debates on the limits of AI-driven personalization and automation in fostering sustainable customer loyalty, while also motivating future theory-refining research.
Related Work
- AI-Driven Personalization and Perceived Value
A substantial body of literature positions AI-driven personalization as a key mechanism through which retailers enhance customer experience and loyalty. Studies consistently report that AI-enabled recommendation systems and data-driven personalization can improve perceived relevance, efficiency, and convenience of retail interactions, thereby strengthening engagement and satisfaction (Shankar, 2018; Davenport et al., 2020). From this perspective, personalization is treated as a value-creation tool that aligns offerings with individual customer preferences at scale.
However, empirical evidence also highlights important limitations to this assumption. Prentice et al. (2020) demonstrate that personalization contributes to engagement primarily when it is perceived as contextually appropriate and complemented by high-quality human service. Similarly, Rosado-Pulido et al. (2023) identify a personalizationâprivacy paradox, whereby increased personalization enhances perceived value while simultaneously intensifying privacy concerns that erode trust and loyalty intentions. These findings suggest that personalization alone is insufficient to guarantee loyalty and may become counterproductive when perceived as intrusive or misaligned with consumer expectations.
Despite these insights, much of the existing literature continues to treat personalization as a predominantly positive driver of loyalty, often assuming linear relationships between personalization intensity and relational outcomes. There is limited empirical work examining situations in which perceived value gains from AI personalization coexist with declining loyalty, particularly in culturally diverse and privacy-sensitive retail environments.
- AI-Based Service Automation and Customer Experience
Beyond personalization, AI-driven service automation, particularly through chatbots and virtual assistants, has been widely studied as a means of improving service quality and operational efficiency. Research in hospitality and retail contexts indicates that AI-based service tools can enhance responsiveness, availability, and process convenience, contributing positively to customer satisfaction and usage intentions (Pillai & Sivathanu, 2020; Ameen et al., 2021).
At the same time, several studies caution that service automation introduces risks related to empathy deficits, inflexibility, and service failure. Pillai and Sivathanu (2020) show that while perceived usefulness and ease of use support adoption, chatbot failures or lack of emotional sensitivity can undermine trust and negatively affect loyalty intentions. Huang and Rust (2018) further argue that AI excels at mechanical and analytical service tasks but lacks emotional intelligence, which remains critical for relational loyalty formation. As a result, fully automated service encounters may weaken emotional bonds even when efficiency improves.
The literature thus reflects a growing consensus that hybrid service models combining AI efficiency with human intervention are more likely to sustain long-term customer relationships. However, empirical studies rarely investigate whether customers who rate AI-based service quality highly also exhibit stronger loyalty, or whether positive service evaluations may coexist with relational disengagement.
- Trust, Privacy, and Ethical Concerns in AI-Enabled Retail
Trust is repeatedly identified as a central mediating mechanism in AI-driven customer experiences. Van der Aa et al. (2021) provide empirical evidence that trust mediates the relationship between AI capabilities and loyalty outcomes, emphasizing that efficiency and personalization gains do not translate into loyalty when trust is compromised. Similarly, Puntoni et al. (2021) highlight consumer concerns related to autonomy, control, transparency, and data usage, which can reduce trust and loyalty intentions despite improvements in convenience.
Retail-focused studies reinforce this view by demonstrating that perceived risk and lack of transparency can offset the benefits of smart retail technologies (Ameen et al., 2021). Davenport et al. (2020) further argue that ethical data practices and transparency are strategic prerequisites for sustainable AI-driven marketing outcomes. Collectively, these studies suggest that trust is not a secondary consideration but a foundational condition for AI-enabled loyalty.
Nevertheless, while trust is frequently theorized as a mediator, empirical findings on its role remain mixed, and many studies assume trust functions uniformly across contexts. There is limited evidence on how trust operates in emerging or rapidly digitalizing markets, where regulatory frameworks, cultural norms, and consumer expectations regarding data usage may differ substantially from Western contexts.
- Gaps and Positioning of the Current Study
Across the literature, three dominant assumptions are evident. First, AI-driven personalization and service automation are generally expected to enhance loyalty through improved value and convenience. Second, trust is acknowledged as a critical mediator, yet often treated as a stable construct rather than a fragile or contested one. Third, most empirical studies focus on mature markets, with limited attention to region-specific dynamics.
This study positions itself at the intersection of these gaps by empirically examining AI-enabled loyalty relationships in the UAE retail context through an exploratory pilot approach. Rather than assuming positive linear effects, the study allows for the possibility of counterintuitive associations between AI-related constructs and loyalty outcomes. By doing so, it responds to calls in the literature to critically examine the limits of AI-driven personalization and automation, particularly in environments where trust, privacy, and human-centric service expectations are salient (Huang & Rust, 2018; Rosado-Pulido et al., 2023).
Methods and Data
This study uses a quantitative, cross-sectional exploratory design to examine how consumers in the UAE perceive AI-enabled retail services and how these perceptions relate to customer loyalty. The work is framed as a pilot study. Its purpose is to surface early directional patterns, assess how the constructs behave in a real survey setting, and identify potential tensions (for example around trust and privacy) rather than to test causal relationships or produce generalizable estimates. A survey was used to capture self-reported perceptions of AI-enabled retail interactions at a single point in time, which fits the studyâs exploratory scope.
- Sample and data collection
Data were collected through a structured online questionnaire administered to adults (18+) residing in the UAE who reported recent exposure to AI-enabled retail services (such as recommendations, automated support, or AI-assisted service features). Due to practical constraints related to access, time, and resources, recruitment followed a non-probability approach combining purposive and snowball sampling. The purposive element ensured relevance (respondents had experience with AI-enabled retail), while snowball sharing helped reach additional participants within similar consumer networks.
The final pilot dataset includes 25 usable responses. Given the sampling approach and small sample size, results are treated as exploratory and indicative. Demographic variables (including age, gender, and shopping frequency) are reported descriptively to provide context, not to claim representativeness of the wider UAE retail population.
Ethical procedures were incorporated into the questionnaire. Participants were informed about the purpose of the study, that participation was voluntary, and that responses were anonymous. No personally identifiable information was collected.
The survey instrument consisted of 20 items organized around the study constructs. Perceptual items were measured using a five-point Likert scale (1 = Strongly disagree to 5 = Strongly agree), while demographic items used categorical response options. The questionnaire captured the following constructs:
- AI awareness: familiarity with AI features used by retailers
- Perceived personalization and perceived value: perceived relevance of recommendations and value of offers
- Service convenience and AI-enabled service quality: perceived ease and usefulness of AI-supported retail services (including automated support where applicable)
- Trust and privacy concern: confidence in retailersâ data handling and perceived privacy-related risk
- Customer loyalty: repurchase intention and advocacy-related intentions
Where constructs were measured using multiple items, composite scores were created by averaging the relevant items. Items reflecting privacy concern were reverse-coded where needed so that higher composite values consistently reflected more positive evaluations (for example higher trust or lower perceived risk). The instrument was developed by mapping items to the study focus and was reviewed for clarity prior to distribution.
Analysis was conducted in SPSS using an exploratory workflow. First, the data were screened and summarized using descriptive statistics to profile the sample and examine the distributions and central tendencies of key constructs. Internal consistency of multi-item composites was assessed using Cronbachâs alpha, with the recognition that reliability estimates can fluctuate substantially in small pilot samples.
Bivariate correlations were then used to examine directional associations between AI-related constructs and customer loyalty. Multiple linear regression was conducted in an exploratory manner to examine the combined association of AI awareness, perceived value, service quality, and trust with loyalty outcomes. Simple group comparisons (t-tests) and descriptive comparisons across age categories were also used to explore whether patterns differed across participant subgroups.
Given the pilot nature of the study, results are interpreted cautiously. The emphasis is placed on directionality, construct behavior, and potential tensions rather than on statistical confirmation. The intent is to inform instrument refinement and guide the design of future larger-scale studies. Given the small pilot sample, all statistical analyses are used strictly for exploratory and descriptive purposes. In particular, the correlation and regression results are retained only as preliminary diagnostic checks to identify possible directional patterns and measurement issues, not as confirmatory evidence of stable construct-level relationships.
Results
The pilot sample consists of 25 UAE retail consumers with recent exposure to AI-enabled retail services. The age distribution is concentrated in younger and middle-aged groups, with 28% aged 18â25, 36% aged 26â35, and 20% aged 36â45, while smaller proportions fall in the 46â55 and 55+ categories. The sample includes 56% male and 44% female respondents. In terms of nationality, the sample reflects the UAEâs demographic diversity, comprising Arab expatriates (48%), Asian expatriates (32%), UAE nationals (12%), and Western expatriates (8%). Most respondents reported frequent retail engagement, with 52% shopping monthly and 24% shopping weekly.
These characteristics indicate that the sample represents active retail consumers with varied demographic backgrounds, while remaining limited in size and representativeness due to the pilot design.
- Descriptive Statistics of Key Constructs
Descriptive analysis shows high awareness of AI in retail contexts (mean = 4.28, SD = 0.84), suggesting that respondents are generally familiar with AI-driven features. Perceptions of personalization and perceived value are moderate, with a personalization mean of 3.32 (SD = 0.85) and offer value mean of 3.76 (SD = 0.93).
Perceived service convenience scores relatively high (mean = 4.16, SD = 0.80), whereas chatbot satisfaction is lower and more variable (mean = 3.20, SD = 1.12), indicating mixed evaluations of AI-based customer service interactions.
Trust-related measures reveal comparatively lower scores. Data trust has a mean of 2.88 (SD = 1.24), while privacy concern is moderate (mean = 3.44, SD = 1.00). After reverse coding privacy concern, the trust composite remains low (mean = 2.72, SD = 0.77), suggesting persistent consumer unease regarding data usage.
Customer loyalty indicators are moderate, with repurchase intention (mean = 3.40, SD = 1.00), brand advocacy (mean = 3.72, SD = 0.89), and an overall loyalty composite mean of 3.56 (SD = 0.67).
Reliability analysis produced negative Cronbachâs alpha values for both the Trust (Îą = â0.165) and Loyalty (Îą = â0.019) composites. These results indicate substantial measurement instability, likely attributable to the very small sample size, reverse-coded items, and potential multidimensionality of the constructs. As such, reliability coefficients are not interpreted as evidence of scale quality but rather as diagnostic indicators highlighting the need for scale refinement and validation in future studies.
Correlation analysis reveals several counterintuitive associations. Service quality is significantly and negatively correlated with loyalty (r = â0.489, p = 0.013). Perceived value also shows a negative association with loyalty (r = â0.375, p = 0.065), approaching conventional significance thresholds. Trust exhibits a negative but non-significant correlation with loyalty (r = â0.190, p = 0.362), while AI awareness is also negatively related to loyalty without statistical significance (r = â0.217, p = 0.298).
These findings indicate that, within this pilot sample, higher evaluations of AI-related service attributes do not correspond to higher loyalty scores.
- Exploratory Regression Analysis
An exploratory multiple regression model was estimated with AI awareness, perceived value, service quality, and trust as predictors of customer loyalty. The overall model is statistically significant (F = 4.224, p = 0.012), explaining 45.8% of the variance in loyalty (R² = 0.458). However, the direction of relationships contrasts with dominant theoretical expectations.
Both perceived value (β = â0.465, p = 0.013) and service quality (β = â0.546, p = 0.006) show significant negative associations with loyalty. AI awareness (β = â0.062, p = 0.729) and trust (β = â0.016, p = 0.927) are not significant predictors. Given the pilot nature of the dataset and measurement limitations, these results are interpreted cautiously as indicative of unexpected relational patterns rather than confirmatory evidence.
- Group Comparisons and Visual Patterns
Independent-sample t-tests indicate no statistically significant gender differences in loyalty or trust. Descriptive comparisons across age groups suggest variability in loyalty, with higher mean loyalty among respondents aged 46â55 and lower mean loyalty among those aged 55+, although small cell sizes limit interpretability. Visual inspection of scatterplots and boxplots further reinforces the observed negative association between perceived value and loyalty and highlights demographic dispersion in loyalty scores.
Discussion
This exploratory pilot study examined whether the commonly assumed positive link between AI-enabled retail experiences and customer loyalty is reflected in consumer perceptions in the UAE context. Overall, the results suggest that this assumption does not necessarily hold. In particular, the negative associations observed between perceived service quality, perceived value, and loyalty contrast with the dominant narrative that AI-driven efficiency and personalization naturally strengthen customer relationships.
One plausible interpretation is that consumers can evaluate AI-enabled retail features positively at a functional level while still remaining weakly attached to the retailer. In other words, higher convenience, smoother service, or more relevant offers may improve the transaction without strengthening relational loyalty. This distinction between functional satisfaction and emotional attachment is consistent with arguments in prior research that AI performs well in analytical and mechanical tasks but may struggle to support the relational and emotional foundations associated with long-term loyalty, especially when human interaction is reduced or perceived as replaceable. In such settings, consumers may acknowledge service improvement but feel less connected to the brand itself.
Trust also appears central to interpreting the findings. Trust-related measures were relatively low in the pilot sample, and trust did not emerge as a positive predictor of loyalty. This pattern suggests that AI-enabled services may be experienced within a broader context of skepticism toward data use and automated decision-making. Even when AI-based features work effectively, concerns about privacy, transparency, and control may inhibit the translation of service convenience into loyalty. Existing literature often treats trust as a necessary condition for AI-driven personalization to generate relational outcomes; without that foundation, efficiency gains may remain transactional rather than loyalty-building.
The UAE retail environment provides a relevant lens for understanding these dynamics. The market combines rapid digitalization and strong retail competition with a culturally diverse consumer base and increasing attention to data protection and ethical technology use. In such a context, AI implementations that are perceived as overly automated, opaque, or intrusive may generate ambivalence or resistance rather than loyalty. The pilot findings therefore support the broader view that AI strategies focused primarily on hyper-personalization or operational efficiency can create relational risk if they are not aligned with customer expectations regarding transparency, choice, and human engagement.
At the same time, these findings must be interpreted carefully. The study is exploratory, based on a very small non-probability pilot sample, and reliability estimates for some composites were unstable. These conditions place clear limits on interpretation and do not support confirmatory claims. In this sense, the study should be read primarily as a pilot and instrument-diagnostic exercise intended to identify early patterns, possible tensions, and areas requiring further measurement refinement. Although negative directional relationships were observed across the exploratory analyses, these should be treated as preliminary signals requiring verification in larger and more stable samples.
Overall, the study contributes early context-specific insight into how AI-enabled retail features may be perceived in relation to loyalty in the UAE, while also highlighting the methodological challenges of examining these relationships in a pilot setting. Rather than concluding that AI-driven personalization or service quality reduces loyalty, the present findings more cautiously suggest that functional AI-enabled improvements may not automatically translate into stronger relational loyalty under conditions of low trust, privacy concern, or limited human engagement. Practically, this points to the importance of hybrid service models, transparent data practices, and explicit trust-building mechanisms alongside AI deployment. From a research perspective, the study motivates larger-scale investigations with refined measures to examine these relationships more robustly.
Limitations and future research
This study is subject to several important limitations that should be considered when interpreting the findings. First, the analysis is based on a very small pilot sample (n = 25) collected using non-probability sampling. As a result, the findings are not intended to be generalizable to the broader UAE retail population. Rather, the sample is used to support an exploratory pilot objective focused on identifying early patterns, potential tensions, and measurement issues that can inform subsequent research design.
Second, the small sample size implies low statistical power, limited precision, and instability in estimated relationships. It also increases the risk that regression-based patterns may reflect sample-specific variation rather than robust underlying associations. For this reason, all inferential results in the study should be interpreted as tentative and descriptive rather than confirmatory.
Third, the cross-sectional design captures customer perceptions at a single point in time. Given the rapidly evolving nature of AI technologies and consumer expectations, longitudinal designs would be better suited to examine how trust, perceived value, and loyalty evolve as customers gain prolonged exposure to AI-enabled retail services. The present results should therefore be interpreted as indicative of contemporaneous associations rather than stable relationships.
Fourth, measurement instability was observed in composite constructs, particularly trust and loyalty. This raises concerns regarding construct consistency in the present pilot dataset and suggests that some items, including reverse-coded items, may require further refinement, testing, or re-specification in future work. The present study therefore serves partly as an instrument-testing stage, helping identify areas where construct definitions, item wording, and dimensional structure require improvement before larger-scale validation.
Future research should build on this pilot by employing larger and more diverse samples to improve robustness and generalizability. Future studies should also refine and validate the measurement instrument through item review, reverse-code verification, scale refinement, and construct validation procedures such as factor analysis where sample size permits. In addition, qualitative methods such as interviews or focus groups could help explain why AI-enabled convenience or personalization may not translate directly into loyalty at the relational level. More broadly, future research should examine trust, perceived intrusiveness, transparency, and human-AI service balance as central mechanisms shaping AI-loyalty relationships in culturally diverse and regulation-sensitive retail environments such as the UAE.
Conclusion
This study offers an exploratory pilot examination of how AI-driven retail features relate to customer loyalty in the UAE. In this pilot sample, the findings suggest that higher perceived AI-enabled service quality and value did not clearly correspond to stronger loyalty. Instead, the results point to a potential disconnect between functional efficiency delivered by AI and the relational foundations of customer loyalty, particularly in contexts where trust, transparency, and human-centric interaction matter.
While preliminary, these findings contribute to ongoing debates on the limits of AI-driven personalization and service automation in retail. The study highlights the importance of aligning AI deployment with trust-building practices and hybrid service models that preserve meaningful human engagement. Future research using larger samples and refined measures can further clarify the conditions under which AI supports, rather than undermines, sustainable customer loyalty.
References
Ameen, N., Tarhini, A., Reppel, A., & Anand, A. (2021). Customer experiences in smart retailing. Technological Forecasting and Social Change, 173, 121130. https://doi.org/10.1016/j.techfore.2021.121130
Bibi, M., Aziz, W., Khan, I. R., & Nadeem, M. S. A. (2025). Twitter sentiment analysis using ensemble of sentic computing and a novel clustering optimisation algorithm based on combined proximity measures. Journal of Information & Knowledge Management, 24(6), 2550072. https://doi.org/10.1142/S0219649225500728
Davenport, T. H., Guha, A., Grewal, D., & Bressgott, T. (2020). How artificial intelligence will change the future of marketing. Journal of the Academy of Marketing Science, 48(1), 24â42. https://doi.org/10.1007/s11747-019-00696-0
Grewal, D., Roggeveen, A. L., & Nordfält, J. (2017). The future of retailing. Journal of Retailing, 93(1), 1â6. https://doi.org/10.1016/j.jretai.2016.12.008
Huang, M. H., & Rust, R. T. (2018). Artificial intelligence in service. Journal of Service Research, 21(2), 155â172. https://doi.org/10.1177/1094670517752459
Pillai, R., & Sivathanu, B. (2020). Adoption of AI-based chatbots for hospitality and tourism. International Journal of Contemporary Hospitality Management, 32(10), 3199â3226. https://doi.org/10.1108/IJCHM-04-2020-0259
Prentice, C., Dominique Lopes, S., & Wang, X. (2020). The impact of artificial intelligence and employee service quality on customer engagement and loyalty. Journal of Hospitality Marketing & Management, 29(7), 803â822. https://doi.org/10.1080/19368623.2020.1722304
Puntoni, S., Reczek, R. W., Giesler, M., & Botti, S. (2021). Consumers and artificial intelligence. Journal of Marketing, 85(1), 7â32. https://doi.org/10.1177/0022242920953847
Rosado-Pulido, J., MuĂąoz-Leiva, F., & LiĂŠbana-Cabanillas, F. (2023). The paradox of personalization and privacy in AI-driven retailing. Journal of Business Research, 156, 113124. https://doi.org/10.1016/j.jbusres.2022.113124
Shankar, V. (2018). How artificial intelligence (AI) is reshaping retailing. Journal of Retailing, 94(4), 1â5. https://doi.org/10.1016/j.jretai.2018.10.002
Van der Aa, Z., Van Kollenburg, G., & Van Loenen, E. (2021). The role of trust in AI-driven customer experiences. Journal of Consumer Behaviour, 20(3), 345â358. https://doi.org/10.1002/cb.1874
