The Effect of Generative AI-Based Digital Marketing Communication Personalization on Customer Trust, Customer Experience, and Repurchase Intention: Evidence from Iranian Online Store Customers

Introduction

Rapid developments in digital technologies and artificial intelligence have changed the logic of marketing and customer experience management. Generative artificial intelligence has become especially relevant because it can produce text, images, recommendations, and conversational responses that are adapted to individual users. Unlike traditional predictive systems, generative AI can support dynamic interaction and content creation, allowing brands to communicate with customers in ways that appear more responsive, personalized, and human-like (Grewal et al., 2025; Huang & Rust, 2025).

In online retail, personalization refers to the adjustment of marketing messages, product recommendations, content, and interaction style according to customer characteristics, preferences, and behavioral data. In a generative AI context, personalization is not limited to repeating previously observed patterns; it can also generate new content and real-time responses that correspond to customers' immediate needs. This capability can reduce cognitive effort, improve perceived relevance, and strengthen the customer's sense that the platform understands individual preferences (Lemmens et al., 2025).

Customer trust is a central condition in this process. AI-based interactions can increase perceived convenience and value, but they can also raise concerns about authenticity, data security, transparency, and accountability. If customers do not trust the platform or the AI-mediated communication process, even a positive online experience may not translate into repurchase intention. Trust is therefore not only an outcome of personalization but also a mechanism that can shape the strength of the relationship between customer experience and future behavioral intention (Belanche et al., 2025; Benk et al., 2025; Huynh & Aichner, 2025; Lu & Zhang, 2025; McGrath et al., 2024).

In the Iranian online retail context, this issue is practically important. Although online stores have grown rapidly, many platforms still rely on relatively simple behavioral personalization. Promotional messages are often designed for broad customer groups and may not sufficiently reflect the unique needs, style, and expectations of individual customers. At the same time, exposure to global platforms and AI tools has increased customer expectations regarding relevance, immediacy, and natural interaction. Similar evidence from digital advertising and e-commerce shows that perceived relevance, usefulness, trust, and personalized recommendations can strengthen purchase and loyalty-related outcomes (An & Ngo, 2025; Hassan et al., 2025).

The present study addresses this gap by examining how generative AI-based personalization of digital marketing communication affects customer trust, customer experience, and repurchase intention among Iranian online store customers. The theoretical basis of the study draws on perceived value theory, which explains customer evaluations in terms of perceived benefits relative to costs; expectation-confirmation theory, which links confirmation of expectations and satisfaction to continued use; diffusion of innovations theory, which explains adoption through perceived innovation attributes such as relative advantage and compatibility; and recent e-commerce customer-experience research, which emphasizes interaction quality, relevance, trust, and perceived usefulness in digital purchase behavior (Hermann & Puntoni, 2024; Muhammad et al., 2024; Nguyen et al., 2025; Oliver, 1980; Pires et al., 2024; Rogers, 2003; Tan et al., 2024; Zeithaml, 1988). Accordingly, the conceptual model of this study was designed based on the theoretical foundations and prior empirical findings. The hypotheses, consistent with the conceptual model, are as follows: H1: generative AI-based personalization of digital marketing communication increases customer trust; H2: generative AI-based personalization improves customer experience; H3: generative AI-based personalization increases repurchase intention; H4: customer experience positively affects customer trust; H5: customer trust positively affects repurchase intention; H6: customer experience positively affects repurchase intention; and H7: customer trust moderates the relationship between customer experience and repurchase intention.

Methods

This study was applied in purpose and descriptive-survey in design. It was conducted cross-sectionally to examine the effects of generative AI-based personalization of digital marketing communication on customer trust, customer experience, and repurchase intention. Structural equation modeling (SEM) was used to test the proposed relationships among latent variables, and the analyses were conducted using SPSS and LISREL 8.8.

The statistical population consisted of customers of online stores in Iran who had prior online purchasing experience. Because a complete sampling frame of online store customers was not available, non-probability convenience sampling was used. The questionnaire link was distributed through social networks and text messages. After screening and removing incomplete responses, 384 valid questionnaires were retained for final statistical analysis.

The data collection instrument was a researcher-developed questionnaire based on the theoretical framework and previous studies. The first section included five demographic questions covering gender, age, education, online purchase experience, and familiarity with online stores. The second section included 39 Likert-type items: 12 items for generative AI-based personalization of digital marketing communication, 8 items for customer trust, 10 items for customer experience, and 9 items for repurchase intention. All items were scored on a five-point Likert scale ranging from 1 = strongly disagree to 5 = strongly agree.

Content validity was reviewed by academic supervisors and management experts. The adequacy of the data for factor analysis was evaluated using the Kaiser-Meyer-Olkin (KMO) statistic. Reliability was assessed using Cronbach's alpha. The measurement and structural models were examined using LISREL fit indices and standardized estimates. The SEM procedure was then used to test the structural paths. The moderation hypothesis was tested by entering the interaction term between customer experience and customer trust into the structural model.

Results

Reliability, Sampling Adequacy, and Normality

KMO values were above .70 for all constructs, indicating adequate sampling adequacy. Cronbach's alpha coefficients were above .80 for all constructs and .968 for the total instrument, indicating strong internal consistency. Kolmogorov-Smirnov results did not indicate substantial univariate normality problems for the observed composite variables.

Descriptive Statistics

The descriptive findings showed that the mean values of all research variables were above the theoretical midpoint of 3. This indicates that respondents generally agreed with the questionnaire items and perceived the constructs as important in online shopping interactions.

Model Fit and Structural Equation Modeling

The SEM results indicated that model-fit assessment was based on the LISREL output available for this manuscript. The measurement models were reported using chi-square divided by degrees of freedom. For the final structural model, the output reported χ² = 919.72 and df = 698; therefore, χ²/df = 1.31. This value is within the commonly accepted range for acceptable model fit and supports interpretation of the structural model together with the other fit indices. The remaining reported indices for the drawn structural model were GFI = .95, AGFI = .98, CFI = 1.00, and RMSEA = .000. Accordingly, the model-fit results indicate acceptable overall fit and should be interpreted alongside the reported standardized path estimates.

Note. The first fit column reports χ²/df. For the drawn structural model, χ² = 919.72 and df = 698; therefore, χ²/df = 1.31.

Hypothesis Testing

The path analysis results showed that all proposed relationships were statistically significant. Generative AI-based personalization of digital marketing communication had a strong positive effect on customer trust (H1: β = .85, t = 3.47), customer experience (H2: β = .79, t = 3.80), and repurchase intention (H3: β = .73, t = 3.80). Customer experience had a positive effect on customer trust (H4: β = .56, t = 4.08), customer trust had a positive effect on repurchase intention (H5: β = .52, t = 3.99), and customer experience had a positive effect on repurchase intention (H6: β = .57, t = 4.78).

To examine the moderating role of customer trust in the relationship between customer experience and repurchase intention, the interaction term between customer experience and customer trust was entered into the structural model. The interaction coefficient was positive and significant (H7: β = .29, t = 4.08), indicating that trust strengthens the effect of customer experience on repurchase intention. In other words, a positive customer experience is more likely to lead to repurchase intention when customer trust is high.

Discussion

The findings show that generative AI-based personalization of digital marketing communication plays an important role in shaping trust, customer experience, and repurchase intention among customers of Iranian online stores. The positive effect of personalization on trust suggests that customers are more likely to trust an online store when its messages and recommendations appear relevant to their personal needs and preferences. In an environment where most interactions are remote and mediated by digital interfaces, the feeling of being recognized by the platform can partly substitute for direct interpersonal interaction and reduce perceived uncertainty. This is consistent with recent evidence that AI-enabled personalization and trust-related mechanisms are central to purchase intention and loyalty in digital environments (An & Ngo, 2025; Hassan et al., 2025).

The effect of personalization on customer experience was also strong. Personalized content, recommendations, and interaction paths can reduce the mental effort required for online decision-making. When customers receive relevant messages and product suggestions, the shopping process becomes more fluent, meaningful, and engaging. This finding supports the view that customer experience in generative AI-mediated environments is not only determined by interface quality but also by the perceived relevance, responsiveness, and appropriateness of the interaction (Huang & Rust, 2025; Nguyen et al., 2025; Pires et al., 2024).

The direct effect of personalization on repurchase intention indicates that personalization does not only influence perceptions and attitudes; it can also shape future behavioral tendencies. When customers believe that an online store understands their preferences and provides useful recommendations, they are more inclined to return to the platform. In this sense, generative AI-based personalization functions as a relationship-building mechanism rather than merely as a promotional tool, consistent with perceived value and repurchase-intention research (Muhammad et al., 2024; Tan et al., 2024; Zeithaml, 1988).

The moderation finding is central to the interpretation of the model. Customer trust strengthened the effect of customer experience on repurchase intention. This means that even when customers have a positive experience, the experience is less likely to convert into stable repurchase intention if trust is low. In generative AI-mediated online retail, trust therefore acts as a bridge between experience and behavioral continuity. This is particularly relevant in markets where customers may be concerned about data privacy, algorithmic transparency, content authenticity, and platform accountability (Benk et al., 2025; Huynh & Aichner, 2025; Lu & Zhang, 2025; McGrath et al., 2024).

The results have practical implications for online retailers. First, personalization should be designed as a transparent and customer-centered process rather than as hidden persuasion. Second, online stores should explain why specific recommendations or messages are offered when possible, because explainability can improve customers' perceived control and trust. Third, managers should treat trust, privacy, and data security as strategic elements of customer experience rather than as purely technical or legal concerns. Finally, generative AI should be used to create real customer value through relevant content, easier decision-making, and more meaningful interaction.

This study has limitations. The data were collected through self-report questionnaires, so customers' reported perceptions may differ from their actual behavior on online platforms. The cross-sectional design also prevents conclusions about changes in trust and experience over time. In addition, the study was conducted in the Iranian online retail context; therefore, cultural and market conditions may limit the generalizability of the findings. Because all variables were measured using the same self-report instrument at one time point, possible common method bias should also be considered. The summarized output available for this manuscript did not include all item-level CFA statistics, such as standardized factor loadings, composite reliability, and average variance extracted; therefore, the measurement results should be interpreted with this reporting limitation in mind. Future research should test the model longitudinally, compare different product categories, and examine how transparency, privacy assurance, and explainable AI features influence the conversion of personalized experiences into repurchase behavior.

Ethics Considerations

Participation in the survey was voluntary. The questionnaire was distributed electronically, and informed consent was obtained before participation. No personally identifiable information was collected or reported in the manuscript, and only completed and valid responses were used in the final analyses. Data were analyzed in aggregate form and used only for academic research purposes. The study involved an anonymous, minimal-risk online survey and followed standard ethical principles for human-subject survey research.

Conflict of Interest

The authors declare that they have no conflict of interest.

Author Contributions

Homa Doroudi contributed to conceptualization, supervision, methodology, manuscript review, and final approval. Seyedeh Aynaz Jabbari contributed to literature review, instrument preparation, data collection, data organization, and drafting of the manuscript. Yasser Heidari contributed to data analysis, interpretation of findings, methodological checking, manuscript editing, and final approval. All authors read and approved the submitted version.

Declaration of Generative AI Use

Generative AI tools were used for language translation, grammar editing, and formatting assistance. No AI tool was used to generate, alter, or fabricate data, statistical outputs, research findings, or citations. The authors reviewed and approved the final manuscript and remain fully responsible for the accuracy, integrity, and content of the work.