University of Isfahan New Marketing Research Journal 2228-7744 15 3 2025 11 22 Explaining a Conceptual Model of Artificial Intelligence Applications in Digital Marketing with an Emphasis on Enhancing Consumer Loyalty: A Mixed-Methods Approach Explaining a Conceptual Model of Artificial Intelligence Applications in Digital Marketing with an Emphasis on Enhancing Consumer Loyalty: A Mixed-Methods Approach 105 132 29826 10.22108/nmrj.2025.145405.3194 FA Afshin Alipour Assistant professor, Department of Business Management, Faculty of Management and Industrial Engineering, Malek Ashtar University of Technology, Tehran, Iran Ali Naeimi Khondabi Master's student, Department of Business Management, Faculty of Management and Industrial Engineering, Malek Ashtar University of Technology, Tehran, Iran Mohammadreza Zolghadr Master's student, Department of Business Management, Faculty of Management and Industrial Engineering, Malek Ashtar University of Technology, Tehran, Iran Journal Article 2025 05 27 This research aimed to design, validate, and elucidate a conceptual model for harnessing Artificial Intelligence (AI) in digital marketing, specifically to enhance consumer loyalty. The study was conducted in 4 structured phases. First, key components were identified through a systematic literature review (meta-synthesis) of 53 academic sources. In the second phase, the fuzzy Delphi method was utilized with 10 industry experts to validate the relevance of the content and achieve expert consensus. The third phase employed Interpretive Structural Modeling (ISM) to analyze and structure the causal relationships among 26 identified components. Finally, MICMAC analysis was used to categorize these components based on their driving power and dependence. The resulting model integrated both technological enablers—such as supervised, unsupervised, and reinforcement learning, Natural Language Processing (NLP), Large Language Models (LLMs), recommender systems, and Graph Neural Networks (GNNs)—and human-centric psychological dimensions, including flow experience, perceived value, satisfaction, trust, and consumer engagement, across 4 hierarchical levels. The findings indicated that foundational elements like “Reinforcement Learning” and “Natural Language Processing” served as primary drivers, while behavioral and attitudinal outcomes, such as “loyalty”, “brand advocacy”, and “eWOM” ranked at the top of the model. The model's innovation lay in its structured synthesis of data-driven AI technologies and human perception layers—a perspective often overlooked in previous frameworks. Practical implications were discussed, providing marketers with guidelines for deploying AI-based tools, such as recommender engines, real-time pricing algorithms, and sentiment analysis through NLP. The study concluded with recommendations for future research on industry-specific applications (e.g., fintech, edtech, tourism) and the ethical considerations surrounding AI-driven marketing decisions. Introduction Rapid advancement of digital transformation has fundamentally altered the modern marketing landscape. With the emergence of sophisticated technologies, such as machine learning, cloud computing, the Internet of Things (IoT), and particularly Artificial Intelligence (AI), marketing has become increasingly data-driven and experience-oriented. As a cornerstone of the fourth industrial revolution, AI enables organizations to automate processes, gain deep insights into consumer behavior, predict preferences, and personalize interactions in real time. This transformation has significantly reshaped how companies engage with consumers, devise strategies, and cultivate long-term loyalty. In today’s highly competitive and fast-evolving market, consumers demand immediacy, relevance, and personalization. AI technologies facilitate companies in meeting these expectations by analyzing vast amounts of consumer data and generating insights that inform tailored marketing efforts. For instance, recommendation systems on platforms like Amazon and Netflix, chatbots utilizing natural language processing, and predictive analytics employed by financial institutions exemplify AI's pervasive influence. Despite this growth, there remains a notable absence of an integrated model that synthesizes these diverse AI applications into a coherent framework while considering psychological, experiential, and ethical dimensions. Current research often focuses on specific AI tools in marketing; some studies examine predictive models for customer churn, while others investigate NLP in sentiment analysis. While these inquiries are valuable, they frequently overlook the broader context—how various AI elements interconnect to impact customer loyalty. Loyalty is a multifaceted concept shaped not only by repeated transactions, but also by attitudinal factors, such as trust, perceived value, and brand advocacy. The role of AI in fostering these deeper loyalty outcomes is yet to be clearly defined. Addressing this gap is crucial both academically and practically. From an academic perspective, it enriches marketing theory by integrating technological and behavioral dimensions. Practically, it provides guidance for practitioners seeking to leverage AI responsibly and effectively. The primary research question guiding this study was: What are the key applications of AI in digital marketing and how can these be organized into a conceptual framework that elucidates their role in enhancing consumer loyalty? By answering this question, the study aimed to advance the theory of AI-enabled marketing and present a structured, practical model that aligned advanced technologies with human-centered values. Materials & Methods Research Design This study employed a qualitative, exploratory, and applied research design aimed at developing a conceptual model. This approach was particularly well-suited for topics that remained underexplored and required the construction of a grounded framework rather than mere hypothesis testing. The methodological process was executed in 3 phases: meta-synthesis, Delphi validation, and structural modeling using ISM and MICMAC techniques. Phase 1: Meta-Synthesis The first stage involved a systematic literature review conducted through the Scopus, ScienceDirect, and Emerald databases. The inclusion criteria specified publications from 2015 to 2025 that were indexed in reputable journals and explicitly focused on AI applications in marketing or consumer behavior. A total of 53 articles met these criteria. Utilizing MAXQDA software, a 3-stage coding process (open, axial, and selective) was implemented, resulting in the extraction of 26 components categorized into technological, experiential, and socio-ethical dimensions. This meta-synthesis ensured comprehensive coverage of both empirical and conceptual contributions. Phase 2: Delphi Method To validate the identified components, the Delphi method was employed with a panel of 10 experts, comprising both academic researchers and senior industry professionals. The Delphi technique was chosen for its effectiveness in achieving consensus on complex, multi-dimensional constructs. Two iterative rounds of surveys were conducted, yielding a high reliability coefficient (Cohen’s Kappa = 0.82) and indicating strong agreement among the experts. The panel confirmed the relevance of the 26 components and suggested two refinements: (1) incorporating “AI-driven responses to competitor strategies” within the context of reinforcement learning and (2) including “sentiment-informed CSR initiatives”. These additions underscored the dynamic and ethical dimensions of AI in marketing. Phase 3: ISM and MICMAC Analyses Interpretive Structural Modeling (ISM) was employed to map the relationships among the identified components and construct a hierarchical model. This approach clarified which elements served as foundational drivers and which were outcomes. Complementing ISM, MICMAC analysis was utilized to classify the components based on their driving and dependence power. The combined analysis revealed 4 categories: driving forces (e.g., reinforcement learning), linkage factors (e.g., transparency), dependent outcomes (e.g., loyalty), and relatively autonomous elements (e.g., multi-sensory engagement). Together, these methodological phases ensured rigor by integrating breadth (literature synthesis), depth (expert validation), and structure (hierarchical modeling). Research Findings Core Components and Hierarchical Layers The validated model comprised 26 components organized into 4 hierarchical layers: Technological Enablers: This layer included supervised and unsupervised learning, reinforcement learning, Natural Language Processing (NLP), Large Language Models (LLMs), generative AI models, recommender systems, and graph neural networks. Together, these elements formed the infrastructural backbone that facilitated advanced data analysis, prediction, and personalization. Consumer Experience and Perception: This layer encompassed constructs, such as flow experience, perceived value, consumer satisfaction, trust, algorithmic transparency, human-like interaction, and multi-sensory engagement. These factors mediated the relationship between technological enablers and outcomes related to loyalty. Socio-Ethical Considerations: This layer was defined by Corporate Social Responsibility (CSR), ethical issues in LLMs, fairness, privacy, and consumer engagement in CSR, reflecting the growing demand for responsible and ethical AI practices. Behavioral Outcomes: At the pinnacle of the model were attitudinal loyalty, behavioral loyalty, electronic word-of-mouth, and brand advocacy, and consumer recommendation intentions— outcomes that organizations valued most. ISM–MICMAC Results The hierarchical analysis yielded the following classifications: Drivers: Reinforcement learning, LLMs, and recommender systems served as critical initiators within the model. Linkage Factors: Transparency, flow experience, and NLP-driven sentiment analysis mediated the relationship between technological enablers and behavioral outcomes. Dependents: Loyalty measures, consumer satisfaction, and electronic word-of-mouth emerged as dependent outcomes. Autonomous Elements: Peripheral factors, such as multi-sensory engagement and AI-assisted user-generated content, exerted comparatively lower influence. This analysis highlighted a clear causal pathway: technological foundations shaped consumer experiences, which were moderated by socio-ethical considerations and, in turn, drove loyalty-related behaviors. Discussion of Results & Conclusion Hypothesis Validation The study provided empirical support for 6 hypotheses: Supervised learning enhances the predictive accuracy of consumer behavior. Unsupervised learning facilitates segmentation and the identification of hidden patterns. Reinforcement learning enables adaptive, real-time decision-making. Flow experience supported by AI positively influences consumer loyalty. AI-driven personalization of perceived value strengthens both attitudinal and behavioral loyalty. CSR initiatives informed by sentiment analysis reinforce brand trust and advocacy. Theoretical Contributions This study contributed to the literature by proposing a holistic conceptual model that integrated AI technologies with consumer psychological constructs and ethical considerations. Unlike prior fragmented research, the model introduced a layered structure that systematically connected technological enablers to loyalty outcomes. This integration enhanced our theoretical understanding of how AI-driven personalization and responsible data practices jointly shaped sustainable consumer relationships. Managerial Implications Recommendation Systems: Organizations should implement advanced AI engines (e.g., matrix factorization and deep learning-based recommenders) to provide highly tailored consumer experiences. Sentiment Analysis: NLP models, such as BERT and GPT-4, can be utilized to decode consumer emotions, thereby informing CSR strategies and enhancing communication effectiveness. Experience Design: AI can facilitate immersive and adaptive digital experiences that foster consumer engagement and promote flow states. Ethical AI Practices: Managers must prioritize transparency, fairness, and privacy in AI applications to ensure long-term consumer trust and loyalty. Limitations and Future Research A key limitation of this study was its reliance on expert judgment. Empirical validation using large-scale consumer datasets is essential to strengthen the robustness of the proposed model. Future research should adopt quantitative approaches, such as Structural Equation Modeling (SEM) or longitudinal designs. Additionally, sector-specific adaptations (e.g., fintech, healthcare, and education) can enhance external validity. Cross-cultural comparisons would further elucidate how cultural contexts moderate AI-driven loyalty formation. Finally, the ethical challenges associated with LLMs—including bias, misinformation, and privacy risks—warrant deeper scholarly investigation. Conclusion This research presented a structured conceptual model that integrated the technological, experiential, and socio-ethical dimensions of AI-driven digital marketing, positioning consumer loyalty as the ultimate outcome. The model offered both theoretical insights and practical guidance, emphasizing the necessity of aligning AI tools with human values and social responsibility. In doing so, it establishes a foundation for future empirical investigations and for responsible application of AI in managerial practice. This research aimed to design, validate, and elucidate a conceptual model for harnessing Artificial Intelligence (AI) in digital marketing, specifically to enhance consumer loyalty. The study was conducted in 4 structured phases. First, key components were identified through a systematic literature review (meta-synthesis) of 53 academic sources. In the second phase, the fuzzy Delphi method was utilized with 10 industry experts to validate the relevance of the content and achieve expert consensus. The third phase employed Interpretive Structural Modeling (ISM) to analyze and structure the causal relationships among 26 identified components. Finally, MICMAC analysis was used to categorize these components based on their driving power and dependence. The resulting model integrated both technological enablers—such as supervised, unsupervised, and reinforcement learning, Natural Language Processing (NLP), Large Language Models (LLMs), recommender systems, and Graph Neural Networks (GNNs)—and human-centric psychological dimensions, including flow experience, perceived value, satisfaction, trust, and consumer engagement, across 4 hierarchical levels. The findings indicated that foundational elements like “Reinforcement Learning” and “Natural Language Processing” served as primary drivers, while behavioral and attitudinal outcomes, such as “loyalty”, “brand advocacy”, and “eWOM” ranked at the top of the model. The model's innovation lay in its structured synthesis of data-driven AI technologies and human perception layers—a perspective often overlooked in previous frameworks. Practical implications were discussed, providing marketers with guidelines for deploying AI-based tools, such as recommender engines, real-time pricing algorithms, and sentiment analysis through NLP. The study concluded with recommendations for future research on industry-specific applications (e.g., fintech, edtech, tourism) and the ethical considerations surrounding AI-driven marketing decisions. Introduction Rapid advancement of digital transformation has fundamentally altered the modern marketing landscape. With the emergence of sophisticated technologies, such as machine learning, cloud computing, the Internet of Things (IoT), and particularly Artificial Intelligence (AI), marketing has become increasingly data-driven and experience-oriented. As a cornerstone of the fourth industrial revolution, AI enables organizations to automate processes, gain deep insights into consumer behavior, predict preferences, and personalize interactions in real time. This transformation has significantly reshaped how companies engage with consumers, devise strategies, and cultivate long-term loyalty. In today’s highly competitive and fast-evolving market, consumers demand immediacy, relevance, and personalization. AI technologies facilitate companies in meeting these expectations by analyzing vast amounts of consumer data and generating insights that inform tailored marketing efforts. For instance, recommendation systems on platforms like Amazon and Netflix, chatbots utilizing natural language processing, and predictive analytics employed by financial institutions exemplify AI's pervasive influence. Despite this growth, there remains a notable absence of an integrated model that synthesizes these diverse AI applications into a coherent framework while considering psychological, experiential, and ethical dimensions. Current research often focuses on specific AI tools in marketing; some studies examine predictive models for customer churn, while others investigate NLP in sentiment analysis. While these inquiries are valuable, they frequently overlook the broader context—how various AI elements interconnect to impact customer loyalty. Loyalty is a multifaceted concept shaped not only by repeated transactions, but also by attitudinal factors, such as trust, perceived value, and brand advocacy. The role of AI in fostering these deeper loyalty outcomes is yet to be clearly defined. Addressing this gap is crucial both academically and practically. From an academic perspective, it enriches marketing theory by integrating technological and behavioral dimensions. Practically, it provides guidance for practitioners seeking to leverage AI responsibly and effectively. The primary research question guiding this study was: What are the key applications of AI in digital marketing and how can these be organized into a conceptual framework that elucidates their role in enhancing consumer loyalty? By answering this question, the study aimed to advance the theory of AI-enabled marketing and present a structured, practical model that aligned advanced technologies with human-centered values. Materials & Methods Research Design This study employed a qualitative, exploratory, and applied research design aimed at developing a conceptual model. This approach was particularly well-suited for topics that remained underexplored and required the construction of a grounded framework rather than mere hypothesis testing. The methodological process was executed in 3 phases: meta-synthesis, Delphi validation, and structural modeling using ISM and MICMAC techniques. Phase 1: Meta-Synthesis The first stage involved a systematic literature review conducted through the Scopus, ScienceDirect, and Emerald databases. The inclusion criteria specified publications from 2015 to 2025 that were indexed in reputable journals and explicitly focused on AI applications in marketing or consumer behavior. A total of 53 articles met these criteria. Utilizing MAXQDA software, a 3-stage coding process (open, axial, and selective) was implemented, resulting in the extraction of 26 components categorized into technological, experiential, and socio-ethical dimensions. This meta-synthesis ensured comprehensive coverage of both empirical and conceptual contributions. Phase 2: Delphi Method To validate the identified components, the Delphi method was employed with a panel of 10 experts, comprising both academic researchers and senior industry professionals. The Delphi technique was chosen for its effectiveness in achieving consensus on complex, multi-dimensional constructs. Two iterative rounds of surveys were conducted, yielding a high reliability coefficient (Cohen’s Kappa = 0.82) and indicating strong agreement among the experts. The panel confirmed the relevance of the 26 components and suggested two refinements: (1) incorporating “AI-driven responses to competitor strategies” within the context of reinforcement learning and (2) including “sentiment-informed CSR initiatives”. These additions underscored the dynamic and ethical dimensions of AI in marketing. Phase 3: ISM and MICMAC Analyses Interpretive Structural Modeling (ISM) was employed to map the relationships among the identified components and construct a hierarchical model. This approach clarified which elements served as foundational drivers and which were outcomes. Complementing ISM, MICMAC analysis was utilized to classify the components based on their driving and dependence power. The combined analysis revealed 4 categories: driving forces (e.g., reinforcement learning), linkage factors (e.g., transparency), dependent outcomes (e.g., loyalty), and relatively autonomous elements (e.g., multi-sensory engagement). Together, these methodological phases ensured rigor by integrating breadth (literature synthesis), depth (expert validation), and structure (hierarchical modeling). Research Findings Core Components and Hierarchical Layers The validated model comprised 26 components organized into 4 hierarchical layers: Technological Enablers: This layer included supervised and unsupervised learning, reinforcement learning, Natural Language Processing (NLP), Large Language Models (LLMs), generative AI models, recommender systems, and graph neural networks. Together, these elements formed the infrastructural backbone that facilitated advanced data analysis, prediction, and personalization. Consumer Experience and Perception: This layer encompassed constructs, such as flow experience, perceived value, consumer satisfaction, trust, algorithmic transparency, human-like interaction, and multi-sensory engagement. These factors mediated the relationship between technological enablers and outcomes related to loyalty. Socio-Ethical Considerations: This layer was defined by Corporate Social Responsibility (CSR), ethical issues in LLMs, fairness, privacy, and consumer engagement in CSR, reflecting the growing demand for responsible and ethical AI practices. Behavioral Outcomes: At the pinnacle of the model were attitudinal loyalty, behavioral loyalty, electronic word-of-mouth, and brand advocacy, and consumer recommendation intentions— outcomes that organizations valued most. ISM–MICMAC Results The hierarchical analysis yielded the following classifications: Drivers: Reinforcement learning, LLMs, and recommender systems served as critical initiators within the model. Linkage Factors: Transparency, flow experience, and NLP-driven sentiment analysis mediated the relationship between technological enablers and behavioral outcomes. Dependents: Loyalty measures, consumer satisfaction, and electronic word-of-mouth emerged as dependent outcomes. Autonomous Elements: Peripheral factors, such as multi-sensory engagement and AI-assisted user-generated content, exerted comparatively lower influence. This analysis highlighted a clear causal pathway: technological foundations shaped consumer experiences, which were moderated by socio-ethical considerations and, in turn, drove loyalty-related behaviors. Discussion of Results & Conclusion Hypothesis Validation The study provided empirical support for 6 hypotheses: Supervised learning enhances the predictive accuracy of consumer behavior. Unsupervised learning facilitates segmentation and the identification of hidden patterns. Reinforcement learning enables adaptive, real-time decision-making. Flow experience supported by AI positively influences consumer loyalty. AI-driven personalization of perceived value strengthens both attitudinal and behavioral loyalty. CSR initiatives informed by sentiment analysis reinforce brand trust and advocacy. Theoretical Contributions This study contributed to the literature by proposing a holistic conceptual model that integrated AI technologies with consumer psychological constructs and ethical considerations. Unlike prior fragmented research, the model introduced a layered structure that systematically connected technological enablers to loyalty outcomes. This integration enhanced our theoretical understanding of how AI-driven personalization and responsible data practices jointly shaped sustainable consumer relationships. Managerial Implications Recommendation Systems: Organizations should implement advanced AI engines (e.g., matrix factorization and deep learning-based recommenders) to provide highly tailored consumer experiences. Sentiment Analysis: NLP models, such as BERT and GPT-4, can be utilized to decode consumer emotions, thereby informing CSR strategies and enhancing communication effectiveness. Experience Design: AI can facilitate immersive and adaptive digital experiences that foster consumer engagement and promote flow states. Ethical AI Practices: Managers must prioritize transparency, fairness, and privacy in AI applications to ensure long-term consumer trust and loyalty. Limitations and Future Research A key limitation of this study was its reliance on expert judgment. Empirical validation using large-scale consumer datasets is essential to strengthen the robustness of the proposed model. Future research should adopt quantitative approaches, such as Structural Equation Modeling (SEM) or longitudinal designs. Additionally, sector-specific adaptations (e.g., fintech, healthcare, and education) can enhance external validity. Cross-cultural comparisons would further elucidate how cultural contexts moderate AI-driven loyalty formation. Finally, the ethical challenges associated with LLMs—including bias, misinformation, and privacy risks—warrant deeper scholarly investigation. Conclusion This research presented a structured conceptual model that integrated the technological, experiential, and socio-ethical dimensions of AI-driven digital marketing, positioning consumer loyalty as the ultimate outcome. The model offered both theoretical insights and practical guidance, emphasizing the necessity of aligning AI tools with human values and social responsibility. In doing so, it establishes a foundation for future empirical investigations and for responsible application of AI in managerial practice.

https://nmrj.ui.ac.ir/article_29826_7ae3c62418b833a1604f021fdab6c6f5.pdf