Abstract
The aim of this study is to provide a framework for implementing AI‑based demand‑driven business intelligence in the apparel industry. In terms of purpose, this research is applicable; and in terms of methodology it follows a mixed‑methods approach (qualitative and quantitative).
The statistical population in the qualitative phase consisted of 17 participants, including university professors, experts, sales specialists, managers in the apparel industry, and artificial intelligence specialists, selected through purposive sampling. In the quantitative phase, the statistical population consisted of 384 individuals, including managers, senior experts, data analysts, supply chain managers, and information technology specialists in the apparel industry, selected by cluster random sampling. Data collection instruments included semi‑structured interviews and questionnaires. For data analysis, MAXQDA 2020 software was employed in the qualitative phase, while SPSS and PLS were employed in the quantitative phase. The results of data coding showed that 675 initial codes were refined to 461 cleaned codes and subsequently categorized into 214 conceptual codes, 45 sub‑categories, 19 main categories, and 6 core categories.
Based on the research paradigm model, causal factors include increasing competition, changes in customer roles and market behavior, rising internal costs, and resource constraints. Continuous learning and monitoring, as well as privacy protection, serve as positive interventions in this process, whereas factors such as lack of artificial intelligence skills, resistance to change, and legal limitations may act as inhibiting interventions. These factors influence strategies such as transformational leadership, organizational memory management, and technological investment. The final model is aligned with the internal and structural challenges faced by businesses.
Introduction
Today, in some market segments, the speed of production cycles has become more important than cost, because shorter production times provide the flexibility required to adapt to rapidly changing fashion trends in these market segments. Accordingly, selecting the most beneficial supply chain strategy has become an established research topic, particularly for fashion products where demand for greater variety exists within short time intervals. At the same time, managing demand and supply simultaneously is not easy, as supply chains have become increasingly complex due to outsourcing processes (Anitha & Neelakandan, 2024).
The apparel industry, characterized by short product life cycles and rapid changes in customer preferences, requires highly flexible supply chains. Fast fashion brands, due to their need for rapid responses to fashion changes, require responsive and flexible supply chains, whereas traditional brands focus more on efficiency and productivity through economies of scale. The sensitivity of retail brands to fashion trends shapes their supply chain strategies. Fast fashion brands require high flexibility and speed in their supply chains in order to respond to market demand in a timely manner. Coordination between demand and supply remains challenging, and achieving mutual benefits for both manufacturers and retailers require precise supply chain management (Swaminathan & Venkitasubramony, 2024).
Choosing between these two strategies requires consideration of consumer purchasing behavior as well as competitors’ strategies. The interaction of these factors leads to complex market structures that require appropriate tools for evaluating strategies. Agent‑based modeling and simulation enable the consideration of individual consumer preferences, social interactions, and competition among producers, thereby illustrating the overall behavior of the system (market) at a macro level (Rand et al., 2018).
This research not only seeks to present a theoretical model but also aims to assist manufacturers and businesses active in the apparel industry in moving beyond traditional approaches toward production aligned with real demand and value creation. Such a model, while preserving cultural standards, can contribute to sustainable growth, social identity formation, and economic development. Considering the above, the present study seeks to answer the main research question: How can a framework for implementing AI‑based demand‑driven business intelligence in the apparel industry be developed?
Theoretical Framework
Demand‑Driven Business Intelligence
Demand‑driven business intelligence refers to an approach to business intelligence that focuses on analyzing and understanding customer needs, preferences, and market behaviors in real time or near real time. This type of business intelligence utilizes both internal and external data to help organizations base their decisions on actual market demand rather than relying solely on historical data (Ben et al., 2019).
Artificial Intelligence
Artificial intelligence (AI) is considered a strategic approach for maximizing the use of technology and market data in order to improve customer experience. Based on previous studies, advanced AI techniques include tools such as email automation, chatbots, and various analytical systems that assist managers in introducing product features to customers on a broad scale without temporal or spatial limitations, thereby creating a better purchasing and consumption experience for customers (Hayatmehr et al., 2026).
Hayatmehr et al. (2026) examined the impact of artificial intelligence applications and intelligent learning on the strategic thinking skills and academic performance of management students, considering the moderating role of individual ethics. The results indicated that the use of AI tools has a positive effect on strategic thinking dimensions (systems thinking, creative thinking, future-oriented thinking, and critical thinking), intelligent learning, and academic performance. The mediating role of strategic thinking—particularly critical thinking and future-oriented thinking—was confirmed in the relationship between AI usage and academic performance, as well as between intelligent learning and academic performance. Additionally, individual ethics not only had a positive effect on intelligent learning but also played a moderating role in the relationship between intelligent learning and strategic thinking (systems thinking, critical thinking, and future-oriented thinking).
Zolghadr et al. (2026) investigated the modeling and validation of the role of artificial intelligence in enhancing the export capabilities of companies in the electronics industry using a mixed‑methods approach. The qualitative results indicated that the model consists of five main categories: causal conditions (ICT infrastructure, data quality, and technical capacity), contextual conditions (supportive policies, international collaborations, and an innovative organizational culture), intervening conditions (sanctions, rapid technological changes, and legal and customs regulatory barriers), strategies (demand forecasting, price optimization, logistics intelligence, and human resource empowerment), and outcomes (competitive advantage, penetration into global markets, increased customer satisfaction, and cost reduction). In the quantitative phase, composite reliability indices were all above 0.7 and convergent validity for most constructs exceeded 0.5. The hypothesis testing results also confirmed all relationships among the model constructs with a significance level of p < 0.001.
Research Methodology
This study is applicable in terms of purpose, and adopts a mixed‑methods approach (qualitative and quantitative) in terms of research design.
In the qualitative phase, the statistical population consisted of 17 university professors, experts, sales specialists, managers in the apparel industry, and artificial intelligence specialists selected through purposive sampling.
In the quantitative phase, the statistical population included 384 individuals consisting of managers, senior experts, data analysts, supply chain managers, and information technology specialists in the apparel industry. These participants were selected by cluster random sampling.
The data collection tools consisted of semi‑structured interviews and a questionnaire.
Research Findings
For data analysis, MAXQDA 2020 was applied in the qualitative phase, while SPSS and PLS were applied in the quantitative phase. The results of the coding process reduced the initial 675 codes to 461 refined codes, ultimately categorized into 214 conceptual codes, 45 sub‑categories, 19 main categories, and 6 core categories.
Based on the study’s paradigmatic model, the causal factors include increased competition, changes in the roles and behaviors of customers and markets, rising internal costs, and resource constraints. Continuous learning and monitoring, as well as the protection of privacy, are considered positive interventions in this process. In contrast, factors such as the lack of AI skills, resistance to change, and legal limitations can act as inhibiting interventions.
These factors influence strategies such as transformational leadership, organizational memory management, and technological investment. The final model is aligned with the internal and structural challenges faced by businesses.
Conclusion
The present study was conducted with the aim of developing a framework for implementing AI‑based demand‑driven business intelligence in the apparel industry. The findings of this research are consistent with the results of previous studies (Hayatmehr et al., 2026; Zolghadr et al., 2026; Ghanbari Gheshlaghi et al., 2025; Liu et al., 2025; Swaminathan & Venkitasubramony, 2024; Anitha & Neelakandan, 2024; Roth, 2024; Hemati, 2024; Kunz et al., 2023; Hajipourfard et al., 2022; Wamba‑Taguimdje et al., 2020).
Swaminathan and Venkitasubramony (2024), through a systematic review and comprehensive analysis of reputable studies, compared statistical methods, machine learning, and deep learning approaches, identifying fundamental challenges in the fashion industry such as short product life cycles, high variety, and demand volatility. Their findings demonstrated that AI models provide higher predictive accuracy in this industry and emphasized the growing necessity of robust validation frameworks within demand‑driven models.
Based on the research results, it is recommended that apparel industries develop the necessary foundation for activating AI‑based demand‑driven business intelligence by establishing core infrastructure, gradually developing knowledge and human capital, redesigning decision‑making structures, and institutionalizing a data‑driven culture. Such a foundation is neither inherently complex nor excessively costly; rather, it can be implemented gradually and aligned with the actual capacities of the national apparel industry.