Unlocking Industry 4.0 technologies adoption in inventory management: empirical evidence from Australian retailers

Industry 4.0, first conceptualised in 2011 by a former SAP manager in Germany, represents a paradigm shift in industrial practices through the adoption of digital and smart technologies that enable decentralised, automated processes within organisations (Kassen, 2022). These technologies facilitate the seamless movement of data and materials without direct human intervention, transforming the way businesses operate across multiple sectors. Although Industry 4.0 emerged initially in the manufacturing domain, its principles have extended to supply chain management, logistics, and retail operations due to their cross-functional nature and the increasing need for efficiency, agility, and responsiveness in complex environments (Tjahjono et al., 2017; Calabrese et al., 2022).

The integration of Industry 4.0 technologies in supply chains offers multiple opportunities. By leveraging advanced digital tools, organisations can optimise process management, forecast demand more accurately, improve supplier selection, and enhance decision-making across interconnected activities. Technologies such as augmented reality (AR), big data analytics, cloud computing, Internet of Things (IoT), radio-frequency identification (RFID), robotics, and artificial intelligence (AI) are increasingly employed to provide real-time visibility, improve operational efficiency, and strengthen competitive advantage (Mendoza et al., 2025; Kamal, 2020). Retailers, in particular, rely on these innovations to manage highly dynamic inventory systems, respond to fluctuating customer demand, and deliver superior service levels in an increasingly competitive marketplace.

Inventory management constitutes a critical area for the application of Industry 4.0. It involves controlling stock levels, optimising warehousing strategies, reducing lead times, and efficiently allocating resources across the supply chain. Effective inventory management not only supports operational efficiency but also contributes directly to customer satisfaction and profitability. Retail organisations face mounting pressure to maintain the right balance between stock availability and cost efficiency, particularly in fast-moving consumer goods and high-demand sectors. These pressures are exacerbated by the volatility of global markets, frequent changes in consumer preferences, and the increasing complexity of supply networks (Singh and Verma, 2018; Shriharsha et al., 2025).

Despite the clear benefits, the adoption of Industry 4.0 technologies in inventory management is accompanied by multiple challenges. Studies indicate that organisations encounter barriers at various levels, including organisational, technological, strategic, legal, and ethical dimensions. Common obstacles include financial constraints, resistance from employees, lack of skilled personnel, inadequate infrastructure, low standardisation of processes, cybersecurity threats, and insufficient policy support from governments (Almada-Lobo, 2015; Nicoletti, 2018). These challenges often interact and amplify one another, making the implementation of smart technologies in inventory systems a complex endeavour. For example, technological advancements such as IoT and AI may require substantial investment in workforce training and data management systems, while regulatory uncertainties and cybersecurity risks further complicate adoption decisions.

Although Industry 4.0 has attracted considerable research interest in recent years, studies specifically focusing on barriers in inventory management systems are limited. Existing literature often addresses general applications or technological benefits without providing an in-depth analysis of the challenges that organisations face in practical implementation (Paul et al., 2019; Yadav et al., 2020, 2022). As a result, there is a critical knowledge gap regarding the nature of these barriers, their interrelationships, and their influence on the adoption of Industry 4.0 technologies in the retail sector.

This study aims to address this gap by investigating the key barriers to implementing Industry 4.0 technologies in retailers’ inventory management systems and exploring the interrelationships between these barriers. Specifically, the study seeks to answer the following research questions:

1. What are the key barriers to implementing industry 4.0 technologies in the retailers’ inventory management systems?

2. What is the interrelationship between the identified barriers and to what extent do these barriers influence one another?

This research will answer the questions by collecting data from experts in the field of inventory management from Australian retailers. The data collection leads to identifying and summarizing key barriers and the interrelationship between these barriers will be discussed using an interpretive structural model (ISM). Besides, a MICMAC analysis is used to determine to what extent these barriers influence one another in the supply chain.

The remainder of this study is summarized as follows. A comprehensive literature review is proposed in section 2 to highlight the knowledge gap. Section 3 proposes a quantitative approach that comprises collecting data, ethical approval, and applying ISM and MICMAC analysis approaches. Section 4 illustrates the collected data and analyses them using a reachability matrix. The results obtained from the ISM, MICMAC, and reachability matrix analysis are summarized in section 5 by assigning the identified barriers to four clusters based on their driving and dependence power. Eventually, Section 6 summarizes the findings and proposes theoretical and managerial implications.

Retailers have started using new strategies to enhance their business models and propose value for stakeholders. Adopting smart technologies is an effective method that can be used to improve the service level amongst supply chain activities. Transparency of products through shipping containers and interoperability of machines and humans based on virtual technologies such as augmented reality (AR), precise optimization algorithms, and AI automation are some instances of applying these technologies in an inventory system. Along with logistic applications, inventory management is also the case for employing industry 4.0 technologies. Currently, most approaches in inventory management are multi-objective optimization models to minimize the total cost of handling inventory and keep the service level at an acceptable value. The transition from traditional to smart inventory management requires huge pools of date, internet connectivity, enterprise software systems, and smart products (Karimi-Nasab and Aryanezhad, 2011).

Uncertain demand is one of the main factors that pose uncertainty in the whole inventory system. Therefore, the inventory system requires processes that can help increase predictability, balance demand fluctuations, and maintain the selling price. Besides, developing transparency through digital ledgers and IoT can help inventory systems to move towards decentralizing the system, decreasing the complexity of decision-making, and remaining flexible (G et al., 2019). However, retailers still confront numerous obstacles in implementing industry 4.0 technologies in their inventory systems effectively. This section proposes a literature review on the potential barriers to this implementation. It starts with identifying the technologies that are applicable in inventory systems and moves toward finding barriers that are obstacles to implementing each technology (Yerpude and Singhal, 2018).

The autonomous vehicle can be considered one of the ideal technologies to provide constant traceability and automation in the inventory system and reduce costs and time spent on material handling (Wankhede and Vinodh, 2022; Piron et al., 2024). One of the instances of applying autonomous vehicles is using installed scanners to follow a pre-defined flight route to read barcodes (Beul et al., 2018). A light detection device can also be attached to a self-positioning drone capable of scanning the environment autonomously. These drones use RFID scanners to identify tags around warehouses and provide inventory reports (Fernández-Caramés et al., 2019). Besides, QR code-based vehicles can also be used to detect items in a warehouse. Combining these vehicles with an ultra-wideband (UWB) network improves the accuracy of this detection significantly. The application of these vehicles mainly results in reducing delivery delays, improving pickup accuracy, increasing productivity, mitigating average operation time, and developing user-friendliness (Macoir et al., 2019).

Big data analytics which is the process of extracting and analysing data from large data pools can also be applied to optimize knowledge creation and enhance decision-making in inventory systems (Tiwari et al., 2018). Managing the safety stock, predicting demand patterns, minimizing costs, and optimizing materials flow are some advantages of using this approach in inventory systems (Wang et al., 2016; Han et al., 2021).

Internet of things (IoT) which is defined as a network of objects accommodating technologies that interact with internal state and external environment includes a range of different physical sensors, processors, and actuators that send data to a virtual platform. Technologies such as RFID, wireless sensor networks (WSN), cloud computing, laser scanners, and intelligent information sensing devices are some of the new advances in the area (Calabrese et al., 2022). Near-field communication (NFC) is also used in inventory systems to enhance the traceability of products over short distances (Abdel-Basset et al., 2018).

Blockchain is a shred time-stamped data ledger that makes the interaction between participants without relying on authorities (Tiwari et al., 2018). This technology is a distributed database among a network of computers that can store and share information electronically. As inventory systems rely on fast information exchange, which is needed for real-time decision-making, blockchain can be used for data authenticity across all parties involved in inventory activities. Blockchain can be combined with autonomous vehicles to assist with the traceability of products and sharing the data in real-time in a safe platform (Kapitonov et al., 2017).

AI can administrate customer data and forecast customer behaviour, provide proactive notification for operators to re-order stock, and assist operators to optimize the re-ordering points (Sustrova, 2016). Artificial neural network which is based on AI can imitate human brain functions and solve the problem of re-ordering and demand forecasting (Paul and Azeem, 2011). Another application of AI which is expert systems addresses material handling patterns and their complexities. In this regard, an organisation’s material requirement planning (MRP) can incorporate expert systems to collect and store data on master production schedules (MPS), order patterns, and inventory replenishment intervals.

Although some barriers were initially identified in manufacturing, logistics, and IT contexts, they are transferable and relevant to retail inventory management. For instance, challenges related to data integration, system interoperability, and workforce skills, while documented in other sectors, also directly affect the implementation of Industry 4.0 technologies in retail inventory systems. These barriers influence how retailers adopt digital tools such as IoT, RFID, and analytics platforms, impacting inventory accuracy, process efficiency, and overall operational performance. By including these barriers, the study ensures a comprehensive exploration of factors that can hinder Industry 4.0 adoption in the retail inventory context.

A study conducted by Deloitte on the challenges and solutions of industry 4.0 in Switzerland depicts a divergence in the current state of adopting industry 4.0 in inventory management and the perceived potential of this segment. Form the industrial experts’ point of view, procurement and manufacturing processes have already undergone a huge transformation. However, with %74 of participants rating, warehousing and inventory management is the segment that benefits from adopting these technologies more than others (Finance, 2015). Despite this significance, there is a lack of literature specifically targeting inventory systems.

Financial constraints are huge problems in implementing new technologies (Horváth and Szabó, 2019). Lack of clarity in cost-benefit analysis and monetary gains within an organisation is the main challenge of this implementation. Besides, a lack of financial resources is an important challenge for companies to overcome (Theorin et al., 2017).

As industry 4.0 technologies are rapid, intensive, and resource-consuming, adequate management skills are required to implement timely decisions to achieve the expected results. In this regard, the management team coordinates the cross-functional collaboration of digitalized process of value chain networks (Ras et al., 2017; Hossain and Thakur, 2020). Moreover, some executives who are not familiar with these technologies become reluctant in adopting them. This resistance to change can arise from a lack of competence or denying the benefits of digital transformation (Gökalp et al., 2017).

Some other organisational challenges such as the feeling of over-supervision, inexplicit values, feeling inadequacy, and high workload concerns can emerge during the adoption of new technologies, and workers might feel micromanaged. Organisational inadaptability might convey the feeling of being under pressure or inadequate (Birkel et al., 2019). Potentially, some of the staff might be afraid of losing their job and organisational power. Disruption of existing jobs and conflict of interests interferes with upcoming changes and requires constant adaptability to solve the challenges (Ito et al., 2021).

Different studies mutually agreed that the major challenges of most organisations in implementing industry 4.0 technologies reside in the lack of a skilled workforce and retraining current staff to fit into new roles (Erol et al., 2016). Multi-criteria analysis on the identified barriers of industry 4.0 specifically points out that there is a shortage of workforce who understand the requirements of supply chain management. IoT also adds more complexity to the data management system which means companies need to retain or hire new employees who can work with the data management tools and networks (Bag et al., 2018).

Implementing industry 4.0 technologies requires high-speed internet connection in different sectors of the organisation to maintain the processing power and address specific computational units. Besides, the lack of unified communication protocols and back-end systems may interfere with the processes of industry 4.0 technologies which will be challenging. Most of these technologies are new and their backup systems need to be maintained to ensure the procedures’ reliability. For some technologies such as autonomous vehicles, the cost of purchasing equipment, low support for end-users, and complex interface affect the technology that is being implemented (Ajmera and Jain, 2019).

Cyber security issues are threats for organisations while industry 4.0 technologies create numerous data transactions. Data security and legal repercussions are the main issues of adopting the technologies in inventory systems. For instance, blockchain users’ information including name, address, and personal details can be publicly accessed due to being visible to all nodes (Cimini et al., 2018). Therefore, major attacks such as denial of service, spoofing, and double spending are common threats. Oracles which are centralized third parties constitute a relationship between blockchain and the real world (Etemadi et al., 2021). In the case of smart contracts, oracles are essential for data collection and there is always a possibility that these oracles consist of criminal individuals or organisations (Fraga-Lamas and Fernández-Caramés, 2019).

Legal complexities in implementing industry 4.0 technologies emphasize data protection regulations when in the private-law domain, liability is significant as contracts can be miscoded and the intended expectations of parties might not be achieved (Omar et al., 2020). In smart contracts, parties need to agree on jurisdiction and governance for dispute resolution and the identity of both parties can also be questioned as most blockchains avoid sharing details (Ho et al., 2021). From the public perspective, blockchains can be used for money laundry purposes to take advantage of pseudonymous involvement (De Giovanni, 2019).

Low degrees of standardization across business processes drastically hinder the further adaptation of industry 4.0 technologies in inventory systems (Raj et al., 2020). Organizations with a lack of digital culture often encounter problems in integrating industry 4.0 technologies due to difficulties in establishing seamless connectivity and integration between machines, workers, and equipment (Rajput and Singh, 2019). A variety of platforms and databases is an issue for data exchanging when developing interoperability (Müller, 2019).

Difficulties in coordination can be classified as one of the significant barriers to adopting smart technologies in inventory systems (Kmiecik, 2022; Rabelo et al., 2002). As processes in an inventory system are becoming more interconnected, coordinating different functions is challenging as most organisations still operate within departments and use hardware and software that is designed for their pre-defined processes.

Industry 4.0 improvements are applied to service-oriented activities; however, some organisations focus on the return on investment and bottom lines which underlies risks that should be taken into account (Cachon and Fisher, 2000). Therefore, high expectations can negatively affect the process of integrating technologies within an inventory system. In this regard, inventory systems can be improved by increasing the collected data volume, synchronizing transactions, and obtaining consensus amongst clients. This can affect IoT devices that require rapid access to computational performance with low computational power (Hamadneh et al., 2021).

In case of new disruptive technologies, governments need to provide the necessary infrastructure for the digital transition and implement policies to control and regulate the systems. Lack of support and policy making affects the rate of implementing industry 4.0 technologies in inventory systems (Horvat et al., 2018). For instance, cold chains in food industries and pharmaceuticals require control systems based on various sensors that measure different environmental factors (Yadav et al., 2022). However, there is little evidence from the governments that try to provide guidelines for implementing smart technologies in inventory management sectors (IBISWorld, 2019).

The sheer volume of heterogeneous data generated through adopting industry 4.0 technologies has been exponentially on the rise since its inception. Data protection, collection, reliability of data, transfer rate, and storage efficiency are relatively faster while using smart methods. However, constant exposure to volatile results has made decision-making strenuous in some cases. The immature status of technology, energy cost, low scalability, and untrusted nodes are some reasons that prevent organisations from investing in industry 4.0 technologies. As an extension of blockchain technology, smart contracts have limitations such as transaction capacity, latency, throughput, and validation protocols (Paul et al., 2019).

The existing literature has been explored to provide a review of current barriers preventing the further implementation of Industry 4.0 technologies in inventory management. The main inventory management-related Industry 4.0 technologies are autonomous vehicles, big data analytics, the Internet of Things (IoT), blockchain, and artificial intelligence (AI). This review was conducted through an examination of peer-reviewed academic publications and industry reports. These sources contributed to identifying 13 common barriers to the adoption of these five key technologies. These barriers include financial constraints, lack of managerial support, organisational inadaptability, skill and training requirements, inadequate infrastructure and facilities, cybersecurity issues, legal complexities, integration and interoperability challenges, lack of coordination and collaboration, unclear perception of benefits, government reluctance, limited investment in research and development (R&D), and technological complications.

To ensure that the barriers included in this study represent the current state of knowledge in the literature, a structured process was followed to identify, synthesise, and consolidate barriers to the adoption of Industry 4.0 technologies in inventory systems. This process comprised three main steps:

• Step 1: Extraction of barriers from the literature: A broad review of peer-reviewed journal articles, conference proceedings, and industry reports was conducted, focusing on the application of Industry 4.0 technologies in supply chain and inventory management. The search targeted the five key technologies most frequently discussed in the context of inventory systems: autonomous vehicles, big data analytics, Internet of Things (IoT), blockchain, and artificial intelligence (AI). All challenges and obstacles mentioned in relation to these technologies were extracted to form an initial long list of potential barriers.

• Step 2: Thematic clustering and categorisation: The extracted barriers were then organised into thematic categories, drawing upon established frameworks in digital transformation and supply chain management (e.g. Ras et al., 2017; Bag et al., 2018; Raj et al., 2020). For example, barriers such as lack of managerial support, resistance to change, and workforce retraining needs were grouped under organisational and human resource barriers. Similarly, issues such as cybersecurity, interoperability, and legal uncertainties were clustered under technological and regulatory barriers. This step ensured that overlapping concepts were merged, while preserving the distinct challenges highlighted in the literature.

• Step 3: Synthesis into common barriers: Following clustering, a consolidation process was undertaken to refine the list into broader but clearly defined categories that captured recurring themes across multiple sources. This synthesis produced the final set of 13 barriers: financial constraints, lack of managerial support, organisational inadaptability, skill and training requirements, inadequate infrastructure and facilities, cybersecurity issues, legal complexities, integration and interoperability challenges, lack of coordination and collaboration, unclear perception of benefits, government reluctance, limited investment in research and development (R&D), and technological complications.

This systematic approach ensured that the identified barriers were not only drawn from diverse sources but also represent the most recurrent and cross-cutting challenges consistently acknowledged across the literature. The 13 barriers therefore provide a comprehensive and integrative framework for examining Industry 4.0 adoption in inventory systems, particularly within the retail sector, where research remains limited compared to manufacturing and logistics.

Despite the breadth of research available, a closer analysis reveals several important gaps. Firstly, while existing studies have extensively discussed individual technologies or focused on a single type of barrier, there is a lack of integrative research that examines multiple Industry 4.0 technologies in relation to inventory management simultaneously—especially in the context of the retail sector. Most of the available literature tends to focus on the manufacturing sector or third-party logistics (3PL) providers, where the technological environment and implementation challenges may differ significantly from those encountered in retail. Consequently, the barriers identified in prior studies may not fully capture the unique complexities and operational dynamics within retail inventory systems.

Secondly, the current literature is predominantly conceptual or based on theoretical frameworks, with limited empirical investigation into how these barriers manifest in real-world inventory operations. This shortfall limits the practical applicability of findings and underscores the need for more industry-specific, evidence-based research. In particular, there is minimal research examining the level of Industry 4.0 readiness among retail organisations, and how this readiness influences the prioritization or mitigation of adoption barriers. Moreover, the interrelationships among barriers—such as how a lack of infrastructure may exacerbate cybersecurity risks or how unclear benefits hinder managerial support—are rarely addressed through structured analytical methods.

These gaps highlight the necessity for empirical studies that not only identify but also map the interconnectivity and relative impact of the various barriers to adopting Industry 4.0 in inventory management. Accordingly, the present study seeks to fill this void by adopting a quantitative approach to data collection. Methods include structured phone interviews, face-to-face discussions, and online meetings with inventory managers and planners. The goal is to elicit insights from practitioners and draw a networked understanding of the barriers, thereby contributing to both practical implementation strategies and theoretical development in this area.

To fill the corresponding knowledge gap, this study adopts an exploratory method which allows us to address the mentioned research question. A joint empirical approach and quantitative methodology are employed in this study in which the quantitative method is mainly used for collecting and analysing data (Percy et al., 2015). The study adopts an exploratory approach, combining empirical and quantitative methods (Percy et al., 2015), with data collected through face-to-face or online interviews to enhance reliability and depth of responses (Merriam and Tisdell, 2015). Participants were Australian retail managers responsible for inventory, logistics, and supply chain operations, selected for their holistic perspective on processes and policies (Kim and Daniel, 2020). Interviews followed a structured protocol, including pre-defined open-ended and pairwise comparison questions to capture the perceived influence of barriers to Industry 4.0 adoption. Example questions included:

1. “Which barriers do you perceive as most critical to adopting Industry 4.0 technologies in your inventory management system?”

2. “How does Barrier A influence Barrier B in terms of operational or strategic impact?”

Ethical approval was obtained from the University of Adelaide, and all participants provided informed consent. Confidentiality was ensured by assigning unique identifiers to each participant, and all data were securely stored.

Australian retailers are considered as the unit of analysis. These units include inventory, logistics, and supply chain managers along with inventory, logistics, and supply chain planners. This study chooses managers as the primary source for observation which aligns with the study conducted by (Kim and Daniel, 2020) due to their holistic perspective on policies and procedures.

Information about retail companies in Australia is freely available through the Australian Bureau of Statistics (ABS) distributing website (Australian Bureau of Statistics (abs.gov.au). On this website, different retailing companies are classified as follows:

1. Household goods retailing

2. Food retailing

3. Cafes and food services

4. Clothing and personal accessories

5. Department stores

6. Other retailing companies

Quota sampling which is a non-probability purposing sampling methodology is selected as the main sampling approach in this study. This methodology which generates a non-random basis for sample selection using sample characteristics is more flexible in selecting a minimum number of participants instead of a fixed number (Saunders, 2014). The selection quota for this study is based on the business size (financial turnover and number of employees) and considered three to five participants from each of the retailing companies which are in the top five in terms of business size. Therefore, selected participants represent the managerial positions that are related to inventory management activities (e.g. directors, executives, etc.). Based on the proposed sampling process, only the companies that demonstrated their push towards implementing industry 4.0 technologies in inventory management are selected. Their performance in this implementation can be verified through accessible performance reports, corporate websites, and third-party reports. Due to creating rapport and having a natural encounter with participants, using face-to-face interviews is believed to be the most effective way to get information. This is significant because this study has empirically collected experts’ perspectives on the implication of industry 4.0 in inventory operations which is a novel phenomenon in retailing industries. Thus, the nature of open-ended nature of semi-structured questions allows more freedom to explore the experts’ opinion.

To design semi-structured interviews (SSIs), the methodology developed by Newcomer et al. (2015) is used in this study. One of the advantages of this semi-structured is that enough time is assigned to each participant and a shorthand system, quotation marks, and audio recording are necessary while taking notes from the interview. The confidentiality of the interviews should be clarified for the participants, and they should be aware of the reason for their participation. Interviewers should understand how questions are developed, and more investigations are needed in which areas. The interviews start with a short introduction to inform the participants about each topic and make them sure that they have enough time to respond to each question. Also, they are reminded that they can leave the interview at any time.

In the next section, five questions are proposed to collect information from the participants based on their perceived knowledge of the subject and the size of operations they conduct. The questions are designed open-ended to let participants provide as much information as they want and compare each barrier to implement industry 4.0 with other barriers through a comparison matrix. Participants separately compare the influence of each barrier with another using a scale of 0–3 in which 0 shows no influence and 3 shows the highest perceived influence. In a case where participants are not sure if two barriers affect each other, they can choose “P” which means a “probability” of influence with an unknown degree. Thirteen comparison matrices are recommended in the questionnaire to find the impact of each barrier on others. Finally, two questions allow participants to give their opinion on the topic. The first one is to find the barriers that have not been identified by participants and the second one measures the degree of impact of barriers on each other.

A well-planned approach is used to improve the quality of outcomes from the interviews. In this regard, four areas of preparation including flow, format, length, and quality become important. This approach helps the interviewees shape a general idea of the problem. None of the participants should feel like they are being interrogated during the interview and interrupting the participants is not acceptable.

The designed questionnaire consists of 13 comparison matrices in which the barriers to employing industry 4.0 technologies in the inventory management sector are discussed. The impact of each barrier on each other is also explored based on the participants’ opinion and five scores measures the dependence level of each of these barriers on others summarized in Table 1.

This approach is performed continuously, and the results are shown in the next section. Eventually, two questions are proposed to probe into the participants’ point of view: the first question indicates the barriers that might have not been identified by this study when investigating the application if industry 4.0 technologies in retail industries on inventory management; and the second question asks about possible solutions that experts offer to overcome these barriers. All interviews were separately transcribed, and the documents were sent to corresponding participants to ensure the accuracy of the collected information.

According to the ambiguity of the barriers, a combination of interpretative structural modelling (ISM) model along with MICMAC analysis is proposed (Janes, 1988). The reason for choosing this research methodology is that this approach is capable of finding a compound association between different elements which can be used for finding a multi-tier structure of these barriers and the interrelationship (Singh et al., 2007). Besides, the ISM method quantifies qualitative data based on the judgement of participants and the final output of this approach is a relationship diagram that answers the first and third research questions. MICMAC analysis is also applied to identify the drive and dependence power of barriers identified in the literature and can be used as a complementary approach for ISM to make barriers evident (Khurana et al., 2010; Kumar and Sharma, 2018).

.ISM is a technique for analysing contextual interrelationships between different elements of a system which is first used by (Warfield, 1974) to break down a problem into smaller subproblems. ISM can be used to examine numerous direct and indirect variables that affect the system and clarify the significance, risks, and order of elements within a system. Participants’ bias is something that is not considered in this method and is a limitation of this approach which can affect the outcome of the analysis. In a study conducted by (Singh et al., 2007), the main features of ISM are described as follows.

1. It is an interpretive technique based on experts’ opinions.

2. It is considered a modelling technique due to using diagrams for representing the relationship and linkages.

3. It simplifies a complex system.

4. Its structural model provides and interprets links and nodes.

Six steps are proposed to use the ISM methodology in practice (Gupta and Jain, 2020).

1. Barrier identification: Barriers to applying industry 4.0 technologies in inventory management and their interrelationship should be identified. Interviews are conducted for this purpose.

2. Developing contextual relationship: A pairwise comparison between identified barriers is proposed. To determine the contextual relationship between two barriers $(i,j)$⁠, four symbols are utilized in this study which can be summarized in Table 2 

1. Building reachability matrix: This matrix is constructed based on SSIM. In this step, the transitivity rule is checked to ensure that the reachability matrix follows a logical pattern of finding relationships among barriers. Thereafter, SSIM is translated into a matrix that undergoes arithmetic functions, and four relationships are turned into a binary matrix which is elaborated as follows.

   • If $(i,j)$ entry in SSIM is $V$⁠, then $(i,j)$ and $(j,i)$ entries in the reachability matrix are 1 and 0, respectively.

   • If $(i,j)$ entry in SSIM is $A$⁠, then $(i,j)$ and $(j,i)$ entries in the reachability matrix are 0 and 1, respectively.

   • If $(i,j)$ entry in SSIM is $X$⁠, then $(i,j)$ and $(j,i)$ entries in the reachability matrix are 1 and 1, respectively.

   • If $(i,j)$ entry in SSIM is $O$⁠, then $(i,j)$ and $(j,i)$ entries in the reachability matrix are 0 and 0, respectively.

2. Determining hierarchy of barriers: Final reachability matrixes are achieved by checking the transitivity rule and the transitivity rule is employed to determine the hierarchy of barriers based on participants’ presumptions. As it is mentioned before, ISM is trying to capture opinions and, in some cases, the participants may not accurately explain the consecutive relationships of barriers. Afterwards, level positioning is used to determine the hierarchy of barriers when antecedents and reachability of barriers are identified based on the final reachability matrix. Barriers on the top level of the hierarchy do not enable other barriers and each time a barrier gets categorized in a level of the hierarchy, it is removed from the next iteration. This process continuous until all barriers are assigned in the hierarchy.

3. Establishing a diagram of nodes and edges after implementing the transitivity rule

4. Transforming the diagram into and ISM model where nodes are replaced by statements.

While ISM represents direct barrier influence analysis, MICMAC analysis presents indirect relationships influencing the barriers. MICMAC analysis develops the driving power and dependence of system elements based on matrix multiplication properties. Matrices are initially squared which leads to obtaining the second matrix and this process iterates $n$ times to create interconnecting variables of $n$ th order. The process gets terminated when a stable state iteration of driving power and dependence power is achieved. The input matrix for MICMAC analysis is called direct matrix $A$⁠, which is created by analysing the final diagram for barriers. After self-multiplying matrix $A$⁠, a saturated state is reached and it this stage, $An$ matrix is the final indirect relationship matrix for MICMAC analysis. The dependence and driving rank of barriers can be obtained from the stabilized MICMAC matrix. This approach uses a simplified axis starting from zero in both $x$ (dependence power) and $y$ (driving power) dimensions to classify barriers into four clusters. The mentioned clusters are summarized as follows.

1. Cluster I. Autonomous barriers: These barriers have less driving power and dependence power, which means they are disconnected from the system.

2. Cluster II. Dependent barriers: These barriers have weak driving power but possess higher dependence. Therefore, it can be assumed that by eliminating barriers they rely on, they will also disappear.

3. Cluster III. Linkage barrier: These barriers have both strong driving power and dependence power. These barriers are unstable and influenced by lower-level barriers.

4. Cluster IV. Independent barriers: These barriers have strong driving power but weak independence. These are key barriers to the system and their lack of reliance on other barriers makes them independent.

In the next section, the process of data collection is elaborated to make use of the two methods mentioned in this section and develop a logical flow to answer all research questions.

The data collection was conducted with four participants, and a profile of each is summarised in Table 3. Initially, 14 companies were identified for the study. Within these companies, 40 candidates were contacted via email or LinkedIn messaging. Seven candidates declined the invitation in response to the first email, two candidates accepted and signed the consent form, and the remaining candidates did not respond. In the second round of contact, 31 candidates were followed up with to arrange meetings. During this period, while conducting the first two discussions, two additional candidates agreed to participate. One candidate requested a two-week delay due to being on annual leave. In total, four candidates ultimately participated in face-to-face or online meetings, rather than through email or other remote methods. Although this number was below the preferred sample size, it was still within an acceptable range for the purposes of this research. The participants represented three companies, each from a different sector within the retail industry. For confidentiality, each participant was assigned a number based on the order of their meetings.

We acknowledge the concern regarding the small sample size. The MICMAC method can be effectively applied with limited data (Kaladharan et al., 2024; Yu et al., 2023) because it relies on expert judgement and quantitative analysis to assess relationships between variables. Its strength lies in systematically identifying key drivers and dependencies within a system, even without large datasets, making it suitable for strategic planning and scenario building in contexts with limited data. We opted for quality over quantity by focusing on a smaller, yet varied, group of experts, allowing us to delve deeper into individual perspectives. By examining both similarities and disparities among these expert viewpoints, we aimed to avoid the pitfalls of relying solely on a consensus-based approach.

To construct the Structural Self-Interaction Matrix (SSIM) and reachability matrix, data from face-to-face or online interviews were systematically analysed. Experts assessed pairwise relationships between identified barriers, indicating directional influences. To enhance consistency and reduce individual bias, all responses were cross-verified across participants, and discrepancies were addressed through iterative clarification until consensus was reached. The SSIM was then converted into the initial reachability matrix following the standard Interpretive Structural Modelling (ISM) procedure, with transitivity checks applied to ensure logical consistency. To further strengthen validity, the emerging model was triangulated with insights from relevant literature, aligning expert judgements with prior empirical findings. These procedures ensured that, even with a relatively small sample size, the development of the SSIM and reachability matrix maintained internal consistency, methodological rigour, and theoretical validity.

The data obtained from the questionnaire is used to find the interrelationship between different barriers. ISM, which is an interactive learning methodology, is used to determine the relationship between different variables by designing a series of graphics and words. Later, the contextual relationship between different barriers is determined by participants. The mentioned relationships are defined as the influence of the barrier $i$ on barrier $j$⁠. These pairwise relationships are built using a structural self-interaction matrix (SSIM), and the influence of barriers is quantified. Different types of barriers to implementing Industry 4.0 in inventory management systems are summarized in Table 4.

Although the study is based on a small number of expert participants, several measures were implemented to enhance the reliability and validity of the findings. Participants were carefully selected from diverse retail sub-sectors and managerial levels to capture a range of perspectives on barriers to Industry 4.0 adoption. Responses for the Structural Self-Interaction Matrix (SSIM) were systematically cross-verified across participants, and any discrepancies were resolved through iterative clarification until consensus was achieved. The ISM and MICMAC procedures were applied rigorously, including transitivity checks and systematic calculation of driving and dependence powers, providing a structured framework for analysing barrier relationships. These steps, along with transparent documentation of the methodology, ensured methodological rigour and internal consistency. The findings are exploratory and intended to provide initial insights, forming a foundation for future studies with larger samples.

SSIM matrix for all participants is shown in Table 5, which shows the interrelationship between different barriers in applying Industry 4.0 in inventory systems.

Transitivity and reachability are two important rules in ISM analysis. In this regard, a reachability matrix is formed to represent how barriers affect one another and how far a barrier reaches others. The main difference between the reachability matrix and SSIM is the numerical binary nature of the reachability matrix, which allows mathematical calculations for the problem. SSIM is translated into binary values based on the transformation rule, which is presented in Table 6.

Another rule applied in this process is the transitivity rule of ISM, which is used to determine the final reachability matrix. Checking for transitivity is important because ISM relies on expert opinions, which may sometimes be inconsistent. The final reachability matrix, after applying the transitivity rule, is calculated for all participants. The results for all participants are presented in  Appendix 1, with an example for Participant 1 shown below in Table 7. The cells marked with stars indicate the entries that changed to 1 as a result of applying the transitivity rule.

The final reachability matrix is the base for drawing the hierarchy diagram of interconnectivity, which is the goal of ISM analysis; however, it is still not clear what barriers are antecedent to other barriers. When doing the level partitioning of the reachability matrix, the antecedent set includes the barrier and barriers that caused this barrier. The intersection set is barriers existing in both the antecedent set and reachability set. This process is repeated until all barriers have been allocated to a level. The level partitioning of the reachability matrix for the first participant is shown in Table 8. This matrix for the other three participants is proposed in  Appendix 2.

Table 9 summarises the number of direct influences reported by each participant for each barrier, capturing the essence of the SSIM and initial reachability matrices. Barriers with a higher number of direct influences, such as BA1 and BA2, indicate strong causal potential, suggesting they act as foundational drivers in the adoption of Industry 4.0 in inventory systems. In contrast, barriers with fewer direct influences appear more dependent, highlighting their reliance on other barriers to manifest their effects. Notably, differences across participant responses reveal variability in how barriers are perceived and prioritised, reflecting potential subjectivity in expert judgement. This variability underscores the need for cautious interpretation, as the hierarchical positioning of barriers may differ depending on individual perspectives rather than representing uniform systemic behaviour.

The obtained partition levels are used to draw ISM diagrams which are represented by arrows, nodes, and barrier numbers. In the elaborated diagram, only the transitive links and their interpretation is crucial are proposed in the diagram. The bottom level of the diagram is composed of the barriers with the highest partition level numbers and the diagram is filled upward until all barriers are stored in a designated level. Finally, for each barrier, the antecedent set is used to configure the inward arrows for each level barrier. The final ISM diagram for the barrier’s relationship determined by participants is drawn in Figure 1.

Participant 1 believes that lack of management support and government reluctance are at the bottom level of the hierarchy, as these two barriers can give rise to other barriers. Similarly, Participants 3 and 4 also identify government reluctance as a foundational barrier, aligning with Participant 1's view. In contrast, Participant 2 considers lack of coordination and collaboration to be the lowest-level barrier in the hierarchy. Therefore, it is important to keep in mind that a barrier that is allocated to the bottom level does not mean that these barriers are the most impactful ones as they may cause other barriers to have more negative impacts in some cases. In this case, MICMAC analysis can be used to determine the driving power of each barrier.

When multiple variables exist in a system, the interrelationship between the variables can be interpreted considering direct and indirect relationships. Analysing direct relationships is performed by the ISM approach, and indirect relationships are analysed by MICMAC analysis to propose a more in-depth analysis of barriers. MICMAC is a method from linear algebra that produces a matrix by multiplying two matrices (Jothimani et al., 2015).

The first step in analysing the indirect relationship between the barriers is to generate a direct relationship matrix “A”, which is derived from the final ISM diagram. Here, the transitive relationship between the barriers is neglected, and diagonal elements of the matrix are changed to zero. The driving power of each barrier is determined by adding the number 1 in rows, and the dependence power is measured by adding the number 1 in columns. In this regard, if variable 1 affects variable 2 and variable 2 affects variable 3, then variable 1 affects variable 3 indirectly. These indirect relationships cannot be signified in the direct relationship matrix A. Therefore, matrix A is squared, and the second order of this matrix is generated. The new matrix is analysed to see if the order for driving power and dependence has changed. If so, this process is repeated, and matrix A is multiplied $n$ times to attain interconnecting variables of $n$ th order. This process stops when the $n$ th matrix approaches a stable state and driving power equals dependence power in the $(n−1)$ interaction and this process is called saturation.

Driving and dependence power scores are the most significant outcomes of MICMAC analysis. Dependence power indicates the degree to which a barrier relies on other barriers to show up. Meanwhile, driving power defines a barrier’s level of impact on other barriers. In order to find the elements of the indirect relationships’ matrix amongst barriers, ISM diagram arrows define the direct reachability for a barrier. Table 10 summarizes the direct relationships between barriers for participant 1. Similar tables are proposed for other participants in  Appendix 2.

The matrix mentioned in Table 10 is multiplied n times until it reaches the saturation state for both driving and dependence powers. The $(n−1)$ matrix which has a stabilized dependence and driving power is proposed in Table 11. Direct dependencies between barriers that are indicated by other participants are summarized in  Appendix 3.

Table 10 is obtained using the ranking of the matrix $An$ dependency and driving power. The total dependence power of each barrier is calculated by summing up different column elements while the driving power is measured by summing up the row elements in Table 11. Thereafter, ranking the impact of barriers can be measured using the total value of driving or dependence power in comparison to other barriers. Indirect relationship matrix $A10$ for the rest of the participants is proposed in  Appendix 4.

Table 12 illustrates that $A10$ have the same ranking for driving power and dependence power which indicates that $A10$ is in the saturated state for indirect matrix $A$⁠. The results for driving power and dependence power for the rest of the participants are proposed in Appendix 7. Then, MICMAC analysis categorizes the barriers which are classified into four clusters as follows.