Exploring capabilities enabled by smart industrial products: a framework for fully autonomous production planning and control Open Access

Industrial systems have undergone profound changes under the Industry 4.0 paradigm, driven by digital transformation and characterized by the convergence of cyber-physical technologies (Ali and Essien, 2025; Houshyar et al., 2025; North et al., 2020). Recent research underscores that the strategic integration of digital technologies is no longer optional but a fundamental requirement for enhancing manufacturing performance and organizational competitiveness (Faisal, 2023; Lu et al., 2025; Miao, 2022; Santos et al., 2024). In this context, Smart Industrial Products (SIPs) have emerged as a key enabler of modern manufacturing by integrating physical and digital components to support automation, operational intelligence, and data-driven decision-making. By enhancing operational visibility and responsiveness to market changes, SIPs strengthen organizational decision-making in increasingly dynamic environments (Kahle et al., 2020; Maia dos Santos et al., 2025; Raff et al., 2020).

SIPs are generally defined across three interconnected dimensions—physical, connected, and intelligent—that together enable the seamless flow and interpretation of information across products, machines, and systems. Equipped with embedded systems, these products autonomously collect, analyze, and share real-time data related to customers, manufactured goods, production processes, and the operating environment (Maia dos Santos et al., 2025; Pissardini et al., 2024). The adoption of SIPs, therefore, creates new opportunities for companies to strengthen their Production Planning and Control (PPC) functions. By enabling smart sensing, decentralized decision-making, and adaptive coordination mechanisms, SIPs support more robust and flexible PPC systems that can respond to frequent demand fluctuations, process disruptions, and changes in resource availability (Borangiu et al., 2014; Oluyisola et al., 2022; Rahmani et al., 2022). However, the mere presence of these technological features does not fully explain how this potential is translated into sustained improvements in PPC performance.

In this regard, understanding the impact of SIPs on PPC requires moving beyond a purely technological description. It requires explaining how these SIPs can enable operational capabilities that shape PPC performance. Grounded in the Resource-Based View (RBV), this study adopts a capability-oriented perspective. Following Callefi et al. (2024a), capabilities are defined as abilities that emerge specifically from the adoption of SIPs within PPC. The RBV emphasizes that competitive advantage stems not merely from possessing resources, but from the firm's capacity to integrate and coordinate digital and physical assets into interdependent capabilities that support superior performance (Amit and Schoemaker, 1993; Barney, 2001). This view challenges the assumption that automation and intelligence naturally follow technology adoption and underscores the need for a deliberate, cumulative development of organizational capabilities. Accordingly, SIPs are understood as enablers that progressively enhance the autonomy, intelligence, and adaptability of PPC systems.

Although prior research has examined SIPs in PPC contexts (Bueno et al., 2020; Oluyisola et al., 2020, 2022), most studies focus on isolated applications, such as production monitoring (Borangiu et al., 2014; Meyer et al., 2009), process optimization (Lenz et al., 2020; Pardo et al., 2020), or material flow management (Thürer et al., 2021). These studies rarely consider how different SIP-enabled capabilities interconnect within PPC. Other contributions, such as Raff et al. (2020), discuss the intrinsic characteristics that make a product “smart” but do not examine how these characteristics translate into coordinated capability systems within PPC. Although Pissardini et al. (2024, 2025) identify a set of SIP-enabled capabilities related to PPC, their analyses remain primarily descriptive, thereby leaving unchallenged the implicit assumption that such capabilities evolve linearly or additively. As a result, the literature tends to treat SIP-enabled PPC as a collection of discrete functionalities rather than as a systemic architecture of interdependent capabilities. This perspective obscures the dependency structures that constrain or enable the progression toward autonomous decision-making in PPC. Without a clear understanding of these dependencies, organizations risk investing in advanced technologies that cannot be effectively operationalized. Consequently, the literature still lacks a structured and theoretically grounded explanation of how SIP-enabled capabilities develop, interrelate, and collectively drive PPC autonomy. In this regard, three research questions guide the study:

To answer these questions, this study adopts a multi-phase methodological approach grounded in the RBV. First, a Systematic Literature Review (SLR), supported by content analysis, is conducted to identify SIP-enabled capabilities in PPC (RQ1). Next, Interpretive Structural Modeling (ISM) and fuzzy cross-impact matrix multiplication applied to classification (MICMAC) analysis are applied to examine the interrelationships among these capabilities and to classify them according to their driving and dependence powers (RQ2). Rather than assuming a predefined or linear progression of capability development, these methods are used to uncover the dependency structures that shape how SIP-enabled capabilities can be effectively combined and developed. The resulting hierarchical framework delineates the developmental pathway through which SIP-enabled capabilities evolve toward PPC autonomy (RQ3).

From a theoretical perspective, this study employs the RBV as an analytical framework to interpret how SIP-enabled capabilities interconnect and reinforce one another in shaping autonomous decision-making within PPC. Rather than extending RBV theory, the study uses it as a lens to analyze the systemic configuration of digital and operational capabilities that underpin Industry 4.0 transformations. In this sense, the proposed framework departs from conventional digital maturity models that prescribe sequential stages of adoption by conceptualizing PPC autonomy as an emergent outcome of interdependent capability layers, rather than as the result of linear technological sophistication. From a practical perspective, the framework provides a prescriptive roadmap for managers to structure digital transformation strategies, prioritize capability development, and align technology investments with progressively higher levels of PPC autonomy.

The remainder of this paper is organized as follows. Section 2 reviews the literature on SIPs and PPC, outlining the theoretical background and identifying the research gap. Section 3 describes the research methodology. Sections 4 and 5 present and discuss the results, culminating in the proposed hierarchical framework. Finally, Section 6 concludes the paper by outlining the theoretical and managerial implications and directions for future research.

This section first reviews the literature on SIPs and their role in PPC, highlighting the current state of research and existing gaps. It then presents the theoretical foundation of this study, based on the RBV and a capability-oriented perspective.

SIPs are widely recognized in the Industry 4.0 literature as key enablers of decentralized and data-driven PPC, due to their embedded sensing, communication, and computational capabilities (Meyer et al., 2009; Porter and Heppelmann, 2015). Prior studies show that SIPs can enhance flexibility, responsiveness, and real-time coordination by integrating product-level intelligence into PPC routines (Borangiu et al., 2014; Oluyisola et al., 2020). As cyber-physical connectivity has matured, research has increasingly shown that SIP-generated data can be embedded in planning and scheduling processes to reduce latency between the execution and decision layers (Rahmani et al., 2022; Thürer et al., 2021). Despite these advances, the dominant stream of research remains technology-centric, often implicitly assuming that higher levels of PPC autonomy naturally emerge from increased data availability, connectivity, or algorithmic sophistication. Consequently, the organizational and capability-related conditions required to translate SIP-enabled technological potential into sustained autonomous decision-making in PPC remain underexplored, highlighting the need for a structured, capability-based perspective.

This technological emphasis is evident in studies that describe transitions from connected to intelligent PPC (Oluyisola et al., 2020) or demonstrate the performance benefits of data-driven architectures (Rahmani et al., 2022), yet stop short of explaining how the underlying capabilities develop or interrelate. Research on optimization, scheduling, and material flow management similarly isolates specific functionalities without addressing how they contribute to the formation of cumulative capability (Pardo et al., 2020; Thürer et al., 2021). Consequently, PPC evolution is often framed as a sequence of functional upgrades, thereby obscuring the possibility that some capabilities may be prerequisites for others, or that advanced forms of autonomy may fail without prior capability consolidation.

Recent contributions incorporating artificial intelligence, machine learning, and digital twins further extend the technological scope of SIP-enabled PPC. For example, Oluyisola et al. (2022) demonstrated that combining SIP-generated data with cloud analytics and ML supports dynamic rescheduling and short-term decision-making, while Padovano et al. (2021) highlighted how digital twins synchronize PPC and maintenance planning to enable prescriptive behavior. Similarly, Cañas et al. (2022) positioned AI, digital twins, and big data analytics as core elements of Smart PPC 4.0. However, these studies tend to reinforce a solution-driven narrative in which autonomy is equated with the deployment of advanced analytical tools, rather than examining how such technologies are embedded within coherent, interdependent capability systems in PPC.

An essential step toward addressing this limitation is provided by Pissardini et al. (2025), who structured SIP-enabled capabilities using ISM and Fuzzy MICMAC. Their study confirms the feasibility of modeling interdependencies among capabilities. However, its cross-functional scope, covering maintenance, logistics, sustainability, and quality, limits its ability to explain how these interdependencies unfold specifically within the PPC function. As a result, the study does not fully capture how capability structures shape the progression toward autonomous planning and control in PPC.

In this regard, the literature reveals not only the absence of PPC-specific capability hierarchies but also a deeper conceptual limitation: autonomy in PPC is predominantly treated as the outcome of technological advancement rather than as an emergent organizational property arising from the accumulation and orchestration of interdependent capabilities. Addressing this limitation requires a structured, theoretically grounded framework that explains how SIP-enabled capabilities interact, evolve, and collectively support increasing levels of autonomy in PPC. This study responds to that need by proposing a hierarchical capability-relationship framework explicitly adapted to SIP-enabled PPC.

The RBV offers a well-established theoretical foundation for understanding how firms achieve competitive advantage through unique configurations of resources and capabilities (Amit and Schoemaker, 1993; Barney, 2001). Central to RBV is the argument that organizational performance depends not merely on the possession of valuable resources, but on the firm's ability to integrate, coordinate, and reconfigure those resources into capabilities that support sustained advantage (Teece, 2018). Accordingly, RBV directs analytical attention to how capabilities are formed through resource combinations and complementarities, making it particularly suitable for studies that seek to explain how operational capabilities are built and combined, rather than how firms intentionally renew their resource base over time, as emphasized in dynamic capability theory (Teece, 2007; Teece et al., 1997). In Industry 4.0 contexts, this distinction is especially relevant, as digital and physical assets—such as connected products, intelligent systems, and data infrastructures—create strategic value only when they are integrated into organizational capabilities that enable learning, adaptation, and effective decision-making (Raff et al., 2020).

Grounded in this RBV logic and building on the capability-oriented interpretation proposed by Callefi et al. (2024b), this study defines capabilities as abilities enabled by one or more technologies. This definition provides the conceptual bridge between resources and PPC outcomes, clarifying that SIP-enabled capabilities are not inherent features of smart products, but organizational outcomes that emerge from how these technologies are configured, governed, and embedded within PPC processes. By adopting this view, the analysis remains focused on the accumulation and integration of operational capabilities, rather than on firm-level sensing, seizing, and transforming processes that are central to dynamic capability theory (Teece, 2007). This perspective directly challenges the implicit assumption, prevalent in much of the PPC literature, that technological sophistication alone is sufficient to generate autonomous behavior.

Following this capability-based interpretation, applying the RBV as an analytical lens conceptualizes autonomy in PPC not as a discrete, standalone capability but as a higher-order outcome arising from the interaction of multiple, interdependent lower-order capabilities. Although related perspectives, such as resource orchestration theory, emphasize managerial actions to structure, bundle, and leverage resources (Sirmon et al., 2011), the present study deliberately shifts the analytical focus from managerial action to capability structure, examining the structural interdependencies among SIP-enabled operational capabilities. This distinction is particularly critical in PPC settings, where decision-making authority, coordination mechanisms, and human intervention remain tightly coupled (Bueno et al., 2025). By linking RBV insights with empirical modeling through ISM and Fuzzy MICMAC, this study moves beyond descriptive accounts of digital transformation. It provides a systemic explanation of how SIP-enabled capabilities evolve, reinforce one another, and shape the pathway toward autonomous PPC. In this sense, PPC autonomy is reframed not as a technological endpoint but as an emergent property of structured capability architecture, offering both theoretical clarity and practical guidance for organizations navigating Industry 4.0 transformations.

Grounded in a capability-oriented perspective based on the RBV theory, this study employs a mixed-methods research design comprising four sequential and complementary stages (Figure 1) that combine qualitative and quantitative techniques to identify, validate, and structure SIP-enabled capabilities within the PPC function. The integration of methods was essential not only because of the exploratory nature of the research problem, but also because understanding PPC autonomy requires uncovering interdependencies among capabilities rather than assessing them in isolation.

The first stage involved conducting an SLR, supported by content analysis, to identify and conceptually delimit SIP-enabled capabilities in PPC, thereby establishing a robust theoretical foundation for the study. This step ensured that capabilities were derived from accumulated scholarly knowledge rather than from predefined assumptions of maturity. The second stage comprised expert validation, integrating insights from both academia and industry to refine the initial list of capabilities and confirm their applicability to real-world PPC contexts. This stage strengthened construct validity and aligned theoretical abstraction with operational relevance. In the third stage, ISM was applied to analyze the contextual and hierarchical relationships among the validated capabilities. ISM was particularly suited to this research objective because it allows the explicit modeling of dependency structures, revealing how lower-level capabilities condition the emergence of higher-order ones. Finally, the fourth stage employed fuzzy MICMAC analysis to quantify the driving and dependence power of each capability. This technique complements ISM by accounting for uncertainty and variation in expert judgments, thereby enabling a more nuanced classification of patterns of capability influence.

The combined use of ISM and fuzzy MICMAC was deliberately chosen over traditional multi-criteria decision-making (MCDM) approaches because it is well-suited to exploratory research domains characterized by incomplete knowledge and strongly interdependent constructs. These characteristics make conventional ranking-based methods less appropriate for the present study. Unlike traditional MCDM methods, which emphasize ranking alternatives, the ISM–fuzzy MICMAC combination focuses on uncovering causal and hierarchical relationships. This focus aligns more closely with the study's theoretical objective of explaining how autonomy in PPC emerges from capability configurations, rather than from isolated evaluations. Previous studies have demonstrated the suitability of this approach for such purposes. As shown by Callefi et al. (2024b) and Bianco et al. (2023b), the ISM–fuzzy MICMAC approach provides a robust analytical framework for examining hierarchical structures in emerging and partially structured domains such as SIP-enabled PPC.

To perform the SLR, this study followed the three steps proposed by Denyer and Tranfield (2009): planning, conducting, and disseminating. This protocol ensured transparency, replicability, and theoretical rigor in identifying SIP-enabled capabilities. During the planning phase, the study's main parameters were defined, including the research questions, time frame, and inclusion criteria. In the conducting phase, objective filters specified in the research protocol were applied sequentially to minimize selection bias. Titles and abstracts were first screened. This step was followed by full-text analysis of the remaining studies. In the dissemination phase, all selected papers were coded, classified, and analyzed using NVivo 13 to extract, compare, and synthesize findings on SIP-enabled capabilities within the PPC function.

The SLR aimed to identify and consolidate the capabilities enabled by adopting SIPs in the PPC function, rather than assuming predefined capability categories. An inductive coding strategy was adopted, allowing capabilities to emerge from the literature. RQ1 guided the selection of the main keywords, which, in turn, supported the development of the search string. The review was conducted across three databases—Web of Science, Scopus, and Engineering Village. These databases were selected for their complementary coverage of the literature on engineering, operations management, and industrial systems. This multi-database strategy reduced the risk of domain bias and aligned with prior systematic reviews in the smart PPC research domain (e.g. Bueno et al., 2020; Rahmani et al., 2022). The research protocol developed to guide the SLR process is presented in Table 1.

The initial search returned 148,398 papers. After applying predefined filters, 8,005 papers remained. Following duplicate removal, 4,877 papers were screened by title and abstract. At this stage, studies were excluded if SIPs were treated solely as technologies, with no implications for PPC decision-making or capability development. This process yielded 167 papers that underwent full-text analysis, resulting in a refined sample of 93 studies. Among these, 81 explicitly addressed SIP-enabled capabilities related to PPC. An additional five papers were identified through snowball sampling, yielding a final sample of 86 studies. Figure 2 summarizes the screening flow, while Table 2 details the inclusion and exclusion criteria.

NVivo 13 was used for coding and data analysis, during which 23 SIP-enabled PPC capabilities were initially identified. To avoid conceptual redundancy while preserving analytical completeness, a cluster analysis based on semantic similarity was conducted following Bianco et al. (2023b). Pearson's correlation coefficients above 0.70 were used as an indicative threshold for potential unification. Importantly, this analysis did not eliminate capabilities a priori. Instead, it supported theory-driven consolidation decisions. Two authors independently reviewed each clustering iteration and engaged in multiple rounds of discussion until complete conceptual agreement was achieved. As a result, the initial list of 23 capabilities was consolidated into 12 distinct and theoretically coherent capabilities. Representative excerpts from the reviewed studies were retained within each code to reinforce conceptual validity and empirical grounding.

The expert validation followed a four-step iterative cycle, adapted from Callefi et al. (2022) and Saunders et al. (2012). This cycle comprised: (1) presentation of the initial capability list derived from the SLR; (2) joint discussion of the semantic clustering results; (3) incorporation of expert feedback through iterative refinement; and (4) final confirmation of the refined capability set. This structured procedure was adopted to enhance traceability, transparency, and methodological rigor, ensuring that the resulting capability set reflected shared expert understanding rather than isolated judgments.

Within this process, experts were treated as informed judges. Their experience-based assessments supported the validation, refinement, and consolidation of capability constructs, consistent with established approaches to expert-based consensus validation (Okoli and Pawlowski, 2004). Accordingly, the expert-based validation stage aimed to refine, clarify, and confirm the relevance of the capabilities identified through the SLR, ensuring conceptual robustness and alignment with both academic reasoning and real-world PPC practice. In line with prior capability-identification studies in Industry 4.0 and PPC contexts (e.g. Callefi et al., 2022; Pissardini et al., 2024), this phase was designed to achieve expert consensus, rather than to measure inter-rater agreement statistically.

Accordingly, five experts with extensive experience in Industry 4.0, SIPs, and PPC were consulted. All experts currently hold academic positions as researchers and professors. In addition, they possess substantial, ongoing industrial experience and are actively engaged as consultants on PPC- and Industry 4.0–related projects. Each expert holds a PhD in Production Engineering and has more than ten years of combined academic and industrial experience. This experience encompasses applied research, industrial digital transformation initiatives, and the implementation of smart manufacturing and PPC systems. This composition was intentionally selected to balance theoretical rigor with practical relevance. Table 3 summarizes the characteristics of the experts involved in the capability validation stage, including their current roles, academic background, and years of experience in PPC or Industry 4.0 contexts. Individual semi-structured interviews were conducted, each lasting approximately 50 min. All interviews were recorded, transcribed verbatim, and systematically coded to ensure transparency and traceability of the validation process.

During the interviews, experts were asked to evaluate (1) the conceptual distinctiveness of each capability, (2) the clarity and adequacy of its definition, and (3) the relevance and applicability of each capability within the PPC context. The evaluation deliberately focused on conceptual validity and construct clarity, rather than on technological feasibility or causal relationships. Following practices adopted in prior capability-based studies (Callefi et al., 2022; Pissardini et al., 2024), consensus was assessed qualitatively through expert convergence. A capability was excluded or merged when at least two experts independently questioned its relevance to PPC or explicitly indicated that it overlapped conceptually with another capability and should therefore be consolidated. Capabilities that did not trigger such objections were retained and refined.

To enhance reliability, two authors independently reviewed all proposed changes and resolved discrepancies through iterative discussion until complete agreement was achieved. As a result of this expert-based validation stage, the initial set of SIP-enabled capabilities identified through the SLR was refined and consolidated into a final set of twelve (12) validated capabilities. These capabilities met the minimum consensus criteria defined in this study and were considered conceptually distinct, clearly defined, and relevant to the PPC context. This validated capability set constitutes the final output of the expert validation stage and serves as the direct input for the subsequent ISM and fuzzy MICMAC analyses.

The ISM approach was used to establish the relationship between the capabilities enabled by SIP adoption within the PPC function. ISM was selected because it explicitly models dependency relationships, enabling the transformation of fragmented expert knowledge into a coherent hierarchical structure. The ISM approach offers a structured and interactive learning process, encapsulating specialized knowledge about a set of elements that describe a system and how those elements are aggregated (Godinho Filho et al., 2025). It transforms unclear, poorly articulated knowledge and mental system models into visual, well-defined systems (Ruiz-Benitez et al., 2017). The ISM approach comprises four steps, as proposed by Farris and Sage (1975) and Warfield (1974), and is detailed below.

The ISM model was developed based on input from a panel of 15 experts with complementary academic and professional backgrounds in Industry 4.0, operations management, and SIP-enabled PPC systems (see Table 4). In accordance with best practices for complex system modeling (Saunders et al., 2012), the panel intentionally combined researchers and practitioners, including professionals with dual academic–practical experience. This composition ensured both theoretical robustness and practical relevance.

The expert panel exhibits a methodologically homogeneous profile with respect to formal education. At the same time, it shows significant heterogeneity in industrial exposure and organizational contexts. Most experts hold a PhD in Industrial or Mechanical Engineering (n = 11), while the remaining panel members hold postgraduate (n = 1) and bachelor's-level engineering qualifications (n = 3). This distribution ensures a shared conceptual foundation in systems thinking and operations management. Regarding professional experience, the panel demonstrates a high level of maturity. Eleven experts report 10–20 years of experience in SIPs and operations management, whereas four senior experts report over 20 years of experience in the field. The experts' backgrounds span key segments of the industrial value chain, including capital goods, automotive, consumer goods, biorefinery, and technology consulting. While organizational experience is predominantly concentrated in large enterprises (n = 12), the inclusion of medium (n = 2) and small companies (n = 1) helps mitigate potential sample bias. This diversity strengthens the reliability and applicability of the ISM results across different industrial settings.

Experts assessed pairwise relationships among the 12 SIP-enabled PPC capabilities identified through the SLR using the standard ISM symbols (V, A, X, O). These symbols were used to indicate the direction of influence between capability pairs (Farris and Sage, 1975; Warfield, 1974). To enhance methodological rigor and reduce subjectivity, a minimum agreement threshold of 75% was adopted for each relationship. Given the absence of a universally prescribed consensus metric in ISM studies, this threshold reflects a context-appropriate strategy aligned with recommendations in MCDM research (Massam, 1988). When divergences occurred, the experts were consulted until convergence was achieved.

Input was collected from a panel of 15 experts with complementary academic and professional backgrounds in Industry 4.0, operations management, and SIPs, ensuring a balance between theoretical rigor and practical relevance, as recommended for complex-system modeling (Saunders et al., 2012).

The validated Structural Self-Interaction Matrix (SSIM) was transformed into the Initial Reachability Matrix (IRM) using the standard binary conversion rules proposed by Rajesh et al. (2013), enabling the formal representation of direct contextual relationships among the identified capabilities. Subsequently, the Final Reachability Matrix (FRM) was derived by applying the transitivity principle. In contrast to the standard ISM procedure, which applies full transitive closure, this study restricts transitivity to first-order relationships only. Accordingly, an indirect relationship i→k was incorporated into the FRM only when a single intermediate capability j existed such that both i→j and j→k were present in the IRM. No further iterative transitivity was applied beyond this single step.

This methodological choice was motivated by the high relational density observed in the system. In such contexts, applying full transitive closure often yields highly saturated reachability matrices and flattened hierarchical structures, reducing analytical tractability and limiting the ISM model's ability to distinguish hierarchical levels and structural roles among elements clearly. Similar methodological choices have been adopted in several established applied ISM studies to preserve interpretability and hierarchical discrimination, particularly in complex systems characterized by dense interconnections (e.g. Barrionuevo et al., 2025; Bianco et al., 2023a; Bianco et al., 2023b; Gan et al., 2018; Poduval et al., 2015).

From a conceptual standpoint, ISM functions as a heuristic and interpretive structuring tool rather than a model of strong causality. While full transitivity maximizes logical completeness, it may introduce higher-order transitive noise in dense systems (Warfield, 1974). Restricting transitivity to first-order relationships, therefore, serves as a structural regularization mechanism, preserving the most cognitively meaningful indirect dependencies and avoiding excessive flattening of the structure.

Based on the resulting FRM, reachability and antecedent sets were derived for each capability following the classical ISM formulation (Warfield, 1974). The intersection of these sets was used to determine hierarchical levels. Capabilities whose reachability and intersection sets coincided were assigned to the same level, removed from subsequent iterations, and the procedure was repeated until all capabilities were hierarchically classified.

The final ISM model was derived as a directed graph representing the structural relationships among the 12 SIP-enabled capabilities in the PPC function. Directed links indicate dependency relationships. Higher-level capabilities were positioned above those on which they depend, in accordance with established ISM modeling conventions (Farris and Sage, 1975). All transitive links were removed, thereby retaining only the most meaningful direct relationships.

While the ISM model evaluates relationships among capabilities in binary terms, the strength of influence between capabilities may vary in intensity, particularly in complex, partially structured systems such as SIP-enabled PPC (Barrionuevo et al., 2025). To capture these variations and complement the hierarchical insights provided by ISM, a fuzzy MICMAC analysis was conducted. This approach enables the assessment of both driving and dependence power under conditions of uncertainty, thereby offering a more nuanced understanding of capability interdependencies.

The fuzzy MICMAC procedure builds upon the Initial Reachability Matrix (IRM) derived from the ISM analysis. First, a Binary Direct Reachability Matrix (BDRM) was obtained by setting all diagonal elements of the IRM to zero, thereby excluding self-influence. The BDRM represents the presence or absence of direct relationships among capabilities. Subsequently, the BDRM was transformed into a Fuzzy Direct Reachability Matrix (FDRM) by fuzzifying the strength of influence between capability pairs. This fuzzification was based on the number of experts agreeing on each directional relationship. Five linguistic terms, ranging from No influence to Very strong influence, were used, with the corresponding numerical values defined in Table 5, following established fuzzy MICMAC procedures (Barrionuevo et al., 2025). Non-zero entries in the BDRM were replaced by their respective fuzzy values to obtain the FDRM.

To account for both direct and indirect relationships among capabilities, the FDRM was iteratively stabilized using fuzzy matrix multiplication until convergence was achieved, that is, until the row and column sums no longer changed. As noted by Mota et al. (2021), various fuzzy composition operators can be employed; in this study, the max–min composition was adopted because it is particularly suitable for identifying dominant minimal-influence paths in complex systems.

Matrix multiplication was then calculated using the rule described below to achieve the fuzzy MICMAC stabilized matrix:

Where X = $[Xik]$ and Y = $[Ykj]$

To perform the multiplication of fuzzy matrices using this operator, MATLAB® was used, and the Fuzzy MICMAC-stabilized matrix was obtained as the output.

From the fuzzy MICMAC-stabilized matrix, the driving power of each capability is calculated by summing the values in its row, indicating the extent to which it influences other variables. Conversely, the dependence power is computed as the sum of the values in each column, indicating the extent to which different elements influence a given capability. Capabilities with high driving power tend to trigger changes in other elements, while those with high dependence power are primarily reactive within the system.

To illustrate these relationships, a four-quadrant diagram was developed to categorize capabilities into four types. Autonomous capabilities exert minimal influence and are minimally influenced, playing an isolated role in the system. Dependent capabilities exhibit high dependence and low driving power, meaning other capabilities strongly influence them but exert limited influence in turn. Linkage capabilities are both influential and dependent, often serving as intermediaries within the system. Finally, driving capabilities exhibit high driving power and low dependence, actively shaping the system's capability structure.

Each method in the previously described mixed-method approach yields distinct yet complementary results. The following subsections detail the outcomes of each methodological step.

The initial phase of this research involved an SLR, which laid the foundation for subsequent steps. The outcomes of this phase also serve as input for the next phase. By analyzing the 86 papers, we derived an initial list of capabilities enabled by adopting SIPs in the PPC function. A total of 23 PPC-enabled capabilities were identified in the literature. To refine this list, a cluster analysis based on semantic similarity was performed. In addition, the definitions for each capability were linked to a code representing that capability. This step reduced the list to 12 capabilities, unifying 11 that, although described differently in the literature, shared the same conceptual meaning. Finally, five academic experts analyzed the refined list, validating the clustering process. The final list of identified capabilities, together with their codes, definitions, and the authors citing each capability, is presented in Table 6.

This phase presents the steps and outcomes of the ISM analysis, defining the structural hierarchy among the PPC-enabled capabilities. All relevant tables are provided in  Appendix 1. The matrix was constructed from expert responses, with the most frequent response (mode) selected for each capability pair, as shown in Table A1 of  Appendix 1. This matrix was then converted to binary form (Table A2) and, after applying the transitivity rule, to a final form (Table A3). Table A4 presents the reachability, antecedent, and intersection sets used to determine the model levels. Based on this structure, a graphical representation of the ISM model was developed and is presented in Figure 3. The twelve PPC-enabled capabilities were organized into five hierarchical levels, discussed in the following section.

This section presents the results of the Fuzzy MICMAC analysis, with all matrices associated with this stage provided in  Appendix 2, and follows the methodological procedures described previously. The analysis begins with constructing the BDRM matrix (Table A5), obtained by setting all diagonal entries of the Initial Reachability Matrix to zero. The FDRM matrix (Table A6) was then generated using the characteristic values associated with the number of experts assigning V, A, X, or O ratings in the SSIM matrix (Table A1 in  Appendix 1). The stabilized matrix was calculated through matrix multiplication, as detailed in the Methodology section, and is presented in Table A7. Based on this stabilized matrix, a four-quadrant diagram was developed to illustrate the driving forces and interdependencies among the identified capabilities. The resulting classification is shown in Figure 4.

Figure 4 presents the results of the Fuzzy MICMAC analysis, in which the twelve SIP-enabled capabilities are positioned within the four conventional quadrants of the driving power–dependence power diagram. As established in the MICMAC methodology, these quadrants are strictly defined by the relative magnitude of driving and dependence power along the two axes and represent distinct structural roles within the system (i.e. driving, linkage, dependent, and autonomous factors). In Figure 4, the elements labeled as clusters were identified through a visual grouping based on the proximity of capabilities in terms of their driving and dependence power values, rather than through a rigid one-to-one correspondence with the quadrants themselves. Accordingly, Cluster 1 is predominantly located within the driving factors quadrant and includes PPC1 and PPC2, which exhibit high driving power and low dependence, indicating their strong influence on other capabilities. Cluster 2 is located within the linkage factors quadrant and comprises PPC8, PPC9, and PPC12, reflecting their strong bidirectional interactions and structurally sensitive role within the system. Cluster 3 also occupies the linkage region but extends toward higher dependence values, encompassing PPC3, PPC5, PPC6, and PPC7, which remain influential but are more strongly conditioned by other capabilities, thus representing a transitional structural position. Finally, Cluster 4 is predominantly positioned within the dependent factors quadrant and includes PPC4, PPC10, and PPC11, whose development depends strongly on the prior consolidation of other capabilities. Thus, while the quadrant structure in Figure 4 follows the conventional Fuzzy MICMAC classification, the clusters highlighted in the figure represent data-driven visual aggregations derived from the distribution of driving and dependence values, rather than strictly algorithmic quadrant assignments. At this stage, no additional interpretive aggregation is introduced beyond the visual identification of these groupings.

Notably, none of the capabilities are classified as autonomous factors, which are defined by simultaneously low driving and low dependence power. Rather than indicating a limitation of the analysis, this absence reflects the high degree of systemic interdependence among SIP-enabled capabilities in PPC. In the context of PPC, no capability operates in isolation: even advanced functions require foundational data, coordination, and integration capabilities to be operationalized. Thus, autonomy in PPC does not emerge from stand-alone capabilities but from the orchestration and mutual reinforcement of multiple interdependent capabilities. The absence of autonomous factors, therefore, reinforces the framework's internal coherence. It supports interpreting PPC autonomy as an emergent, system-level property rather than as an isolated capability.

Based on the results presented in the previous sections, Figure 5 illustrates the proposed framework by depicting the structural hierarchy of SIP-enabled capabilities within the PPC function. Grounded in the RBV, the framework conceptualizes autonomy in PPC not as a standalone capability, but as an emergent outcome of the coordinated integration, orchestration, and reconfiguration of SIP-enabled resources into organizational capabilities. The consolidation of results from ISM and Fuzzy MICMAC was intentionally adopted to capture complementary but distinct structural dimensions of the system. While ISM reveals the hierarchical dependency structure among capabilities, indicating which configurations must precede others, Fuzzy MICMAC provides additional insight into the systemic positioning and relative structural influence of capabilities within that hierarchy. Integrating these two perspectives enables a more comprehensive representation of how capability configurations evolve and interact. Accordingly, the four layers identified—Connected, Transparent, Autonomous Decision-Making I, and Autonomous Decision-Making II—represent interdependent structural configurations of capabilities that must be developed in relation to one another, as inferred from the joint interpretation of hierarchical dependencies (ISM) and influence–dependence tendencies (Fuzzy MICMAC) derived from expert-based judgments. These layers do not constitute predefined maturity stages or a normative sequence of adoption. Instead, they emerge inductively from systematically elicited expert knowledge, reflecting how practitioners and scholars collectively perceive the structural logic underpinning autonomy in PPC.

From this perspective, higher-order capability configurations cannot be effectively developed or operationalized without the prior consolidation and stabilization of lower-layer configurations, not because of a prescriptive roadmap, but due to the structural constraints and enabling conditions embedded in the system. Progression toward autonomy in PPC is therefore not driven by sequential technological upgrades, but by the accumulation, alignment, and reinforcement of interdependent capability layers. Each layer establishes the organizational and informational conditions required for the emergence of subsequent layers, such that autonomy materializes only when these configurations are coherently aligned. This reinforces the view of PPC autonomy as a systemic, capability-based phenomenon, rather than the outcome of a linear maturity trajectory or a direct result of isolated technology adoption.

The Connected level represents the foundational resource configuration of a SIP-enabled PPC system, enabling basic connectivity among manufacturing resources within a broader Internet of Things (IoT) network. As shown in Figure 5, this level comprises PPC1 (Sensing integration) and PPC2 (Interoperability of smart resources). Both capabilities occupy the lowest positions in the ISM hierarchy (Figure 3). They are classified as driving factors in Cluster 1 of the fuzzy MICMAC analysis (Figure 4), indicating high driving power and low dependence on other capabilities.

From an RBV standpoint, firms achieve competitive advantage not simply by possessing digital resources, but by deliberately deploying and combining them into organizational capabilities that coordinate, integrate, and reconfigure those resources over time. Within the PPC function, PPC1 and PPC2 serve as foundational resource-integrating capabilities, transforming otherwise isolated SIP-related resources into a coherent operational infrastructure. PPC1 enables the systematic capture and sharing of real-time production data, establishing the informational backbone of PPC and creating the conditions for data-driven decision-making that enhances operational efficiency and product quality (Borangiu et al., 2014; Porter and Heppelmann, 2015). PPC2 complements this foundation by ensuring interoperability among heterogeneous manufacturing resources, allowing organizations to leverage analytics to identify patterns that would otherwise remain inaccessible due to data volume and complexity (Meyer and Wortmann, 2010). By supporting plug-and-play integration of SIPs, this capability facilitates the incorporation of new resources. It mitigates coexistence challenges between legacy and smart assets, thereby creating the structural conditions required for higher-order integrative capabilities and, ultimately, for autonomous decision-making in PPC (Abramovici et al., 2017; Leitão et al., 2015; Oluyisola et al., 2022).

Critically, the Connected level does not generate autonomy in itself. Instead, it establishes the necessary conditions for subsequent capability development by integrating digital and physical resources into a coherent infrastructure. Consistent with RBV logic, these foundational capabilities underpin all higher-order SIP-enabled capabilities in PPC, making investments at this level a prerequisite for progression toward adaptive and autonomous control (Meyer et al., 2009; Pissardini et al., 2024; Thürer et al., 2021).

Building on the Connected level, the Transparent level consolidates basic connectivity into system-wide visibility and coordination capabilities. At this stage, SIP-enabled resources are no longer merely connected; instead, they are deliberately deployed to enable organizations to observe, interpret, and coordinate production states across the manufacturing system via machine-to-machine communication. This level comprises PPC8 (Machine-to-Machine integration), PPC9 (Enhanced data collection), and PPC12 (Real-time operational visibility), which occupy a lower-intermediate position in the hierarchy (Figure 5), consistent with their placement in the ISM model (Figure 3) and their classification in Cluster 2 of the fuzzy MICMAC analysis (Figure 4).

PPC8 is widely recognized as a foundational capability for decentralized coordination in PPC, as it integrates sensing and interoperability resources into a unified operational network that enables real-time monitoring of machine status, maintenance conditions, and workloads (Antons and Arlinghaus, 2022a; Porter and Heppelmann, 2015). PPC9 extends this coordination logic by ensuring the systematic collection and communication of high-quality process data, supporting centralized planning combined with decentralized execution (Keshavarzi and Van Den Hoek, 2019; Lenz et al., 2020). PPC12 consolidates these data flows into real-time operational visibility, thereby enhancing responsiveness to disruptions and enabling proactive maintenance and exception handling (Lei et al., 2023).

From an RBV perspective, the Transparent level exemplifies resource orchestration, whereby previously integrated digital resources are transformed into coordinating capabilities that convert dispersed data into shared operational knowledge (Oluyisola et al., 2022). While these capabilities significantly improve planning accuracy and decision quality (Rahmani et al., 2022; Thürer et al., 2021), they primarily support interpretation and coordination rather than autonomous execution. Transparency thus represents a transitional configuration: organizations gain situational awareness and improved control, but remain dependent on human judgment to enact decisions. This observation reinforces that visibility is a necessary but insufficient condition for autonomy in PPC.

At Autonomous Decision-Making Level I, autonomy begins to directly shape the PPC function itself, although human operators on the shop floor largely retain supervisory and intervention roles. This level comprises PPC3 (SIP/Production line auto-reconfiguration), PPC5 (Full cyber-physical-social integration), PPC6 (Remote process control), and PPC7 (Self-optimizing decision performance). These capabilities occupy an upper-intermediate position in the hierarchy (Figure 5), aligned with their placement in the ISM model (Figure 3) and their classification in Cluster 3 of the fuzzy MICMAC analysis (Figure 4), reflecting a higher degree of dependence on lower-level capabilities.

Structurally, this level marks the transition from coordination to adaptive decision-making. PPC3 enables SIPs to autonomously reconfigure production lines by leveraging real-time data, analytics, and digital twins to respond dynamically to performance deviations (Zhang et al., 2020). PPC5 extends this adaptive logic by integrating physical, cyber, and social dimensions, enhancing contextual awareness and coordination across systems and stakeholders (Kahle et al., 2020; McFarlane et al., 2013). PPC6 further operationalizes decision authority by enabling remote configuration and control of production processes, increasing flexibility and safety in dynamic environments (Porter and Heppelmann, 2015). PPC7 embeds learning mechanisms that enable SIPs to improve decision quality over time, drawing on decentralized, experience-based planning principles, such as pheromone-based coordination in adaptive holonic control architecture (ADACOR) architectures (Leitão et al., 2005; Valckenaers et al., 1999).

Within the RBV framework, these capabilities support adaptive and reconfigurative behavior by enabling sensing, learning, and operational adjustment in response to environmental changes. However, autonomy at this level remains essentially embedded within cyber-physical routines rather than entirely reshaping organizational roles and governance structures. This explains their bridging position in the hierarchy: they depend on prior connectivity and transparency while simultaneously creating the conditions for more profound socio-technical transformation in PPC (Derigent et al., 2021).

The highest level of the framework comprises PPC4 (Workforce reconfiguration), PPC10 (Adaptive and resilient production), and PPC11 (Complexity management across product–service offerings). At this stage, decisions generated by SIP-enabled systems are directly implemented on the shop floor, thereby closing the loop among planning, execution, and organizational learning. These capabilities occupy the top of the hierarchy (Figure 5), corresponding to their placement in the ISM model (Figure 3) and their classification in Cluster 4 (dependent capabilities) of the fuzzy MICMAC analysis (Figure 4).

A central insight emerging from the RBV-driven analysis concerns the dependent positioning of PPC4. Importantly, PPC4 does not refer to the direct management or expansion of human resources, but rather to the organizational ability to dynamically reconfigure team-level resources in response to frequent changes in production conditions. This capability presupposes the existence of stabilized data flows, reliable real-time visibility, and autonomous decision routines generated by SIP-enabled systems (Borangiu et al., 2014; McFarlane et al., 2013). As such, workforce reconfiguration is not a primary driver of autonomy. Instead, it emerges as an outcome of system-level maturity, once other enabling capabilities have been established and operationalized. This explains why PPC4 appears as a dependent capability in both the ISM and fuzzy MICMAC analyses.

PPC10 captures the system's ability to sustain performance under disturbances by leveraging decentralized and autonomous control, thereby enhancing resilience and scalability in volatile environments (Meyer et al., 2009). PPC11 extends autonomy beyond the factory by enabling firms to manage complex product–service ecosystems, as illustrated by digital platforms such as Siemens MindSphere and performance-based service models supported by high levels of data integration (Petrik and Herzwurm, 2019; Siemens, 2023).

From the RBV lens, the capabilities at this level constitute higher-order dynamic capabilities that support strategic renewal. At this stage, autonomy in PPC is no longer a technological feature but an organizational condition that emerges from the coordinated orchestration of SIP-enabled capabilities across planning, execution, and human systems. Together, these interdependent capabilities represent the culmination of the capability-building trajectory, translating SIP-enabled technological potential into sustained competitive advantage.

To further clarify how the proposed framework can be used in practice, this section presents an illustrative application of the structural hierarchy of SIP-enabled capabilities in a PPC context. This illustration follows the logic of a vignette-based explanation, understood as a concise and theory-informed narrative that depicts a plausible managerial situation to facilitate understanding of complex frameworks, without constituting an empirical case or an additional methodological step. As discussed by Bueno et al. (2023), Vignette-based illustrations are beneficial for explicating how conceptual artifacts and frameworks may be applied in real settings, supporting sensemaking while avoiding methodological inflation or claims of empirical generalization.

Consider a manufacturing firm operating in a high-mix production environment, where frequent disturbances, machine variability, and demand fluctuations challenge traditional PPC practices. Initially, the firm deploys SIPs with sensing and connectivity features, establishing the Connected level through sensing integration (PPC1) and interoperability of smart resources (PPC2). As these foundational capabilities stabilize, the organization advances to the Transparent level by implementing machine-to-machine integration (PPC8), enhanced data collection (PPC9), and real-time operational visibility (PPC12). At this stage, planners can monitor system states and coordinate responses more effectively, while decision authority remains predominantly human.

Building on this visibility, the firm introduces adaptive decision routines associated with Autonomous Decision-Making Level I, including SIP-enabled line reconfiguration (PPC3), cyber-physical-social integration (PPC5), remote process control (PPC6), and self-optimizing decision performance (PPC7). As these capabilities mature, the organization transitions to Autonomous Decision-Making Level II, where SIP-generated decisions are directly enacted on the shop floor, workforce roles are reconfigured (PPC4), and the system exhibits adaptive and resilient behavior (PPC10) while managing increasing product–service complexity (PPC11). This illustration demonstrates how the framework can be used as a capability-based roadmap, enabling managers to assess current configurations, identify gaps in foundational layers, and avoid premature investments in advanced autonomy. In doing so, it reinforces the RBV view that autonomy in PPC emerges from the coherent orchestration of interdependent capabilities rather than from linear adoption of technology.

Recently, SIPs have been increasingly adopted, supporting organizations in transforming how they produce goods and services. At the same time, organizations are required to acquire capabilities that directly and positively impact the PPC objectives. Accordingly, this research identified 12 capabilities enabled by adopting SIPs in the PPC function. This study established a structural hierarchy among these capabilities through an integrated multimethod approach comprising an SLR, expert interviews, ISM, and fuzzy MICMAC. The following subsections present the theoretical and practical contributions of this research, along with its limitations and suggestions for future studies.

This study advances the theoretical understanding of SIPs in PPC by interpreting their role through the RBV. First, it consolidates and defines twelve SIP-enabled capabilities within PPC, thereby addressing the fragmentation of prior research, which has predominantly examined isolated technological applications such as monitoring, optimization, or reconfiguration. By systematizing these capabilities, the study establishes more precise conceptual boundaries and a foundation for theory development and future empirical operationalization in PPC research.

Second, the proposed framework contributes theoretically by explicating the structural interdependencies among SIP-enabled capabilities. Anchored in the RBV, it conceptualizes SIP-enabled PPC as a layered configuration in which foundational capabilities (e.g. sensing integration and interoperability) enable higher-order dynamic capabilities (e.g. self-optimizing decision performance and adaptive production). This perspective clarifies how autonomy in PPC emerges through the accumulation and orchestration of capabilities, rather than through linear technological advancement alone.

Third, the integration of ISM and fuzzy MICMAC analysis provides analytical grounding for the capability hierarchy by revealing patterns of driving and dependence among capabilities. The resulting clusters highlight the interplay between foundational, coordinating, and adaptive capabilities, reinforcing the RBV view that competitive advantage stems from the systematic alignment of resources, rather than from their isolated possession (e.g. Oluyisola et al., 2022; Pissardini et al., 2024; Rahmani et al., 2022).

Finally, the theoretical implications extend beyond PPC by offering an alternative to stage-based views of digital transformation. In contrast to digital maturity models that frame autonomy as the outcome of predefined adoption stages, this study theorizes autonomy in PPC as a structural property emerging from inter-capability dependencies. This contribution provides a transferable analytical lens for examining digital transformation across manufacturing contexts and informs future research on the relationship between capability configurations and organizational performance.

The proposed framework provides significant managerial value by translating intricate linkages among SIP-enabled capabilities into practical guidance for practitioners. It enables managers to evaluate their organization's existing PPC configuration and prioritize capability enhancement based on the hierarchical dependencies identified by the model. By clarifying these relationships, the framework helps organizations avoid hasty investments in advanced capabilities before core capabilities are stabilized. For example, an organization seeking to deploy autonomous scheduling (PPC7) must first ensure dependable sensor integration (PPC1) and interoperability among intelligent resources (PPC2), thereby mitigating the risk of resource misallocation and enhancing performance.

Implementing the framework requires its integration into capability evaluation and investment strategy processes. Companies may align current PPC-related technologies and processes with the twelve capabilities described in this study and assess their maturity using straightforward, capability-specific metrics. At the Connected level, managers may evaluate maturity using criteria such as sensor coverage rate or data latency across production resources. A prevalent drawback at this stage is the degradation of data structures, which can compromise transparency and limit the effectiveness of future investments in autonomous decision-making. This diagnosis facilitates the creation of focused roadmaps that prioritize interoperability enhancements in legacy systems before implementing advanced decision-support or optimization technologies.

The framework additionally facilitates performance management, risk minimization, and organizational alignment. Managers can use it to establish KPIs across various capability layers, track progress over time, and anticipate implementation issues arising from interdependencies among capabilities. Furthermore, by emphasizing the socio-technical ramifications of SIP adoption, the framework promotes the proactive management of workforce competencies, governance frameworks, and cybersecurity issues alongside technological investments. Organizations can enhance cumulative capability growth, bolster operational resilience, and consistently align their PPC systems with evolving strategic objectives by adhering to a hierarchical progression from connectivity to transparency and, ultimately, to autonomous decision-making.

This study identified twelve capabilities enabled by the adoption of SIPs in the PPC function. Although the capabilities were derived through an SLR, and efforts were made to minimize subjectivity by involving two researchers in the screening process, it is not possible to eliminate subjectivity from the analysis. The SLR has inherent limitations, including the databases used to identify papers, the publication period, the types of materials selected, and other factors that may have led to the exclusion of relevant studies, as well as additional capabilities. Notwithstanding these limitations, the ISM approach also entails subjectivity and potential bias in expert judgments, reflecting their respective perspectives and fields of expertise. Additionally, because the output of ISM serves as input to the fuzzy MICMAC analysis, such subjectivity may be propagated to subsequent stages.

Future research could explore the applicability of the developed model across a diverse range of industries, including but not limited to automotive, electronics, and pharmaceuticals, and identify sector-specific and production-system-specific characteristics. Such comparative analyses could shed light on sectoral variations in the impact of SIPs on PPC capabilities, providing valuable insights for industry-specific strategies and empirically validating or challenging the framework presented here. In addition, conducting longitudinal studies over extended periods could help assess the evolution of SIP-enabled capabilities in the PPC function. Researchers could track changes in capabilities (including the emergence of new ones), adoption trajectories, performance outcomes over time, and challenges associated with each capability level, thereby providing a dynamic perspective on the influence of SIPs on PPC. Finally, future studies could investigate the integration of emerging technologies, such as AI, ML, and blockchain, with SIPs in the context of PPC, assessing how these technologies enhance or complement the identified capabilities. Such research could open new avenues for both theoretical development and practical implementation in the field of SIP-enabled PPC.