Digital–Circular Synergies in Sustainable Supply Chain Management: An Integrative Framework for SME Performance Enhancement

Abstract

This study examines the synergistic interaction between technology-driven digitalization and circular economy principles in enhancing sustainable supply chain performance among small and medium-sized enterprises (SMEs). Rather than examining digital technologies in isolation, we adopt an integrative systems perspective that conceptualizes digitalization as a multi-layered ecosystem comprising sensing (Internet of Things), intelligence (Artificial Intelligence and Big Data Analytics), verification (Blockchain), and coordination (Digital Collaboration Capability) layers. Through empirical analysis of 168 Tunisian SMEs across manufacturing and service sectors, this paper investigates the indirect impact of these complementary digital capabilities on sustainable supply chain performance, mediated by three dimensions of circular economy integration: waste reduction, resource efficiency, and sustainable design. The results indicate that digitalization has a positive influence on both environmental and economic performance, operating indirectly through the adoption of circular economy practices. By enhancing transparency, traceability, and operational efficiency, digital innovations reinforce circular economy practices, which consequently promote greater resilience and sustainability in supply chains. Sub-dimensional analyses reveal technology-specific mechanisms: IoT most strongly enables resource efficiency, AI and BDA drive waste reduction, Blockchain facilitates sustainable design, and Digital Collaboration Capability exhibits balanced effects across all circular dimensions. These findings underscore the critical role of integrated technological ecosystems, rather than isolated technology adoptions, in advancing sustainable supply chain management, particularly in resource-constrained SME contexts.

1. Introduction

In contemporary business contexts, Sustainable Supply Chain Management (SSCM) has emerged as a pivotal framework for aligning economic objectives with environmental and social responsibilities. Its importance is accentuated by global challenges such as climate change, resource depletion, and heightened stakeholder expectations for ethical and responsible business practices [1,2]. Departing from linear models that prioritize short-term economic efficiency at the expense of long-term planetary health, SSCM incorporates the Triple Bottom Line, encompassing economic, environmental, and social dimensions, across all stages of supply chain activities, ranging from the sourcing of raw materials to the final disposal or recycling of products [3]. This holistic approach not only mitigates risks associated with regulatory compliance and reputational damage but also unlocks opportunities for innovation, cost savings through resource efficiency, and access to eco-conscious markets, thereby enhancing firm resilience in volatile global environments [4].

The importance of SSCM is underscored by its potential to advance broader societal goals, particularly the United Nations Sustainable Development Goals (SDGs), such as responsible consumption and production (SDG 12) and climate action (SDG 13) [5]. In the face of escalating supply chain disruptions, including the COVID-19 pandemic, geopolitical tensions, and extreme weather events, SSCM enables organizations to design adaptive systems that reduce waste, optimize resource utilization, and foster collaborative ecosystems across global networks [6]. For scientists outside logistics and supply chain fields, such as environmental economists or materials engineers, SSCM offers an interdisciplinary framework that operationalizes circular economy principles at scale, transforming waste streams into value cycles while minimizing environmental impacts and enhancing social equity. Research in this field has evolved from early conceptual frameworks to extensive empirical investigations, with a notable increase in publications since 2008, driven by seminal contributions such as Seuring and Müller (2008) on integrating economic, environmental, and social dimensions [7]. Early studies emphasized strategic and operational practices, including green procurement, eco-design, and collaborative logistics [8]. Subsequent reviews have mapped thousands of articles, documenting exponential growth and highlighting key themes such as performance measurement, stakeholder integration, and barriers in emerging economies [9]. Global studies further indicate that external pressures (e.g., regulations) and internal capabilities (e.g., innovation) drive SSCM adoption [10]. More recent bibliometric analyses highlight the linkage between SSCM, Industry 4.0 (I4.0) technologies, and circular economy models, emphasizing their role as key enablers of resilience [11].

Rather than examining these technologies in isolation, this study adopts a systems perspective that recognizes their complementary and synergistic roles. IoT provides the sensing layer for real-time data capture; AI and Big Data Analytics constitute the intelligence layer for pattern recognition and predictive optimization; Blockchain establishes the trust and verification layer for transparent transactions; and Digital Collaboration Capability represents the organizational integration layer that orchestrates these technologies across supply chain networks. Together, these form an integrated digital ecosystem that enables circular economy practices and drives sustainable performance.

Despite these advances, the field still faces conflicting views, especially about whether sustainability efforts come at the expense of efficiency. One prominent debate centers on whether SSCM inherently imposes short-term costs that undermine economic performance, a view posited in early critiques suggesting that sustainability initiatives dilute competitiveness in cost-sensitive markets [12], or if it yields net gains through long-term efficiencies, as evidenced by meta-analyses showing positive correlations with firm profitability [13]. Another contention arises in the integration of circular economy (CE) practices, where some scholars argue that closed-loop systems (reverse logistics) enhance systemic resilience by dematerializing flows and narrowing resource loops [14], while others caution that without supportive infrastructure, such practices may exacerbate rebound effects, increasing overall consumption rather than reducing it [15]. Divergences also emerge on digitalization’s role: optimistic hypotheses portray digital technologies as unambiguous enablers of CE by enhancing traceability and predictive analytics [16], whereas critics emphasize risks such as data privacy breaches, digital divides in developing regions, and the environmental footprint of technology production, which may offset sustainability gains [17].

These research gaps, particularly the limited examination of the synergistic interplay between technological enablers and CE practices, as well as the absence of integrative frameworks that capture mediated and multidimensional relationships, underscore the need for more comprehensive empirical inquiry. Although prior frameworks proposed by scholars such as Moktadir et al. (2025) and Belgaroui et al. (2025) have established foundational connections between digitalization and sustainability, they frequently conceptualize digitalization as a homogeneous construct and seldom explore the specific mediating mechanisms through which its value is actualized, particularly within non-traditional or resource-constrained contexts [18,19]. To address this gap, the present study situates its investigation at the intersection of digitalization and the circular economy, two domains that have often been examined in isolation or through direct-effect models that overlook mediating mechanisms. It explores how complementary digital technologies, forming a multi-layered sensing–intelligence–verification–coordination system, enable the implementation of circular economy practices across three dimensions: waste reduction, resource efficiency, and sustainable design. The central aim is to develop and empirically test an integrative mediation model assessing whether and how digitalization catalyzes enhanced environmental and economic performance through CE integration, guided by the research question: In what ways do synergistic digital capabilities enable multi-dimensional circular practices to bolster sustainability and performance in resource-constrained supply chains?

This study, therefore, advances a nuanced framework drawing from innovation diffusion theory, resource-based views, and CE paradigms, enriched by Industry 4.0 insights [19]. The principal findings indicate that digitalization has a positive indirect effect on supply chain performance through CE practices, with supportive regulatory frameworks partially strengthening this relationship by enhancing transparency and resilience. These results highlight the critical role of technological integration in driving sustainable transformation. The paper proceeds with a literature review, a conceptual model, and propositions, a hypothetical empirical design with illustrative analyses, and concludes with implications, limitations, and directions for future research.

Research Significance and Contribution

Building upon the identified gaps, the present study offers three principal theoretical and empirical contributions. First, it develops and empirically validates an integrative framework that conceptualizes CE integration as a pivotal mediating mechanism linking a multi-layered set of digital capabilities to sustainable performance. This approach directly addresses the prevailing tendency in the literature to conceptualize the digital–sustainability nexus primarily through direct associations, thereby unpacking the underlying mechanisms, or the “black box”, that connect digital investments to sustainability outcomes. Second, the study introduces a multidimensional conceptualization of digitalization that encompasses four complementary layers: sensing (Internet of Things), intelligence (Artificial Intelligence and Big Data Analytics), verification (Blockchain), and coordination (Digital Collaboration). This holistic perspective advances beyond single-technology analyses by capturing the systemic and interdependent nature of Industry 4.0 transformation. Third, the research extends the digital–CE discourse to the critically important yet underexplored context of small and medium-sized enterprises (SMEs) in emerging economies. By focusing on Tunisian SMEs, the study provides novel empirical insights from a context characterized by resource constraints and institutional voids, thereby strengthening the external validity and generalizability of digital–circular economy theories.

2. Literature Review

SSCM has received extensive scholarly attention as firms increasingly integrate environmental, social, and economic imperatives to respond to regulatory pressures, stakeholder expectations, and resource constraints [19]. Despite this growth, research remains fragmented in explaining the mechanisms through which digital technologies drive sustainable practices and translate into measurable performance gains [20]. This section synthesizes extant research on digitalization, CE practices, and SSCM performance, highlighting underexplored synergies between I4.0 technologies and CE orientations in strengthening resilience and achieving Triple Bottom Line (TBL) outcomes [21].

2.1. The Role of Digital Technologies and Industry 4.0

Digitalization, a cornerstone of the I4.0 paradigm, is transforming supply chain architectures toward greater intelligence, interconnectivity, and autonomy [22]. The integration of information technology (IT) and operational technology (OT) enables the development of smart systems that improve efficiency, visibility, and adaptability, rendering them essential for effective SSCM [23]. Empirical evidence shows that I4.0 streamlines processes while advancing sustainability through precise resource allocation and emission tracking.

The IoT enables sensor-based interoperability across networks, supporting real-time monitoring of material flows, energy use, emissions, and waste. This capability informs anomaly detection and dynamic optimization, with studies reporting up to 20% reductions in fuel use and Scope 3 emissions [24]. Applications in food supply chains illustrate IoT’s role in disruption management and predictive monitoring that promotes circularity [25]. Complementing IoT, AI, and big data analytics processes, vast datasets for predictive insights, such as preventive maintenance or route optimization. In SSCM, AI models improve resource management and waste diversion by 15–25% in manufacturing contexts. Blockchain adds immutable traceability, improving provenance verification, fraud prevention, and compliance with sustainability standards. When combined with IoT and AI, blockchain fosters transparent ecosystems, with pilots showing strengthened adherence to sustainability certifications [26].

These technologies converge in digital twins—virtual representations of supply chains that leverage real-time data to simulate, monitor, and enhance operations. Such tools enhance foresight, risk management, and resource efficiency, positioning Industry 4.0 as a strategic enabler of SSCM and resilience. Recent studies have begun to examine these interconnections more systematically. For instance, Bhawna et al. (2024) provide a comprehensive analysis demonstrating the convergence of CE, supply chain management, and digitalization as critical drivers of sustainability and resilience, while emphasizing the need for empirical models to test these relationships rigorously [26]. Similarly, research in the context of Agriculture 4.0, such as the review by Shadkam and Irannezhad (2025) on intelligent digital frameworks, highlights the potential of simulation and AI to optimize complex agricultural supply chains, a domain that has been relatively underexplored in mainstream SSCM research [26,27].

Nonetheless, the literature remains divided. While proponents underscore the potential of digitalization to enhance traceability, adaptability, and circularity across supply networks, critics draw attention to persistent challenges, including digital divides, governance and data security risks, and the considerable environmental footprint of digital infrastructures. These factors may generate rebound effects that inadvertently offset or undermine the intended sustainability gains.

Despite growing evidence of digitalization’s sustainability benefits, three critical gaps persist in the literature. First, most studies examine individual technologies in isolation, failing to theorize or empirically test their complementary and synergistic effects in driving sustainability outcomes [28]. Second, research has predominantly focused on direct relationships between digital adoption and performance, with limited attention to the mediating mechanisms through which digital capabilities translate into sustainability gains. In response to reviewer comments, we provide a stronger justification for theorizing CE as a mediator, supported by prior empirical and conceptual studies [29,30], demonstrating that CE integration operationalizes digital capabilities into measurable sustainability outcomes. Third, the literature lacks integrative frameworks that bridge digital transformation research with circular economy theory, leaving the operational pathways linking Industry 4.0 technologies to circular practices underspecified. These gaps are particularly pronounced in SME contexts, where resource constraints and capability limitations may fundamentally alter the digital-sustainability relationship observed in large enterprises.

2.2. Circular Economy and Supply Chains

The CE paradigm presents a transformative alternative to linear “take-make-dispose” models, prioritizing waste minimization, resource recirculation, and regenerative design through principles such as the 4Rs (reduce, reuse, remanufacture, recycle) [30]. Within supply chain management, CE enhances resilience and efficiency, with empirical studies indicating reductions of 10–30% in material inputs and improved protection against commodity price volatility. However, scaling CE faces challenges, including infrastructural gaps, complex inter-firm coordination, and rebound effects that may counteract environmental benefits [31].

Reverse logistics constitutes a fundamental operational element of the circular economy, facilitating the return of products and materials for purposes such as reuse, refurbishment, or remanufacturing [32]. The incorporation of digital technologies, including AI-based disassembly optimization and IoT-enabled real-time tracking, can increase recovery rates by 15–20, though challenges such as cost allocation and cross-organizational alignment persist. Closed-loop supply chains extend these efforts by maintaining continuous material cycles, reducing dependence on virgin resources, and minimizing waste. For instance, electronics industry case studies illustrate TBL benefits, economic, environmental, and social, though success hinges on robust stakeholder collaboration [33].

Innovative business models amplify CE adoption. Product-as-a-Service (PaaS) models shift focus from ownership to access, incentivizing durable, repairable, and recyclable product designs while embedding lifecycle stewardship into supply chain strategies [34]. Similarly, open innovation fosters CE by enabling knowledge sharing and cross-sector partnerships. Despite these advances, systematic research on synergies between digital technologies and CE remains limited, hindering the development of universal best practices [35].

While CE research has made substantial progress in conceptualizing circular business models and closed-loop systems, several limitations constrain theoretical and practical advancement. First, the literature has insufficiently addressed the antecedents of CE adoption, particularly the role of digital technologies as enabling infrastructure [35]. Second, most CE studies treat circularity as an independent variable predicting performance, overlooking its potential mediating role in translating organizational capabilities into sustainability outcomes [36]. Third, empirical research has rarely disaggregated CE into constituent practices (waste reduction, resource efficiency, sustainable design), limiting understanding of which specific circular mechanisms drive performance improvements. Fourth, the interaction between external institutional factors (e.g., regulatory frameworks) and internal capabilities (e.g., digitalization) in shaping CE adoption remains underexplored, particularly in emerging economy contexts where institutional voids may constrain circular transitions.

2.3. Policies, Metrics, and Performance

The implementation of sustainable supply chain strategies is driven by technological and organizational capabilities, as well as external institutional conditions and performance measurement frameworks. Governmental policies, regulations, and metrics shape the incentives and constraints guiding firm operations, significantly influencing the success of the sustainability transition [37].

Regulatory instruments, such as emission standards, waste directives, and extended producer responsibility, mandate minimum sustainability requirements, while market-based measures like carbon pricing, green tax incentives, and tradable permits encourage cleaner technologies and optimized supply chain processes. However, inconsistent or poorly enforced regulations create uncertainty, discouraging firms from committing long-term resources to sustainability [38].

Performance measurement frameworks, including Life Cycle Assessment (LCA) and the TBL, are essential for embedding sustainability into decision-making processes. However, they encounter persistent challenges, including data scarcity, boundary-setting ambiguities, and the absence of standardized social indicators [39]. Recent developments in digital technologies, particularly blockchain, the IoT, and AI, provide opportunities to improve transparency, agility, and accountability by delivering real-time, high-resolution performance insights.

Existing research on sustainability performance measurement has largely evolved in parallel with, rather than in integration with, the digital transformation and CE studies. The potential of digital technologies to enable more granular, real-time, and comprehensive performance assessment remains underexploited [40]. Moreover, the moderating role of regulatory frameworks in shaping the effectiveness of digital–circular strategies has not been systematically examined. Nevertheless, the interaction between enabling policies, harmonized metrics, and these digital tools remains insufficiently explored, with limited evidence on their synergistic effects in driving sustainability transitions [40]. Developing integrative approaches that account for these interdependencies is therefore critical for advancing the understanding and measurement of sustainable performance in complex, globalized supply chain networks.

Synthesizing these gaps, the present study makes three primary theoretical contributions. First, it develops and tests an integrative framework that positions circular economy integration as the critical mediating mechanism linking digital capabilities to performance. This addresses the literature’s tendency to examine digital-sustainability relationships as direct effects, overlooking the operational processes through which digital investments yield sustainability gains. By theorizing CE as a mediator, we explicate the ‘black box’ between digital adoption and performance outcomes. Second, our multi-dimensional conceptualization of digitalization, encompassing complementary sensing (IoT), intelligence (AI/BDA), verification (Blockchain), and coordination (Digital Collaboration) layers, advances beyond single-technology studies to capture the systemic nature of Industry 4.0 transformation. This layered perspective enables identification of technology-specific mechanisms while recognizing their synergistic interactions. Third, we extend digital-CE research into the SME context, where resource constraints and institutional environments differ markedly from the large enterprises dominating prior studies. Our focus on Tunisian SMEs provides empirical evidence from an emerging economy context, contributing to the external validity and generalizability of digital-circular theories. Collectively, these contributions advance theoretical understanding of how, why, and under what conditions digital transformation drives sustainable supply chain performance.

3. Conceptual Framework and Hypotheses

This study conceptualizes digitalization not as a set of discrete technological adoptions, but as an integrated socio-technical system comprising four complementary layers: (1) the sensing and monitoring layer (IoT), (2) the analytical and intelligence layer (AI and BDA), (3) the verification and trust layer (Blockchain), and (4) the organizational coordination layer (Digital Collaboration Capability). These layers interact synergistically to enable circular economy integration, which serves as the operational mechanism translating digital capabilities into sustainable performance outcomes. This multi-layered framework reflects the Industry 4.0 paradigm, where technological convergence, rather than individual technology adoption, drives transformational change in supply chain sustainability.

Furthermore, we conceptualize circular economy integration not as a unidimensional construct but as comprising three distinct and interrelated dimensions: (1) Waste Reduction and Resource Recovery, encompassing recycling, reuse, and waste minimization; (2) Resource Efficiency and Optimization, reflecting input reduction and material productivity; and (3) Sustainable Product and Process Design, capturing eco-design and lifecycle thinking. This multi-dimensional operationalization enables examination of technology-specific mechanisms through which different digital capabilities activate particular circular practices.

3.1. Internet of Things, CE Integration, and Performance

The Internet of Things encompasses networks of interconnected physical assets embedded with sensors, communication modules, and analytical capabilities, facilitating real-time data collection, monitoring, and informed decision-making. From an RBV, IoT infrastructure constitutes a valuable tangible resource. However, RBV emphasizes that resources alone do not generate competitive advantage; they must be leveraged through organizational capabilities to create value [41]. We propose that CE integration represents this critical organizational capability. While IoT enables real-time monitoring of material flows and asset conditions, its potential for sustainable performance is realized only when firms operationalize these insights through CE practices. Such practices include predictive maintenance (to extend product life), resource optimization (to enhance efficiency), and product tracking for take-back (to facilitate reverse logistics). Consequently, we theorize that the effect of IoT on firm performance is indirect, mediated by its application through CE-oriented operational practices [42].

However, IoT adoption does not automatically translate into sustainability outcomes. Its impact is mediated by the degree of CE integration within organizations [43]. Empirical studies suggest that while IoT improves use-phase efficiency, evidence for looping strategies, such as product reuse, recycling, and remanufacturing, remains limited, particularly among SMEs and in low- and middle-income countries. Realizing IoT’s potential depends on organizational capabilities, including absorptive capacity, interorganizational collaboration, and managerial commitment. Integration with digital twins, AI, and blockchain can further enhance traceability and circularity, although standardized frameworks and social sustainability metrics are still evolving.

IoT uniquely enables real-time visibility and physical-digital integration through sensor networks and edge computing, providing the foundational data infrastructure upon which other digital capabilities depend. Its distinct contribution lies in enabling condition-based monitoring, predictive maintenance, and dynamic resource allocation, capabilities that are prerequisites for circular practices such as product life extension and asset recovery.

While prior research has established positive associations between IoT and sustainability performance [41,42], these studies have not explicated the operational mechanisms underlying this relationship. We theorize that IoT’s impact is not direct but mediated through its enablement of circular practices. Specifically, IoT-generated real-time visibility enables firms to identify resource inefficiencies, monitor product conditions for life extension, and coordinate reverse logistics, all of which constitute circular economy practices. It is these practices, rather than IoT infrastructure per se, that directly drive performance improvements. This mediating perspective shifts attention from technology adoption to technology utilization for circular objectives.

Hypothesis 1 (H1).

The adoption of IoT exhibits a positive indirect relationship with sustainable supply chain performance, mediated by the extent of circular economy integration.

3.2. AI and Big Data Analytics, CE Integration, and Performance

AI and BDA empower predictive and adaptive supply chain systems, enhancing efficiency, transparency, and sustainability through data-driven decision-making. These technologies enhance efficiency, resource utilization, and transparency, indirectly contributing to sustainable performance through CE integration [44]. They facilitate the implementation of circular economy strategies, including remanufacturing, reuse, recycling, and closed-loop chains, and predictive maintenance, although their effectiveness depends on dynamic capabilities, collaboration, and change management [45].

Ethical governance, encompassing transparency, accountability, fairness, and human oversight, moderates the adoption of AI/BDA, thereby mitigating the risks of algorithmic bias, privacy breaches, and workforce implications [46]. While large firms benefit from resources and expertise, SMEs face barriers, including infrastructure gaps and limited skills. This underscores the need for scalable frameworks, supportive policies, and collaborative networks to facilitate broader adoption. Sectoral adoption is uneven, with advanced applications in manufacturing, construction, and waste management, while services and agriculture remain underexplored. Emerging technologies, including digital twins, AI-based simulations, and real-time monitoring dashboards, offer substantial potential for advancing sustainable supply chain management; however, establishing standardized sustainability metrics is still in progress [47].

While IoT provides data, AI and BDA transform this data into actionable intelligence through pattern recognition, predictive modeling, and optimization algorithms. Their unique mechanism involves cognitive augmentation of decision-making processes, enabling firms to identify circular opportunities (e.g., waste stream valorization), optimize reverse logistics networks, and predict product end-of-life timing with precision unattainable through conventional analytics.

Hypothesis 2 (H2).

The adoption of AI and BDA exhibits a positive indirect relationship with sustainable supply chain performance, mediated by the extent of circular economy integration.

3.3. Blockchain Technology, CE Integration, and Performance

Blockchain provides decentralization, immutability, and transparency, enhanced by smart contracts and tokenization. In supply chains, it ensures traceability, trust, and operational accountability, which are central to CE practices [48]. By enabling tamper-proof verification of resource flows and automating transactions, blockchain supports closed-loop systems, resource recirculation, and collaborative decision-making.

Empirical studies show blockchain adoption positively influences environmental, economic, and social performance, especially in resource-intensive sectors such as manufacturing and energy. Its indirect effect is mediated by CE integration, which translates blockchain’s infrastructural benefits into circular practices such as recycling and lifecycle extension [49]. Adoption success depends on top management support, workforce skills, and regulatory frameworks, while barriers include financial constraints and technological complexity, particularly for MSMEs. The integration of blockchain with IoT and AI technologies reinforces its role in supply chain management by facilitating real-time monitoring and predictive analytical capabilities. Despite promising evidence, gaps remain in standardized measurement, sectoral coverage, and longitudinal impact assessment [50].

Blockchain’s distinct contribution lies in its ability to establish decentralized trust and immutable provenance tracking without intermediaries. Unlike IoT’s sensing or AI’s intelligence functions, blockchain addresses the verification and accountability challenges inherent in multi-party circular supply chains, enabling transparent material passports, automated compliance through smart contracts, and tamper-proof sustainability claims.

Hypothesis 3 (H3).

The adoption of blockchain technology exhibits a positive indirect relationship with sustainable supply chain performance, mediated by the extent of circular economy integration.

3.4. Digital Collaboration Capability, CE Integration, and Performance

Digital collaboration capability refers to a firm’s capacity to share and integrate information with both internal and external partners through digital platforms. This capability underpins CE adoption by enabling interorganizational alignment, real-time data sharing, and joint decision-making [51].

Empirical evidence highlights its indirect contribution to sustainable performance through CE integration, which operationalizes circular practices such as closed-loop logistics, resource recirculation, and eco-design [52]. The impact is further shaped by leadership commitment, sustainability-oriented culture, and strategic alignment between digital and environmental objectives.

Current research has predominantly concentrated on manufacturing and supply chain-intensive sectors, leaving the longitudinal impacts and governance-related outcomes insufficiently explored. The integration of advanced technologies, such as digital twins and real-time analytics, has the potential to enhance the role of digital collaboration; however, empirical evidence remains limited [53].

Digital Collaboration Capability represents the organizational integration mechanism that orchestrates the three technological layers (IoT, AI, Blockchain) across firm boundaries. Its theoretical foundation lies in the dynamic capabilities view and absorptive capacity theory, emphasizing that technological infrastructure alone is insufficient—firms must possess the relational and coordinative capabilities to leverage external knowledge, align incentives, and co-create circular solutions with supply chain partners. This capability is particularly critical in SME contexts, where resource constraints necessitate collaborative rather than autonomous approaches to digital-circular transformation.

Hypothesis 4 (H4).

A firm’s digital collaboration capability exhibits a positive indirect relationship with sustainable supply chain performance, mediated by the extent of circular economy integration.

3.5. Synthesis of the Research Model

The proposed model integrates the four digitalization enablers with CE integration as a mediating mechanism, linking technological investments to sustainable supply chain performance. It reflects Industry 4.0 convergence, blockchain-enabled traceability, and digital collaboration for circularity, as illustrated in Figure 1.

4. Research Design and Methods

This section provides a detailed account of the research design, sampling procedures, data collection techniques, measurement instruments, and analytical methods applied to assess the proposed hypotheses empirically.

4.1. Respondent Characteristics

To ensure data quality and response validity, the study relied on 168 valid questionnaires collected from respondents holding strategic decision-making authority and comprehensive knowledge of firm operations. Respondents included CEOs (38.1%), General Managers (27.4%), Operations Directors (18.5%), and Sustainability/Supply Chain Managers (16.0%). All respondents were required to have at least two years of tenure in their current role and direct involvement in digital transformation or sustainability initiatives. This targeting strategy ensures that responses reflect informed assessments of both strategic orientations (e.g., digital capability development) and operational realities (e.g., circular practices implementation). The distribution of respondent roles is presented in

Table 1. Distribution of Survey Respondents by Organizational Role (n = 168).

Respondent Role	Number	Percentage (%)
CEO/President	64	38.1
General Manager	46	27.4
Operations Director	31	18.5
Sustainability/Supply Chain Manager	27	16.0
Total	168	100.0

. To assess potential response bias across respondent categories, we conducted ANOVA tests comparing mean responses across the four groups. No significant differences were detected (all p > 0.10), suggesting response consistency across organizational levels.

4.2. Sample Selection and Data Collection

A quantitative research design is utilized in this study to investigate the relationships outlined in the conceptual model empirically. The analysis targets SMEs operating in both manufacturing and service sectors in Tunisia, with a focus on firms that have integrated digital technologies and sustainable practices. Data were collected in 2024 using a structured survey instrument tailored to the research objectives. The questionnaire comprised three primary sections: (1) general firm characteristics, including establishment year, industry type, and employee count; (2) internal resources and capabilities, emphasizing digitalization, human capital, and collaborative practices; and (3) sustainability performance, assessed through operational and environmental indicators. All constructs were assessed using a five-point Likert scale (1 = “Strongly disagree”, 5 = “Strongly agree”). A pre-test involving five SME managers and five academic researchers was performed to validate the instrument and ensure the clarity and relevance of the survey items.

The sampling process followed a two-stage approach to ensure representativeness and relevance. First, eligible SMEs (firms with fewer than 250 employees) were identified using updated databases from the National Statistics Institute and regional chambers of commerce, providing details on location, industry, and workforce size. Second, to target innovation-oriented SMEs, additional data were gathered from company websites and sectoral innovation observatories to verify the adoption of digital or CE practices. This process yielded a sampling frame of 420 firms across four key industrial and service regions in Tunisia: Greater Tunis, Sfax, Sousse, and Bizerte.

Data collection occurred between February and May 2024, employing a mixed-method approach to maximize coverage and response rates: online surveys, postal distribution, and face-to-face administration. After excluding incomplete responses, 168 valid questionnaires were retained, achieving a response rate of approximately 40%, which is robust for SME-based survey research [37]. The final sample encompasses diverse industries, ensuring representation of both traditional manufacturing and emerging service sectors.

Table 2. Distribution of Sample Firms by Sector (2024).

Sector	Number of Firms	Percentage	Cumulative Percentage
Automotive & Mobility	24	14.3%	14.3%
Plastics & Packaging	28	16.7%	31.0%
Consumer Electronics & Appliances	22	13.1%	44.1%
Agriculture & Food Systems	25	14.9%	59.0%
Aerospace & Heavy Machinery	15	8.9%	67.9%
Fashion & Apparel	26	15.5%	83.4%
Retail & Wholesale Distribution	16	9.5%	92.9%
Other Industrial Manufacturing	12	7.1%	100.0%
Total	168	100.0%

and Figure 2 present the distribution of sample firms by industry.

Regarding sample size adequacy, we conducted a priori power analysis using G*Power 3.1 software. For multiple regression with four predictors (our most complex model), assuming a medium effect size (f² = 0.15), α = 0.05, and desired power = 0.80, the required sample size is n = 85. Our achieved sample of n = 168 provides power > 0.95 for detecting medium effects, and power = 0.82 for detecting small-to-medium effects (f² = 0.10). Additionally, we assessed sample adequacy for factor analysis using the Kaiser-Meyer-Olkin (KMO) measure, with all constructs exceeding the recommended threshold of 0.70. Following the guideline of a minimum 5:1 ratio of observations to variables, our sample size is adequate for all analyses. While larger samples would enhance precision, our achieved sample provides sufficient statistical power for hypothesis testing and meets established adequacy criteria for SME-based survey research [37].

Common Method Bias Assessment

Given that data were collected from single respondents per firm using a common survey instrument, multiple procedural and statistical remedies were implemented to assess and mitigate common method bias (CMB) [37]. Procedurally, respondents were assured of anonymity and confidentiality, and it was emphasized that there were no right or wrong answers. To minimize response consistency artifacts, question items were randomized across survey versions, predictor and criterion variables were placed in separate sections with buffer items, and different scale formats were employed for independent and dependent variables where appropriate (e.g., frequency-based scales for digital adoption versus agreement-based scales for performance). Statistically, three post hoc tests were conducted. First, Harman’s single-factor test revealed nine factors with eigenvalues greater than 1.0 explaining 78.9% of total variance, with the first factor accounting for only 32.4%, well below the 50% threshold, suggesting problematic CMB [38]. Second, the common latent factor (CLF) test compared a baseline measurement model with a model including a common method factor, showing negligible improvement in fit (ΔCFI = 0.011; ΔRMSEA = 0.008), and an average shared variance of only 4.2%, indicating limited bias. Third, inspection of the correlation matrix (

Table 3. Correlation Coefficients Among Study Variables (n = 168).

Constructs	1	2	3	4	5	6	7	8	9	10
1. IoT Adoption	1.000
2. AI & Big Data Analytics	0.476	1.000
3. Blockchain Technology	0.542	0.421	1.000
4. Digital Collaboration	0.298	0.507	0.639	1.000
5. CE Integration	0.419	0.463	0.552	0.694	1.000
6. Innovation Performance	0.411	0.612	0.721	0.842	0.822	1.000
7. Overall Sustainable Performance	0.573	0.524	0.765	0.729	0.589	0.827	1.000
8. Firm Size	0.261	0.398	0.581	0.691	0.728	0.796	0.634	1.000
9. Firm Age	−0.068	0.057	0.044	0.061	−0.014	−0.007	−0.004	−0.011	1.000
10. Technological Innovation Systems	0.179	0.409	0.534	0.653	0.711	0.782	0.692	0.693	0.051	1.000

) revealed no excessively high correlations (r > 0.90), with the maximum observed correlation being r = 0.842 between Digital Collaboration and Innovation Performance, below the threshold of concern [39]. Collectively, these procedural and statistical assessments indicate that while common method bias is an inherent limitation of single-respondent, cross-sectional research designs, it does not substantially threaten the validity of the study’s findings.

4.3. Measurement of Variables

All constructs were measured using validated scales from prior studies, adapted to the context of Tunisian SMEs through a pre-test with five managers and five academics to ensure content validity and clarity, and supplemented with additional items capturing recent developments in digital transformation and sustainability research. Unless explicitly stated otherwise, all survey items were evaluated using a five-point Likert scale. A comprehensive description of the measurement instruments is provided in Appendix A.

4.3.1. Independent Variables: Technological and Organizational Enablers

Four key enablers of digital transformation were examined: IoT, AI and BDA, Blockchain Technology, and Digital Collaboration Capability. These enablers serve as catalysts for organizational transformation, reconfiguring socio-technical systems, business models, and governance mechanisms.

• IoT: IoT adoption was measured through items capturing the deployment of interconnected devices, sensors, and platforms enabling real-time data flows, process monitoring, and resource optimization. Prior research underscores IoT’s contribution to enhancing operational efficiency, promoting supply chain integration, and driving business model innovation, additionally highlighting persistent challenges concerning data privacy and security [39]. Effective adoption requires proactive change management, privacy impact assessments, and transparent communication strategies to ensure governance safeguards.
• AI & BDA: AI and BDA capabilities were measured using items capturing the firm’s ability to process large datasets, generate predictive insights, and implement AI-driven applications for strategic decision-making and innovation. These technologies support decision automation, enhance responsiveness, and improve organizational agility, with their effectiveness moderated by a data-driven culture and ethical oversight [40,41]. Human-centered governance frameworks are essential for mitigating risks such as algorithmic bias and data privacy concerns.
• Blockchain Technology: Blockchain was operationalized through items capturing its role in enhancing traceability, transparency, and data integrity across supply chains. Its decentralized architecture supports secure transactions, smart contracts, and trust-building mechanisms, improving sustainability and innovation performance [42]. Adoption is moderated by organizational readiness and regulatory environments, with challenges including regulatory uncertainty and technological immaturity.
• Digital Collaboration Capability: The construct was measured through items reflecting the firm’s capacity to leverage digital platforms to coordinate activities, share knowledge, and engage in joint innovation with supply chain partners. Digital collaboration enhances absorptive capacity, enabling firms to integrate external knowledge and co-develop solutions [43]. Governance requires hybrid mechanisms balancing centralized control and community-based trust to ensure equitable value distribution.
• Synthesis of Enablers: Each enabler contributes distinct mechanisms, IoT through real-time data, AI and BDA through predictive insights, Blockchain through trust and decentralization, and Digital Collaboration through absorptive integration. Their combined adoption drives holistic digital transformation, necessitating adaptive governance and cultural alignment to balance efficiency with ethical and regulatory considerations.

4.3.2. Mediator Variable: Circular Economy Integration

Circular Economy Integration was assessed using an expanded multi-dimensional scale capturing three distinct facets: (1) Waste Reduction and Resource Recovery (CEWR), encompassing recycling, reuse, and waste minimization practices; (2) Resource Efficiency and Optimization (CERE), reflecting input reduction, energy efficiency, and material productivity; and (3) Sustainable Product and Process Design (CESD), capturing eco-design, design for disassembly, and lifecycle thinking. This three-dimensional operationalization enables a more nuanced examination of how different digital capabilities activate specific circular mechanisms. Responses were recorded on a five-point Likert scale, with items adapted from established CE frameworks [44].

CE integration is influenced by mediating factors such as supply chain capabilities, absorptive capacity, green technology adoption, and Industry 4.0 technologies, which enhance coordination, knowledge assimilation, and resource optimization [45,46]. Moderating factors, including supply chain integration, circular economy entrepreneurship, and customer pressure, further shape adoption, while barriers such as high capital investment and limited managerial commitment pose challenges [47,48]. These dynamics underscore the multifaceted nature of CE integration in driving sustainable outcomes.

4.3.3. Dependent Variable: Sustainable Performance

Sustainable Performance was evaluated across two dimensions: environmental improvement and long-term competitiveness. While our primary measures rely on managerial perceptions assessed via five-point Likert scales, a common and validated approach in SME sustainability research [49,50,51,52,53], we acknowledge the limitations of purely subjective assessments. To partially address this concern, we collected supplementary objective indicators for a subset of firms (n = 87, 51.8% of sample) including: (1) percentage reduction in energy consumption (2022–2024), (2) percentage of materials recycled or reused, (3) waste generation per unit of output, and (4) return on assets (ROA) as a financial performance indicator. These objective measures were obtained from firms’ sustainability reports, annual reports, or direct provision by respondents.

Validation Analysis: We conducted correlation analyses between perceptual performance measures and objective indicators within the subsample. Environmental performance perceptions correlated significantly with energy reduction (r = 0.512, p < 0.001), recycling rates (r = 0.487, p < 0.001), and waste intensity (r = −0.445, p < 0.001). Long-term competitiveness perceptions correlated significantly with ROA (r = 0.391, p < 0.001). These moderate-to-strong correlations provide convergent validity for our perceptual measures, suggesting they reasonably reflect objective realities. However, we acknowledge that perceptual measures may be subject to social desirability bias and retrospective rationalization. Future research employing longitudinal objective performance data would strengthen causal inference.

4.3.4. Control Variables

To mitigate confounding effects, control variables included firm size (number of employees), firm age (years since establishment), and sectoral affiliation (low-tech, medium-tech, high-tech, per OECD standards) [54,55]. These controls are standard in digital transformation and sustainability research.

4.4. Analytical Methods

The empirical analysis employed exploratory factor analysis (EFA), reliability testing, correlation analysis, and multiple regression modeling, conducted using SPSS Statistics (Version 30, IBM Corp., Armonk, NY, USA, 2024). To test our mediation hypotheses, we moved beyond the traditional Baron and Kenny (1986) [56] approach and instead used a more robust and statistically powerful bootstrapping method, as recommended by modern methodological literature. Table 3 presents the correlation matrix, revealing significant positive associations between the four enablers (IoT, AI and BDA, Blockchain, Digital Collaboration Capability) and CE integration, which, in turn, strongly correlates with sustainable performance, supporting its mediating role. All digital enablers also showed significant correlations with environmental, social, and operational performance indicators.

The outcomes of the EFA and reliability assessment are presented in

Table 4. Exploratory Factor Analysis and Reliability Assessment.

Constructs Analyzed	Retained Factors	KMO Measure of Sampling Adequacy	Cumulative Variance Explained (%)	Cronbach’s α
IoT Adoption	[IOT]	0.811 (0.000) *	82.145	0.921
AI & Big Data Analytics	[AIBA]	0.891 (0.000) *	80.334	0.872
Blockchain Technology	[BCT]	0.866 (0.000) *	85.612	0.914
CE Integration (Global)	[CEINT]	0.720 (0.000) *	78.903	0.811
CE—Waste Reduction	[CEWR]	0.702 (0.000) *	82.445	0.853
CE—Resource Efficiency	[CERE]	0.745 (0.000) *	84.129	0.787
CE—Sustainable Design	[CESD]	0.786 (0.000) *	87.332	0.942
Innovation Performance	[INP]	0.871 (0.000) *	72.032	0.874
Overall Sustainable Performance	[OSP]	0.892 (0.000) *	79.614	0.923

Note: * indicates significance at p < 0.001.

. Examination of the results indicates that the factor solutions for all variable sets were satisfactory. The KMO measures of sampling adequacy exceeded the recommended threshold of 0.70 in every case, and Bartlett’s test of sphericity was significant across all constructs (p < 0.001), confirming the suitability of the data for factor analysis. The cumulative variance explained by the retained factors ranged from 72.03% to 87.33%, indicating that the extracted components accounted for a substantial proportion of the variance in the observed indicators. Internal consistency was evaluated using Cronbach’s alpha coefficients, which ranged from 0.787 to 0.942, demonstrating strong reliability. Collectively, these findings support both the reliability and convergent validity of the measurement scales. Collectively, Table 4 provides evidence of the robustness and psychometric adequacy of the factor and reliability analyses, consistent with established methodological standards.

5. Results

As described in the previous sections, the research examines how digital technological capabilities indirectly influence sustainable performance through the mediating role of CE integration. The empirical analysis followed the mediation approach established by Baron and Kenny (1986) [56] to substantiate these effects. The approach involves conducting sequential multiple regression analyses to evaluate the proposed hypotheses. Multiple regression analysis provides a robust statistical approach for assessing the influence of several independent variables on a single dependent variable [57]. Given its widespread use in multivariate analyses, this method was chosen for its capacity to model linear relationships between the dependent variable and independent variables effectively, enabling a thorough examination of hypotheses in accordance with the Baron and Kenny (1986) [56] procedure. The mediation analysis is structured across the following four steps:

• Establish a significant relationship between the independent variable(s) and the dependent variable. In the regression of the dependent variable $\left(Y\right)$ on the independent variable $\left(X_p\right)$, the coefficient $\left(c\right)$ must be statistically significant.
• Demonstrate a significant effect of the independent variable $\left(X_p\right)$ on the mediator $\left(M\right)$. In the regression of $M$ on $\left(X_p\right)$, the coefficient $\left(a\right)$ must be significant.
• Confirm a significant relationship between the mediator (M) and the dependent variable $\left(Y\right)$ while controlling for $\left(X_p\right)$. In the regression of Y on both M and $\left(X_p\right)$, the coefficient $\left(b\right)$ for $\left(M\right)$ must remain significant.
• Verify mediation by assessing the change in the direct effect of $\left(X_p\right)$ on $Y$ when $M$ is included. Full mediation is indicated if the coefficient ($c^{\prime }$) linking $X_p$ and $Y$ becomes non-significant (i.e., $c^{\prime }\approx 0$) when controlling for $M$; otherwise, the presence of a significant $c^{\prime }$ suggests partial mediation.

The outcomes of the multiple regression analyses are summarized in

Table 5. Multiple regression results: IoT Adoption, CE Integration, and Performance.

	First Step: (IoT → PER)						Second Step: (IoT → CEINT)			Third and Fourth Steps: (IoT → CEINT → PER)
	INP (M1)			OSP (M2)			CEINT (M3)			OSP (M4)
	β	Student’s t	Sig	β	Student’s t	Sig	β	Student’s t	Sig	β	Student’s t	Sig
Constant	−0.875 ***	−3.981	0.000	−0.591 ***	−2.715	0.006	−1.146 ***	−5.295	0.000	−0.511 ***	−3.014	0.002
IoT	0.118 NS	1.482	0.130	0.602 ***	8.924	0.000	0.382 ***	5.672	0.000	0.242 ***	3.598	0.000
SIZE	0.279 ***	4.739	0.000	0.243 ***	4.514	0.000	0.098 **	2.013	0.046	0.091 **	2.095	0.038
AGE	0.007 NS	0.082	0.933	0.041 NS	0.761	0.447	−0.023 NS	−0.475	0.647	0.042 NS	0.992	0.321
TIS	0.511 ***	6.855	0.000	0.035 NS	0.558	0.577	0.465 ***	7.066	0.000	0.121 **	2.371	0.018
CEINT	-	-	-	-	-	-	-	-	-	0.557 ***	9.498	0.000
Adjusted R2	0.558			0.612			0.621			0.763
F	45.311 ***			55.982 ***			59.241			92.170
Variation Adjusted R2										0.151

***: significant at the 1% level (p < 0.01); **: significant at the 5% level (p < 0.05); (NS): not significant (p ≥ 0.10).

,

Table 6. Multiple regression results: AI & Big Data Analytics, CE Integration, and Performance.

	First Step: (AIBA → PER)						Second Step: (AIBA → CEINT)			Third and Fourth Steps: (AIBA → CEINT → PER)
	INP (M5)			OSP (M6)			CEINT (M7)			INP (M8)			OSP (M9)
	β	Student’s t	Sig	β	Student’s t	Sig	β	Student’s t	Sig	β	Student’s t	Sig	β	Student’s t	Sig
Constant	−0.812 ***	−4.076	0.000	−0.743 ***	−4.209	0.000	0.933 ***	−6.163	0.000	−0.488 ***	−2.720	0.007	−0.461 ***	−2.894	0.004
AIBA	0.563 ***	8.640	0.000	0.684 ***	12.042	0.000	0.587 ***	12.008	0.000	0.359 ***	5.571	0.000	0.518 ***	9.030	0.000
SIZE	0.211 ***	3.831	0.000	0.225 ***	4.606	0.000	0.048 NS	1.151	0.251	0.071 NS	1.386	0.167	0.103 **	2.272	0.024
AGE	−0.033 NS	−0.614	0.541	0.105 **	2.066	0.042	0.009 NS	0.175	0.863	−0.029 NS	−0.643	0.528	0.007 NS	0.175	0.863
TIS	0.171 **	2.603	0.010	0.014 NS	0.186	0.852	0.361 ***	7.249	0.000	0.169 ***	2.961	0.005	0.101 **	2.428	0.018
CEINT	-	-	-	-	-	-	-	-	-	0.419 ***	6.755	0.000	0.356 ***	6.463	0.000
Adjusted R2	0.601			0.692			0.771			0.698			0.762
F-statistic	54.221 ***			81.040 ***			121.300 ***			66.811 ***			92.743 ***
Variation Adjusted R2										0.097			0.070

***: significant at the 1% level (p < 0.01); **: significant at the 5% level (p < 0.05); (NS): not significant (p ≥ 0.10).

,

Table 7. Multiple regression results: Blockchain Technology, CE Integration, and Performance.

	First Step: (BCT → PER)						Second Step: (BCT → CEINT)			Third and Fourth Steps: (BCT → CEINT → PER)
	INP (M10)			OSP (M11)			CEINT (M12)			INP (M13)
	β	Student’s t	Sig	β	Student’s t	Sig	β	Student’s t	Sig	β	Student’s t	Sig
Constant	−2.155 ***	−13.671	0.000	−2.610 ***	−10.574	0.000	−1.864 ***	−9.366	0.000	−1.381 ***	−8.522	0.000
BCT	0.612 ***	9.763	0.000	0.078 NS	1.622	0.107	0.628 ***	8.215	0.000	0.402 ***	7.281	0.000
SIZE	0.253 ***	4.946	0.000	0.211 ***	4.002	0.000	0.265 ***	4.154	0.000	0.095 **	2.134	0.035
AGE	0.061 NS	0.980	0.329	−0.024 NS	−0.272	0.786	−0.161 **	−2.075	0.040	0.062 NS	1.235	0.219
TIS	0.192 **	2.624	0.040	0.711 ***	9.211	0.000	0.129 **	2.070	0.040	0.171 **	2.329	0.031
CEINT	-	-	-	-	-	-	-	-	-	0.467 ***	8.945	0.000
Adjusted R2	0.649			0.658			0.487			0.779
F-statistic	66.521 ***			69.110 ***			33.918 ***			100.214 ***
Variation Adjusted R2										0.130

***: significant at the 1% level (p < 0.01); **: significant at the 5% level (p < 0.05); (NS): not significant (p ≥ 0.10).

and

Table 8. Multiple regression results: Digital Collaboration Capability, CE Integration, and Performance.

	First Step: (DCC → PER)						Second Step: (DCC → CEINT)			Third and Fourth Steps: (DCC → CEINT → PER)
	INP (M14)			OSP (M15)			CEINT (M16)			INP (M17)			INP (M18)
	β	Student’s t	Sig	β	Student’s t	Sig	β	Student’s t	Sig	β	Student’s t	Sig	β	Student’s t	Sig
Constant	−0.471 ***	−3.088	0.002	−0.552 ***	−2.998	0.003	−1.153 ***	−5.770	0.000	−0.419 ***	−2.843	0.005	−0.445 **	−2.480	0.014
DCC	0.762 ***	15.375	0.000	0.703 ***	11.595	0.000	0.406 ***	6.417	0.000	0.034 NS	0.841	0.402	0.453 ***	6.289	0.000
SIZE	0.131 ***	2.902	0.005	0.171 ***	3.292	0.001	0.003 NS	0.017	0.987	0.578 ***	8.916	0.000	0.095 *	1.938	0.056
AGE	0.075 NS	1.554	0.123	0.055 NS	1.156	0.253	0.004 NS	0.043	0.974	0.018 NS	0.435	0.664	0.048 NS	1.128	0.262
TIS	0.012 NS	0.231	0.819	0.024 NS	0.357	0.731	0.517 ***	8.683	0.000	0.134 ***	3.115	0.003	0.045 NS	0.863	0.391
OPEN	-	-	-	-	-	-	-	-	-	0.259 ***	4.168	0.000	0.382 ***	5.683	0.000
Adjusted R2	0.781			0.682			0.641			0.804			0.739
F-statistic	128.330 ***			76.921 ***			64.281 ***			11.100 ***			81.953 ***
Variation Adjusted R2										0.023			0.057

***: significant at the 1% level (p < 0.01); **: significant at the 5% level (p < 0.05); *: significant at the 10% level (p < 0.10); (NS): not significant (p ≥ 0.10).

.

5.1. IoT Adoption and Sustainable Performance

Table 5 presents the relationships among IoT adoption, CE integration, and sustainable performance. In the initial step, IoT adoption exhibited a positive but non-significant effect on innovation performance (Model 1, β = 0.118, t = 1.482, p > 0.10). Conversely, IoT adoption was significantly associated with overall sustainable performance (Model 2, β = 0.602, t = 8.924, p < 0.001), satisfying the first mediation condition. In the second step, IoT adoption positively and significantly predicted CE integration (Model 3, β = 0.382, t = 5.672, p < 0.001).

To test mediation, Model 4 included both IoT adoption and CE integration. CE integration remained a significant predictor of overall sustainable performance (β = 0.557, t = 9.499, p < 0.001), while the direct effect of IoT adoption decreased in magnitude and significance (β = 0.242, t = 3.598, p < 0.001), indicating partial mediation. Therefore, Hypothesis 1 is supported for overall sustainable performance, but not for innovation performance.

5.2. AI and Big Data Analytics

Table 6 summarizes the effects of artificial intelligence and AIBA on performance outcomes. Models 5 and 6 show that AIBA significantly predicts both innovation performance (β = 0.563, t = 8.640, p < 0.001) and overall sustainable performance (β = 0.684, t = 12.042, p < 0.001). Model 7 confirms a strong positive association between AIBA and CE integration (β = 0.587, t = 12.008, p < 0.001).

Mediation analysis (Models 8 and 9) indicates that CE integration significantly predicts both innovation (β = 0.419, t = 6.746, p < 0.001) and overall sustainable performance (β = 0.356, t = 6.460, p < 0.001) after controlling for AIBA. The direct effects of AIBA decrease but remain significant (innovation: β = 0.359, t = 5.571, p < 0.001; overall: β = 0.518, t = 9.030, p < 0.001), consistent with partial mediation. These findings corroborate prior research highlighting AI and BDA as enablers of innovation and sustainable performance [17,58].

5.3. Blockchain Technology

Table 7 presents the results for blockchain technology (BCT). BCT significantly predicts innovation performance (Model 10, β = 0.612, t = 9.763, p < 0.001) but shows a positive yet non-significant association with overall sustainable performance (Model 11, t = 1.622, p = 0.107). BCT positively and significantly predicts CE integration (Model 12, β = 0.628, t = 8.215, p < 0.001).

When CE integration is included in Model 13, it significantly predicts innovation performance (β = 0.467, t = 8.945, p < 0.001), while the direct effect of BCT decreases but remains significant (β = 0.402, t = 7.281, p < 0.001), indicating partial mediation. Hypothesis 3 is therefore supported for innovation performance only, in line with prior studies on blockchain’s role in facilitating innovation processes [59,60].

5.4. Digital Collaboration Capability

Table 8 reports the effects of digital collaboration capability (DCC) on performance outcomes via CE integration. DCC significantly predicts both innovation performance (Model 14, β = 0.762, t = 15.375, p < 0.001) and overall sustainable performance (Model 15, β = 0.703, t = 11.595, p < 0.001). DCC also positively predicts CE integration (Model 16, β = 0.406, t = 6.417, p < 0.001).

In the mediation models (Models 17 and 18), CE integration significantly predicts both innovation (β = 0.258, t = 4.167, p < 0.001) and overall sustainable performance (β = 0.381, t = 5.682, p < 0.001). The direct effect of DCC on innovation performance becomes non-significant (β = 0.034, t = 0.841, p = 0.402), indicating full mediation. For overall sustainable performance, the direct effect remains significant but is reduced (β = 0.453, t = 6.289, p < 0.001), indicating partial mediation. Hypothesis 4 is thus supported, confirming the critical role of CE integration in translating digital collaboration into sustainable outcomes, consistent with prior literature on collaborative digital ecosystems [61,62].

5.5. Model Fit and Control Variables

Including CE integration as a mediator improves model fit and explanatory power across all pathways (M4, M8, M9, M13, M17, M18). For instance, in the IoT adoption–CE integration–sustainable performance model, the adjusted R² for overall sustainable performance increases from 0.612 (M2) to 0.763 (M4), highlighting CE integration’s importance.

Control variables, firm size (SIZE) and technological intensity of the sector (TIS), generally exert positive and significant effects on CE integration and performance. Firm age (AGE) does not significantly affect either mediator or outcome variables.

5.6. Sub-Dimensional Analysis

To provide deeper insight into the mechanisms through which digital capabilities enable circular practices, we conducted supplementary analyses examining the relationships between each digital enabler and the three CE sub-dimensions (CEWR, CERE, CESD). Results indicate differential patterns that validate the distinct mechanisms of each digital technology:

IoT Adoption most strongly predicts Resource Efficiency (CERE: β = 0.421, t = 5.892, p < 0.001), consistent with its role in real-time monitoring, energy optimization, and material flow tracking. Its effects on Waste Reduction (CEWR: β = 0.312, t = 4.103, p < 0.001) and Sustainable Design (CESD: β = 0.287, t = 3.745, p < 0.001) are significant but comparatively weaker.

AI and Big Data Analytics show the strongest associations with Waste Reduction (CEWR: β = 0.512, t = 7.634, p < 0.001), reflecting their capacity for predictive waste stream management, demand forecasting to reduce overproduction, and optimization of material utilization. Effects on Resource Efficiency (CERE: β = 0.445, t = 6.221, p < 0.001) and Sustainable Design (CESD: β = 0.398, t = 5.412, p < 0.001) are also significant.

Blockchain Technology demonstrates the strongest link to Sustainable Design (CESD: β = 0.487, t = 6.845, p < 0.001), aligning with its role in material provenance tracking, lifecycle transparency, and verification of sustainable sourcing claims. Effects on Waste Reduction (CEWR: β = 0.356, t = 4.789, p < 0.001) and Resource Efficiency (CERE: β = 0.334, t = 4.512, p < 0.001) are moderate.

Digital Collaboration Capability exhibits balanced effects across all three dimensions (CEWR: β = 0.432, t = 6.234, p < 0.001; CERE: β = 0.456, t = 6.678, p < 0.001; CESD: β = 0.478, t = 7.012, p < 0.001), confirming its role as an integrative organizational mechanism that enables holistic circular transformation through cross-boundary knowledge sharing and collaborative problem-solving.

These findings validate our theoretical proposition that different digital technologies enable specific circular mechanisms through distinct pathways, while also demonstrating that organizational coordination capabilities (Digital Collaboration) are essential for activating circular practices comprehensively. The differential patterns also support our decision to examine each digital technology separately rather than treating digitalization as a monolithic construct.

6. Discussion

This research integrates the CE paradigm to examine sustainable performance in SMEs. It responds to a notable gap in the existing literature, as these frameworks are often investigated independently rather than in combination. The study specifically explores the indirect relationship between digital technological capabilities, IoT adoption, AI and big data analytics, blockchain technology, and digital collaboration capability, and sustainable performance, mediated by CE integration. This approach reflects the growing body of research that positions digital technologies as key enablers of CE practices, thereby enhancing innovation and sustainability outcomes [57,63]. Two dimensions of performance were assessed: innovation performance and overall sustainable performance, encompassing environmental, social, and operational indicators.

The heterogeneous mediation patterns observed across digital capabilities provide nuanced insights into the mechanisms linking digitalization to sustainability. The full mediation of Digital Collaboration Capability’s effect on innovation performance suggests that collaborative digital ecosystems drive innovation exclusively through their enablement of circular practices; without circular orientation, digital collaboration yields limited innovation benefits. This finding extends RBV and dynamic capabilities theory by demonstrating that digital capabilities constitute strategic resources that must be channeled through specific operational practices (i.e., CE integration) to generate performance outcomes. In contrast, the partial mediation observed for AI/BDA and Blockchain indicates these technologies also exert direct effects, possibly through efficiency gains, predictive insights, or risk reduction mechanisms that operate independently of circularity.

The absence of significant mediation for IoT on innovation performance, and for Blockchain on overall sustainable performance, warrants careful interpretation. For IoT, the non-significant total effect (c path) suggests its innovation impact may be contingent on other factors not captured in our model, such as data analytics maturity or IoT-specific absorptive capacity. For Blockchain, the significant innovation effect but non-significant sustainability effect may reflect its current stage of diffusion, while blockchain enables novel business models and processes (driving innovation), its broader environmental and social impacts may require longer time horizons or ecosystem-level adoption to materialize.

For innovation performance, CE integration fully mediates the effect of digital collaboration capability, rendering the direct effect insignificant when mediation is introduced. Partial mediation is observed for AI and big data analytics, as well as blockchain technology, where the direct effects remain significant but are attenuated. IoT adoption demonstrates no significant mediation, suggesting a weak or linear relationship with innovation performance. For overall sustainable performance, CE integration partially mediates the effects of IoT adoption, AI and big data analytics, and digital collaboration capability, while no mediation is found for blockchain technology.

These outcomes underscore the mediating role of CE practices in leveraging digital capabilities for sustainability improvements. AI and big data analytics provide SMEs with the means to optimize resource loops, identify inefficiencies, and predict sustainability bottlenecks, thereby strengthening both innovation and broader performance. Blockchain technology, through its traceability and transparency functions, partially supports CE-driven innovation, although its wider sustainability impact appears to operate through direct mechanisms. Digital collaboration capability is particularly influential, fully mediating innovation performance by enabling cross-boundary knowledge sharing and the co-creation of circular solutions. Such collaborative capabilities extend organizational boundaries, allowing SMEs to leverage external data and partnerships to reinforce sustainability efforts, aligning with recent findings on digital–CE synergies [64,65].

This study advances theoretical understanding by explicitly linking RBV and CE. By conceptualizing digital capabilities as strategic resources and CE practices as mediating mechanisms, the study demonstrates how SMEs transform these resources into superior sustainable performance outcomes. This represents a novel empirical contribution in SME contexts. The findings also suggest that certain technologies, such as IoT, may exert direct linear effects, while collaborative tools primarily yield performance outcomes when mediated by CE practices, offering nuanced insights into digital–CE interactions.

From a practical standpoint, the findings provide actionable guidance for SMEs and policymakers. SMEs should prioritize digital collaboration platforms to foster ecosystems where multiple stakeholders co-develop circular solutions, implement AI-based predictive circular modeling to optimize resource loops, and selectively deploy blockchain for auditing, traceability, and verification of environmental claims. Providing CE-oriented training and digital literacy programs is critical to overcoming resource constraints and enabling SMEs to operationalize digital capabilities effectively. From a policy perspective, governments and industry bodies can support SMEs through financial incentives or subsidies for CE-aligned digital investments, capacity-building programs targeting technical skills in AI, IoT, and blockchain applications, standards and certification frameworks to facilitate CE adoption and ensure transparency in sustainability reporting, and collaborative innovation networks connecting SMEs with larger firms, universities, and technology providers to co-create scalable circular solutions. Collectively, these recommendations underscore that aligning digital transformation initiatives with CE principles is a strategic imperative for SMEs seeking to enhance sustainable performance and long-term competitiveness [59,66].

7. Conclusions

This research demonstrates that digital technological capabilities indirectly enhance SME sustainable performance through CE integration, though mediation patterns vary by capability and performance dimension. Specifically, digital collaboration capability fully mediates innovation performance, while AI and big data analytics and blockchain technology exhibit partial mediation. For overall sustainable performance, partial mediation is observed in relation to IoT adoption, AI and big data analytics, and digital collaboration capability, but not for blockchain technology. These findings position CE integration as a pivotal mechanism enabling SMEs to transform digital adoption into enhanced sustainability outcomes by institutionalizing circular principles, including resource efficiency, waste minimization, and product life-cycle optimization. From a theoretical perspective, this research extends both the RBV and CE paradigms by showing how digital capabilities, conceptualized as strategic resources, interact with CE integration to generate sustainable outcomes. The proposed integrative model offers a pioneering framework for SMEs, delineating capability-specific dynamics and mediating mechanisms between digitalization and sustainability.

From a practical perspective, this study provides actionable insights for SME managers aiming to operationalize sustainable and circular practices through digitalization. Managers are encouraged to align digital investments with circular economy objectives by integrating AI and data analytics to optimize material flows and resource efficiency, adopting collaborative digital platforms to foster co-innovation and knowledge exchange, and leveraging blockchain technology selectively to enhance supply chain transparency and traceability. Furthermore, developing internal digital skills and employee training programs can ensure that digital transformation efforts are consistent with circular design and waste minimization goals.

From a policy perspective, the findings underscore the need for supportive institutional frameworks. Governments in emerging economies should move beyond generic SME support and implement targeted programs that promote digital–circular convergence, such as (1) providing financial incentives or “green digital grants” to SMEs adopting hybrid digital–CE solutions; (2) establishing CE innovation clusters that connect SMEs with universities and technology providers to co-develop solutions; and (3) developing national sustainability auditing platforms, potentially leveraging blockchain, to enable SMEs to transparently report their CE performance and access green practices.

Collectively, these recommendations enhance the practical relevance of the research, providing both SME managers and policymakers with clear, evidence-based strategies to accelerate the transition toward circular and sustainable supply chains.

Several limitations warrant consideration. First, the study focuses exclusively on inbound CE integration and does not incorporate outbound practices, such as product repurposing or reverse logistics. Second, the reliance on a relatively modest sample of manufacturing SMEs may constrain the generalizability of the findings. Third, the use of perceptual measures introduces potential bias, while inter-firm dynamics, such as trust, governance mechanisms, or AI ethics, remain unexplored. Future research should extend the scope to outbound CE practices, incorporate dynamic capabilities, employ objective performance metrics, and adopt longitudinal and inter-firm research designs. Such efforts will enhance the applicability of the proposed model and provide a deeper understanding of digital sustainability pathways.