Today, companies are being called upon to embrace digitalization and address pressing environmental and social challenges. In doing so, they are increasingly integrating sustainability priorities and objectives into their innovation and digital transformation processes. The concept of sustainability-oriented innovation capability (SODC) has recently emerged to describe the ability of organizations to effectively integrate, build, and reconfigure resources to meet sustainability criteria in product design and proactively adopt digital technologies for green operations. Despite the progress in the literature on SODC, knowledge is still limited and further research is needed to improve the understanding of its antecedents. This study sheds some light on this by investigating three classes of factors that capture the different and complementary dimensions of digitalization (integrating knowledge management)—namely, the infrastructural, informational, and analytical: digital platforms capability (DPC), digital knowledge management (DKM), and big data management capability (BDMAC). We propose that these three dimensions influence the SODC of startups, and we identify the ultimate effects of these specific capabilities on their sustainability-oriented innovation outcomes, namely green product innovation performance (GPIP). Given the exploratory nature of this research, a partial-least squares approach to structural equation modeling was adopted. Our findings reveal that, unlike DKM, both DPC and BDAMC play a key role in influencing SODCs which, in turn, enable startups to improve their GPIP. The findings provide insights for founders and managers in their resource allocation strategies and for policymakers to support startups in developing these capabilities.
It is crucial for modern organizations to leverage new digital technologies to navigate continuously uncertain and changing environments in the current global situation (Santisteban et al., 2021; Sultana et al., 2022; Troise et al., 2022). Advances in digital or smart technologies, along with the associated digitalization processes and the exploitation of these new technologies, are in fact making a significant contribution to companies across a wide range of sectors in building or improving their agility (Bresciani et al., 2022; Ferraris et al., 2022). This is crucial to the survival, competitiveness, and improved performance of these companies, especially younger and smaller ones. While companies are called to embrace digital transformation, they must also face the urgent need to address current environmental and social challenges (Jung et al., 2025; Raman et al., 2024). Companies undergoing this transformation must effectively consider integrating new, more sustainable practices into their businesses and harnessing new technologies and innovations toward better and more sustainable operations (Hanelt et al., 2016; Phillips, 2018; Salamzadeh et al., 2022).
Over the years, the potential arising from the convergence of digitalization and sustainability has attracted the attention of a growing number of scholars, practitioners, and policymakers. Generally speaking, these represent two of the most debated topics in the literature of recent decades and are usually considered two independent fields of research; however, the connection between them and their prospective effects on achieving sustainable development have recently stimulated academic debate. Scholars have emphasized the importance of this nexus between corporate digitalization and sustainable development for reshaping business models and value creation processes, as well as balancing short- and long-term benefits while enabling the adoption of sustainable practices (Hellemans et al., 2022; Pizzi et al., 2021; Pricopoaia et al., 2025; Salamzadeh et al., 2022). Despite this evidence, companies are actually having difficulty translating the potential of new technologies into tangible sustainability results and aligning digital strategies with sustainability goals (Grybauskas et al., 2022; Guandalini, 2022; Hellemans et al., 2022; Meinhold et al., 2025). Similarly, organizations face recurring challenges and barriers (e.g., value-based ones) in sustainability-oriented alliances and innovation management activities, particularly in integrating stakeholder values into their innovation activities (Ivanov, 2025; Santa‐Maria et al., 2025).
Knowledge management (KM) and innovation processes or systems represent key drivers for the sustainable development of organizations and their effective exploitation of new technologies. KM represents a fundamental mechanism for companies to capture and share knowledge, thus facilitating decision-making and digital knowledge processes (Lim & Hwang, 2024; Martínez-Navalón et al., 2023). Innovation processes should drive companies to integrate sustainable priorities and goals into their digital transformation process to align with market demands (Cabrilo et al., 2024; Dzhunushalieva & Teuber, 2024; Shahzad et al., 2022). Following this perspective, several scholars have proposed the concept of sustainability-oriented innovation capability (SODC) as a valuable ability that organizations need to develop to incorporate sustainability and/or environmental criteria into product/service design or to proactively adopt digital technologies/tools for green operations, and ultimately gain long-term competitive advantages (Alkaraan et al., 2024; Dangelico et al., 2017).
The current literature on SODC is still developing (Alkaraan et al., 2024; Ortiz-Avram et al., 2024) and leaves significant gaps in this emerging field of research in relation to the antecedents of such capability—that is, its enabling factors in terms of capabilities or resources (Dangelico et al., 2017). Several scholars have attempted to explore the complex configurations of these factors that can lead organizations to improve their SODCs (Dangelico et al., 2017; van de Wetering et al., 2017). For example, Van de Wetering et al. (2017) argued that the increase in an organization’s SODCs is driven by three factors related to the environment, IT flexibility, and cross-border collaboration.
In this research, we have tried to contribute to the current debate on the interaction between digitization and KM and their role as potential strategic levers to be exploited by companies for their sustainable innovations (He et al., 2024; Vo-Thai & Tran, 2025). We therefore investigated three new classes of factors that have the potential to increase organizations’ SODCs, given their potential complementarity: digital platforms capability (DPC), digital knowledge management (DKM), and big data management capability (BDMAC); these are known to capture the different dimensions of digitalization while incorporating KM—namely the infrastructural, informational, and analytical dimensions, respectively (Hellemans et al., 2022; Lopes et al., 2017; Wang et al., 2023).
Anchored by two well-established theories, namely the resource-based view (RBV) and dynamic capability (DC) theory, we have developed a conceptual model that proposes these three dimensions as antecedents of the SODCs of startups, and highlights the ultimate effects of these capabilities on their sustainability-oriented innovation outcomes—namely, green product innovation performance (GPIP). Drawing on these theories, our study sought to shed light on how startups can pursue sustainability and related innovation outcomes through digitalization and KM. In doing so, we adopted the theoretical framework for SODC proposed by Dangelico et al. (2017) and empirically tested our conceptual research model with the relationships discussed above. We adopted a quantitative research design and resorted to the partial-least squares approach to structural equation modeling (PLS-SEM).
This study specifically focused on startups in the European context, a less explored type of venture in the current literature to explore these relationships. These companies face relevant challenges in balancing survival or development with sustainability (Jung et al., 2025) and are currently exploring new innovative digital technologies and KM practices to face these challenges and improve their survival in a competitive and turbulent environment. Our study thus provides an interesting contribution to the literature by exploring a little-researched case of ventures and proposing, as well as testing a new conceptual model that includes DPC, BDAMC, and DKM as driving factors behind SODCs. Our findings highlight that DPC and BDAMC are key drivers for startups in achieving higher levels of SODCs and these, in turn, improve their GPIP.
The remainder of this article is structured as follows: first, we introduce the theoretical background and develop the research hypotheses. Next, we present the research design, including data sampling and measures. We then report the methodology and display the results, which form the basis for the discussion and conclusions in the final section. A specific list of acronyms used in this paper is provided in Appendix 1.
Theoretical background and hypothesis developmentTheoryThis study examines how firms’ digitalization and KM capabilities translate into sustainability-oriented innovation outcomes by drawing on the RBV (Barney, 1991) and extending this perspective through the DC framework (Teece et al., 1997). RBV posits that firms achieve sustained competitive advantage through the possession and deployment of resources that are valuable, rare, inimitable, and non-substitutable (Barney, 1991). Organizations create economic value when such resources are mobilized in ways that competitors cannot easily replicate (Barney & Wright, 1998; Makadok, 2001; Newbert, 2008). From this perspective, digital assets—including digital platforms, big data analytics tools, and KM systems—constitute strategic resources with the potential to create and capture value.
Despite its strengths, RBV has been criticized for adopting a relatively static orientation, which overlooks how firms renew, develop, and reconfigure their resource base in dynamic and uncertain environments (Dubey et al., 2019; Kero & Bogale, 2023). To address this limitation, RBV is complemented in this study by the DC framework, which highlights the organizational ability to integrate, build, and reconfigure resources in response to environmental change (Teece, Pisano, & Shuen, 1997). In this sense, DCs extend and refine RBV by addressing its rigidity problem and shifting the focus from resource possession to resource orchestration (Ambrosini & Bowman, 2009; Eisenhardt & Martin, 2000; Mishra et al., 2022).
Within this theoretical context, digitalization and KM emerge as critical enablers of firms’ DCs. Digital platforms and business analytics tools provide the technological and organizational conditions that strengthen sensing, seizing, and reconfiguring processes, thereby enhancing firms’ capacity to generate innovation and maintain competitiveness (Gao & Sarwar, 2022; Mikalef & Pateli, 2017; Xiao et al., 2020). At the same time, knowledge has long been recognized as a dominant factor in the creation and development of firms’ DCs, serving as the foundation for sensing opportunities, integrating insights, and reconfiguring resources (Eisenhardt & Martin, 2000; Li et al., 2025; Santoro et al., 2021; Tseng & Lee, 2014).
Scientific studies have also increasingly emphasized the relevance of DCs in advancing sustainability-oriented outcomes and shown how they enable firms to align the deployment of resources with ecological and social objectives (Dangelico et al., 2017; Russo, 2003; Wu et al., 2012). This growing body of research highlights that DCs sustain competitiveness in turbulent environments as well as play a pivotal role in embedding sustainability into innovation processes. Building on this theoretical foundation, the present study adopts the DC framework as the mechanism through which digital and KM resources, conceptualized within the RBV, are mobilized and transformed into tangible sustainability outcomes.
Literature review and conceptual modelThe pursuit of sustainability through digitalization has emerged as a critical frontier in the research of academics and practitioners. Digital transformation offers organizations unprecedented opportunities to align economic performance with environmental and social imperatives, thereby redefining pathways toward sustainable development (Hellemans et al., 2022; Pricopoaia et al., 2025). However, the academic debate is only beginning to uncover the complex mechanisms linking digitalization to sustainability outcomes.
Empirical evidence on how digital tools and technologies influence environmental and social performance remains limited and fragmented (Anastasiadou et al., 2021; Broccardo et al., 2023). Nishant et al. (2020) provide a systematic literature review examining how artificial intelligence technologies can contribute to achieving sustainability outcomes, while also identifying the key challenges and opportunities associated with their adoption. In addition, the role of digital platforms, the Internet of Things (IoT), and sensor technologies has been also examined for their potential to enhance sustainability-related outcomes (Perotti et al., 2024). These technologies can be seen as part of the infrastructural dimension of digitalization. Other research streams have investigated the informational and analytical facets of digitalization and their influence on firms’ sustainability performance. For instance, Wamba et al. (2017) identified big data analytics as a critical organizational capability that fosters sustainable competitive advantage, whereas Martínez-Navalón et al. (2023) have provided empirical evidence of a strong and positive relationship between DKM practices and firms’ sustainability outcomes.
Within this research context, the present study seeks to advance the understanding of how digitalization supports sustainability-oriented innovation by addressing the fragmentation that still characterizes the digital–sustainability nexus. Existing research has mainly examined isolated technologies or individual organizational capabilities, providing only a partial view of how different digital dimensions interact to enhance sustainability performance. To fill this gap, this study adopts an integrative perspective that conceptualizes digitalization as a multidimensional construct encompassing three complementary yet distinct dimensions: infrastructural, informational, and analytical. Together, these dimensions capture the technological, knowledge-based, and data-driven aspects through which digital transformation can foster sustainability-oriented outcomes. More specifically, this study develops a theoretical model that examines how these three digitalization dimensions influence the SODCs of startups and how these latter ultimately impact the sustainability-oriented innovation outcomes of these new ventures, namely their GPIP.
Three antecedent capabilities are identified as particularly relevant in the digital–sustainability nexus. First, DPC refers to an organization’s ability to use advanced digital tools and platforms as competitive instruments (Ahmed et al., 2022; Cenamor et al., 2019; Troise et al., 2023). Digital platforms represent strategic resources that allow firms to achieve sustainable competitive advantages (Sakas et al., 2014). Recent studies have highlighted their strategic and transformative role in fostering sustainable entrepreneurship (Hadizadeh et al., 2024). Their inclusion in the model reflects the growing recognition that platformization is central to leveraging ecosystem-level innovation for sustainability (Chen et al., 2021; Gao et al., 2022).
Second, BDAMC refers to a firm’s ability to systematically manage information technology (IT) resources and analytics routines in line with business priorities (Wamba et al., 2017). Sustainability challenges are inherently data-intensive and require the ability to process large and complex datasets to address environmental issues, improve energy efficiency, and promote sustainable practices (Wu et al., 2016). By transforming data into actionable insights, BDAMC provides the analytical foundation that can be leveraged as a source of sustainable competitive advantage (Wamba et al., 2017).
Third, DKM captures the implementation of information and communication technology (ICT) to manage organizational knowledge and facilitate digital knowledge processes (Martínez-Navalón et al., 2023; Shaher & Ali, 2020). DKM ensures that sustainability-related knowledge circulates across organizational boundaries, thereby enhancing firms’ absorptive capacity (Rodríguez-Gonzàlez et al., 2023) and organizational learning (He et al., 2024), both of which are essential for sustainable development (Wang et al., 2022).
These three antecedents are emphasized because they jointly capture the infrastructural (digital platforms), informational (DKM), and analytical (big data analytics) dimensions of digitalization, which have been identified as among the dimensions the most directly tied to sustainability-oriented innovation (Alkhatib, 2024; Aziz et al., 2024; Barnes et al., 2022; Hellemans et al., 2022; Kolk & Ciulli, 2020; Lopes et al., 2017; Martínez-Navalòn et al., 2023; Shehzad et al., 2024; Wang et al., 2023). The model then positions SODCs as a key predictor of innovation outcomes of startups. SODCs are defined as a firm’s ability to integrate, build, and reconfigure resources to embed sustainability into new product development in response to environmental and market changes (Dangelico et al., 2017).
Conceptualized as a second-order construct, SODCs consist of external resource integration (ERI), internal resource integration (IRI), and resource building and reconfiguration (RBR). ERI refers to the ability to exchange and incorporate sustainability knowledge from suppliers, customers, regulators, and other stakeholders. IRI reflects the capacity to coordinate environmental knowledge across departments and functions. RB represents the development of new sustainability competencies and the restructuring of existing resources to respond to ecological demands. This multidimensional capability provides the organizational mechanism through which digital platforms, data analytics, and KM can be transformed into sustainability-oriented innovation. Without such dynamic capabilities, digitalization efforts may remain confined to operational efficiency rather than driving genuine green innovation (Hein et al., 2026).
Finally, the outcome of interest is GPIP, which refers to the environmental dimension of product innovation performance, including the development of new products that embody attributes such as energy saving, pollution prevention, waste recycling, or eco-design (Chen et al., 2006). GPIP reflects a crucial aspect of companies’ sustainability, as it directly addresses the growing demand for environmentally friendly products while supporting long-term competitiveness (Albort-Morant et al., 2016).
Hypothesis developmentDPC and SODCsDPC refers to an organization’s ability to leverage advanced digital tools and platforms as competitive instruments (Ahmed et al., 2022; Troise et al., 2023). Prior research has shown that DPC positively influences innovation, competitive advantage, and overall firm performance (Cenamor et al., 2019; Mikalef & Pateli, 2017). However, scholars also argue that the impact of DPC on firm outcomes is not necessarily direct but rather operates through the development of DCs (Kroh et al., 2018; Ravichandran, 2018). In this view, digital platforms provide the technological and organizational conditions that enhance sensing, seizing, and reconfiguring processes and resources, thereby enabling firms to build the DCs required for sustained competitiveness and innovation (Xiao et al., 2020).
Supporting this perspective, Mikalef and Pateli (2017) demonstrate through multiple case studies of European firms that IT capability, defined as the ability to mobilize and deploy IT-based resources, plays a fundamental role in fostering DCs, specifically by enabling sensing, seizing, and reconfiguring processes. Additionally, Chen et al. (2014) provide quantitative evidence that the DC of the business process fully mediates the relationship between IT capabilities and firm performance, further confirming that IT-related resources influence outcomes primarily through the development of higher-order capability. While their study refers to IT capabilities more broadly, digital platforms can be conceptualized as a contemporary and concrete manifestation of such capabilities, functioning as a form of IT infrastructure that facilitates integration and reconfiguration (Sedera et al., 2016).
In line with this reasoning, Li et al. (2022) show also that internet platforms are a form of digital platform that strengthens information sharing within firms (supporting internal integration) and across firms (supporting external integration), while also improving the allocation of traditional resources (supporting reconfiguration). Collectively, these findings confirm that digital platforms provide the infrastructural and organizational conditions for building DCs. Building on this stream of research, it can be argued that DPCs are equally critical for the development of SODCs, which specifically focus on integrating environmental knowledge across organizational and inter-organizational boundaries and reconfiguring resources to embed sustainability principles into product innovation (Dangelico et al., 2017). As such, DPC is expected to play a decisive role in strengthening SODCs, which in turn drive sustainability-oriented innovation outcomes.
- •
H1: DPC positively influences SODCs
Big data analytics capability (BDAC) is broadly defined as the competence to provide business insights using data management, infrastructure, and personnel knowledge to transform business into a competitive force (Wamba et al., 2017). A growing body of research has demonstrated that BDAC enhances DCs by enabling firms to sense opportunities, seize resources, and reconfigure processes in data-intensive environments. For example, Mikalef et al. (2019), using survey data from 175 chief information officers and IT managers in Greek firms, have shown that BDAC enables organizations to generate insights that strengthen their DCs. In a complementary multiple case study of 27 European firms, Mikalef et al. (2021) further confirm BDAC as a strong enabler of DC development. More recently, Wu et al. (2024), in a large-scale empirical study of 352 firms, reinforced this positive relationship between BDAC and DCs, while other studies also provide consistent evidence of this link (Pedota, 2023; Yoshikuni et al., 2023).
While BDAC is commonly conceptualized as a multidimensional construct, this study focuses on BDAMC, which is defined as a firm’s ability to handle routines in a structured (rather than ad hoc) manner to manage IT resources in accordance with business needs and priorities (Wamba et al., 2017). BDAMC represents the managerial and strategic dimension of BDAC. Because BDAMC is one of the central dimensions of BDAC, its effect can be expected to follow the same logic as overall BDAC. Specifically, BDAMC ensures that analytics insights are effectively mobilized to guide sensing, integration, and reconfiguration processes at the organizational level. The positive impact of BDAMC on DCs has been further empirically confirmed by Gao and Sarwar (2022) in a study involving 149 firms. Building on this body of research, the present study extends these findings to the sustainability domain. Just as BDAC and, in turn, BDAMC have been shown to foster DCs, it is argued here that BDAMC also plays a decisive role in developing SODCs, thus enabling firms to sense sustainability opportunities, integrate sustainability knowledge across organizational boundaries, and reconfigure resources for eco-innovation (Dangelico et al., 2017). Accordingly, the following hypothesis is proposed:
- •
H2: BDAMC positively influences SODCs.
KM has long been defined as the formalized approach to managing the creation, transfer, retention, and utilization of an enterprise’s explicit and tacit knowledge assets (Liebowitz & Wilcox, 1997; O’Leary, 1998). The positive relationship between KM capabilities and firms’ DCs is well established. Over the years, scholars have emphasized knowledge as a dominant factor and primary mechanism for the creation and development of DCs (Eisenhardt & Martin, 2000; Iris & Vikas, 2011; Li et al., 2025; Santoro et al., 2021; Tseng &Lee, 2014). Knowledge resources enable firms to sense environmental shifts, seize emerging opportunities, and reconfigure resources in line with market and technological changes.
In the digital economy, however, the nature of KM is evolving (Li et al., 2025). Rapid access to information and knowledge has become vital for organizational competitiveness (Mizintseva & Gerbina, 2018), while new digital technologies enable firms to apply knowledge more rapidly and adaptively (Zbuchea & Vidu, 2018). This shift has given rise to DKM, which is defined as the ability to implement and use ICT to effectively manage organizational knowledge (Martínez-Navalón et al., 2023) and to facilitate knowledge processes digitally (Shaher & Ali, 2020). DKM reflects the digitalization of knowledge flows, thus ensuring that relevant knowledge is collected, codified, shared, and applied across organizational boundaries.
Building on the well-established link between KM and DCs, DKM can therefore be seen as a key enabler of DC development in environments shaped by digital technologies. Extending this reasoning to the sustainability domain, DKM is expected to play a pivotal role in the development of SODC by making it possible to absorb sustainability-related knowledge from external stakeholders (external resource integration), diffuse it across internal functions (internal resource integration), and embed it into processes of resource building and reconfiguration. Accordingly, it is assumed:
- •
H3: DKM positively influences SODCs.
The existing literature on DCs has long highlighted their central role in enhancing organizational performance (Aftab et al., 2024; Albort-Morant et al., 2016; Khan et al., 2020; Laaksonen & Peltoniemi, 2018; Teece, 2007; Teece et al., 1997). More recently, scholars have extended this perspective to examine how DCs intersect with issues of sustainability. For example, Russo (2003) applied the DCs framework to show how the adoption of new process standards strengthens firms’ capacity to improve environmental performance by reducing toxic emissions. Similarly, Bhupendra and Sangle (2015) emphasized the contribution of DCs to pollution prevention. Chakrabarty and Wang (2012) investigated how organizations mobilize DCs to design and implement sustainability practices, while Mousavi et al. (2018) further demonstrated the positive influence of firms’ DCs on sustainability-driven innovation.
Reinforcing the strong link between DCs and sustainability, Wu et al. (2012) introduced the concept of sustainable DCs, and later, Dangelico et al. (2017) refined this idea by formulating the concept of SODC as the firm’s ability to integrate, build, and reconfigure competences and resources to embed environmental sustainability into new product development to respond to changes in the market. According to Dangelico et al. (2017), firms that continuously exchange and integrate sustainability-related knowledge and competences both with external partners and internally, while simultaneously creating environmental knowledge and reconfiguring firm resources to address environmental sustainability challenges, are better positioned than their competitors to develop green product innovations that succeed in the marketplace. Based on these findings, the following hypothesis is proposed:
- •
H4: SODC positively influences GPIP
The proposed hypotheses are represented in the conceptual model shown in Figure 1.
MethodologyData and sample
The units of analysis for this study are startups in the European context. These companies are registered in the specific Startup Business Registers in their respective countries of origin. For example, in the case of Italy, which is the country most represented in the sample, there is a specific register dedicated to innovative startups (https://startup.registroimprese.it/). Data were collected through a structured questionnaire; the survey process was conducted in two phases. In the first phase, a pilot study was carried out with eight participants to test the clarity and reliability of the questionnaire. Specifically, we involved the CEOs of eight startups that we knew from previous studies and that were eligible to be included in our research. We administered the questionnaire to these participants to check its readability, clarity, and understanding of the various terminology, thus allowing us to make any necessary changes to the questions to improve the final questionnaire before its distribution (Podsakoff et al., 2012; Ruel et al., 2016). Feedback was carefully reviewed and incorporated in its final version to enhance clarity and comprehensibility.
The second phase involved the administration of the finalized questionnaire, which was distributed between July 23 and September 5, 2025. Anonymity was assured to participants to mitigate social desirability bias. To minimize common method bias, the order of the items was randomized and a dedicated message was sent to participants explaining the academic purposes of the research (Cimino et al., 2025a; Kline et al., 2000, Podsakoff et al., 2003). In total, 298 responses were collected. Following a rigorous screening process, removing cases with uniform response patterns, incomplete questionnaires with missing values, and incoherent answers, 273 valid responses were kept and formed the final dataset. This sample size exceeds the minimum thresholds recommended in the literature (Barclay et al., 1995; Kock & Hadaya, 2018). The demographic characteristics of the startups included in the final sample are reported in Table 1.
The quality of the responses was verified by checking for potential insufficient effort responses (IER) bias (Huang et al., 2012). We first tested the systematic IER and checked for the presence of long strings—that is, sequences of responses in the same category from an individual. In our case, no such problems (systematic IER) emerged because no response category had a single string longer than five. Second, we checked the random component, and the Mahalanobis distance was adopted. This second test also ruled out any problems (p-value of having an outlier in the dataset was above the suggested threshold of 0.01) and therefore confirmed that there was no risk of random IER in our dataset.
Non-response bias was also tested, particularly given the nature of the survey (cross-sectional and online). Levene’s test was used, and early and late respondents were compared. Analysis of this two-wave procedure (Armstrong & Overton, 1977) showed that no significant differences were found between respondents, thus ruling out potential problems of non-response bias in this study. A paired t-test on the latent variables was conducted, and the results showed no statistical differences between these two groups (significance level 0.05).
MeasuresPreviously validated measurement scales were employed to assess the proposed theoretical model. All measurement items were sourced from prior studies and adjusted, when necessary, to match the contextual setting of this research. Respondents evaluated all items using a 7-point Likert scale, ranging from 1 (strongly disagree) to 7 (strongly agree). The model consists of four constructs in total: three independent variables, one mediating construct, and one dependent variable. The independent variables are DPC, BDAMC, and DKM. DPC was measured using a 7-item scale adapted from Troise et al. (2023); BDAMC was assessed with a 21-item scale developed by Wamba et al. (2017); and DKM was captured through a 6-item scale from Cabrilo et al. (2024). The mediating construct, SODC, was conceptualized as a second-order construct consisting of three first-order dimensions: ERI (4 items), IRI (3 items), and RBR (6 items). The items for these measures were adapted from Dangelico et al. (2017). Finally, the dependent variable, GPIP, was measured with a 4-item scale adapted from Chen et al. (2006). In total, the survey instrument included 51 items, the details of which are presented in Table 2.
ResultsData analysisThe PLS-SEM approach was used to assess the structural model and test the proposed hypotheses in this study, given its exploratory nature (Becker et al., 2023; Hair et al., 2017). PLS-SEM represents a suitable technique given its flexibility while still guaranteeing robustness and dealing with small sample sizes, as in our case (Hair et al., 2017, 2019, 2021; Willaby et al., 2015). The analyses were carried out with the last version of the SmartPLS 4.0 software.
Given the cross-sectional approach used for the survey and the potential problems related to the presence of common method bias (CMB), we controlled this through the full collinearity variance inflation factors (VIFs) approach (Hair et al., 2019). The results showed that our model had acceptable full VIFs, in our case, lower than 3.6, and thus below the threshold of 5. We also tested for the presence of CMBs by adopting Harman’s single-factor test (Podsakoff et al., 2003) to avoid potential effects on subsequent analysis. In this test, the extraction of the first factor is done using principal axis factoring. In our case, the value (first factor) was well below the conventional thresholds (50% of the total variance). Hence, CMB was not a problem in our model and did not influence the subsequent analysis. We performed PLS-SEM following the two main steps proposed in the literature—namely, the evaluation of the quality of the measurement model and that of the structural model’s predictive power (Hair et al., 2017; Henseler et al., 2009).
Measurement modelFirst, we evaluated the measurement model—that is, the consistency of the indicators and constructs were assessed through the adoption of Cronbach’s alpha and composite reliability. Second, the convergent validity was evaluated by adopting both average variance extracted (AVE) and loadings, followed by the evaluation of the discriminant validity by adopting the heterotrait–monotrait (HTMT) method and the cross-loadings criterion.
As reported in Table 3, both Cronbach’s alpha and composite reliability values exceeded the traditional threshold of 0.7 proposed in the literature (Hair et al., 2017), thus highlighting acceptable reliability. The convergent validity of the model was also ensured. All the AVE values were above the threshold of 0.50 (Hair et al., 2017), while the loadings were greater than 0.7 except for five items related to BDAMC and two related to DPC (see Appendix 2); however, these items exceeded 0.6 and are therefore considered acceptable for exploratory studies. (Hair et al., 2011). As shown in Appendix 2, all the indicators have loadings that exceed the cross loadings on the other constructs.
Finally, the discriminant validity assessed through the HTMT method, which highlighted that all the values are below the traditional threshold of 0.9 proposed in the literature, as shown inTable 4 (Henseler & Fassott, 2010; Henseler et al., 2015), thus confirming the discriminant validity of this model.
Characteristics of the sample.
Source: authors’ elaboration
Constructs and items description.
| Construct | Item-ID | Item description | Source |
|---|---|---|---|
| Digital Platform Capability(DPC) | DPC-1 | Our platform easily accesses data from our partners’ IT systems | Troise et al. (2023) |
| DPC-2 | Our platform has the capability to exchange real-time information with our partners | ||
| DPC-3 | Our platform easily aggregates relevant information from our partners’ databases (e.g. operating information, business customer performance, cost information etc.) | ||
| DPC-4 | Our platform is easily adapted to include new partners | ||
| DPC-5 | Our platform can be easily extended to accommodate new IT applications or functions | ||
| DPC-6 | Our platform employs standards that are accepted by most current and potential partners | ||
| DPC-7 | Our platform consists of modular software components, most of which can be reused in other business applications | ||
| Big Data Analytics management capabilities(BDAMC) | BDAMC-1 | We continuously examine innovative opportunities for the strategic use of business analytics | Wamba et al. (2017) |
| BDAMC-2 | We enforce adequate plans for the utilization of business analytics | ||
| BDAMC-3 | We perform business analytics planning processes in systematic ways | ||
| BDAMC-4 | We frequently adjust business analytics plans to better adapt to changing conditions | ||
| BDAMC-5 | When we make business analytics investment decisions, we estimate the effect they will have on the productivity of the employees' work | ||
| BDAMC-6 | When we make business analytics investment decisions, we project how much these options will help end users make quicker decisions | ||
| BDAMC-7 | When we make business analytics investment decisions, we estimate whether they will consolidate or eliminate jobs | ||
| BDAMC-8 | When we make business analytics investment decisions, we estimate the cost of training that end users will need | ||
| BDAMC-9 | When we make business analytics investment decisions, we estimate the time managers will need to spend overseeing the change | ||
| BDAMC-10 | In our organization, business analysts and line people meet regularly to discuss important issues | ||
| BDAMC-11 | In our organization, business analysts and line people from various departments regularly attend cross-functional meetings | ||
| BDAMC-12 | In our organization, business analysts and line people coordinate their efforts harmoniously | ||
| BDAMC-13 | In our organization, information is widely shared between business analysts and line people so that those who make decisions or perform jobs have access to all available know-how | ||
| BDAMC-14 | In our organization, the responsibility for analytics development is clear | ||
| BDAMC-15 | We are confident that analytics project proposals are properly appraised | ||
| BDAMC-16 | We constantly monitor the performance of the analytics function | ||
| BDAMC-17 | Our analytics department is clear about its performance criteria | ||
| BDAMC-18 | Our company is better than competitors in connecting (e.g., communication and information sharing) parties within a business process | ||
| BDAMC-19 | Our company is better than competitors in reducing cost within a business process | ||
| BDAMC-20 | Our company is better than competitors in bringing complex analytical methods to bear on a business process | ||
| BDAMC-21 | Our company is better than competitors in bringing detailed information into a business process | ||
| Digital Knowledge Management(DKM) | DKM-1 | Our company uses information communication technology (ICT) to enable efficient information search and discovery | Cabrillo et al. (2024) |
| DKM-2 | Our company uses ICT in internal communication throughout the organization | ||
| DKM-3 | Our company uses ICT to communicate with external stakeholders | ||
| DKM-4 | Our company uses ICT to analyze knowledge and make better decisions | ||
| DKM-5 | Our company uses ICT to collect business knowledge related to its competitors, customers and operating environment | ||
| DKM-6 | Our company uses ICT to develop new products and services with external stakeholders | ||
| Green Product Innovation Performance (GPIP) | GPIP-1 | Our company chooses the materials of the product that produce the least amount of pollution for conducting the product development or design | Adapted from Chen et al. (2006) |
| GPIP-2 | Our company chooses the materials of the product that consume the least amount of energy and resources for conducting the product development or design | ||
| GPIP-3 | Our company uses the fewest amount of materials to comprise the product for conducting the product development or design | ||
| GPIP-4 | Our company would circumspectly deliberate whether the product is easy to recycle, reuse, and decompose for conducting the product development or design | ||
| Sustainability-Oriented Dynamic Capability (SODC)* | |||
| External Resource Integration (ERI) | ERI-1 | Our company integrates customers’ requirements about products’ environmental performance | Adapted from Dangelico et al. (2017) |
| ERI-2 | Our company integrates knowledge on environmental impact of products during customers’ use | ||
| ERI-3 | Our company integrates suppliers’ knowledge and competencies on environmental impact of components or materials | ||
| ERI-4 | Our company integrates suppliers’ knowledge and competencies on environmental impact of production processes | ||
| Internal Resource Integration (IRI) | IRI-1 | Within our company there is collaboration between specialized environmental unit (e.g. environmental sustainability managers, environmental sustainability unit) and design function/department | Adapted from Dangelico et al. (2017) |
| IRI-2 | Within our company there is collaboration between specialized environmental unit (e.g. environmental sustainability managers, environmental sustainability unit) and production function/department | ||
| IRI-3 | Within our company there is collaboration between specialized environmental unit (e.g. environmental sustainability managers, environmental sustainability unit) and marketing function/department | ||
| Resource Building and Reconfiguration (RBR) | RBR-1 | Our company has hired/hires environmental specialists (e.g. experts on life cycle assessment (LCA) and Design for the Environment (DfE)) | Adapted from Dangelico et al. (2017) |
| RBR-2 | Our company has trained/trains (e.g. through attendance at conferences, workshops, courses) product development team members to upgrade their environmental knowledge and competencies | ||
| RBR-3 | Our company has trained/trains (e.g. through attendance at conferences, workshops, courses) R&D staff to upgrade their environmental knowledge and competencies | ||
| RBR-4 | Our company has strengthened/strengthens environmental R&D (e.g. increasing the scope, increasing investments) | ||
| RBR-5 | Our company has reconfigured/reconfigures organizational structure to focus on environmental sustainability (e.g. creating a new division, reconfiguring product lines) | ||
| RBR-6 | Our company has reconfigured/reconfigures product development teams to include environmental specialists | ||
Source: authors’ elaboration
Construct reliability and validity.
Source: authors’ elaboration
The hypotheses proposed in our research model in Figure 1 were tested by performing the PLS model, and a bootstrapping technique, based on 5,000 re-samples, was applied to ensure the robustness and significance of results. We examined the structural model through path modeling and the related relationships between the model’s constructs. Path coefficients are reported in Table 5. As shown in this table, our results support H1, H2, and H4, while H3 was not supported. Specifically, both BDAMC (β = 0.758, p < .001) and DPC (β = 0. 249, p < .05) have positive and statistically significant effects on SODC, while DKM has a relatively small and negative value that is, however, not statistically significant. The results also showed that SODC (β = 0.377, p < .001) is positively linked to GPIP. We also analyzed the R² to assess the predictive power of the constructs (Chin, 1998; Hair et al., 2017, 2021); substantial predictive power was found for SODC (R² = 0.756) while weak (very low) predictive power was found for GPIP (R² = 0.142).
Path coefficients.
| HPs | Path | Β | Sample mean (M) | SD | T_stat | 2.50 CI | 97.50 CI | Support |
|---|---|---|---|---|---|---|---|---|
| H1 | DPC ➔ SODC | 0.249⁎⁎ | 0.247 | 0.114 | 2.178 | 0.114 | 0.114 | YES |
| H2 | BDAMC ➔ SODC | 0.758⁎⁎⁎ | 0.761 | 0.096 | 7.873 | 0.096 | 0.096 | YES |
| H3 | DKM ➔ SODC | -0.155 | -0.152 | 0.075 | 2.077 | 0.075 | 0.075 | NO |
| H4 | SODC ➔ GPIP | 0.377⁎⁎⁎ | 0.395 | 0.080 | 4.710 | 0.080 | 0.080 | YES |
Source: authors’ elaboration
Note:
The results provide support for H1, thus confirming that DPCs positively influence SODC. This finding is consistent with the broader body of research showing that IT-related capabilities represent a fundamental enabler of DCs. Prior studies have emphasized that the influence of IT capabilities on firm performance is rarely direct but operates through the development of higher-order capabilities such as sensing, seizing, and reconfiguring (Kroh et al., 2018; Ravichandran, 2018). Digital platforms can thus be seen as a contemporary manifestation of IT capabilities (Sedera et al., 2016), providing the technological and organizational infrastructure that supports integration and reconfiguration processes.
By extending these insights to the sustainability domain, our study demonstrates that DPC are also critical for the development of SODC. Startups that effectively leverage digital platforms are better able to exchange and integrate sustainability-related knowledge across internal and external boundaries, as well as reconfigure resources to embed environmental considerations into product innovation (Dangelico et al., 2017). This result reinforces earlier findings that platforms enhance information sharing and resource allocation (Li et al., 2022), but it also goes further by showing that these platform-enabled processes directly contribute to SODCs.
Importantly, this finding is particularly suitable in the context of our sample of Italian innovative startups, which are firms in their first years of life. In fact, at this early stage, startups are typically characterized by severe resource constraints, limited formal structures, and a strong dependence on external networks for survival and growth. In such conditions, digital platforms represent a cost-effective and scalable alternative to costly in-house infrastructures, as they are often accessible through freemium or pay-per-use models that reduce upfront investment. What makes the difference, however, is the development of DPC—namely, the ability to leverage these digital platforms as competitive instruments. Through DPC, very young firms are able to coordinate activities, access complementary resources, and connect with partners and markets that would otherwise be unattainable. Acting as substitutes for the organizational routines and infrastructures of more mature firms, DPCs enable startups to externalize resource-intensive activities and focus their scarce resources on building sustainability-oriented dynamic capabilities. Consequently, the positive relationship between DPC and SODC in our study illustrates that very young startups can rely on platformization as a strategic mechanism to overcome their liabilities due to newness while building the dynamic capabilities required to integrate sustainability into their organizational processes.
The results also provide strong support for H2, showing that BDAMC has a significant and positive effect on SODC. This highlights the importance of analytics management as a strategic and managerial capability that enables organizations to align data-driven insights with business priorities (Wamba et al., 2017). By establishing structured routines for handling data and IT resources, startups can continuously mobilize analytical knowledge to support sensing, integration, and reconfiguration processes.
Our findings are consistent with prior studies that link analytics management to dynamic capability development (Gao & Sarwar, 2022) and extend this evidence to the sustainability domain. In particular, BDAMC provide the mechanisms through which sustainability-related data can be systematically processed and translated into actionable knowledge. This allows startups to identify opportunities, integrate environmental knowledge across functions, and reallocate resources to address ecological and market demands. The strong effect of BDAMC on SODC in our sample can be explained by the fact that very young firms often lack established routines and accumulated organizational experience. Operating under high uncertainty and resource scarcity, these firms benefit greatly from systematic and data-driven decision-making. Unlike KM processes, which typically rely on formal structures and stable routines, analytics management can deliver immediate and objective guidance even in the absence of organizational maturity. This makes BDAMC a particularly powerful driver of SODC in the context of startups.
The results show that H3 is not supported, which indicates that DKM does not exert a significant influence on SODC. This contrasts with the extensive body of literature that has consistently highlighted KM as a critical enabler of DCs (Eisenhardt & Martin, 2000; Iris & Vikas, 2011; Li et al., 2025; Santoro et al., 2021;Tseng & Lee, 2014). Several interrelated explanations can be advanced for this finding. First, the difference may arise from the qualitative nature of DKM compared to other digital capabilities. DKM embodies a process-oriented and internally focused capability, emphasizing knowledge codification, sharing, and storage through ICT systems (Martínez-Navalón et al., 2023). In contrast, capabilities such as DPC and BDMAC are inherently action-oriented and externally oriented, enabling rapid sensing, ecosystem engagement, and data-driven responsiveness. In a startup context, characterized by time pressure, agility, and a premium on execution, these externally directed digital capabilities may contribute more directly to dynamic reconfiguration processes than the more foundational, knowledge-structuring functions of DKM. Thus, DKM’s effect may be more indirect and long-term, supporting organizational learning trajectories that only materialize once a certain degree of structural and process maturity is achieved.
Second, startups often operate under conditions of low organizational formalization and emergent routines, which constrains their ability to benefit from the digitalization of knowledge processes. DKM relies on stable information architectures, codified routines, and repository-based learning systems, which are costly to establish and maintain. Early stage ventures, in contrast, depend largely on tacit and experiential learning, with knowledge embedded in the founders’ intuition, team interactions, and real-time experimentation. Attempting to digitalize such informal flows too early may disrupt their agility rather than enhance their adaptability, resulting in an immaturity paradox, where the technological sophistication of DKM exceeds the organizational absorptive capacity needed to exploit it effectively.
Third, this finding may reflect a temporal or sequencing effect in the development of dynamic capabilities. From a capability evolution perspective (Teece, 1997), firms often build sensing and seizing capabilities before developing reconfiguration capabilities that depend on systematic knowledge integration. In this evolutionary logic, DKM may become a second-order enabler of SODCs only once startups reach a certain stage of growth, stability, and resource endowment.
Finally, moving on to the last hypothesis, the analysis confirms H4, showing that SODCs are a significant driver of GPIP. This result is in line with prior research that positions DCs at the core of firms’ ability to enhance organizational performance under changing environmental and market conditions (Albort-Morant et al., 2016; Teece, 2007; Teece et al., 1997). Extending this reasoning to sustainability, scholars have emphasized how SODCs enable firms to transform environmental knowledge and stakeholder inputs into eco-innovations (Dangelico et al., 2017). Our findings provide empirical support for this view and show that startups that succeed in integrating, building, and reconfiguring resources around sustainability are also those that achieve stronger outcomes in green product innovation. This relationship is particularly meaningful for innovative startups whose survival often depends on differentiation and legitimacy. In this setting, developing SODC is not only beneficial but often essential: it provides a systematic way to embed sustainability in product development despite resource constraints. Green innovations become a way for startups to stand out in crowded markets, respond to regulatory and societal expectations, and attract critical external support such as investment and partnerships. By cultivating SODCs, startups gain the flexibility to continuously reconfigure limited resources toward sustainability goals, thus turning constraints into opportunities for eco-innovation.
The positive effect of SODC on GPIP in our study reflects how DCs oriented toward sustainability act as both a shield and a lever. As a shield, SODCs protect very young firms from the liabilities of newness by providing legitimacy and credibility with external stakeholders and by offering structured ways to deal with uncertainty and resource scarcity. As a lever, they allow startups to reconfigure scarce resources toward sustainability goals and to differentiate themselves in markets where environmental performance is increasingly valued. Through this dual function, SODC mitigate vulnerabilities typical of early stage firms but also create avenues for competitive positioning in sustainability-driven niches.
Implications for theory and practiceOur study responds to some of the calls for more research to investigate the antecedents of SODC (see, among others, Dangelico et al., 2017) and to improve our collective understanding of the relationships between digitalization, KM, and sustainability innovation and outcomes. Together, the findings advance the literature on digitalization, KM, and sustainability-oriented innovation in four ways. First, this study refines DC theory and clarifies the heterogeneous nature of digital antecedents. The results reveal that digitalization should not be treated as a monolithic construct uniformly fostering capability development. Rather, its impact depends on the functional orientation of digital capabilities. Infrastructural (DPC) and analytical (BDAMC) capabilities directly enhance the sensing, integrating, and reconfiguring processes underlying SODCs, while DKM does not exert a significant effect. This finding challenges the prevailing assumption in the DC literature that digitalization inherently strengthens dynamic capabilities (Gao & Sarwar, 2022; Mikalef & Pateli, 2017; Xiao et al., 2020). Instead, it suggests a more accurate interpretation: for startups, only digital capabilities that are action-oriented and outward-looking, enabling data-driven sensing and ecosystem coordination, translate effectively into sustainability-oriented capabilities.
Second, the study extends the integration of RBV and DC by introducing organizational maturity as a boundary condition for the resource-capability conversion process. Platforms and analytics can act as substitutes for missing routines and structures in early stage firms (Corvello et al., 2024), thus supporting the formation of SODCs despite resource scarcity (Cimino et al., 2025b), whereas DKM requires codified processes and stability that startups typically lack (Prashantham & Kumar, 2019). This insight contributes to theory-building by linking the effectiveness of digital resources to firms’ developmental stage, thus suggesting that the path from digitalization to sustainability capabilities is evolutionary rather than immediate.
Third, and more broadly, this study addresses a key limitation in the existing literature—its fragmented and technology-specific approach to the digitalization-sustainability nexus (Anastasiadou et al., 2021; Broccardo et al., 2023)—by simultaneously examining three distinct yet interrelated digital dimensions (infrastructural, informational, and analytical). This holistic perspective advances theory by illustrating how these dimensions jointly underpin the development of SODCs, ultimately offering a more integrated conceptual understanding of the mechanisms through which digitalization supports sustainable innovation. In doing so, the study bridges previously disconnected theoretical streams on digital platforms, big data analytics, and DKM to provide a unified lens on how digital resources interact to enable sustainability outcomes.
Finally, the study enriches the research stream on SODCs (Dangelico et al., 2017) by examining their role in the context of early stage startups. In this context, SODCs operate as both a shield and a lever—that is, they mitigate the liabilities of newness by providing legitimacy and structure, while simultaneously enabling green product innovation and differentiation in sustainability-driven niches.
Beyond advancing theory, the findings also have important practical implications. The results indicate that DPC and BDAMC are the most effective levers. First, investing in DPC enables startups to build the flexible infrastructure needed for coordination, ecosystem engagement, and sustainable opportunity sensing. Rather than broad or generic digital investments, startups should prioritize modular, cloud-based, and interoperable platforms that support collaboration with partners and customers and facilitate access to external knowledge and resources. Second, strengthening analytics practices allows startups to ground sustainability decisions in evidence rather than intuition. Managers should focus on lightweight but systematic analytics routines (defining a few critical sustainability indicators and monitoring them through dashboards), while cost-effective solutions (including open-source tools) can help balance technological sophistication with startups’ resource constraints. For policymakers, the results highlight the need to create conditions that enable startups to acquire and deploy these two types of digital capabilities. Policy efforts will be most effective when they: (a) support access to scalable digital infrastructures, thus lowering entry barriers for startups; (b) strengthen digital and analytical skills related to sustainability through targeted training programs; and (c) promote collaborative environments, such as digital innovation hubs or sustainability-focused accelerators, where startups can access data, expertise, and potential partnerships.
Limitations and future research directionsThis exploratory research has some limitations. First, to derive causations, we used cross-sectional data; this, as know, has some limits that could be solved in a subsequent research based on a longitudinal research design approach. Second, we focused on startups in the European context. Despite the vibrant and dynamic context explored on this continent, our findings are subject to restrictions in terms of generalization, as they cannot be extended to other countries outside Europe, particularly startups with different characteristics. This leads to a future research opportunity—that is, to analyze other countries and continents to compare the results. It would also be interesting to fully understand whether and to what extent cultural differences can have a certain impact. Third, the limited knowledge of SODCs and the absence of similar studies that adopt new related and emerging constructs (Dangelico et al., 2017) does not allow us to compare our findings with previous research. Fourth, in light of our findings, future studies could further investigate the role of DKM, including, for example, whether it can act as a moderator or mediator in this model or in a more complex one. Finally, to measure innovation outcomes of startups we used only one measure, namely GPIP; however, in addition to product innovation, future research could use other measures (e.g., process innovation) to better frame and enrich the conceptual model and more thoroughly measure the impacts on innovation performance.
CRediT authorship contribution statementJintao Lu: Writing – review & editing, Writing – original draft, Visualization, Validation, Supervision, Software, Investigation, Formal analysis, Conceptualization. Cheng Chen: Writing – review & editing, Writing – original draft, Visualization, Validation, Software, Resources, Methodology, Investigation, Data curation, Conceptualization. Antonio Cimino: Writing – review & editing, Writing – original draft, Visualization, Validation, Software, Resources, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Ciro Troise: Writing – review & editing, Writing – original draft, Visualization, Validation, Supervision, Software, Resources, Methodology, Investigation, Formal analysis, Data curation, Conceptualization.
No conflict of interest exits in the submission of this manuscript, and manuscript is approved by all authors for publication.
This work was supported by the 2025 Annual Ideological and Political Work Research Projects of Shanxi under Grant [25SXSZ0215], University-Industry Collaborative Education Project under Grant [250705458160033], The Postgraduate Education and Teaching Reform Project of Shanxi under Grant [2024JG166], The Teaching Reform and Innovation Research Project of Taiyuan University of Science and Technology under Grant [JG2023056].
List of Acronyms.
Source: authors’ elaboration
Cross Loadings.
Source: authors’ elaboration
