Sustainable Product Innovation in SMEs: The Role of Digital–Green Learning Orientation and R&D Ambidexterity

Abstract

As digitalization and environmental sustainability advance globally, small and medium-sized enterprises (SMEs) are facing transformative pressures as well as emerging opportunities. Rapid digital innovation promotes intelligent production, cost reduction, efficiency gains, and improved management practices, while green development mandates emphasize energy conservation, emissions reduction, and sustainable supply chains. Amid concurrent digital and green transformations, SMEs are leveraging digital technologies to bolster green learning and enhance sustainable product development. This study investigates the digital–green learning orientation (DGLO) and its influence on ambidextrous research and development (R&D) capabilities, which in turn shape sustainable product development performance (SPDP). Drawing on survey data from 306 SMEs in eastern and southern China, multiple regression analysis was employed to assess the relationships between DGLO, ambidextrous R&D capabilities, and SPDP. The findings reveal that DGLO significantly enhances SPDP. Moreover, DGLO promotes SPDP through both exploitative and exploratory R&D capabilities, with each playing a complementary role.

1. Introduction

As digitalization and greening converge globally, SMEs are facing transformation pressures alongside new opportunities. Rapid digitalization enables SMEs to adopt intelligent production modes, optimize structures, and streamline value chains, thereby reducing costs and improving efficiency. Simultaneously, the global pursuit of carbon neutrality compels enterprises to prioritize sustainability through clean production, energy conservation, and green supply chains. In navigating this “dual transformation,” SMEs must confront challenges like technological upgrades, financial investments, and talent shortages while integrating their operations into digital–green innovation systems to enhance resource allocation and environmental performance. The integration of green principles with digital capabilities strengthens competitiveness and facilitates energy savings, emission reductions, and efficiency improvements in product development, fostering continuous advancements in sustainable production performance.

Given the complexity and uncertainty of digital–green transformation, relying solely on external technology acquisition or short-term investments is inadequate for improving SPDP. Achieving sustainable product development through digital–green transformation requires enhancing internal learning capabilities in parallel with technological implementation. Accordingly, the DGLO has emerged as a crucial internal driver helping SMEs to achieve transformational and innovative breakthroughs. Early research on learning orientation emphasized organizational knowledge acquisition and absorptive capacity [1]. With the integration of green principles and digital technologies, the concept of a DGLO has been introduced. A DGLO emphasizes continuous knowledge accumulation, enhanced employee cognition, and learning behaviors directed toward digital–green convergence. It creates an internal mechanism where green awareness and digital literacy evolve synergistically [2,3]. Chanias et al. (2019) argued that a digital strategy constitutes a dynamic process combining “learning” and “practice,” underscoring the adaptive interplay between formulation and implementation [4]. In the context of digital–green convergence, organizational ambidexterity is essential for balancing flexibility and efficiency. Zhao et al. (2024) noted that integrating digital and physical systems requires enterprises to pursue both exploratory and exploitative innovation [5]. This dual approach allows them to adapt to dynamic environmental shifts while meeting internal demands for resource integration.

In summary, researchers have not adequately explored the performance implications of DGLO in terms of three critical aspects: (a) the internal mechanisms of SMEs remain understudied, as most studies primarily examined the concept and antecedents of digital–green transformation using qualitative methods; (b) although previous studies have linked digital–green transformation to financial and innovation performance, its impact on SPDP remains underexamined; and (c) the pathways linking digital transformation to performance remain unclear, leaving underlying mechanisms largely unexplored.

This study examines how Chinese SMEs enhance SPDP during digital–green transformation by adopting a learning orientation that integrates digitalization and greening. Driven by a DGLO, SMEs leverage digital technologies to reconfigure green organizational capabilities, thereby addressing the tensions and coordination challenges between digitalization and greening in R&D transformation. A DGLO facilitates the deep integration of digital and green principles at the cognitive, technological, and process levels, fostering continuous learning and knowledge renewal, in turn allowing the development of ambidextrous innovation abilities. Exploitative innovation capacity involves the intensive application and optimization of existing resources, processes, and knowledge, thereby consolidating competitive advantages along established technological and market paths. In contrast, exploratory innovation capacity breaks path dependence by allowing one to acquire and apply new knowledge and technologies, driving organizational reconfiguration and renewal. The synergy between research and development activities constitutes R&D ambidexterity, a mechanism that reconciles efficiency and flexibility while managing conflicts across technological trajectories [6]. Due to restricted resources and capacity-building needs, SMEs must enhance their exploratory and exploitative R&D capabilities during digital–green transformation. This ambidexterity is essential for improving knowledge and capabilities, acting as the key mechanism by which a DGLO boosts SPDP.

This study investigates SMEs in eastern and southern China to elucidate how a DGLO influences SPDP through digital enablement. Drawing on survey data from 306 firms, we employ multiple regression analysis to assess DGLOs’ impact on enhancing ambidextrous R&D capabilities, encompassing both exploitative and exploratory innovation. The study also examines how these capabilities enhance SPDP, thereby clarifying the interconnections among DGLO, R&D capabilities, and SPDP.

The theoretical contributions of this study are threefold. (a) It extends the theoretical framework of digital–green transformation by addressing a gap in existing research, which has primarily examined strategic trajectories and capability building regarding large firms through qualitative methods. It reveals how a DGLO improves new-product development (NPD) performance by augmenting ambidextrous R&D capacity, thus enriching our understanding of innovation performance from a learning-oriented perspective. (b) This study differs from prior studies by focusing on SPDP as the main outcome variable, extending performance outcomes in digital–green studies beyond financial or general innovation measures. (c) We apply organizational-learning and ambidexterity theories within the digital–green context. Establishing a theoretical link among learning orientation, capability building, and performance outcomes extends the relevance of learning orientation and R&D ambidexterity to integrated transformation contexts while also advancing knowledge of capability evolution mechanisms in digital–green transformation.

2. Literature Review and Research Hypothesis

2.1. DGLO

DGLO is a fusion of digital learning orientation and green learning orientation, integrating concepts, tools, and capabilities to combine digital and green dimensions within learning logic. Unlike a digital learning orientation, which emphasizes digital knowledge and information system capabilities, a DGLO incorporates green principles to avoid resource wastage and environmental impacts from the unchecked pursuit of efficiency and scale. It requires enterprises to possess technical agility and align digital capabilities with green values, thereby achieving synergy between digital empowerment and environmental goals. In contrast to a green learning orientation, a DGLO goes beyond adopting green concepts and environmental responsibilities by promoting the use of digital tools to strengthen information perception, cross-border knowledge integration, and technology transformation, making green strategies more practical and scalable. A DGLO also differs from market and entrepreneurial orientations. A market orientation centers on customer demands, whereas a DGLO integrates environmental responsibility and digital empowerment into organizational learning logic, enabling proactive management of environmental risks and regulatory constraints [7]. An entrepreneurial orientation emphasizes innovation and risk-taking; a DGLO also values innovation but stresses the integration of green concepts and digital technologies to sustain innovation while reducing ecological and resource risks [8]. Thus, a DGLO addresses immediate pressures and nurtures long-term competitiveness. It helps enterprises adapt to policy, market, and environmental challenges and develop lasting skills in information perception, knowledge integration, and technology transformation. For SMEs with limited resources, technology, and manpower, a DGLO is vital. While large enterprises can often afford to undergo independent green transformations, SMEs face high costs. A DGLO provides a cost-effective and flexible path for knowledge enhancement, supporting capability development through incremental learning, diversified knowledge acquisition, and resilience in uncertain environments. It enables SMEs to balance exploration and exploitation, strengthen R&D ambidexterity, and advance green technologies, product development, and green business models under the dual-carbon strategy and digital transformation.

Based on the above analysis, Figure 1 presents the conceptual framework of DGLO.

2.2. DGLO and SPDP

A DGLO has the potential to drive continuous improvement in information- and knowledge-intensive enterprises by enhancing the integration of digital technologies and green concepts. This can lead to increased efficiency in resource utilization, improved knowledge sharing, and overall advancement in SPDP across R&D, production, and management processes. For instance, sustainable product development teams can utilize three-dimensional computer-aided design tools for virtual prototype design, incorporating green principles early in the design phase to optimize performance. Furthermore, by using knowledge management platforms and desktop sharing technologies, teams can engage in real-time cross-regional collaboration, facilitating the integration of green concepts into R&D efforts. Additionally, employing an artificial-intelligence-based project management system enables dynamic progress monitoring, risk prediction, and optimal resource allocation, ensuring that the influence of digital technologies is pervasive in the design-to-implementation process. This approach will enhance product innovation capabilities while advancing environmental sustainability [9,10]. In manufacturing contexts, an increasing number of enterprises are leveraging intelligent manufacturing, big data, cloud computing, and social platforms to promote sustainable product development, addressing the challenges posed by digitalization and environmental responsibilities [11,12]. Systematic research in the textile industry indicates that digital transformation fosters greater consumer involvement in green product R&D and enables enterprises to utilize big data on consumers’ environmental behaviors to transition R&D decision-making towards a “human–data interaction” model, thereby enhancing the innovation performance of sustainable products [13,14]. Recent studies underscore the significant role of artificial intelligence: Silva et al. (2024) demonstrated that frequent use of digital tools in new-product development can notably enhance innovation performance, a result largely attributed to substantial investments in digital–green learning [15]. Zhang et al. (2023) supported this notion through a survey of 35 Chinese listed companies, indicating that digitalization can bolster enterprises’ information technology capabilities in NPD, thereby effectively promoting SPDP [16]. In summary, as a strategic approach, a DGLO integrates digital technologies with green development imperatives, heightening organizational awareness of digital–green synergy, optimizing resource allocation efficiency, and enhancing the quality of green innovation. Building upon this foundation, we posit the following hypotheses.

H1: 

The enterprises’ degree of DGLO has a positive impact on SPDP.

2.3. The Mediating Role of R&D Ambidexterity

Organizational ambidexterity theory categorizes R&D activities into exploitation and exploration [17]. Exploitation refines established paradigms through efficiency improvements, standardization, or stricter process control, whereas exploration creates new paradigms or applies existing ones in novel ways to generate value [18,19]. R&D ambidexterity is an organization’s capacity to balance the two activities, and in the context of DGLO’s impact on SPDP, digital–greening empowers organizations to achieve this balance, which is essential for managing the challenges posed by digital–green transformation. R&D exploration capability (RDEC) involves innovative experiments characterized by high uncertainty and long-term returns and is aimed at identifying technological paths and market opportunities that shape future growth [20]. During digital–green transformation, exploration helps firms overcome path dependence on existing technologies and markets to capture unmet sustainable demands [21]. SMEs are often more adaptable and agile in disruptive innovation than large firms, benefiting from streamlined hierarchies, concise decision-making, and less entrenched operations [22], which favor early entry into green markets. Through continuous digital knowledge acquisition, green technology absorption, and cross-domain integration [23,24], DGLO enables SMEs to recognize emerging trends and experiment with novel digital methods and technologies in R&D, fostering innovation in processes and methodologies [25]. RDEC, driven by digital–green learning, supports forward-looking transformation in product design, manufacturing, and business models, leading to sustainable products that combine environmental benefits with intelligent features. Digitalization also dismantles rigid structures, promoting distributed and decentralized collaboration [26] while also reshaping resource allocation through data analysis and virtual simulation, reducing sunk costs and improving efficiency [27]. Based on this reasoning, the following hypothesis is proposed.

H2a: 

RDEC plays a mediating role in the relationship between DGLO and SPDP.

R&D exploitation capability (REXC) refers to the enhancement of existing capabilities, technologies, and processes through refinement, integration, and expansion in R&D, thereby improving organizational performance [28,29]. Optimizing and standardizing current R&D processes enable employees to use accumulated knowledge and resources efficiently, improving the efficiency and quality of NPD [30]. When adopting a DGLO, enterprises integrate digital tools with green technology expertise to upgrade traditional processes through intelligent production lines, improving energy efficiency and resource utilization. Incorporating intelligent detection and energy-monitoring systems into production lines reduces energy waste and enhances green performance without altering core product structures [31]. Enhancements involve gradual optimization of products and processes, such as transforming traditional products into green and intelligent products that remain energy-efficient and environmentally friendly across their life cycles, rather than creating disruptive new ones [31]. Examples include adding an intelligent energy efficiency management module to photovoltaic systems to increase conversion efficiency or embedding big-data monitoring in irrigation equipment to improve water use. REXC advances SPDP through refined application and continuous improvement of existing technologies. Therefore, we propose the following hypothesis.

H2b: 

REXC plays a mediating role in the relationship between DGLO and SPDP.

2.4. The Interaction Effect Between RDEC and REXC

Studies on organizational ambidexterity and performance reveal conflicting views. Some emphasize their positive effects, arguing that exploitation and exploration complement each other by allowing the sharing and expansion of resources such as knowledge and information, thereby enhancing performance [19,32,33,34]. Organizations can implement ambidexterity by separating activities across units, time, or space or fostering a culture and processes that accommodate contradictions and cultivate employees’ ambidextrous thinking [17,18,35]. In information systems research, IT use is also categorized into exploration-oriented and exploitation-oriented types, which together balance innovation with efficiency and improve performance [36,37]. Other scholars contend that ambidextrous capabilities may hinder performance because of substitution effects. Resource constraints lead to competition between activities, organizational inertia causes capability lock-in, and structural or cultural conflicts create tension, resulting in a “capability trap” [38,39,40,41]. Empirical studies have confirmed this risk: Su et al. (2022) showed that ambidexterity weakens the adaptive benefits of exploration for SMEs, while Li et al. (2022) identified an inverted U-shaped interaction between exploration and exploitation [34,42]. SMEs, however, with flat structures, rapid decision-making, and adaptable cultures, can manage ambidexterity more effectively than large corporations with rigid hierarchies and entrenched routines [43]. In the context of digital and green transformation, a DGLO offers SMEs a pathway allowing them to balance exploitation and exploration by optimizing existing processes and resources while enhancing exploratory innovation through digital knowledge and green technology absorption. Firms like BYD illustrate this dual approach in the new-energy bus sector, advancing power battery and intelligent control technologies while optimizing assembly processes and supply chains through digitalization, thereby boosting efficiency, innovation, and environmental performance. Thus, digital–green learning enables enterprises to harmonize R&D exploitation and exploration, enhancing SPDP.

H3: 

REXC and RDEC complementarily enhance SPDP in SMEs.

Specifically, REXC positively moderates the link between RDEC and SPDP, while RDEC similarly moderates the relationship between REXC and SPDP.

The theoretical model pertaining to this study is shown in Figure 2.

3. Methodology

3.1. Data Collection and Participants

In this study, we utilized a questionnaire survey method to gather data, targeting enterprises experienced in applying digital technologies and SPD. The specific questions assessed whether enterprises have adopted a DGLO and promote sustainable practices. Data collection occurred in two phases: April to August 2024 and April to June 2025. The questionnaires were distributed via three main channels: online platforms; on-site distribution among MBA students, supplemented by electronic distribution through MBA alumni email groups; and distribution by alumni in Beijing, Shanghai, Guangdong, and Zhejiang, who distributed paper questionnaires locally and collected responses online via personal networks. The choice of this sample is justified by two considerations. First, SMEs in eastern and southern China, where digital infrastructure and industrial clusters are highly developed, face significant pressure regarding digital and green transformation, making them a representative group for this study. Second, the typical characteristics of SMEs—limited resources and organizational flexibility—are also prevalent in other emerging and developed economies, supporting the generalizability of the findings.

An analysis of variance conducted on data from the three sources and two phases revealed no significant differences, allowing for a combined analysis. Of the 700 questionnaires distributed, 426 were returned, corresponding to a response rate of 60.8%. After screening, 120 questionnaires were deemed invalid due to incomplete responses or evident similarities, leaving 306 valid questionnaires. The characteristics of the sample are detailed in

Table 1. Sample distribution.

Variable	Items	N	%
Industry	Mechanical and Equipment Manufacturing	75	24.50
	Electronics and Information Manufacturing	96	31.37
	Chemical Engineering and Materials Manufacturing	72	23.52
	Food and agricultural product processing and manufacturing industry	37	12.09
	Others	26	8.49
Size	Less than 100	19	6.20
	101–300	77	25.16
	300–500	139	45.42
	Over 500	71	23.20
Age	Less than 3	38	12.41
	3–5	102	33.33
	7–10	92	30.06
	Over 10	74	24.18

.

3.2. Measurement

To ensure this study’s reliability and validity, we employed established scales from prior research. The DGLO scale, comprising six items, is based on the work of Al Halbusi (2023) and Yin et al. (2024) [2,44]. The R&D ambidexterity scale, with ten items, measures two dimensions, namely, REXC and RDEC, drawing from Im and Rai (2008, 2014) [45,46]. Each dimension contains five items. The SPDP scale, with four items, was developed in the studies by Chen et al. (2006) and Al Halbusi et al. (2023) [47,48]. All these scales use a 5-point Likert format. Detailed items are listed in Appendix A. The control variables include industry type, enterprise size, and years since establishment, with industry type coded as a dummy variable, using “others” as the reference group.

4. Results

4.1. Reliability and Validity Tests

We employed Cronbach’s α to assess variable reliability. The results indicate that the Cronbach’s α coefficients for each variable and its dimensions (refer to

Table 2. The results of the reliability and convergent validity tests.

Variable	Items	Factor Loading (λ)	Cronbach’s α	AVE	CR
DGLO	DGLO1	0.842	0.919	0.695	0.920
	DGLO2	0.772
	DGLO3	0.756
	DGLO4	0.750
	DGLO5	0.863
	DGLO6	0.878
RDEC	RDEC1	0.826	0.884	0.593	0.879
	RDEC2	0.782
	RDEC3	0.754
	RDEC4	0.763
	RDEC5	0.722
REXC	REXC1	0.832	0.888	0.634	0.896
	REXC2	0.782
	REXC3	0.818
	REXC4	0.714
	REXC5	0.830
SPDP	SPDP1	0.746	0.834	0.660	0.884
	SPDP2	0.770
	SPDP3	0.711
	SPDP4	0.748

) exceeded 0.70, surpassing the acceptable threshold of 0.60. Additionally, the composite reliability (CR) for each variable was above the critical value of 0.70, confirming the scale’s reliability.

This study primarily assesses content and discriminant validity. For content validity, the measurement scales for DGLO, R&D ambidexterity, and SPDP are based on established scales by both domestic and international scholars, ensuring robust content validity. Discriminant validity was evaluated through confirmatory factor analysis, with the results presented in Table 2. The standardized factor loadings (λ) for DGLO, RDEC, REXC, and SPDP exceed 0.60, surpassing the 0.50 threshold. Additionally, the average variance extracted (AVE) for each variable exceeds 0.60, with square roots above 0.70, all surpassing the correlation coefficients with other factors. Thus, the scale demonstrates strong discriminant validity.

4.2. Descriptive Statistical Analysis

All the variables exhibit VIF values below the critical threshold of 10, indicating there is no significant multicollinearity. We conducted a non-response bias test by comparing early and late respondents and found no significant differences in firm age, firm size, or other characteristics at the 0.05 significance level, indicating that the relationships between variables are not substantially affected. Therefore, non-response bias is unlikely to have a major impact on the study’s conclusions. Additionally, the Harman’s single-factor test for common method bias revealed a maximum single-factor variance of under 40%, suggesting there is no substantial bias. The correlation coefficients among variables (refer to

Table 3. Correlation and descriptive statistics.

Variable	1	2	3	4	5	6	7
IT	1
ES	−0.033	-
YE	−0.038	−0.04	-
DGLO	−0.012	−0.194 **	0.137 *	0.812
RDEC	0.022	−0.158 **	0.113 *	0.328 **	0.770
REXC	−0.04	−0.265 **	0.135 *	0.551 **	0.441 **	0.796
SPDP	0.04	−0.180 **	0.142 *	0.434 **	0.415 **	0.480 **	0.812
Mean	1.92	2.56	2.03	3.75	3.49	3.38	3.67
SD	0.74	0.61	0.73	0.97	1.04	0.94	0.95

Note: * denotes a 5% significance level, ** denotes a 1% significance level; IT and ES refer to industry types; YE indicates the year an enterprise was established; DGLO stands for digital–green learning orientation; RDEC represents R&D exploration capability; REXC signifies R&D exploitation capability; and SPDP indicates sustainable product development performance. The definitions hold for the tables below as well.

) are all below 0.7. Notably, there are significant positive correlations between DGLO, RDEC, REXC, and SPDP, aligning with the research hypothesis.

4.3. Hypothesis Testing

4.3.1. Main Effect Test

Table 4. Results of the regression analysis on the relationship between DGLO and SPDP.

Variable	SPDP
	M1	M2
IT	0.039 (0.698)	0.045 (0.869)
ES	−0.173 (−3.09)	−0.097 (−1.853)
YE	0.137 (2.442)	0.085 (1.639)
DGLO		0.404 ** (7.648)
R2	0.052	0.206
Adj-R2	0.043	0.196
F	5.553 **	19.580 **

Note: ** represents the significance level of 1%, and the values in parentheses are T-values. The same applies hereinafter. F denotes the F-statistic for the overall significance of the regression model. The definitions hold for the tables below as well.

illustrates that, when controlling for industry type, firm size, and year of establishment in Model 1 (M1), the inclusion of the independent variable, DGLO, significantly enhances SPDP (β = 0.404, p < 0.01). This finding confirms H1.

4.3.2. Mediating Effect Test

The mediating effect was evaluated using the step-by-step regression method. After controlling for industry type, enterprise nature, and scale, as shown in

Table 5. Regression analysis of the relationship between DGLO and R&D ambidexterity.

Variable	RDEC		REXC
	M3	M4	M5	M6
IT	0.021 (0.375)	0.025 (0.468)	−0.044 (−0.805)	−0.037 (−0.792)
ES	−0.153 (−2.71)	−0.097 ** (−1.753)	−0.261 (−4.743)	−0.165 ** (−3.431)
YE	0.107 (1.899)	0.069 ** (1.26)	0.123 ** (2.241)	0.058 (1.212)
DGLO		0.300 ** (5.402)		0.511 ** (10.569)
R2	0.037	0.122	0.088	0.335
Adj-R2	0.072	0.110	0.124	0.326
F	3.848 **	10.450 **	9.662 **	37.828 **

, M4 and M6 reveal DGLO significantly enhances RDEC and REXC (β= 0.300, p < 0.01; β = 0.511, p < 0.01). As shown in

Table 6. Regression analysis of the relationship between ambidextrous R&D capabilities and SPDP.

Variable	SPDP
	M7	M8
IT	0.031 (0.598)	0.059 (1.181)
ES	−0.114 * (−2.181)	−0.054 (−1.038)
YE	0.096 (1.836)	0.081 ** (1.592)
RDEC	0.385 ** (7.313)
REXC		0.457 ** (8.699)
R2	0.195	0.243
Adj-R2	0.185	0.233
F	18.259 **	24.115 **

, M7 and M8 demonstrate that RDEC and REXC have positive effects on SPDP (β= 0.385, p < 0.01; β= 0.457, p < 0.01). Subsequently, the combined influence of DGLO and R&D ambidexterity on SPDP was assessed. In

Table 7. Mediating-effect test.

Variable	SPDP
	M9	M10
IT	0.037 (0.761)	0.057 (1.163)
ES	−0.069 (−1.372)	−0.042 (−0.830)
YE	0.065 (1.310)	0.066 (1.327)
DGLO	0.316 (5.998)	0.234 (3.960)
RDEC	0.292 (5.594)
REXC		0.333 (5.548)
R2	0.281	0.280
Adj-R2	0.269	0.268
F	23.499 **	23.370 **

, M9 and M10 show that incorporating RDEC and REXC into M2 maintains the significant impact of DGLO on SPDP (β= 0.316, p < 0.01; β= 0.234, p < 0.01), albeit with a reduced effect (0.404→0.316; 0.404→0.234). RDEC and REXC partially mediate the relationship between DGLO and SPDP, supporting H2a and H2b.

We employed the Bootstrapping method, with 500 iterations, to assess the mediating effects of RDEC and REXC.

Table 8. Bootstrap regression analysis under the mediating effect of R&D ambidexterity.

Specification	Estimate	95% Confidence Interval
		Percentage
		LL	UP
DGLO → RDEC → SPDP	0.087	0.038	0.148
DGLO → REXC → SPDP	0.170	0.091	0.261

presents the findings, revealing that both abilities significantly mediate DGLO’s influence on SPDP, as the confidence interval excludes zero. Thus, H2a and H2b are confirmed.

4.3.3. Test of the Interaction Effect Between RDEC and REXC

We utilized stepwise regression to assess how the interaction between RDEC and REXC influences SPDP. As detailed in

Table 9. Test of the interaction effect between RDEC and REXC.

Variable	SPDP
	M11	M12	M13	M14	M15
EOT	0.039 ** (0.698)	0.031 ** (0.598)	0.059 (1.181)	0.050 (1.017)	0.043 ** (0.905)
ES	−0.173 (−3.090)	−0.114 (−2.181)	−0.054 (−1.038)	−0.044 (−0.864)	−0.054 (−1.090)
YE	0.137 (2.442)	0.096 (1.836)	0.081 (1.592)	0.067 (1.366)	0.059 (1.223)
DORR		0.385 ** (7.313)		0.243 ** (4.471)	0.237 (4.437)
DORI			0.457 ** (8.699)	0.354 ** (6.327)	0.395 ** (7.058)
DORR × DORI					0.179 ** (3.648)
R2	0.052	0.195	0.243	0.290	0.320
Adj-R2	0.043	0.185	0.233	0.278	0.307
F	5.553 **	18.259 **	24.115 **	24.507 **	23.478 **

, the analysis proceeded as follows: (a) M11 included control variables, namely, enterprise type, scale, and age; (b) M12 and M13 added RDEC and REXC to M11, respectively; (c) M14 incorporated both abilities simultaneously, revealing positive impacts on SPDP (RDEC: β= 0.243, p < 0.01; REXC: β= 0.354, p < 0.01); and (d) M15 introduced the interaction term based on M14, showing a significant effect on performance (β = 0.179, p < 0.01), indicating a complementary enhancement in SPDP.

5. Conclusions and Implications

5.1. Conclusions and Discussion

Firstly, this study demonstrates that a DGLO significantly enhances the SPDP of SMEs, aligning with the established notion that a learning orientation fosters knowledge accumulation and technology integration and enhances innovation outcomes [49]. In contrast to prior studies, which focused solely on either digital or green strategies individually [50,51], this research introduces the concept of a DGLO within the framework of digital-enabled green value creation, thereby extending the discourse on digital sustainability. Additionally, it addresses the limitations of prior discussions on the “green innovation bubble” or “low-quality green innovation” by highlighting how a DGLO mitigates the risk of prioritizing quantity over quality in green innovation efforts, particularly through digital empowerment [52]. The study specifically targets resource-constrained SMEs lacking adequate capital, skilled staff, and technological infrastructure. Consequently, the positive impact of a DGLO is validated under stringent resource limitations, offering valuable insights akin to an academic “stress test.” This finding suggests that a DGLO is not only relevant to larger enterprises with more abundant resources but may also yield greater benefits through economies of scale, intersectoral collaborations, and technology investments. For instance, large enterprises could strategically establish digital–green knowledge platforms guided by DGLO principles to facilitate cross-sector knowledge exchange and international green research-and-development partnerships, thereby overcoming sectoral barriers and historical dependencies [53].

Secondly, a DGLO indirectly improves the SPDP of SMEs via RDEC. Some studies have emphasized the importance of performance uncertainty in exploration [54], yet RDEC positively affects SPDP with a one-year lag. RDEC fosters the accumulation of novel knowledge and technologies and drives the integration of green concepts and digital technologies into NPD, enhancing carbon reduction, energy efficiency, and resource utilization. With adaptable structures and limited path dependencies, SMEs translate exploratory R&D into green innovation practices [55]. RDEC serves as a link between a DGLO and SPDP, showing how a DGLO stimulates learning potential in exploratory endeavors and how SMEs leverage digital tools to pursue green innovation under resource constraints. In industries with low technological intensity or insufficient green demand, the conversion rate of exploratory R&D outcomes may decline, weakening the impact of a DGLO. This boundary condition highlights the role of a DGLO by providing a pathway through which resource-constrained enterprises can overcome technical and market barriers through intensive learning and knowledge integration.

Thirdly, a DGLO indirectly facilitates SPDP improvement through REXC. Studies indicate that REXC enhances green product development efficiency, resource integration, and application effects, broadening insights into sustainable product development [22,56]. Exploitative R&D strengthens green innovation by enabling enterprises to use digital technologies for data processing, prediction accuracy improvement, cross-boundary collaboration, and the alignment of R&D processes with green objectives through traceability and transparency. REXC boosts efficiency and resource utilization, mitigates environmental risks, and elevates products’ environmental value. A DGLO provides a feasible R&D pathway for SMEs with limited resources and capabilities, supporting progress in green R&D under digital conditions. The mechanism may be constrained in contexts lacking institutional support or digital infrastructure, consistent with Attanasio and Battistella’s (2025) findings on the dynamic nature of stakeholder participation in sustainable business models. The boundary conditions of a DGLO highlight that its effectiveness depends on enterprises’ internal learning mechanisms and the sustained engagement and integration of external stakeholders for authentic digital–green collaborative innovation [57].

Lastly, this study reveals that RDEC and REXC positively influence SPDP and exhibit a synergistic relationship in the digital domain. This finding contributes to empirical research on the link between organizational ambidexterity and performance and explains why SMEs are more adept at balancing exploration and exploitation, given their modest size, limited resources, and lower organizational inertia, which reduce ambidextrous conflicts. The contemporary digital landscape, marked by speed, novelty, and change, requires enterprises to enhance both exploration and exploitation simultaneously to adapt to dynamic markets and technologies [58]. The results indicate that integrating ambidextrous R&D capabilities provides SMEs with a critical opportunity to enhance their SPDP while also introducing a contingency perspective to the ongoing “complementary-substitutive” debate. For large enterprises with complex hierarchies and rigid structures, the effectiveness of a DGLO in balancing exploitation and exploration may be limited. SMEs and large enterprises, however, face similar digital challenges: rapidly evolving markets and technologies demand a balance of exploration and exploitation to maintain competitiveness, while organizations must also reconcile short-term efficiency with long-term innovation to improve their SPDP. The complementary mechanism identified here clarifies SMEs’ successful practices and offers theoretical insights for large enterprises seeking to overcome inertia, optimize resource allocation, and stimulate innovation in the digital era.

5.2. Practical Implication

Firstly, a DGLO offers a pathway allowing SMEs to overcome resource constraints. These enterprises often struggle with limited capital, talent, and technology, placing them at a disadvantage in terms of sustainable product development. By adopting a digital–green learning approach, they can harness external platforms and ecosystems to address internal shortcomings. Platforms like Alibaba’s Business Operating System and Haier’s COSMOPlat Industrial Internet Platform provide cost-effective and efficient digital–green solutions. This model of external learning and resource integration enables enterprises to rapidly acquire knowledge and experience in greening and digitalization, enhancing their SPDP despite resource limitations. Managers should proactively establish digital–green learning mechanisms and engage with external digital–green ecosystems to transcend traditional development constraints.

Secondly, in the digital–green transformation, SMEs grapple with balancing “stable improvement” and “breakthrough innovation.” A DGLO fosters the development of ambidextrous R&D capabilities, integrating exploration and exploitation through systematic learning and knowledge absorption. Utilizing digital tools and data-driven methods, enterprises can create new products and technologies while optimizing existing processes, creating a complementary innovation mechanism. Managers must cultivate ambidextrous thinking in employees through digital empowerment, such as big data analysis and intelligent manufacturing systems, to blend exploitation with exploration. This will enable organizations to achieve dynamic equilibrium with limited resources, thereby enhancing their SPDP.

Lastly, the digital era’s demands for “speed, novelty, and change” require SMEs to learn rapidly and adapt flexibly. A DGLO fosters continuous improvement in green products and processes, enabling enterprises to more precisely identify energy-saving and emissions reduction opportunities, optimize green designs, and shorten R&D cycles through big data analysis, AI predictions, cloud computing, and real-time industrial Internet monitoring. By leveraging these digital technologies, enterprises can markedly enhance their green R&D efficiency, product quality, and environmental friendliness while reducing their energy consumption and waste emissions. This holistic approach can boost market recognition throughout the product lifecycle. Consequently, SMEs strengthen compliance and environmental responsibility and simultaneously secure sustainable competitive advantages by pursuing efficient, low-carbon, and differentiated product development. This underscores that in the context of digital–green transformation, SMEs can significantly enhance SPDP and establish a long-term growth trajectory by leveraging learning and ambidextrous innovation capabilities.

5.3. Research Deficiencies and Future Prospects

This study analyzes how digital transformation impacts SMEs’ NPD performance. While it expands on digital transformation and NPD, certain limitations remain. Primarily, the sample is geographically and industrially restricted, focusing on enterprises in eastern China, which may not represent other regions or industries. Future research should be extended to different industries and regions to validate and broaden the conclusions. Additionally, this study regards SPDP as a single-dimensional concept, despite the scholarly consensus that it is a complex, systemic innovation involving various internal elements of sustainable value creation. Future studies may refine SPDP into multiple dimensions, treating it as a multidimensional construct for further investigation and adopting diverse methodological approaches to reveal its underlying mechanisms more comprehensively. Lastly, while this study examines SPDP through the lens of DGLO, the specific processes involved require further exploration via case studies.