Enterprise Openness and Open Innovation Performance: The Dual Mediation of Knowledge Management Capability and Organizational Learning

Abstract

Against the backdrop of sustainability, open innovation has been widely adopted by various businesses. Within scholarly communities, how corporate openness affects innovation performance has also become a focal topic. Nevertheless, existing literature has not carried out an in-depth exploration of the intrinsic mechanisms that link openness to open innovation performance. To illustrate how corporate openness affects open innovation performance, this research intends to incorporate knowledge management capability and organizational learning as dual mediating variables. This study utilizes several research methods, including SEM analysis and Bootstrap testing, to confirm the relevant hypotheses. The findings reveal that openness and open innovation performance are mediated by external and internal knowledge management capability. In the same vein, the relationship between the two is also mediated by explorative learning and exploitative learning. Furthermore, two dual-chain mediating paths enable openness to improve open innovation performance, namely “external knowledge management capability—explorative learning” and “internal knowledge management capability—exploitative learning”. By establishing a chain mediation mechanism of “capability-learning-performance”, this study delivers a more holistic theoretical structure for deciphering the mechanisms that shape open innovation performance, in turn propelling theoretical advancements within this domain.

1. Introduction

In the context of sustainability, which is marked by critical worldwide challenges like climate change, resource scarcity, and social inequity that transcend the capabilities of individual organizations, open innovation emerges as an indispensable driver of meaningful progress [1]. Sustainability is rooted in balancing environmental health, social equity and long-term economic resilience, and it demands breakthroughs in green technologies, circular business models and inclusive practices. These breakthroughs often require diverse expertise, shared resources and cross-stakeholder alignment, and open innovation excels in all these areas [2]. By breaking down organizational silos, open innovation enables firms to collaborate with academia, NGOs, policymakers and even competitors. Through this collaboration, they can co-develop and scale sustainable solutions [3]. Moreover, open innovation ensures sustainability solutions are not just technically viable but also socially inclusive. It engages local communities, marginalized groups and end-users in co-design processes to address on-the-ground needs. Without open innovation, the slow pace of isolated R&D, limited access to specialized knowledge and misalignment with diverse stakeholder needs would severely hinder progress toward sustainability goals. In short, open innovation turns the complexity of sustainability challenges into a collective strength, making it a critical enabler of a more resilient, equitable and low-carbon future [4]. As a result, a proliferating number of enterprises are adopting open innovation to drive knowledge sharing and resource integration, which in turn enhances their innovation capabilities. Without in-depth research on open innovation within sustainability contexts, organizations are likely to become trapped in the predicament of advancing in isolation and at a sluggish pace. Such efforts, ultimately, will struggle to address the urgency and complexity inherent in global sustainable development challenges. It is precisely for this reason that scholars are growing increasingly focused on the research field of open innovation.

Openness, a metric that measures an enterprise’s level of open innovation, was first introduced by Laursen and Salter [5]. They defined it as ‘the number of distinct external knowledge sources utilized by a firm in its innovation processes. By enhancing their openness, enterprises can more effectively acquire novel knowledge and complementary resources, better identify customer preferences and needs, and mitigate investment risks—thereby facilitating smoother innovation processes [6,7]. In the ensuing years, numerous studies have examined the impacts of openness on corporate innovation performance [8,9]. Scholarly work to date has approached the openness-innovation performance linkage from diverse methodological and theoretical standpoints, yielding mixed results that warrant further investigation [10,11]. Furthermore, despite growing evidence linking openness to innovation performance, theoretical understanding of the pathways through which openness operates remains limited.

Knowledge management capability denotes a systematic capacity to create, disseminate, utilize knowledge assets for developing core competencies and sustaining competitive advantage [12,13,14]. Prior research has demonstrated that knowledge management capability serves a pivotal function in open innovation [15,16,17,18]. Organizational learning denotes an enterprise’s ability to acquire, generate, disseminate, and synthesize knowledge while dynamically adjusting its conduct to strengthen competitive advantage [19,20]. Organizational learning also serves a critical function in facilitating open innovation [21]. Enterprises can optimize knowledge resource utilization through two distinct learning approaches—explorative and exploitative learning—thereby enhancing innovation performance [22,23].

In summary, while existing studies have independently examined the direct effects of openness and other relevant variables on enterprise innovation performance, the underlying mechanisms and mediating pathways through which these factors operate remain underexplored in the current literature. Additionally, existing studies predominantly treat knowledge management capability and organizational learning as unitary constructs, failing to deconstruct their constituent dimensions to systematically examine the specific pathways through which they influence enterprise innovation performance.

Therefore, within the open innovation paradigm, this study introduces knowledge management capability and organizational learning—two openness-related factors—as dual mediating variables, constructing a conceptual model to elucidate the mechanism through which enterprise openness influences innovation performance. Building on this framework, we operationalize knowledge management capability through its internal and external dimensions, and organizational learning through its dual modes of exploitative and explorative learning. By systematically integrating these constructs, we elucidate the interconnected nature of knowledge management and organizational learning processes. This approach clarifies the sequential mediating pathway through which enterprise openness enhances open innovation performance.

This study makes threefold theoretical contributions: First, grounded in open innovation theory, this research explores how knowledge management capability and organizational learning function as mediating factors in the association between corporate openness and innovation performance. By elucidating these mechanisms, our work offers a novel analytical framework for understanding openness-innovation dynamics and an extension of core propositions in open innovation theory. Second, this study extends the knowledge-based view by differentiating between external and internal knowledge management capabilities and examining their distinct mediating roles in translating openness into open innovation performance. This dimensional approach provides a more nuanced understanding that extends beyond the unitary treatment of knowledge management in existing literature. Third, this study distinguishes between explorative and exploitative learning, analyzing their differential roles in mediating how openness enhances open innovation performance, which bridges organizational learning theory with open innovation theory and offers a more integrated theoretical perspective.

Below is the organized structure of the subsequent parts of this research: First, by referencing previous scholarly studies, the research proposes the hypotheses. Second, the research method is elaborated, including data collection, sample, and variable measurement. Next, an empirical analysis of the theoretical model is conducted to test the aforementioned hypotheses, followed by a discussion of the analysis results, which encompasses main findings, theoretical implications, practical implications, as well as research limitations and prospects. Finally, the study’s conclusions are summarized.

2. Hypothesis Development

2.1. Openness and Open Innovation Performance

Openness was first proposed by Laursen and Salter [5] and was defined as the quantity of external knowledge sources that firms draw upon in their innovation processes. By increasing openness, enterprises can foster stronger collaborative ties with diverse external partners to enhance innovation outcomes [24]. Openness serves as a robust measure of external collaboration, capturing both the breadth and depth in organizational knowledge acquisition [5]. Consequently, openness could enhance innovation performance through two distinct dimensions: breadth and depth [8].

The impact of the breadth of openness on innovation performance stems from its ability to enrich external knowledge inflows [25]. By expanding their scope of external engagement, firms gain access to more diverse knowledge domains, which both broadens their knowledge base and increases opportunities for novel knowledge recombination—ultimately strengthening innovation capabilities [26,27]. Depth of openness represents the intensity of an organization’s external collaborative engagements, which critically influences both the quality and stability of assimilated knowledge [28]. Firms exhibiting greater depth of openness could establish sustained, resource-intensive partnerships with knowledge collaborators, mitigating opportunism in knowledge exchange while enabling more effective tacit knowledge transfer and strategic deployment of core knowledge assets—collectively fostering enhanced innovation performance [29,30].

Rauter et al. [24] demonstrated that open collaboration with universities, NGOs, and intermediaries significantly strengthens sustainable innovation performance. Consistent with this, Ahn et al. [31] established that increasing openness develops dynamic capabilities, which consequently enable organizations to integrate novel knowledge and achieve sustained performance gains. Thus, we propose:

H1: 

Openness exerts a significant positive impact on open innovation performance.

2.2. The Mediating Effect of Knowledge Management Capability

Knowledge management capability denotes a capacity to obtain, generate and deploy knowledge-based resources across functional domains, thereby facilitating the generation of novel knowledge [32]. Under the circumstance of open innovation, enterprises facilitate knowledge sharing and flows through R&D collaborations and co-creation alliances with external organizations, as well as through internal innovation activities among team members and across departments. Consequently, Knowledge management capability exerts a pivotal influence on boosting enterprises’ innovation performance [33,34]. Knowledge management capability is classified into two facets: external and internal knowledge management capability [35]. This paper investigates how knowledge management capability, from these two perspectives, acts as an intermediary in linking openness to firms’ innovation performance.

The former comprises three components: knowledge absorption, connection, and desorption. The intermediary effect it exerts on the connection between corporate openness and firms’ innovation performance functions via these three pathways [36].

Firstly, knowledge absorption capability denotes an ability to obtain, integrate, and apply information and data [37,38]. In the data era, where data is a core strategic asset driving value creation, cross-organizational collaboration, and iterative innovation, greater innovation openness lets firms absorb not only substantial external knowledge but also high-quality, diversified data resources. Unlike traditional static information, this real-time, scenario-specific external data captures emerging opportunities more accurately, directly inspiring employees’ creative thinking beyond conventional frameworks and helping generate new market-aligned ideas. Later, integrating this absorbed knowledge and data into new product development optimizes the innovation workflow, significantly reduces trial-and-error costs, and thus effectively enhances firms’ innovation performance [10,18].

Secondly, knowledge connection capability reflects an organization’s capacity to systematically integrate external knowledge resources with its internal knowledge base [39]. Greater openness facilitates (a) the acquisition of novel external knowledge, (b) the redeployment of existing knowledge assets, and (c) the identification of synergistic opportunities for knowledge recombination. This process ultimately strengthens interorganizational knowledge linkages [40].

Finally, knowledge desorption capability represents an organization’s capacity to externally commercialize internal knowledge assets for value creation [41]. While firms acquire substantial external knowledge through open innovation, they can leverage their knowledge desorption capability to strategically transfer non-core knowledge resources to collaborators (e.g., research institutions, even competitors), thereby generating innovation revenues and facilitating market expansion [42].

Kashosi et al. [43] demonstrated that openness positively influences innovation performance by enhancing knowledge absorption capabilities. Through a multi-country follow-up study of manufacturing, service and retail firms across 11 Sub-Saharan African nations, Medase and Abdul-Basit [44] established that the efficient application of diverse external knowledge channels constitutes a critical factor in enhancing innovation capabilities during open innovation endeavors. Thus, we advance the following:

H2a: 

External knowledge management capability functions as a mediator in the association between openness and open innovation performance.

Internal knowledge management capability comprises three core dimensions: knowledge creation, transformation, and innovation. The intermediary effect it exerts on the connection between corporate openness and firms’ innovation performance also functions via these three pathways [36].

Firstly, knowledge creation capability reflects a capacity to assimilate and synthesize knowledge [30]. When internal knowledge stocks are limited and lack diversity, firms struggle to generate novel insights solely from existing resources [45]. Higher openness levels thus become critical, enabling organizations to cultivate external linkages that fuel knowledge creation and enhance innovation performance [30,46].

Secondly, knowledge transformation capability denotes an organization’s capacity to internally generate knowledge or leverage employees’ accumulated expertise [47]. Enhanced openness facilitates access to domain-specific prior knowledge, enabling more efficient knowledge retention and reactivation. This process enriches the organizational knowledge base, thereby strengthening the foundation for innovation performance [18].

Finally, knowledge innovation capability represents an organization’s capacity to acquire, assimilate, and integrate knowledge for creative recombination, ultimately transforming novel insights into new products, processes, and services. This capability enhances strategic agility, enabling firms to rapidly adapt their technological innovation trajectories in response to market dynamics, thereby improving innovation performance [48].

Wu and Gao [49] empirically demonstrated that internal knowledge integration significantly enhances open innovation performance in firms drawing on dynamic capability theory. Khraishi et al. [50] established that strengthening internal knowledge creation capabilities equips firms to efficiently obtain and integrate supplier knowledge, thereby generating innovation benefits. Therefore, we propose:

H2b: 

Internal knowledge management capability functions as a mediator in the association between openness and open innovation performance.

2.3. The Mediating Effect of Organizational Learning

Organizational learning enables firms to dynamically leverage both external and internal capabilities in response to evolving business environments. This adaptive process enhances innovation performance through effective knowledge absorption and utilization [51,52,53]. March [54] theorized organizational learning as a dual-dimensional construct comprising explorative learning and exploitative learning. Openness could influence firm innovation performance through these dual organizational learning mechanisms.

(1)

The mediating effect of explorative learning

Enhanced openness creates opportunities for firms to cultivate explorative learning, which centers on novel knowledge acquisition and embodies an experimental orientation. This learning mode enables organizations to overcome path dependence and achieve innovation by providing essential knowledge assets and fostering an entrepreneurial mindset [55]. Explorative learning enhances the ability to reconfigure existing knowledge while lowering the cost of integrating diverse knowledge for organizational innovation. Through continuous knowledge absorption and systemic integration, enterprises can develop and maintain a dynamic knowledge base. By strategically applying this knowledge to optimize management practices, they ultimately enhance their innovation performance [56,57].

Kim et al. [58] verified, via a study of 254 South Korean manufacturing firms, that explorative learning acts as a mediator connecting environmental dynamism to innovation performance. Zhang et al. [59] confirmed open innovation enhances sustainable competitive advantage by strengthening explorative learning capabilities. Therefore, we propose:

H3a: 

Explorative learning functions as a mediator in the association between openness and open innovation performance.

(2)

The mediating effect of exploitative learning

Exploitative learning, characterized by its cost-efficiency, allows organizations to further develop their established knowledge base while systematically integrating newly acquired external knowledge. This dual process of knowledge refinement and application reduces operational errors, accelerates development cycles, and enhances product development efficiency [23]. Particularly in open innovation contexts, where market awareness is crucial, exploitative learning plays a pivotal role in aligning technical capabilities with market demands. By effectively identifying and deploying valuable knowledge assets across production practices, it serves as a critical mechanism for driving innovation while optimizing resource utilization [55,60]. The positive effect of open innovation on enterprises’ strategic innovation is mediated by exploitative learning, as verified by Mirza et al. [61], through their analysis of 330 Pakistani pharmaceutical companies. Therefore, we propose:

H3b: 

Exploitative learning functions as a mediator in the association between openness and open innovation performance.

2.4. The Serial Mediation of Knowledge Management Capability and Organizational Learning

Within the open innovation paradigm, external knowledge management capability and explorative learning jointly serve as critical mediating mechanisms through which openness influences enterprise innovation performance. Specifically, when enterprises acquire substantial heterogeneous knowledge from external environments by increasing openness, the external knowledge management capability of enterprises exerts a crucial function in coordinating and integrating relevant resources [62]. This capability enables enterprises to effectively screen, categorize, and store external knowledge resources. Concurrently, explorative learning facilitates the absorption and transformation of new knowledge, internalizing external knowledge into organization-specific innovative resources and capabilities.

Together, external knowledge management capability and explorative learning drive enterprises to engage in proactive R&D activities and technological exploration, thereby significantly improving innovation performance [63]. Given the foregoing analysis, we put forward:

H4a: 

External knowledge management capability and explorative learning sequentially function as serial mediators in the association between openness and open innovation performance.

Internal knowledge management capability and exploitative learning collectively constitute another critical pathway through which openness influences open innovation performance. By enhancing their openness levels, enterprises establish in-depth interactions with external entities such as suppliers, customers, and industry competitors, thereby gaining precise insights into market dynamics and industrial demands. This heightened market sensitivity drives enterprises to develop demand-oriented internal knowledge management mechanisms, encompassing the systematic creation, transformation, and application of knowledge.

In this process, exploitative learning plays a pivotal role in knowledge reconstitution. Through in-depth mining and reorganization of existing knowledge systems, enterprises can effectively integrate new knowledge generated by internal knowledge management with their current technological foundations [64]. This knowledge integration process enables enterprises to dynamically adapt to changes in market and technological environments, ultimately transforming knowledge advantages into market-competitive new products and technologies, thereby significantly enhancing innovation performance [55]. Building on the above analysis, we put forward the following:

H4b: 

Internal knowledge management capability and exploitative learning sequentially function as serial mediators in the association between openness and open innovation performance.

3. Methods

3.1. Data and Sampling

This research primarily collects the required data through questionnaires. The research subjects are targeted at technology-based enterprises, as these enterprises are more prone to partake in open innovation endeavors and do so with higher frequency. Prior to the formal survey, a pilot study was conducted with 10 enterprises to pretest the questionnaire. Based on the pretest results, certain questions were refined to enhance clarity, precision, and cultural adaptation to the Chinese context. The formal questionnaire is composed of three sections. The first section briefly introduces the main content and purpose of the survey to respondents and obtains their informed consent for participation. The second section collects basic information about the enterprises where the respondents work, such as enterprise ownership type and sales revenue, etc. The third section serves as the core part of the questionnaire, including survey items related to the variables involved in the theoretical model of this study. The measurement scales for these variables will be elaborated in the following text.

The formal questionnaire was administered through two primary methods. First, face-to-face distribution was conducted, during which respondents received a brief introduction to the survey’s purpose and key concepts before participation. A total of 600 questionnaires were distributed through this method, with 486 returned. After excluding responses with missing data or evident response bias, 429 valid questionnaires were retained for analysis. Second, the questionnaire was distributed via online channels. The electronic survey was disseminated through social networks, including friends, relatives, and colleagues, with key concepts clearly explained within the questionnaire itself. Through this method, 352 questionnaires were collected. After excluding responses with completion times under 120 s, 315 valid questionnaires were retained. Combined with the face-to-face survey results, this yielded a final dataset of 744 valid responses.

We implemented an analysis of variance to examine potential differences between the datasets collected through the two different channels. The results indicated no statistically significant intergroup differences between the two data sources. Furthermore, the exploratory factor analysis revealed that scale variables loaded on multiple distinct factors, with the unrotated first principal component explaining only 26.37% of the variance (below the 50% threshold) [65]. The results obtained indicate common method bias did not pose a significant issue for the dataset.

The sample distribution is illustrated in

Table 1. Sample distribution.

	Characteristic	Number	Percent (%)
Ownership type	State-owned	316	42.5
	Private-owned	247	33.2
	Joint Ventures	128	17.2
	Foreign-owned	53	7.1
Sales revenue (CNY)	<5,000,000	48	6.5
	5,000,000–50,000,000	229	30.8
	50,000,000–300,000,000	373	50.1
	>300,000,000	94	12.6
Workforce size	<50	78	10.5
	50–300(including)	192	25.8
	300–1000(including)	318	42.7
	>1000	156	21.0
R&D intensity	<1%	139	18.7
	1–2%(including)	217	29.2
	2–5%(including)	296	39.8
	>5%	92	12.4

.

3.2. Measurement

Drawing on well-established measurement tools from existing literature, our scales were modified slightly to fit the current research context. A seven-point Likert scale was adopted to measure the variables.

The measurement scale for openness was adapted from the work of Ahn et al. [66], consisting of four items. External knowledge management capability was measured using scales derived from Forés and Camisón [67] as well as Mudambi and Tallman [68], with a total of nine items. The measurement scale for internal knowledge management capability was adapted from the work of Huang et al. [69], which included nine items. Explorative and exploitative learning were evaluated relying on the measurement put forward by Zhao et al. [70], each comprising five items. Open innovation performance was measured using a four-item scale adapted from Huang et al. [71]. Complete scale details are provided in the Appendix A.

Additionally, this study controlled for several potential confounding variables that may influence open innovation performance, including enterprise ownership type, sales revenue, workforce size, and R&D intensity.

4. Empirical Analysis

4.1. Correlation Analysis

Table 2. Correlation analysis and descriptive statistics.

Variable	1	2	3	4	5	6
1. Openness	0.854
2. EKMC	0.602 ***	0.849
3. IKMC	0.340 **	0.370 **	0.805
4. Explorative learning	0.576 **	0.557 **	0.282 *	0.865
5. Exploitative learning	0.466 **	0.531 **	0.206 *	0.546 **	0.884
6. OIP	0.590 ***	0.615 ***	0.334 **	0.571 **	0.507 **	0.867
Mean	4.550	4.694	5.158	4.762	4.152	4.997
Standard deviation	1.300	1.592	1.312	1.469	1.609	1.630

Note: *** p < 0.01, ** p < 0.05, * p < 0.1; EKMC = External knowledge management capability, IKMC = Internal knowledge management capability, OIP = Open innovation performance, the same below; The diagonal values represent the square root of variable’s average variance extracted.

revealed there are significant correlation coefficients among the core variables, which provided preliminary support for our hypotheses. Furthermore, all inter-variable correlations were below the threshold of 0.7, suggesting no substantial multicollinearity concerns that would compromise subsequent analyses [72].

4.2. Reliability and Validity Test

As a statistical technique, confirmatory factor analysis is extensively employed across social sciences and business studies. It works by evaluating whether the data collected aligns with a pre-specified theoretical model of latent variables. This process helps researchers confirm if their measurement scales are reliable and valid for assessing the intended constructs [73,74]. Cronbach’s α was employed to evaluate the scale’s reliability.

Table 3. Reliability and validity test.

Variable	CR	AVE	Cronbach’s α	Factor Loading
Openness	0.915	0.730	0.880	0.761
				0.827
				0.911
				0.910
EKMC	0.958	0.720	0.913	0.852
				0.826
				0.899
				0.784
				0.899
				0.761
				0.913
				0.835
				0.852
IKMC	0.943	0.648	0.830	0.793
				0.724
				0.770
				0.829
				0.834
				0.862
				0.837
				0.781
				0.806
Explorative learning	0.937	0.749	0.896	0.865
				0.879
				0.909
				0.845
				0.827
Exploitative learning	0.947	0.781	0.879	0.875
				0.886
				0.912
				0.905
				0.839
OIP	0.938	0.751	0.939	0.866
				0.908
				0.862
				0.824
				0.870

shows that each variable exhibited a Cronbach’s α exceeding 0.8., indicating high internal consistency [73].

The concept of validity refers to how accurately a measurement tool assesses the particular construct or concept it is meant to measure, ensuring the results reflect the true nature of the variable being studied. Validity can be categorized into several key types [75]. Next, the validity of the scale will be tested sequentially.

Content validity refers to the degree to which a measurement tool thoroughly encompasses all relevant domains, dimensions, or constituent parts of the construct it aims to measure, and is often verified through expert judgment or theoretical alignment [75]. The measurement scale employed in this study has been extensively used and validated in prior research, ensuring strong content validity.

Convergent validity evaluates the extent to which different items or indicators designed to measure the same latent construct yield highly correlated results, confirming they capture the same underlying concept [75]. All items achieved standardized factor loadings in excess of 0.7, with composite reliability (CR) values surpassing 0.9 and AVE values exceeding 0.7. In accordance with Hair et al. [73], these findings verify that the measurement scale possesses robust convergent validity.

Discriminant validity examines the degree to which a construct is empirically distinct from other related but different constructs, ensuring its indicators do not overly correlate with indicators of unrelated constructs. As Table 2 shows, the correlation coefficients all fell below the square root of the AVE. In accordance with the criteria established by Hair et al. [73], the results verify adequate discriminant validity for the scale.

4.3. Hypothesis Test

4.3.1. Structural Equation Modeling Analysis

Structural equation modeling is an advanced statistical framework to test the relationships between latent variables and observed variables, while also assessing the overall fit of theoretical models to empirical data [76]. The analysis was conducted using Mplus 7.0. The fit indices for the model are listed below: χ²(339) = 591.94, χ²/df = 1.746, RMSEA = 0.052, CFI = 0.903, TLI = 0.912, and SRMR = 0.065. All fit indices met established thresholds, indicating an acceptable model fit.

In structural equation modeling, a path coefficient quantifies the intensity and direction of the linear association between variables connected by a “path” in the theoretical model, with positive/negative values indicating positive/negative associations and larger absolute values representing stronger effects. The results of the path coefficients are presented in Figure 1.

A significant positive association between openness and open innovation performance was identified, thus supporting H1. Furthermore, openness showed strong positive effects on external knowledge management and internal knowledge management. Regarding organizational learning, a positive association was found between openness and explorative learning, as well as with exploitative learning.

The analysis revealed significant positive effects on open innovation performance from multiple antecedents. External knowledge management capability demonstrated most pronounced influence, followed by internal knowledge management capability which showed a marginally significant positive effect. Regarding organizational learning mechanisms, both explorative learning and exploitative learning contributed positively to open innovation performance, with exploitative learning showing greater statistical significance.

The structural equation modeling results further revealed significant cross-dimensional relationships. External knowledge management capability showed a strong positive association with explorative learning, while internal knowledge management capability demonstrated a significant, though weaker, relationship with exploitative learning. These path coefficients provide preliminary evidence supporting the proposed mediation effects in our conceptual model.

4.3.2. Mediating Effect Test

Then, we conducted bootstrap analyses (N = 5000 samples) to further examine the mediating effects. The bootstrapping results, presented in

Table 4. Bootstrap test of mediating effect.

	Path	Estimate	p-Value	95% CI
				Lower	Upper
Direct effect	Openness→OIP	0.124	0.058	0.049	0.162
Indirect effect	Openness→EKMC→OIP	0.526	0.000	0.443	0.609
	Openness→IKMC→OIP	0.053	0.079	0.026	0.061
	Openness→Explorative learning→OIP	0.033	0.086	0.012	0.067
	Openness→Exploitative learning→OIP	0.087	0.000	0.047	0.127
	Openness→EKMC→Explorative learning→OIP	0.067	0.077	0.005	0.096
	Openness→IKMC→Exploitative learning→OIP	0.007	0.028	0.001	0.012
Total indirect effect	-	0.773	0.000	0.65	0.794
Total effect	-	0.897	0.000	0.738	0.915

Note: CI, confidence interval.

, confirm statistically significant mediation effects for the hypothesized pathways.

The bootstrapping test revealed that external knowledge management capability transmits a significant indirect effect of openness on open innovation performance (β = 0.526, 95% CI [0.443, 0.609]). The finding verifies external knowledge management capacity plays a notable mediating role in this association, thus validating H2a.

Bootstrap analysis revealed a notable indirect impact of openness on open innovation performance via internal knowledge management capability (β = 0.053, 95% CI [0.026, 0.061]). This finding demonstrates internal knowledge management capability acts as a statistically significant mediator in this association, thus supporting H2b.

The bootstrap analysis also revealed two significant mediation pathways: (1) Explorative learning significantly mediated the association between openness and open innovation performance (β = 0.033, 95% CI [0.012, 0.067]); (2) Exploitative learning similarly demonstrated a significant mediating effect (β = 0.087, 95% CI [0.047, 0.127]). Both findings support our hypothesized mediation effects, confirming H3a and H3b.

Moreover, bootstrap-based analysis uncovered a notable sequential mediation pathway that operates through external knowledge management capability and explorative learning (β = 0.067, 95% CI [0.005, 0.096]). This demonstrates that openness enhances open innovation performance first by developing external knowledge management capability, which in turn facilitates explorative learning. These results provide support for H4a regarding the continuous mediating effect of this dual-path mechanism. The bootstrap analysis identified a significant but relatively small sequential mediation effect (β = 0.007, 95% CI [0.001, 0.012]), While modest in magnitude, this chained mediation pathway was statistically significant, providing support for H4b.

4.4. Robustness Test

To validate the robustness of our structural equation modeling and bootstrapping results, we conducted complementary hierarchical regression analyses. Hierarchical regression analysis is a statistical method where predictor variables are entered into a regression model in pre-specified, sequential blocks, allowing researchers to isolate and test the incremental contribution of each block to explaining variance in the dependent variable [77].

As presented in

Table 5. Regression analysis on direct effects.

Variables	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
Ownership type	−0.195 ***	−0.059 ***	−0.073 ***	−0.158 ***	−0.060 **	−0.071 ***
Sales revenue	0.035	0.103	0.065 ***	0.014	0.022	0.024
Workforce size	0.265 ***	0.191	0.171 ***	0.231 ***	0.142 ***	0.217 ***
R&D intensity	0.127 ***	0.033	0.015	0.082 ***	0.013	0.036
Openness		0.663 ***
EKMC			0.762 ***
IKMC				0.290 ***
Explorative learning					0.637 ***
Exploitative learning						0.578 ***
R2	0.104	0.503	0.649	0.184	0.469	0.413
F	30.024 ***	209.973 ***	382.378 ***	46.725 ***	182.691 ***	145.994 ***

Note: *** p < 0.01, ** p < 0.05.

and

Table 6. Regression analysis on indirect effects.

Variables	Model 7	Model 8	Model 9	Model 10	Model 11	Model 12
Ownership type	−0.058 ***	−0.053 **	−0.022	−0.013	−0.037 *	−0.008
Sales revenue	0.077 ***	0.093 ***	0.073 ***	0.080 ***	0.064 ***	0.071 ***
Workforce size	0.168 ***	0.184 ***	0.142 ***	0.179 ***	0.142 ***	0.172 ***
R&D intensity	0.010	0.022	−0.005	−0.002	−0.009	−0.011
Openness	0.168 ***	0.632 ***	0.447 ***	0.505 ***	0.121 ***	0.479 ***
EKMC	0.635 ***				0.532 ***
IKMC		0.097 ***				0.087 ***
Explorative learning			0.381 ***		0.225 ***
Exploitative learning				0.363 ***		0.360 ***
R2	0.659	0.511	0.591	0.603	0.686	0.609
F	333.610 ***	180.612 ***	249.581 ***	261.593 ***	322.299 ***	230.153 ***

Note: *** p < 0.01, ** p < 0.05, * p < 0.1.

, the regression models specified open innovation performance as the dependent variable. First, control variables such as enterprise ownership type, sales revenue, workforce size, and R&D intensity were incorporated to form Model 1. Second, independent variables including openness etc., were introduced sequentially to develop Models 2–6. Then, based on Model 2, external and internal knowledge management capability, explorative learning, and exploitative learning were successively added to construct Models 7–10. Finally, Models 11 and 12 were built.

After hierarchical regression analyses, it was found that the regression results consistently supported our initial findings from both the SEM analysis and bootstrapping tests. This convergence of evidence across different analytical approaches strengthens confidence in the robustness of our study’s conclusions.

5. Discussion

5.1. Main Findings and Theoretical Implications

This study reveals three key findings:

First, the findings of this study highlight that enterprise openness does not directly translate into improved open innovation performance; instead, it relies on knowledge management capabilities as a critical mediating bridge. Specifically, external knowledge management capability, together with internal knowledge management capability, functions as a statistically significant mediator in the association between openness and open innovation performance. These two types of capabilities collectively form a vital transmission channel, enabling enterprises to effectively leverage the opportunities brought by openness and convert scattered internal and external knowledge resources into tangible improvements in open innovation performance.

Prior research has established the positive association between knowledge management capability and innovation performance [17]. Yu et al. [18] further differentiated knowledge management capability into internal and external dimensions, demonstrating their distinct impacts on innovation outcomes. This study advances the literature in two key ways: First, by conceptualizing knowledge management capability as a dual-dimensional mediator (internal and external) between openness and open innovation performance, we address a critical gap in existing research that has predominantly examined direct effects. Second, our findings theoretically enrich both knowledge management and open innovation literature by elucidating the underlying mechanisms through which openness translates into innovation performance.

Second, this study confirms a key point. Explorative learning and exploitative learning are two distinct yet complementary organizational learning approaches. Both function as significant mediating factors in the association between enterprise openness and open innovation performance. Specifically, explorative learning focuses on exploring new knowledge domains, testing novel ideas, and adapting to emerging market needs. Exploitative learning emphasizes refining existing knowledge, optimizing established processes, and enhancing the efficiency of current practices. Each of these two learning approaches plays a pivotal role. It helps translate the potential of enterprise openness into tangible improvements in open innovation performance.

Chan et al. [78] demonstrated organizational learning exerts a direct influence on organizational performance. This study extends prior research by incorporating the open innovation context into the organizational learning framework. Specifically, it examines how organizational learning mediates the link between openness and the performance of open innovation. By doing so, this research not only deepens but also expands existing organizational learning theories and their applications in innovation research.

Third, the study further identifies that enterprise openness enhances open innovation performance through two distinct, non-overlapping chain mediation pathways. The first pathway operates via external knowledge management capability, which enables firms to capture and integrate valuable external knowledge. This acquired knowledge then fuels explorative learning, where enterprises experiment with new ideas and expand into uncharted knowledge domains, ultimately driving improvements in open innovation performance. The second pathway relies on internal knowledge management capability, which helps organize and activate a firm’s internal knowledge reserves. This internal knowledge supports exploitative learning, where enterprises refine existing processes and optimize current practices, to boost open innovation performance. These two pathways are complementary: while the first leverages external resources to explore new innovation directions, the second taps into internal strengths to deepen and optimize existing innovation efforts, together forming a comprehensive mechanism through which openness translates into better innovation outcomes.

Zhang et al. [1] developed a research model in which ambidextrous organizational learning mediates open innovation and sustainable competitive advantage. This study systematically integrates knowledge management capability with organizational learning, establishing their interconnected relationship. Through rigorous examination, it investigates the mechanisms by which openness enhances enterprise innovation performance via knowledge management capability and organizational learning pathways. By translating macro-level strategic concepts into actionable internal processes, this research makes significant theoretical and empirical contributions to understanding how openness influences open innovation performance.

5.2. Practical Implications

This study not only contributes significant theoretical value, but also provides practical managerial guidance for firms.

First, the empirical analysis results indicate openness can enhance enterprises’ open innovation performance. Therefore, enterprises should systematically improve their organizational openness, particularly in the context of sustainable development. Specific implementation strategies include: (1) Build a multi-tiered external interaction network: Co-led by R&D and marketing departments, enterprises should attend at least one provincial-level or higher university-industry-research alliance meeting quarterly and join 1–2 vertical industry technology forums annually. They should also select 3–5 core upstream and downstream enterprises to establish an “innovation resource sharing pool”, clarify technology patent and talent sharing scopes, and improve resource allocation efficiency directly [25]. (2) Establish a standardized information sharing mechanism: Coordinated by brand and R&D departments, enterprises should release a corporate innovation white paper semi-annually and hold an open technical seminar quarterly. An “innovation collaboration section” should be set up on the official website to update cooperation needs and achievements in real time, enhancing stakeholder trust and collaboration intention conversion [23]. (3) Cultivate an open and inclusive organizational culture: Led by HR departments, enterprises should develop an “external knowledge transformation incentive system”, hold monthly cross-department/enterprise learning exchanges, and require R&D personnel to participate in at least two external knowledge learning activities and submit one transformation application plan yearly to boost the organization’s dynamic innovation capabilities directly.

Second, research findings show that internal and external knowledge management capabilities play significant mediating roles in open innovation, so enterprises need to build a comprehensive knowledge management system to systematically develop and strengthen both capabilities. For external knowledge management capability, enterprises should assign a cross-functional team to manage external sources: this team participates in 2–3 annual industry-university-research alliance workshops, conducts 1–2 quarterly targeted technology acquisition assessments on complementary-tech startups, and issues weekly market intelligence reports covering core competitors to capture cutting-edge knowledge [44]. Additionally, enterprises should set up a dedicated knowledge review committee to define criteria, conduct monthly reviews, and only integrate knowledge meeting at least 70% of criteria to ensure quality and applicability [46].

For internal knowledge management capability, enterprises should designate the human resources department and IT department as joint leads to drive implementation with clear quarterly milestones: the IT department should develop a corporate knowledge repository and an expert directory by the end of the first quarter to institutionalize knowledge management infrastructure [29]; the HR department should design knowledge contribution metrics—requiring each employee to upload at least 2 practical knowledge documents quarterly—and integrate these metrics into 20% of annual performance evaluation scores to enforce consistent knowledge sharing; additionally, the HR department should launch a monthly “Knowledge Sharing Reward Program” and establish cross-departmental learning communities that hold biweekly online meetings to motivate employee participation in knowledge creation and diffusion [50].

Third, enterprises must establish a practical ambidextrous learning mechanism to balance explorative and exploitative learning [79,80]. On one hand, enterprises should launch specific innovation incentive programs, such as setting up cross-departmental “innovation micro-teams” with clear project cycles, allocating a portion of the annual R&D budget as “failure-tolerant funds” and organizing regular “technology scanning workshops” with external experts to guide teams in exploring emerging technologies relevant to core business.

On the other hand, enterprises need to focus on digitizing and operationalizing existing knowledge. This can be achieved by building a company-wide searchable “knowledge database”—where each department is required to update practical case studies monthly. Meanwhile, enterprises must mandate quarterly “skill-transfer sessions”, led by top performers, to teach colleagues proven methods. To dynamically balance the two learning approaches, enterprises should adopt a quarterly adjustment framework: every three months, the strategy team is required to review the proportion of revenue from new products and the operational cost reduction rate, and then adjust resource allocation and activity frequency accordingly to ensure both learning approaches play an effective role [61].

5.3. Research Limitations and Prospects

This study systematically investigates the underlying mechanisms through which openness affects open innovation performance, yielding important theoretical insights while acknowledging several limitations that warrant further exploration.

First, regarding variable selection, this research primarily focuses on core constructs—openness, knowledge management capability, and organizational learning—while overlooking potential boundary conditions. For instance, internal governance mechanisms (e.g., internal control efficiency) and external competitive environments (e.g., market dynamism, industry rivalry) may serve as critical moderators in these relationships. Under varying boundary conditions, different categories of knowledge management capabilities and organizational learning might demonstrate divergent effects on open innovation performance—a topic that also deserves further investigation.

To address this limitation, future studies should systematically identify key boundary variables (via literature reviews and enterprise interviews)—including the aforementioned internal governance, external competition, as well as firm size, technology intensity, and policy support level, and then select diverse enterprise samples through stratified sampling, integrate these variables into empirical models as moderators for testing, and analyze their specific impact direction. Ultimately, this clarifies the core mechanism’s application scope across contexts, making conclusions more targeted and practically relevant for enterprises.

Second, methodologically, this study utilizes cross-sectional survey data, which involves collecting information on all research variables at one time point. A key limitation of this data type is its failure to capture how variables evolve over time or the dynamic, time-dependent interactions between them—for instance, it cannot track how changes in enterprise openness may sequentially influence knowledge management capabilities and organizational learning, nor how these shifts collectively shape open innovation performance over months or years. Consequently, this constraint prevents the research from unraveling the dynamic evolutionary process among openness, knowledge management capabilities, organizational learning, etc. It also weakens the study’s ability to verify the potential causal sequence between variables, as cross-sectional data only reflects correlations at a specific moment rather than establishing the direction or timing of causal relationships.

To address this limitation, future research could adopt longitudinal tracking data, which involves collecting data from the same sample of enterprises at multiple time points over an extended period. For example, longitudinal data could clarify whether improved knowledge management capabilities precede enhancements in open innovation performance, or if changes in organizational learning exert a mediating function over different stages. By doing so, future studies would provide more robust, time-sensitive insights into the mechanism linking openness to open innovation performance, thereby addressing the static limitation of the current cross-sectional design.

In addition, this study also has the following limitations that merit consideration. First, it exhibits contextual boundedness: the findings derived from Chinese enterprises may lack generalizability to other institutional environments or geographic contexts. Second, although firm size and other covariates were incorporated into the model as control variables, the potential moderating role of these firm-specific heterogeneous characteristics in shaping the relationships between the core variables was not empirically examined. Third, the current study primarily focuses on investigating linear associations among variables. In terms of future research, exploring potential non-linear relationships between variables could provide more nuanced insights into the underlying mechanisms.

6. Conclusions

Overall, the research conclusions of this paper can be summarized into the following three main points. First, both external and internal knowledge management capabilities significantly mediate the relationship between enterprise openness and open innovation performance. Second, explorative and exploitative learning play a significant mediating role in the association between openness and open innovation performance. Third, openness enhances open innovation performance through two distinct chain mediation pathways: (1) external knowledge management capability and explorative learning; (2) internal knowledge management capability and exploitative learning.

Clarifying the mechanism through which openness influences open innovation performance holds profound significance for sustainability and sustainable innovation. For sustainable innovation, which often demands cross-boundary collaboration (e.g., with green tech startups, research institutions, or local communities) and the integration of diverse knowledge (such as circular economy principles or low-carbon technology insights). Understanding how openness drives open innovation performance eliminates the “black box” between collaborative efforts and tangible sustainable outcomes. Specifically, this clarity helps firms identify which links in the mechanism—whether leveraging external knowledge management to absorb sustainable best practices, using explorative learning to develop breakthrough eco-friendly solutions, or relying on exploitative learning to scale existing green processes—are critical for translating openness into sustainable innovation. Without this understanding, firms may engage in unfocused open initiatives that waste resources and fail to advance sustainability goals. By contrast, a defined mechanism empowers firms to design targeted strategies. For example, prioritizing the “external knowledge management–explorative learning” pathway to adopt cutting-edge renewable energy technologies, or strengthening the “internal knowledge management–exploitative learning” pathway to optimize existing manufacturing processes for lower carbon emissions. In turn, this boosts the efficiency and influence of firms’ sustainable innovation efforts, enabling them to advance toward a greener and more prosperous future.