Introduction

The complex product systems (CoPS) is a high-cost, engineering and information technology-intensive, and custom product with a large number of subsystems and components (Hansen and Rush, 1998). Its complexity is manifested in the number of custom components, the breadth of knowledge and skilled required, the degree to which production involves new knowledge, and other key product dimensions (Hobday, 1998). The CoPS is significantly different from mass-manufactured products in that it is more customized, has less market demand, and involves multi-disciplinary fields. It is the high-tech means of production supporting production and services, and constitutes the “technological asset pillar” of modern economy. For instance, high-speed railway is a very large and complex engineering system generally customized by the government, involving railway construction, train manufacturing, information collection, regulation and control, and other technologies. It has tens of thousands of components, and radiates thousands of supporting firms. The development of the CoPS is accompanied by a large number of innovations, which further gradually form industry standards or play an important role in the formulation of industry technical standards, such as the development of 5 G networks and the development of C919 aircrafts. Therefore, the CoPS is often shared by a multi-functional innovation team across organizations. The innovation team, which is a typical complex innovation network organization, integrates the main business functions of multiple organizations and needs to collaborate in the innovation of the CoPS (Hobday et al., 2000). At the same time, the innovation process is highly dependent on partners, each of whom plays an important role, which may be constantly changing with the development of artificial intelligence (Gunawan et al., 2002; Yu et al., 2024). However, from the perspective of structural embeddedness, the positions of partners in the organizational structure and the resources obtained by innovation subjects are different (Granovetter, 1992). They will affect the innovative development of innovation subjects. Therefore, according to the resource endowment of innovation subjects, how to choose right partners and form unique organization patterns to achieve their own innovation strategy is the goal of this work.

In the previous literature on the innovation of the CoPS, some on organization patterns focus on organizational forms of the whole innovation team, such as multi-party alliances and hierarchical structures, emphasizing the collaboration between CoPS leaders and their partners. They lack the discussion on how other innovation subjects choose organization patterns. Some focus on organizational relationships between two or three members of the team, such as co-developers, complementors, and competitors in the supply chain or value chain, emphasizing the resource integration of some stakeholders by innovation subjects. They lack the discussion on the interaction of other members and the induction of organization patterns. There are different types of team members in the innovation process of the CoPS, and each team member has unique resources (Hong and Su, 2008). An appropriate organization pattern enhances the flexibility of connections among team members in interface management, integration management, and knowledge management, and improves the sensitivity and response capability of the environment, thus creating and utilizing different innovation opportunities (Pan and Si, 2009). Besides, it also helps innovation subjects obtain specialized benefits from direct partners and gain access to advanced technology of the CoPS (Gholz et al., 2018; Moody and Dodgson, 2006).

In the innovation process of the CoPS, the choice of organization patterns becomes the decision of whether innovation subjects can promote innovative development. In general, a successful organization pattern needs to satisfy the core competitiveness of innovation subjects, the partners that can provide complementary advantages, and the common strategic goal (Bateman and Snell, 2013). In the existing literature, factors related to core competencies of innovation subjects, such as integration capability (ICP) and modularity capability (MCP), as well as inter-organizational relationships, such as internal cooperation (ICO), external coordination (ECO), and technological dependence (TDP), have been explored (Rajala et al., 2019; Zhou et al., 2020; Hou et al., 2024; Ethiraj and Posen, 2013). However, there are two limitations. One is that it focuses on the “net effect” of a single factor, while ignoring the “joint effect” among multiple factors. The choice of organization patterns is a complex causal problem that is affected by many factors such as core competencies and inter-organizational relationships. The “net effect” of a single factor cannot be explained clearly by traditional regression analysis and structural equation analysis (Fiss, 2011). The other fails to identify the “interactive relationship” among different factors. That is, the interaction of multiple factors may produce substitutive effect or complementary effect. It is also something that traditional regression analysis and structural equation analysis cannot solve. The two aspects have not been explored in existing studies. fsQCA approach is considered to be an effective way to explore “joint effect” and “interactive relationship” (Ragin, 2008). It can analyze the substitution and the complementarity of multiple factors by exploring multiple causal relationships (Zhang et al., 2019). By explaining the “joint effect” of various factors and the “interactive relationship” among various factors, fsQCA can not only fully understand the complex nature that innovation subjects realize innovative development, but also provide solutions for them to choose complementary or substitutive conditions.

Therefore, this work attempts to address the following questions. How many types of organization patterns there be? Which organization pattern should be chosen for different types of innovation subjects? Which factors are substitutive or complementary in organization patterns? In order to explore these issues in depth, we take 184 firms participating in the Internet of Things (IoT) product system, which is a type of the CoPS, as a sample. Through applying fsQCA approach, we explore the “joint effect” of five factors on innovation performance (IPF), including ICP, MCP, ICO, ECO, and TDP, as well as the “interactive relationship” among the five factors, so as to choose appropriate organization patterns for innovation subjects.

The remainder of the paper is structured as follows: In the next paragraph, we present the theoretical background. Then, remaining sections explain the method used, tests, and results. Finally, we come up with organization patterns, conduct case analysis, and put forward the conclusion and implications.

Theoretical backgrounds

The choice of organization patterns of innovation subjects is affected by factors from different dimensions. According to the dynamic capability theory, the competitive advantage of a firm comes from the combination of capabilities and resources that can be developed, deployed, and protected (Teece et al., 1997). Due to the high uncertainty, high risk, and ambiguity of client demands in the CoPS, the innovation of the CoPS may last for a long time (Hobday, 1998). It requires that capabilities of innovation subjects need to be constantly updated to achieve consistency with the changing environment (Teece et al., 1997). Ravishankar and Pan (2013) found that the development of dynamic capabilities was closely related to core competencies of the innovation team of the CoPS, supplemented by effective resource sharing and integration. Hou et al. (2024) found that the development of dynamic capabilities was based on core competencies of innovation subjects of the CoPS, supplemented by effective integration of external resources. According to the characteristics of internal resources, the selective utilization and integration of external resources is the internal embodiment of dynamic capabilities of innovation subjects (Xu, 2019). Inter-organizational relationships in the innovation team are complex and rich intangible resources. Therefore, this work follows the explanatory framework of the dynamic capability theory of Ravishankar and Pan (2013) and Hou et al. (2024), including core competencies of innovation subjects and their inter-organizational relationships. This section discusses core competencies and inter-organizational relationships.

Core competencies

Module development and system integration are two important stages in the innovation of the CoPS (Hong and Su, 2008). The innovation team includes many integrators and subcontractors. Therefore, core competencies of innovation subjects can be divided into ICP and MCP (Ravishankar and Pan, 2013). ICP is the ability to integrate information, technology, and other resources. MCP is the ability to independently develop and produce modules. Both ICP and MCP affect IPF that additional benefits to the project is brought through the application of new knowledge and technology on top of the original technology of the CoPS (Hou et al., 2024). Strengthening ICP and MCP consolidates team positions of innovation subjects and improves their market sensitivity.

Unlike economies of scale, ICP of innovation subjects in the innovation process of the CoPS requires special management skills and strategies that can focus more on bidding, design, and development (Hobday, 1998). It needs at least three basic capabilities, such as understanding client needs, internal cross-functional knowledge coordination, and the integration of external knowledge resources, and further determines choices of market positions, partners, and outsourcing of innovation subjects (Nightingale, 2000; Pan and Si, 2009). With the formation of the supply chain, ICP of innovation subjects can determine the use scope of high-volume components of the CoPS by changing design architecture, and gain innovation capability by integrating these components into customized functions (Hobday, 1998; Crespin-Mazet et al., 2019). It can also obtain horizontal technology spillovers from the supply chain (Gholz et al., 2018). Kone Corporation has enabled the effective integration of modules through innovative combinations such as the organizational collaboration of supply chain redesign and software interfaces (Rajala et al., 2019). Besides, the completed projects also enable innovation subjects to acquire the organizational skills to manage such innovation networks (Naghizadeh et al., 2017).

In the innovation process of the CoPS, MCP of innovation subjects is a strategic capability based on organizational resources. It facilitates the dynamic reconfiguration, reorganization, and deployment of resources, as well as the prediction and rapid response to client needs, greatly enhancing the resource base of innovation subjects and maintaining their competitive advantages (Ravishankar and Pan, 2013). Innovation subjects can autonomously shift initial innovation requirements from one type of modules to another with a more competitive advantage (Crespin-Mazet et al., 2019). However, the uncertainty of MCP in the innovation process also depends on the interdependencies among components of the CoPS (Ethiraj and Posen, 2013). Therefore, modular design needs to be complemented by practical and compatible design elements that emphasize system integration, and builds the value proposition of MCP based on a market-driven vision, thus enabling innovation subjects to justify their new extended solution concepts (Tee et al., 2019; Rajala et al., 2019). Based on it, innovation subjects use the new domain knowledge to transform the exiting technology or product development into the corresponding modules of the target CoPS according to the overall view and the needs of module functions (Hong and Su, 2008; Holmqvist and Persson, 2003).

Inter-organizational relationships

Network organization is actually a collaborative relationship network among many firms, clients, institutions, etc. It reflects the technical professionalism of each functional structure, the reaction of the market to the product structure, and the balance and flexibility of the organizational structure (Bateman and Snell, 2013). The innovation team of the CoPS is a typical complex innovation network organization. From the perspective of innovation network, the organizational structure of the innovation network of the CoPS can be divided into core network and peripheral network. Core network members are innovation subjects, responsible for the bidding, design, and development of products, and embedded in the relationship network. Peripheral network members include the government, clients, scientific research institutions, responsible for providing technical support and supervision to core network members and assisting them to achieve innovation. They are scattered around core network members. Based on the embeddedness theory, innovation subjects need to find appropriate network embeddedness in order to improve economic efficiency and competitiveness (Uzzi, 1996). Watts (1999) proposed that network embeddedness was based on the mutual dependence of partners in each other’s resources and the structural path dependence formed through the relationship history. In this work, interdependences are embodied in ICO and ECO, and path dependence is embodied in TDP. ICO refers to non-technical cooperation among core network members, such as strategic planning and benefic allocation. ECO refers to the consensus reached between core network members and peripheral network members on technical requirements, such as client requirements for product performance and the introduction and absorption of intellectual capital. TDP refers to the fact that core network members have to be dominated by foreign products or technologies when domestic quality requirements are lacking or not met. Moreover, in terms of the spatial agglomeration of economic activities, ICO and ECO are local network embeddedness (home country), while TDP is hyperlocal network embeddedness (foreign country) (Bathelt et al., 2004). ICO and ECO reduce the adverse impact of technological specialization and innovation risks of innovation subjects, and make ICP and MCP reinforce each other, so as to carry out innovation activities (Tee et al., 2019; Hou et al., 2024). TDP constrains the choice set of innovation efforts of innovation subjects and influences their dynamic capabilities and innovation activities (Ethiraj and Posen, 2013).

The intensification of social division of labor leads to the further deepening of technological specialization of innovation subjects. When various innovation subjects are connected together to develop CoPS, the network created is clearly cooperative and allows information to flow among them (Snow, 2015). The flow of information enables innovation subjects to have a unified goal for solutions, not only reducing the information asymmetry among them, but also enhancing the market channel and supply (Gholz et al., 2018; Rajala et al., 2019). In addition, ICO strategically effectively alleviates downstream competition, promotes participants to share risks and profits, reduces technical risks, and provides long-term stability (Ding et al., 2019; Zhou et al., 2020). Zhou et al. (2020) found that weak dominant manufacturers cooperated with strong suppliers through revenue-sharing contracts based on relationship-specific investments, and ultimately improved their own technical capabilities. Therefore, ICO can yield good results in strategic planning, benefic sharing, and risk taking. When the cooperation relationship achieves a good match, dynamic capabilities of different innovation subjects will be supplemented and strengthened accordingly, and their IPF will also be realized (Hobday et al., 2000).

With the rapid development of technology, the strategic advantages of innovation subjects increasingly need the integration of external activities and technologies (Teece et al., 1997; Franca et al., 2022). ECO is an important indicator for innovation subjects to improve dynamic capabilities and IPF (Hou et al., 2024). At the bidding stage of the CoPS, innovation subjects need to listen to the experience of clients and eliminate the mismatch between their technical knowledge and ideas of clients, so that their dynamic capabilities resonate with innovation needs of clients and prevent design failures (Hansen and Rush, 1998; Crespin-Mazet et al., 2019; Nightingale, 2000; Rozesara et al., 2023). It continues long after the product is delivered (Rajala et al., 2019). The technical supervision of the government is also very important, regulating technological boundaries of innovation subjects and leading their technological directions. In particular, when the government has the dual role of supervisor and client, the government has a significant influence on innovation subjects. When Jiang and Lyu (2019) studied the technological catch-up of high-speed railway industry, they found that there were a series of processes of co-evolution between the government and firms, which superimposed industrial levels and formed an advanced system integration structure, thus realizing technological catch-up. In addition, universities, scientific research institutions, and expert teams play an important role in providing innovation subjects with some basic or related technology support when they lack core competencies or have technical disagreements with the government/clients (Hong and Su, 2008; Naghizadeh et al., 2017).

The further expansion of the global supply chain encourages innovation subjects to devote themselves to the business they are good at, and other businesses are carried out in the form of procurement, outsourcing, and processing on behalf of others, forming a path dependence over the years. TDP is one of them in essence, and reflects the technological weakness of local innovation subjects. Although it not only reduces the research and development costs, but also assumes huge risks, especially in the face of foreign unilateral sanctions, such as technology blockade and trade sanctions (Hong and Su, 2008; Naghizadeh et al., 2017). TDP may lead to the delay of hardware or software integration, and cause the phenomenon of the monopoly of foreign firms or downstream competition (Hansen and Rush, 1998; Zhou et al., 2020). When certain events occur, or in order to prepare for a rainy day, it becomes an opportunity to move away from TDP. The implementation of the strategy of local innovation subjects to get rid of TDP can reduce the negative impact of the change of CoPS architectures, significantly improve their own IPF, and gradually find out innovation patterns that conform to their own development path (Ethiraj and Posen, 2013; Hong and Su, 2008). The bankruptcy of a major UK subcontractor provided an opportunity for Australia to make greater use of developed satellite technology (Moody and Dodgson, 2006).

In conclusion, the desire to achieve the innovative development of innovation subjects through developing dynamic capabilities is the result of many factors. The core competencies can be divided into ICP and MCP, due to the types of innovation subjects, and the inter-organizational relationships can be divided into ICO, ECO, and TDP, due to the network embeddedness. However, it is an open issue how many organization patterns these five factors can form to achieve innovative development. Therefore, fsQCA approach is applied in this work to explore the joint effects of the five factors on the choice of organization patterns, and to reveal possible interactive relationships among these factors. The frame diagram of the work is shown in Fig. 1.

Fig. 1
figure 1

The frame diagram.

Full size image

Research method

Research design

fsQCA, which explores multiple causality as a whole by examining sufficient and necessary subset relationships between prerequisites and outcomes, is a configuration analysis approach based on fuzzy sets (Rihoux and Ragin, 2009). From the perspective of system, the configuration theory holds that there are complex causal and nonlinear relationships among elements in a system, that is, the path from a certain starting point to a certain end point of a system is not unique, and some antecedent variables can be substituted with each other (Katz and Kahn, 1978; Fiss, 2007). Compared with traditional analysis, fsQCA can find configuration relationships among multiple factors; compared with other analytical approaches for testing configuration relationships, it can effectively identify the interdependence, configuration equivalence, and causal asymmetry among conditions; compared with other types of QCA approaches, it is adept at handling varying degrees of condition membership. (Rihoux and Ragin, 2009; Zhang et al., 2019; Ruan et al., 2024; Huang et al., 2024; Malik et al., 2024). Besides, fsQCA has the dual advantage of quantitative analysis and qualitative analysis. It can not only analyze large sample cases, but also carry out systematic analysis of conditional configuration of a certain case to make up for the shortcomings of universality (Du and Jia, 2017; Song and Lu, 2017). The choice of organization patterns is a complex causal problem affected by many factors, and team members have different positions and roles in complex innovation networks, so the path to achieve the innovation of the CoPS is often not the only one. Therefore, we adopt fsQCA approach to take relationships among five key factors, containing ICP, MCP, ICO, ECO, and TDP, as configurations of the realization of IPF, and determine causal relationships among them from the perspective of configuration, so as to propose various organization patterns for realizing the innovative development of innovation subjects. fsQCA analysis is mainly divided into three steps. Namely, data calibration, necessity analysis, and sufficiency analysis.

Sample and data

With the progress of science and technology, the new generation of information technology industry has become an important part of the fourth industrial revolution and has been listed as one of key national support objects in China. As one of the typical representatives of the new generation of information technology industry, the IoT industry is developing very rapidly and has developed to various fields, such as aquaculture and logistics (Chen et al., 2023; Lu et al., 2024; Shen et al., 2024). The IoT product system, which is a type of the CoPS, is a custom engineering system composed of a large number of hardware and software, such as information perception and communication network. Through deep and non-linear interaction among a wide range of cross-domain technologies, it can carry out data visualization, automatic control, and even efficiency improvement to realize ubiquitous connection between people and things, and also requires a lot of innovation. For instance, the IoT product system for transportation integrates radio frequency, car networking, and other technologies. It provides traffic information services and travel planning for citizens to achieve smart transportation. The road planning, pavement structure, and vehicle condition information of each city are of difference, and the IoT product system for transportation needs to be customized and researched, taking into account local circumstances. Therefore, according to random sampling, we developed a sample of 184 firms that were involved in the IoT product system (for the details of the survey, see Supplemental material). The total sample includes a broad spectrum of the IoT industry, including software and application developers, telecommunication equipment manufacturers, terminal equipment manufacturers, operation and service providers, network operators, and system integrators.

Calibration

Central to fsQCA approach is calibration procedure and truth table analysis. Calibration procedure is a transformation process that converts conventional measures into fuzzy sets (Ali et al., 2016). Fuzzy sets can be regarded as continuous variables calibrated to denote membership in a well-defined set (Rihoux and Ragin, 2009). We adopt the direct calibration method of Ragin (2008). There are three substantive meaningful thresholds for the calibration process: full membership, the cross-over point, and full non-membership, which correspond to membership scores of 95%, 50%, and 5%, respectively. The cross-over point refers to the point of maximum ambiguity in the assessment of whether a case is more in or out of a set (Ragin, 2008). Besides, the procedure is based on data values of the sample and the knowledge about the context (Ragin, 2008). Given Chinese culture and language expression habits, the sample means of the constructs are large and scattered (in Supplemental material). Therefore, conditions and outcomes are calibrated to fuzzy sets using the procedure adopted by Chu et al. (2019). Before the calibration procedure, the Likert scale is converted into a fuzzy set by calculating the mean of each construct. Then, according to the 7-point Likert scale, full membership, the cross-over point, and full non-membership are set as 7, the mean value, and 1, respectively.

Results

The necessity analysis

The main principle dominating the technical aspect of fsQCA is the examination of set-theoretic relationships between causally relevant conditions and clearly specified outcomes, and these set-theoretic relationships are then interpreted on terms of necessity and/or sufficiency (Schneider and Wagemann, 2010). Therefore, necessary conditions should be checked before performing truth table analysis. From the perspective of the set theory, a necessary condition is generally viewed as a superset of outcomes (Rihoux and Ragin, 2009). Consistency is an important criterion for the measurement of necessary conditions. Conventionally, a condition or a combination of conditions is called “necessary” or “almost always necessary”, if the consistency score exceeds the threshold of 0.90 (Schneider et al., 2010). fsQCA 3.0 is used to calculate conditions of consistency and analyze the truth table. As shown in Table 1, the consistency scores of all conditions are less than 0.90, indicating that there is no or no necessary condition for achieving IPF.

Table 1 The necessity analysis.
Full size table

Sufficiency analysis

Unlike necessity analysis, configuration analysis attempts to reveal sufficiency analysis of outcomes resulting from different configurations composed of multiple conditions (Zhang et al., 2019). From the perspective of the set theory, a combination of sufficient conditions constitutes a subset of outcomes (Rihoux and Ragin, 2009). In terms of fuzzy sets, a fuzzy subset relation exists when membership scores in one set (a combination of conditions) are consistently less than or equal to their membership scores in another (outcomes) (Ragin, 2006). During truth table analysis, consistency and the minimum acceptable frequency of cases are very important. Consistency refers to the extent to which one set (a combination of conditions) is included in another (outcomes), and the minimum acceptable frequency of cases refers to which combinations of conditions are relevant (Rihoux and Ragin, 2009). They resemble the notion of significance in statistical models (Schneider and Wagemann, 2010). Due to large sample number as well as raw consistency, we set the lowest acceptable consistency for solutions at ≥0.90 and set the minimum acceptable frequency of cases at three (Fiss, 2011; Park et al., 2020). Subsequently, due to a fault in the PRI (proportion reduction in inconsistency) consistency between 0.82 and 0.73, the critical value of PRI consistency is set to 0.75. It is an alternative consistency measure, and eliminates the effect of having the same members in both the set (outcomes) and its complementary set (Park et al., 2020).

Truth table analysis yields complex solutions, parsimonious solutions, and intermediate solutions in total (Rihoux and Ragin, 2009). Fiss (2011) put forward the consist of the last two solutions to bring out core and peripheral conditions, associated with outcomes. Core conditions, which are part of both parsimonious and intermediate solutions, have a strong causal relationship with outcomes, whereas peripheral conditions, appearing only in the intermediate solutions, present a weaker relationship with outcomes (Ali et al., 2016). Table 2 shows four different configurations of conditions for achieving IPF. Following the presentation form of results by Fiss (2011), black circles indicate the presence of conditions, and circles with “×” indicate their absence. Large circles indicate core conditions, small ones indicate peripheral conditions, and blank spaces indicate “don’t care”. The coverage measure assesses the extent to which a configuration covers cases of outcomes in Table 2 (Park et al., 2020). Besides, there are two prime implicants in truth table analysis, i.e., “ICO*ECO” and “~MCP*ECO* ~ TDP”. “ICO*ECO”, as core conditions, are more practical. Hence, we reject the prime implicant “~MCP*ECO* ~ TDP”. Among them, “~” stands for “NOT”, and “*” stands for “AND”.

Table 2 Configurations for achieving IPF.
Full size table

As shown in Table 2, there are four configurations to achieve IPF, indicating that there are multiple and concurrent ways to achieve IPF. The consistency of these configurations is more than 0.95, overall solution consistency is also 0.94, and the overall solution coverage is 0.85, indicating that the four configurations contain the majority of cases in the sample and have high explanatory power. From the perspective of configuration types, configuration 1 contains ICP, MCP, and ICO, among which MCP and ICO are core conditions, and ICP is a peripheral condition. Configuration 2a contains ICP, MCP, and ECO, among which ICP and ECO are core conditions, and MCP is a peripheral condition. Configuration 2b contains ICP, ICO, and ECO. They are all core conditions. Configuration 3 contains ~MCP, ICO, ECO, and ~TDP, among which ICO and ECO are core conditions, and ~MCP and ~TDP are peripheral conditions. The raw coverage and the unique coverage of configuration 3 are both minimal, and only 3 cases are covered (unique coverage). From a single condition, ICO and ECO appear most frequently, and at least one of them, as a core condition, exists in the four configurations. It also confirms that ICO and ECO are important external resources for the innovative development of innovation subjects. The frequencies of ICP and MCP are in the middle, but as core conditions appear in the three configurations, accounting for the vast majority of cases. It also proves that ICP and MCP are important core competencies to achieve the innovative development of innovation subjects. TDP appears lowest frequently, and only ~TDP, as a peripheral condition, exists in one configuration. It indirectly reflects the weak or even negative influence of TDP on the realization of the innovative development of innovation subjects. From the perspective of the relationship between configurations, MCP in configuration 2a and ICO in configuration 2b can be replaced with each other to a certain extent. In view of the above complex relationships, this work will give an in-depth interpretation of the four configurations through case studies below.

Robustness test

In order to verify the stability of the results, we use two methods: changing the measurement method and adjusting the consistency level (Zhang et al., 2019). There are two ways to change the measurement. One is to adopt the calibration procedure of Coduras et al. (2016), that is, according to the 7-piont Likert scale, full membership, the cross-over point, and full non-membership are set as maximum value, mean value, and minimum value, respectively. The other is the calibration procedure partially adopted by Tho and Trang (2014), that is, according to the 7-piont Likert scale, full membership, the cross-over point, and full non-membership are set as 6, mean value, and 3, respectively. Adjusting the consistency level sets the minimum acceptable consistency of solutions to ≥0.95. We refer to the two criteria for robustness of QCA results proposed by Schneider and Wagemann (2012), namely, set relation states of different configurations and the difference of fitting parameters of different configurations. It can be found that the results remain robust (for the details of the robustness test, see Supplemental material).

Organization pattern and case analysis

There are four configurations that can realize the innovation of the IoT product system. According to core conditions, correlation, and element categories in the four configurations, they are summarized into three organization patterns, that is, the modular-oriented organization pattern corresponds to configuration 1, the integration-oriented organization pattern corresponds to configuration 2a and configuration 2b, and the relationship-oriented organization pattern corresponds to configuration 3.

Modular-oriented organization pattern

The core conditions of the modular-oriented organization pattern are MCP and ICO. The organization pattern is suitable for innovation subjects who are good at MCP, and typical representatives are technology start-ups. MCP, together with ICO, can affect IPF. ICO includes strategic planning, benefit distribution, etc. among innovation subjects. While it does not provide technical guidance to innovation subjects, ICO helps them understand technical visions of innovation research, such as benefit objectives, efficiency objectives, attractiveness, specificity, and infrastructure clarity, which are measures of early success (Reid et al., 2015). In addition, the innovation uncertainty of the IoT product system is strong, the ambiguity of client needs is high, and the development cycle is relatively long, resulting in the possibility that the task may be adjusted and changed at any time. ICO can reduce risks involved and help innovation subjects to change their innovation strategies at any time. All these illustrate the unique role of ICO in task design or task adjustment for innovation subjects who excel in MCP. ICP as a peripheral condition indicates that it is necessary to develop MCP of innovation subjects from a holistic perspective. Because the core of modular architectures is a holistic view, better system engineering and planning capability may be required (Holmqvist and Persson, 2003; Ulrich, 1995).

J company is a technology start-up transforming from industrial automation to industrial IoT. It is based on its own industrial automation knowledge and skills, combined with distributed control systems, programmable controllers, cloud services, and other technical means to develop industrial IoT controllers. At the beginning of its involvement in the industrial IoT field, J company provided products and corresponding technical services for integrated firms or platform firms, acting as an equipment manufacturer assisting them to achieve value, in exchange for its own interests. During the cooperation period, it gradually understood a series of issues such as price, brand influence, technological advancement, and policy barriers through communication with them. Besides, there was a certain cycle and uncertainty in the switching and introduction of new products. J company needed to solve the trust problem through the stakeholders, and then participated in the cooperation with the original equipment manufacturers to solve technical problems for clients (Jin, 2019). It can be seen that for innovation subjects who have expertise in certain skills or professional areas, especially start-ups, it is far from enough to rely on excellent product technology as their core conditions. On one hand, it is necessary to cooperate with mature brand firms to obtain a wide range of client resources to enhance the influence of the industry. On the other hand, it is necessary to expand the relevant technology and knowledge through the integrity of the module architecture. The two maintain and enhance each other in order to improve dynamic capabilities and then create a steady stream of IPF for start-ups.

Integration-oriented organization pattern

The integration-oriented organization pattern includes two configurations, and the core conditions for both configurations are ICP and ECO. The organization pattern is suitable for innovation subjects who are good at ICP, and typical representatives are system integrators. Both ECO and ICP can affect IPF. In the innovation process of the IoT product system, innovation subjects can not only integrate the customized needs of clients and technical support of scientific research institutes into one to enhance their competitiveness, but also create the customized needs of clients with their excellent ICP. At the same time, the macro control of technology by the government and supervisory agencies creates a relatively specific environment for innovation subjects and reduces the uncertainties brought by the environment. All these indicate that ECO plays an important role in the innovation process for innovation subjects who are good at ICP. MCP as a peripherical condition and ICO as another core condition can be replaced under certain conditions, indicating that the effects of the two are relatively equivalent, both for innovation subjects to understand all aspects of knowledge and technology system deeply. They facilitate the decomposition of innovation tasks to complete innovation tasks.

Sanchuan is a water metering function service and smart water overall solution provider. It has set up academicians and postdoctoral workstations, and has the ability to lead the technological progress and product innovation of the industry. In 2017, Sanchuan cooperated with China Academy of Information and Communications Technology, Huawei Group, China Telecom, and China Mobile to launch Narrow Band IoT (NB-IoT) smart water meter system in Yingtan, and developed non-magnetic, photoelectric ultrasonic NB-IoT smart water meters according to client needs and environmental requirements. Yingtan government is the national pilot, which is the national information city, the national smart city, and the national telecom universal service, for the development of NB-IoT smart water meter system to provide a stable environment. Meanwhile, Yingtan government signed a series of cooperation agreements with China Unicom, China Mobile, China Telecom, Huawei, and ZTE, etc., to build NB-IoT communication systems, providing the network foundation for NB-IoT smart water meter system (Xu, 2017). It can be seen that for innovation subjects who can provide a series of solutions, especially in the subdivision of the IoT, government policies and support and the network foundation of network operators play a fundamental role, giving innovation subjects space to play. Besides, customized needs of clients and environmental constraints also encourage innovation subjects to expand overall solutions through innovation and enhance their own competitive strength.

Relationship-oriented organization pattern

The core conditions of the relationship-oriented organization pattern are ICO and ECO. The organization pattern is suitable for innovation subjects who are good at information channels, and typical representatives are network operators. From an architectural point of view, the IoT can be divided into perception layer, transport layer, and application layer. The transport layer is the most indispensable part of the IoT product system. It is a new communication and network architecture, which can integrate a variety of existing communication and network technologies and their evolution, and adapt to various perceptual modes, analytic architectures, and network computing processing available in the future (Ning and Xu, 2010). In the innovation process of the IoT product system, ICO through strategic planning not only makes basic common technology and knowledge of two sides interconnect, integrates hardware with software, facilitates innovation subjects to extend to the perception layer and application layer, but also creates and enriches the related facilities and services of the IoT platform. In terms of ECO, the technology promotion of the government, requirements of clients for the platform interface, and the technical support provided by scientific research institutes promote innovation subjects to continuously expand relevant information channels and improve the technical level of the IoT platform.

Wuxi government and Jiangsu Telecom carried out in-depth cooperation in the construction and application promotion of “Smart Wuxi” to comprehensively promote the development of “Smart Wuxi”. Jiangsu Telecom actively participated in the top-level design of the architecture and the operation management of application projects, conducted project constructions in the fields of industry, agriculture, transportation, water conservancy, etc., provided information solutions based on cloud computing and IoT technology, and formed IoT collaborative innovation symbiont with the upstream of industrial chain, the government intermediary end, the academic research end, and the client end (Shao and Lyu, 2016). It can be seen that innovation subjects who can provide information channels are in an important position due to their previous technical experience and capital accumulation. They can make full use of information and knowledge from all sides to continuously improve their dynamic capabilities and achieve IPF.

Through the analysis of the three organization patterns above, we can find that the vast majority of firms, which participated in the development of the IoT product system, are based on ICP or MCP as their core competencies. By further integrating core competencies with external resources, namely ICO and ECO, to enhance their dynamic capabilities, the firms reduce the uncertainty of innovation risks and ensure the smooth progress of innovation tasks. The absence of TDP as a core condition also illustrates the current scientific and technological strength of Chinese IoT firms to a certain extent. Based on the analysis of organization patterns and cases above, corresponding summaries are shown in Table 3.

Table 3 Organization patterns, their corresponding configuration views, and typical cases.
Full size table

Conclusion and implications

Conclusion

The innovation of the CoPS usually accomplished by a multi-functional innovation team across organizations, which is a typical complex innovation network organization. How to choose partners suitable for their own innovative development and form a unique organization pattern is very important for innovation subjects of the CoPS. Based on the dynamic capability theory, we take 184 firms participating in the IoT product system as a sample, and explore the “joint effect” of five factors, containing ICP, MCP, ICO, ECO, and TDP, on IPF and interactive relationships among the five factors. The conclusion is as follows:

The realization of the innovative development of the CoPS is the result of various factors, and any single factor is neither sufficient nor necessary for IPF. Meanwhile, there are four different ways, a total of three organization patterns can achieve IPF, and each way is composed of multiple factors. In the modular-oriented organization pattern, MCP and ICO are core conditions, and ICP is a peripheral condition. The integration-oriented organization pattern contains two configurations, the common feather of which is that ICP and ECO are core conditions, while MCP and ICO can replace each other under certain conditions. In the relationship-oriented organization pattern, ICO and ECO are core conditions, and ~MCP and ~TDP are peripheral conditions. In addition, in the three organization patterns, ICP and MCP as core capabilities, ICO and ECO as external resources, are important support for innovation subjects of the CoPS to achieve innovative development.

Theoretical implications

This work extends the dynamic capability theory into the theoretical system of the innovation of the CoPS. It provides some insights for the research of enterprise strategic management. Firstly, this work expands the perspective of innovation subjects by shifting the focus of the innovation of the CoPS from the innovation team or leader to the members of the innovation team. We not only divide innovation team resources to promote the innovative development of the CoPS into core competencies of innovation subjects and their inter-organizational relationships, but also further divide core competencies and inter-organizational relationships in a fine-grained way. It covers the vast majority members of the innovation team. Secondly, there are four configurations to realize the innovative development of innovation subjects of the CoPS, and there are complementary and substitute relationships among the factors, thus deepening interactive relationships based on the dynamic capability theory. By interacting with ECO, ICP can continuously expand horizontally on the basis of original resources, so that innovation subjects are always in an open and diversified state. By interacting with ICO, MCP can be embodied in the vertical development of heterogeneous resources, so that innovation subjects are always in a constructive state (Holmqvist and Persson, 2003). The mutual substitution of MCP and ICO under certain conditions is the unification of strategic planning to reduce potential technical risks. This work provides a reference for innovation subjects of the CoPS to optimize resource allocation, and is also a supplement to the interpretation of the interactive relationships based on the dynamic capability theory from Hou et al. (2024). Thirdly, due to the limitation of cross-sectional data in previous literature, the choice of organization patterns of innovation subjects of the CoPS has not been further revealed. This work is a new attempt to explore it by using fsQCA approach based on the dynamic capability theory. The vast majority of innovation subjects can promote their own dynamic capability development by integrating ICP/MCP with ICO/ECO based on their own advantages, but combinations of core competencies and inter-organizational relationships are also different in different organization patterns. It not only expands the idea of Bateman and Snell (2013), but also provides new insights into how different innovation subjects within the CoPS team achieve innovative development. Therefore, this work is an important step forward in advancing research approaches, providing a robust and universal explanation of the innovation trends of the CoPS, and fills the current gap in understanding how different innovation subjects of the CoPS find organization patterns that are suitable for their own innovative development based on the dynamic capability theory.

Managerial implications

Our work also has some practical implications. Firstly, innovation subjects highlighted by ICP can promote innovative development from two aspects according to their own resource endowments. On one hand, innovation subjects can use information technology to establish innovation platforms, data retrieval tools, for optimizing resource allocation and consolidating core competencies. On the other hand, they should pay attention to strengthening the coordination with the government, clients, and research institutes, such as grasping the leading and supervisory policies issued by the government, finding out the ambiguous needs of clients, and jointly developing with research institutes in commissioning, collaboration, and establishment ways. The knowledge system and technical architecture of the entire CoPS tasks are gradually mastered in a three-dimensional cycle, thus expanding the boundaries of technical knowledge and enriching product solutions. Secondly, as for innovation subjects highlighted by MCP, in addition to increasing investment in the research and development of key technologies, more importantly, they can gradually maintain long-term cooperative relationships with other enterprises through strategic convergence, interest bundling, technology vision, to solve trust problems, enhance industry influence, and further strengthen the connection among various modules and enrich the types of modules. Thirdly, as for innovation subjects balanced of ICP and MCP, they can carry out project constructions by co-building knowledge sharing platform, information exchange platform, and innovation service platform with many other members. They can also integrate independent innovation and open innovation, and rely on hard power and soft support, so as to achieve the goal of broadening the supply chain field horizontally and expanding the industry field vertically. Fourthly, the government is both the policy maker and the main client of the CoPS. Therefore, the government should lead innovation subjects to constantly carry out technological innovation and iteration in terms of technology, management, and environment. It also could create an environment conducive to innovation output and stable operation of innovation subjects in terms of market, finance, technology, and other policies.

Limitations and future study directions

The study has some limitations. Firstly, although the explanatory framework of the dynamic capability theory used in this work covers several important influencing factors in the relationship between core competencies and inter-organizational relationships that have been tested by many studies, there are still other factors that have not been considered. For example, the information transparency of CoPS design (Tee et al., 2019), and design dependence among innovation subjects (Ethiraj and Posen, 2013). The latter is also worth studying in order to improve the innovation efficiency of innovation subjects and expand their innovation output. Therefore, future research can be further analyzed from other perspectives. Secondly, the data in this study is based on respondents’ understanding of the entire innovation process of the CoPS. However, the innovation of the CoPS can occur throughout the lifecycle of a project, with different members, innovation tasks, and so on at each stage. Therefore, future research can be carried out according to each stage of the lifecycle such as bidding, design, and development. Thirdly, this work chooses the sample data of Chinese IoT industry. As the IoT industry is one of the typical representatives of the new generation of information technology industry, it can reflect the new generation of information technology industry to a certain extent, but the conclusion may not be well applicable to other industries. In addition, the economic and cultural differences between China and foreign countries also affect the applicability of the conclusion to a certain extent. Therefore, future research can further explore the choice of organization patterns of innovation subjects of the CoPS in the new generation of information technology industry in other geographical regions, so as to provide a more comprehensive explanation for the innovation development of the CoPS.