Configurational Drivers of Digital Technology Adoption in Manufacturing SMEs: Evidence from Chile

Introduction

Digital transformation is reshaping industrial competitiveness and creating new opportunities for regional development, particularly for small and medium-sized enterprises (SMEs) (Fiorini et al., 2022).

This study identifies the configurations of factors that explain the depth of digital technology adoption in manufacturing SMEs. Rather than being driven by a single determinant, digital transformation follows different pathways, suggesting that firms may require differentiated support mechanisms.

The analysis is based on a survey of 52 manufacturing SMEs participating in the FIC-R Digital Transformation Program for Manufacturing SMEs, funded by the Regional Government of Bío-Bío, Chile. Surveys were conducted between 2023 and 2024 among owners and managers with some prior awareness of Industry 4.0 technologies.

Given the sample size and the configurational nature of the phenomenon, fuzzy-set Qualitative Comparative Analysis (fsQCA) was employed (Ragin et al., 2017). This approach allows the identification of multiple combinations of conditions leading to similar outcomes and offers an alternative to traditional linear approaches (Pappas et al., 2021).

The study contributes to a still limited body of research combining the Technology–Organization–Environment (TOE) framework and fsQCA to analyse digital transformation in SMEs. While most existing evidence comes from Europe and China, this research provides evidence from a Latin American industrial region characterized by technological lagging and strong manufacturing traditions.

The findings indicate that technological knowledge, access to ICT professionals, organizational culture, and leadership motivation are the most relevant conditions explaining higher levels of digital technology adoption, whereas the environmental conditions analysed play a more limited role. The study contributes to the literature on SME digital transformation and provides evidence for the design of more effective public policies.

Literature Review

The literature on technology adoption includes several models that identify the key factors that influence the incorporation of new technologies within companies. Among the best known are the Theory of Reasoned Action (Fishbein et al., 1975), the Technological Acceptance Model (TAM) and the Theory of Planned Behavior (Ajzen, 1991; Gentry et al., 2002). In this context, two integrative models stand out that synthesize multiple theoretical approaches to technological adoption.

A first model is the Unified Theory of Technology Acceptance and Use (UTAUT), developed by Venkatesh et al. (2012), which identifies several key determinants of technology adoption: performance expectations, effort expectancy, social influence, facilitating conditions, hedonic motivation, price value, and habit. Although this model has been widely used, it focuses mainly on individual decision-making processes, particularly in the context of the adoption of information and communication technologies (ICTs).

A second model, better suited to analysing technological adoption within firms, is the Technology–Organization–Environment (TOE) framework, developed by Tornatzky et al., (1990) and later expanded by Rogers (1995). This model identifies three major groups of factors: (i) technological factors, including perceptions of relative advantage, compatibility, and technological challenges; (ii) organizational factors, such as firm size, leadership support, and absorptive capacity; and (iii) environmental factors, including the level of competition, uncertainty, and the availability of external support. Given its focus on organizational technology adoption, the TOE framework is particularly suitable for analysing the factors that influence the adoption of Industry 4.0 technologies in manufacturing firms.

Under the TOE framework, several studies have analysed the adoption of digital technologies in manufacturing firms. The main findings of this literature include the following:

• Firm size determines the capacity to adopt and absorb new technologies, as it conditions the financial and administrative resources available to acquire and adapt them (Ingaldi et al., 2020; Kiraz et al., 2020). At the same time, firm size may limit operational scale, which can either encourage or discourage the adoption of new technologies.

• Internet connection quality—specifically speed and stability—emerges as a critical factor in the adoption of digital technologies (Wang et al., 2023).

• A workforce with digital skills significantly increases the likelihood of adopting and advancing digital technologies, particularly in areas such as technology scouting, decision-making, system integration, and maintenance (Almeida et al., 2020; Cabrera-Sánchez et al., 2019). This capability enables firms to reach more advanced levels of technological adoption (Gatica et al., 2022).

• Leadership capable of envisioning technological adoption pathways increases the likelihood of successful digital transformation in SMEs (Motta et al., 2019; Maggi et al., 2020). This leadership facilitates the scaling of digital technologies across different layers of organizational integration (Battistoni et al., 2023).

• Organizational culture also plays a determining role in the adoption of digital technologies. Firms that have previously experienced innovation processes tend to show greater flexibility when incorporating new digital tools (Gatica, 2024; Horváth et al., 2019; Chauhan et al., 2021; Marrucci et al., 2025). In this regard, Li et al., (2023) conclude that strategic alignment between technological and organizational factors is essential for the success of digital transformation in SMEs. The relationship between technologies, tasks, people and organisational structures plays a key role in adoption processes and determines the technical and social benefits of new technologies (Margherita et al., 2021).

• Technological knowledge is another critical factor in technology adoption. It facilitates the assessment of relative advantages, improves the understanding of potential challenges associated with adoption, and helps firms address compatibility issues between new and existing technological systems (Marrucci et al., 2025).

Recent studies combining the TOE framework and fsQCA to analyse digital transformation remain relatively limited, particularly in the context of SMEs. Most of the available evidence comes from China and Europe, while studies focusing on manufacturing SMEs in Latin America remain scarce. Xu (2025) notes that only a limited number of studies have applied the TOE framework and fsQCA to identify the configurations that explain digital transformation. Among the existing literature, five studies are particularly relevant to the objectives of this research (see Table 1). Overall, the available evidence suggests that digital transformation is explained by different combinations of technological, organizational, and environmental factors rather than by isolated variables. Studies conducted in manufacturing firms and SMEs identify multiple configurations associated with digital transformation, digital innovation, and digital servitization (Li et al., 2023; Wang et al., 2023; Zhang et al., 2024; Xu, 2025; Marrucci et al., 2025).

However, empirical evidence from manufacturing SMEs in Latin America remains limited. Therefore, this study seeks to contribute to the existing literature by identifying the configurations that explain the depth of digital technology adoption among manufacturing SMEs in an industrial region of Chile.

The TOE framework suggests that the adoption of digital technologies in firms results from the interaction between technological, organizational and environmental factors. However, these factors rarely operate independently. Instead, different combinations of conditions may lead firms to similar levels of digital technology adoption. Codara et al., (2023), drawing on a study of multiple cases, conclude that there are different paths to digitalisation, emphasising in particular the alignment between organisational capabilities and the strategies adopted as a key factor in explaining organisational resilience.

In this context, configurational approaches provide an appropriate methodological perspective to analyse these relationships. Fuzzy-set qualitative comparative analysis (fsQCA) allows the identification of different combinations of conditions that explain the presence of a specific outcome. This approach is particularly suitable for analysing SMEs, where multiple pathways may lead to successful digital transformation.

In summary, the literature suggests three main conclusions. First, most empirical evidence on digital transformation and configurational approaches is concentrated in China and Europe. Second, digital transformation in SMEs is not explained by a single factor, but by different combinations of technological, organizational, and environmental conditions. Third, the TOE framework provides a suitable analytical perspective for examining these configurations and understanding the factors that explain different levels of digital technology adoption.

Methodology

Given the configurational nature of digital technology adoption and the relatively small sample size, fsQCA is considered an appropriate methodological approach because it allows the identification of different combinations of conditions leading to similar outcomes (Zhang et al., 2024). Due to the limited sample size (n = 52), structural equation modelling and regression analysis were considered unsuitable.

The surveys were conducted between 2023 and 2024 and applied to SME owners or managers. The FIC project included manufacturing companies from the metalworking and woodworking sectors in the Bío-Bío Region, grouped into two main associations: the Association of Metalworking Companies (AGMET) and the Association of Woodworking SMEs (Pymemad). Together, both associations comprise a total of 85 member companies.

Based on the total population, a sample size of 71 firms would be required under conventional probabilistic assumptions (95% confidence level and 5% margin of error). However, given the exploratory and configurational nature of the study, together with the characteristics of fsQCA, the final sample of 52 firms was considered adequate for the analysis. No data loss occurred during the information processing stage. The surveys were conducted by a professional specialising in digitalisation and were administered to the manager or owner of the company. Most of the surveys were carried out in urban areas, although some may also operate in rural areas.

The analysis was conducted using fuzzy-set variables, which allow the identification of different degrees of membership within a dataset. Variables were classified according to their membership values: 1 indicates full membership in a condition, 0 indicates no membership, and 0.5 represents the point of maximum ambiguity (Saridakis et. Al, 2022).

The following steps were implemented:

• Calibration of variables. Variables were calibrated using three thresholds: 0.95 (full membership), 0.50 (crossover point), and 0.05 (non-membership). The calibration procedure was based on the maximum, mean, and minimum values of each variable. All calibrations were performed using the algorithm implemented in the fsQCA software. The empirical calibration values for each variable are reported in Table 3.

• Necessary condition analysis. For each variable, consistency and coverage indicators were calculated. Consistency measures the degree to which a causal condition (X) is a subset of the outcome (Y), while coverage indicates the empirical relevance of a consistent set (Kent, 2008). In the analysis, no variable reached the threshold of a necessary condition (>0.9), and consistencies above 0.8 were considered for further analysis (Ragin et al., 2017).

• Truth table construction. A truth table was constructed presenting all possible configurations of conditions explaining the level of adoption. The total number of configurations corresponds to 2k, where k represents the number of analysed variables. To reduce the complexity of the table, the default criteria provided by the fsQCA software were applied. Given the sample size (N = 52), a frequency threshold of one case was adopted. Using a higher threshold would have excluded empirically relevant configurations and substantially reduced configurational diversity (Ragin et al. 2017). For each model, parsimonious, intermediate, and complex solutions were estimated. The intermediate and complex solutions converged, showing the same configurational paths, solution coverage, and consistency. Therefore, the complex solution was reported.

• Solution analysis. A standard analysis was performed to generate complex, intermediate, and parsimonious solutions. To identify whether a condition is central or peripheral within a configuration, the parsimonious solution was used. Central conditions are represented by large black circles (●), peripheral conditions by small black circles (●), and the absence of a condition by empty circles (●).

• Graphical validation of the solution. For the selected configuration, the estimated solution was calculated based on the calibrated variables. An XY Plot was then generated using the fsQCA software to compare the observed outcome (X) and the estimated outcome (Y), allowing the dispersion around the diagonal line to be observed.

To quantify the quality of the estimation, correlation coefficients (Spearman’s Rho and Kendall’s Tau) were calculated between the observed and estimated values using the Gretl econometric software. Table 2 presents the indicators used in the analysis, based on the information obtained from the 52 surveyed firms. The dependent variable in the analysis is the level of adoption, defined as a rate per firm based on the presence (1) or absence (0) of the following technologies: production planning software, ERP systems, internal and external communication platforms, machine learning tools, CAD systems, robots, CNC machines, PLC systems, and renewable energy technologies.

$$\text{Adoption level by firm}=(\text{Σ adoption by technology}/\text{Σ maximum score})$$ (1)

In the analysis, new renewable energy technologies were incorporated. There is an interesting synergy between digital technologies and the green economy (Luo et al., 2022), where timely information can support corporate management practices that are more environmentally responsible. Moreover, there is a growing set of public policies aimed at promoting the development of renewable energy technologies.

Before implementing the qualitative comparative analysis, a preliminary review of the data was conducted. This included analysing the distribution of technologies across the firms, the average level of technological knowledge, and a non-parametric correlation matrix using Kendall’s Tau and Spearman coefficients.

Prior Data Analysis

Table 4 shows that internal and external communication technologies are the most widely adopted digital tools among the surveyed firms, being present in 90.4% of the cases. Production planning and scheduling software ranks second (59.6%), followed by Computer-Aided Design (CAD) technologies (51.9%). Intermediate adoption levels are observed for Enterprise Resource Planning (ERP) systems (32.7%), Computer Numerical Control (CNC) technologies (26.9%), and Programmable Logic Controllers (PLCs) (21.2%). More advanced Industry 4.0 technologies remain less common, with machine learning applications present in 17.3% of firms, industrial robots in 9.6%, and renewable energy technologies in only 7.7% of the surveyed companies. Overall, the results suggest that manufacturing SMEs have achieved higher adoption rates for communication and operational management technologies, while more sophisticated digital technologies remain at an early stage of diffusion.

A second dimension analysed was the average level of technological knowledge. For each technology, respondents indicated whether their level of knowledge was none (0), low (1), medium (2), or high (3). Figure 1 presents the average scores obtained for each technology.

Higher levels of knowledge are observed for external communication (2.6), internal communication (2.5), and production planning software (2.0), which is consistent with their relatively high adoption rates. Intermediate levels are found for CAD technologies (1.5), renewable energy solutions (1.3), ERP systems (1.3), and CNC technologies (1.3). Finally, PLCs (1.0), machine learning (0.9), and robotics (0.8) show the lowest knowledge levels. Overall, the results suggest a positive relationship between technological knowledge and adoption, although some technologies, such as renewable energy, exhibit higher levels of awareness than actual implementation.

Finally, a non-parametric correlation matrix is presented (Table 5). In general, no coefficients above 0.8 are observed, indicating the absence of collinearity problems between the variables. In addition, the Variance Inflation Factor (VIF) indicator was calculated using the Gretl econometric software, confirming that all values are below the threshold of 10.

In the initial analysis, the variables Level of Knowledge (ccnoc), Organizational Culture (ccultu), Internet Quality (cinte), Presence of ICT Professionals (cprotic), and Number of Workers (ctrab) present Tau and Rho coefficients above 0.3 with the dependent variable Level of Adoption (cniadop). The variable most strongly associated with adoption is the firm’s technological knowledge level, with coefficients of Rho = 0.68 and Tau = 0.54.

Results

The objective of this study is to identify the factors that explain the deepening of digital technology adoption. Specifically, the research seeks to answer the following question: What are the configurations of factors that explain the depth in the adoption of new digital technologies in manufacturing SMEs?. This study aims to provide a comprehensive understanding of the factors that foster technological synergies within manufacturing SMEs in the Bío-Bío Region of Chile.

First model

Examining Table 7, the complex solution identifies five relevant configurations, presenting a consistency of 0.84 and coverage of 0.80, which can be considered acceptable. The solution consistency is above the consistency cutoff threshold (0.84 > 0.83).

Complex solution.

Model 1: cniadop = f( cconoc, cultu, cprotic, cinte, motvb)

Algorithm: Quine-McCluskey

Calibrated variable	Abbreviation	C.1	C. 2	C. 3	C. 4	C. 5
Internet quality	Cinte	○	○		●	●
Motivation	Cmotvb	●	●	●	○	●
ICT Professionals	Cprotic		●	●	●	○
Culture	Ccultu	●		●	○	○
Knowledge of technologies	Cconoc	○	●	●	○	●
Raw coverage		0.36	0.37	0.60	0.18	0.21
Unique coverage		0.04	0.01	0.27	0.08	0.01
Consistency		0.84	0.90	0.92	0.87	0.90
Frequency cutoff = 1
Consistency cutoff = 0.83
Solution coverage = 0.80
Solution consistency = 0.84

Source: Authors’ elaboration using FSQCA software

Note 1: Black circles indicate a very high presence of a condition and white circles indicate a very low presence (i.e., absence) of a condition. Blank spaces indicate “I don’t care.” The size of the black circle is associated with the central condition from the parsymonic solution.

Note 2: C=Configurations or Combinations

A general review shows that internet quality is not a relevant variable in explaining the adoption of Industry 4.0 technologies. In configurations 1 and 2 it appears with a low presence, in configuration 3 it does not appear, and in configurations 4 and 5 it appears only as a peripheral variable.

This suggests that internet access operates as a minimum enabling condition, but its presence alone cannot explain the technological scaling of SMEs.

Regarding the other variables, leadership motivation appears in configurations 1, 2, 4 and 5, although mainly as a peripheral condition.

In contrast, the central variables explaining the depth of adoption are:

• access to ICT professionals

• organizational culture supportive of technological change

• technological knowledge.

The presence of ICT professionals appears as a central condition in configurations 2, 3 and 4. Similarly, technological knowledge appears as a central condition in configurations 2, 3 and 5. Organizational culture appears as a central condition in configurations 1 and 3.

Among the identified configurations, configuration 3 presents the strongest explanatory power, with a raw coverage of 0.60 and a consistency of 0.92. Its unique coverage (0.27) is also higher than the other configurations in the first model.

In summary, the combination of ICT professionals, an organizational culture supportive of digital technologies, technological knowledge, and leadership motivation explains the adoption of Industry 4.0 technologies in manufacturing SMEs.

Configuration 3 from the first model was further analysed using an XY Plot, where the observed level of adoption is represented on the Y-axis and the proposed configuration on the X-axis (Figure 2). The results show that most cases fall within the model, indicating a consistent configuration with strong empirical coverage.

The analysed cases show a positive correlation between the proposed configuration and the observed adoption levels (Spearman Rho = 0.57; Kendall Tau = 0.44). However, the dispersion of the data indicates that the configuration tends to slightly underestimate the observed adoption levels while maintaining a consistent positive trend.

Second Model

In the second model, internet quality was removed and the number of workers variable was incorporated, based on the results obtained in the necessary condition analysis.

The consistency of the solution (0.81) is equal to the established threshold and therefore can be considered acceptable. Similarly, the coverage of 0.80 indicates an acceptable, although limited, explanatory capacity.

Table 8 shows that leadership motivation becomes a central variable in configurations 1, 2, 3 and 4. In configuration 5, however, motivation appears as an absent condition.

Complex solution

Model 2: cniadop = f(cmotvb, cprotic, ccultu, cconoc, ctrab)

Algorithm: Quine-McCluskey

Calibrated variable	Abrevia-tura	C.1	C. 2	C. 3	C. 4	C. 5
Motivation	Cmotvb	●	●	●	●	○
ICT Professionals	Cprotic	○		●	●	●
Culture	Ccultu		○		●	○
Knowledge of technologies	Cconoc	○	●	●		○
Number of workers	Ctrab	○	○	○	●	●
Raw coverage		0.43	0.25	0.54	0.37	0.14
Unique coverage		0.04	0.014	0.11	0.11	0.04
Consistency		0.78	0.86	0.91	0.92	0.99
Frequency cutoff = 1
Consistency cutoff = 0.81
Solution coverage = 0.80
Solution consistency = 0.81

Source: Authors’ elaboration using FSQCA software

Note 1: Black circles indicate a very high presence of a condition and white circles indicate a very low presence (i.e., absence) of a condition. Blank spaces indicate “I don’t care.” The size of the black circle is associated with the central condition from the parsymonic solution.

Note 2: C=Configurations or Combinations

Regarding the number of workers, this variable appears as absent in configurations 1, 2 and 3, and present in configurations 4 and 5. Overall, the variable does not appear to play a decisive role in explaining the adoption process.

In this model, organizational culture and technological knowledge appear as peripheral variables, which contrasts with the first model, where they played a central role.

In contrast, the presence of ICT professionals remains a central variable in both models, appearing in configurations 3, 4 and 5, which demonstrates a strong consistency in the results.

Among the configurations identified in the second model, configuration 3 again provides the best explanation of the adoption process, with a raw coverage of 0.54, unique coverage of 0.11, and consistency of 0.91.

This configuration suggests that the combination of leadership motivation, access to ICT professionals, technological knowledge, and the absence of a large workforce explains the depth of digital technology adoption.

Figure 3 presents the XY Plot for configuration 3 of the second model. In this case, the observed adoption level is located on the X-axis and the configuration on the Y-axis.

The dispersion of the points around the diagonal line is greater than in the first model, indicating a lower explanatory quality. This is confirmed by the correlation coefficients (Rho = 0.39; Kendall Tau = 0.30), which are lower than those obtained in the first model.

Overall, the results highlight the importance of internal organizational capabilities in explaining the depth of digital technology adoption. In particular, access to ICT professionals and technological knowledge appear as the most consistent explanatory conditions across the models.

Discussion of results

This section is divided into two parts. First, the main findings are discussed in light of the theoretical framework and previous literature. Second, the implications of the results for public policy are examined.

Final conclusions

The results show that higher levels of digital technology adoption among manufacturing SMEs are explained by combinations of technological and organizational capabilities, particularly technological knowledge, access to ICT professionals, organizational culture, and leadership motivation.

Consistent with previous studies applying the TOE framework and fsQCA (Li et al., 2023; Wang et al., 2023; Zhang et al., 2024; Xu, 2025; Marrucci et al., 2025), our findings confirm that fuzzy-set qualitative comparative analysis is a useful methodological approach for studying digital transformation in SMEs. This type of analysis allows researchers to identify combinations of factors that are not mutually exclusive, providing a closer representation of the organizational complexity involved in the adoption of new technologies.

Given the limitations of our research, future studies should deepen the understanding of the interactions between the factors identified in this study. In particular, it would be valuable to carry out dynamic analyses comparing different patterns of technology adoption over time (Song et al., 2023). Similarly, the present study was based on a relatively small sample of 52 firms belonging to two trade associations. Future research could expand the analysis to other industrial sectors, firms of different sizes, and additional Latin American regions with mature industrial bases facing technological challenges.

Finally, it is important to highlight the strategic role of motivated business leaders in initiating what may be described as a “virtuous cycle of digital transformation.” Such a cycle may involve the incorporation of professionals specialized in digital technologies, the development of deeper organizational knowledge of emerging technologies, and the strengthening of an organizational culture oriented toward digital innovation.