Several stages of digitization precede the digital transformation stage. Bican and Brem (11) pointed out that the digital transformation process has three different stages, namely, digitization, digital optimization, and digital transformation. The first stage is “digitization.” In the digital transformation process, industries that use a high proportion of paper-based management methods start using information tools and technologies to build software and hardware systems and environments for information management. In the operation process, they record all processes using digital tools and generate information that can be stored and managed. In the second stage, “digital optimization,” enterprises can use the information accumulated by digital tools to improve the operation process, improve production quality and the management process, enhance operation efficiency, and improve the internal system. In the third stage, “digital transformation,” digital technologies are integrated and applied to various operational functions of the enterprise, and the overall process of the enterprise is changed owing to the introduction of information technology. According to a survey on digitization for small- and medium-sized enterprises, most small- and medium-sized enterprises in Taiwan are still in the digitization or digital optimization stage. There is still a long way to go to enter the digital transformation stage. The enterprise must clearly understand its current situation to make the best investment on the road to digital transformation (12).

Komninos et al. (13) pointed out that digital transformation can be described as the process of sustainable digital development. Overall, we can summarize two developmental characteristics of digital transformation. The first developmental characteristic is from the perspective of applications and strategic insights. In terms of enterprises’ strategic insights, digital transformation emphasizes client-driven strategies. In addition to the application of digital technology, more emphasis is placed on organizational change across departments. The second developmental characteristic is from the perspective of its wider impacts. The proliferation and use of digital technologies drives systematic restructuring at the economic, institutional, and social levels. Comparisons are made among corporate strategy, the digitization process, the three stages of digital transformation, and the impact of digital technology.

Mihardjo et al. (14) discussed the digital transformation of the manufacturing industry in the past, in the Industry 4.0 era. The digital transformation includes customer journey analysis, personalized experience, etc. Based on the one-stop supply chain from raw materials, production, to finished products and the use of emerging digital technologies, enterprises can make more real-time dynamic adjustments. In the overall supply chain, big data analysis is used to establish a forecast plan and technologies are used to connect the data from the upstream and downstream of the supply chain to achieve precise control of production and inventory. Therefore, warehousing and logistics costs can be reduced, as can the risk of out-of-stock, supply–demand imbalance, and uncertainty and overall operational efficiency can be improved.

This study takes the dimensions proposed by Mihardjo et al. (14) as the basis of the value chain of the motherboard industry and focuses on the value activities in the manufacturing process, which are used as the dimensions and factors of the subsequent maturity assessment. The following are the descriptions and assessment factors of digitization, digital optimization, and digital transformation.

The cluster at this stage is mainly related to strategic prioritization, flexible work, and management support of digital transformation. Basic digital services for existing products and customer experiences across multiple channels are initiated. Employees are familiar with existing digital products. The internal information technology staff ensures the availability of relevant digital technologies and maintains an up-to-date infrastructure. Digitization has become a priority on the strategic agenda. Digital transformation projects are supported and prioritized by senior and middle management.

At this stage, digital innovation has a significant effect on product innovation, both at the strategic and internal levels. The importance of innovation strategies is emphasized by explicitly promoting digitalization and systematically evaluating potential of new technologies. Suitable conditions for innovation are created by strengthening digital competencies, collaborating more strongly with the internal information technology department, and liaising with external partners.

The items in this stage mainly belong to the field of culture and expertise, but they also fall under organization and transformation management. In stage 3, the focus appears to be on validating organizational innovations in the digitization stage and their profound influence on the changes within the organization. Important capabilities in the company culture are proactive error management, the communication of learning from failed projects, and a willingness to take risks. As a company undergoes more radical change, it needs to define roles and responsibilities for all processes related to digital transformation, as well as devise a strategic plan for the transformation process that the company is willing to follow.

This is demonstrated by user participation in innovation processes, personalization of customer experiences, and the collection and consideration of customer data when designing interactions. This stage emphasizes open innovation by engaging users, personalizing customer experiences and processes based on usage data, and improving the process by establishing measurable goals. This stage is considered a “user-centric detailed process.”

The items with the highest difficulty metric are clustered in Stage 5. These items are related to the use of advanced data analysis technologies for expenditure planning. Only advanced companies use them appropriately for decision support or product development. Prerequisites for the implementation of a data-driven business are internal expertise for data utilization, appropriate technological infrastructure, and data governance across different business units. This is the most advanced stage in the maturity model.

Sjödin et al. (33) categorized digital maturity levels from the perspective of smart factories and categorized the key activities that support the development of smart factories by maturity level to create a smart factory maturity model.

This maturity level is highly correlated with understanding the technological requirements for a smart factory concept and developing a vision for connecting various systems. This vision creates the foundation for groundwork and smart factory implementation.

At this stage, organizations must create models for structured data collecting and sharing to facilitate the development of improved data management practices and processes that enable efficient storage and utilization of the increasing amount of production data being collected.

This maturity level yields the beneficial effects of collecting and communicating data. In this stage, organizations build competencies for real-time process analytics and optimization. The focus shifts toward benefiting from the data and system.

As the factory reaches the apex of the maturity model, continuing the focus on smart and predictable production requires continuous innovation and improvement. Efforts to build predictability in manufacturing make it increasingly possible to know what to expect, leading to greater production reliability and profits. This stage includes the potential of developing processes for utilizing data analytics and visualization for real-time decision-making and clarifies how visual representations of activities in the factory help key decision-makers adapt to the need for adjustment and improvement. Another process development in this stage involves creating proactive processes for predicting, forecasting, and planning future production. Production activities should be planned in a proactive environment with a focus on predicting future requirements.

This study explores the maturity stages of the digital transformation of Taiwan’s motherboard industry. We integrated them into a digital transformation maturity assessment model based on relevant literature. Digital transformation is divided into three stages, namely, digitization, digital optimization, and digital transformation. Following the creation of the assessment model, this study summarizes the value chain of the motherboard industry. According to the definition provided in the literature, relevant characteristics and dimensions are used as the basis for the evaluation criteria of fuzzy AHP. This study summarizes 7 dimensions and 21 evaluation criteria, as shown in Table 2. Then, this study clarifies the evaluation criteria by comparing them with industry data, and it selects and revises the criteria that align more with the current situation.

In this study, the scoring criteria used in the maturity assessments of digitization in the first stage and digital optimization in the second stage are shown in Table 3.

In this study, the minimum thresholds of assessment scores change with the core of the industry. For example, the motherboard industry must achieve a minimum threshold of four points in the manufacturing dimension and a minimum threshold of three points in the remaining dimensions. The assessment thresholds are shown in Table 4.

Saaty (42) proposed an AHP to analyze decision-making problems with complex and multiple evaluation criteria under uncertainty and find consistency in the chaotic decision-making process. Saaty (43) stated that AHP obtains the weight and execution priority of each criterion by constructing pairwise comparison matrices, comparing the criteria in the upper level with each criterion in the lower level individually, and identifying the degree of influence of one level on the other. Traditional AHP has the problem of ambiguity. Buckley (44) combined AHP with fuzzy theory and proposed the fuzzy analytic hierarchy process (FAHP).

Subsequently, many scholars used FAHP to conduct various studies. Cheng et al. (45) combined FAHP and Fuzzy Delphi to study enterprises’ choice of Fourth-Party Logistics (4PLs) in uncertain and complex business environments. Borjalilu and Ghambari (46) combined AHP and fuzzy theory to help factory maintenance staff determine the optimal maintenance strategy for each component of different equipment.

Establish the criteria for the study based on the literature review. Form an expert group and utilize the industry experience and suggestions of experts to adjust the dimensions and criteria, thereby revising the formal questionnaire of this study.

Establish a pairwise comparison matrix A and use each level as a benchmark to calculate the weight values of the evaluation criteria A_1, A_2, ⋯, A_n in the next level after the pairwise comparison. Each level’s relative importance of criteria can be expressed as a_ij (i, j = 1,2,⋯,n). Subsequently, place the comparison results of the evaluation criteria on the upper right of the main diagonal of matrix A and place the reciprocals of the upper right values on the lower left of matrix A. Because the main diagonal is the self-comparison (i = j), the criterion value is one. The comparison matrix is shown in Equation (1).

where a_ij 1/aji [   1             a12         ⋯       a1n1/a12      1             ⋯       a2n    ⋮            ⋮           ⋱       ⋮1/a1n    1/a2n     ⋯     1]

Owing to the ambiguity of human decision-making, we define fuzzy semantic variables so experts can compare and assign scores to criteria in pairs. In addition to accurately reflecting the implied meaning of semantic variables, the degree of influence between criteria can also be evaluated. The triangular fuzzy numbers corresponding to the FAHP semantic variables are shown in Table 5.

After constructing the pairwise comparison matrix, we can use the eigenvalue method to obtain the eigenvector w_i or priority vector. Most matrices are inconsistent matrices. Referring to the four approximation methods of eigenvectors proposed by Saaty and Vargas (47), we use the row normalization method based on the row-wise mean, as shown in Equation (2), to increase the accuracy of the eigenvector calculation results. We then use the obtained eigenvectors to calculate the maximum eigenvalue λ_max, as shown in equation (3).

Because the values in the pairwise matrix are subjective scores assigned by experts, large amounts of evaluation criteria and varying levels may cause inconsistency between experts’ judgments on different criteria. Therefore, it is necessary to conduct a consistency check to judge whether the scores assigned by experts are within a reasonable margin of error. We can use a consistency index (CI) and a consistency ratio (CR) to check the consistency of the weights.

CI: Referring to Saaty (42), when CI = 0, the expert’s judgments are completely consistent. When CI > 0.1, the judgments are completely inconsistent. When CI < 0.1, the judgments are within the permissible error range. It is calculated using Equation (4).

CR: Saaty (42) mentioned that the CI values (that is, the value of n) generated by different levels also differ. We adopt the CR to judge whether the matrix is consistent (as shown in Equation 5) for the matrices with the same value of n. When CR < 0.1, it means that there is an acceptable level of consistency. If not, we should re-examine the correlation between levels and criteria.

To obtain the relative fuzzy weights of the overall evaluation criteria, Dubois and Prade (48) used the minimum (L_i), median (M_i), and maximum (R_i) values of each criterion in the questionnaire to construct and calculate triangular fuzzy numbers h represents number of experts, i represents number of evaluation criteria, and n represents the total number of experts, as shown in Equations (6)–(8).

Step 6: Normalize the triangular fuzzy number at each level

We should normalize the triangular fuzzy number obtained in Step 5 for a more rigorous and accurate calculation result. nL_i, nM_i, and nR_i represent the normalized triangular fuzzy numbers, as shown in Equations (911).

Because the triangular fuzzy number is not a clear numerical value, the normalized triangular fuzzy number needs to be defuzzified to facilitate the subsequent ordering of weights. Tzeng and Teng (49) indicated that the center of area (COA) method does not consider the preference of expert scores for defuzzification, finds the best non-fuzzy performance value (BNP), and converts each element into a weight, as shown in Equation (12). However, to ensure the defuzzified weight (BNP_i) of each level adds up to one, we should repeat the normalization process to obtain the final weight (NW_i) of each dimension and criterion, as shown in Equation (13).

After calculating the final weight of each criterion at each level in step 7, we should connect the weight of each dimension and criterion in the different levels to calculate the relative weight of the experts’ selection criterion, as shown in Equation (14). NW_j represents the weight of the jth criterion in the third level under the first level (target level), NW_i represents the weight of the ith dimension in the second level under the first level, and NW_ij represents the weight of the jth subcriteria in the third level under the ith criterion in the second level.

Through the above calculation of hierarchical connection, we can obtain the weight of each subcriterion of the overall hierarchy evaluation structure and prioritize important evaluation criteria.

This study uses G company as a case study for digital transformation maturity assessment. The company was established in 1986 and it mainly provides products such as home and business computers, computer peripheral components, host servers, and expansion devices. It is a leader in the motherboard industry. In the past years, it focused more on the research and development of key technologies and continued to increase its brand awareness and sales of products such as servers, AIoT (AI + IoT), notebook computers, monitors, and gaming peripherals. Under the impact of COVID-19 and given the fact that consumer experience has become the mainstream of the market, G company is following technology trends such as AI, edge computing, and virtual services, creating more efficient products, providing industry-leading products and services, providing cloud and 5G services for customers, and actively moving toward digital transformation and innovation.

In this study, according to the results of the three experts’ questionnaires in the pilot test, we calculated the average score of the digitization and digital optimization stages. An average score of one indicates that the degree of consensus is deficient. An average score of two indicates a low degree of consensus. An average score of three indicates a medium degree of consensus. An average score of four indicates a high degree of consensus. Finally, an average score of five indicates that the degree of consensus is exceptionally high. The pilot test results of the experts are shown in Table 6.

Regarding the digitization assessment, the average scores of all the 21 evaluation criteria in the 7 dimensions are equal to or above 3 in the evaluation of digital factors. This means that the experts reached a consensus on the evaluation criteria. Then, we calculated the average scores of the evaluation criteria of digital optimization, and all the average scores were equal to or above three. This means that the experts reached a consensus on the evaluation criteria of the digital optimization stage. In addition, 11 experts from the motherboard industry selected relatively objective factors of the digital transformation of the motherboard industry. This study compared the relative importance of two factors through the pairwise comparison of factors, integrated expert opinions to construct a fuzzy positive reciprocal matrix, calculated the fuzzy weights and normalized weights of factors at each level, and ranked them in the order of importance.

The analysis results in Table 7 show that manufacturing (0.2345) is the most critical evaluation dimension among the digital transformation maturity dimensions of the motherboard industry, followed by research and development design, customer demand, logistics warehousing, and relationship maintenance, after-sales service, and procurement management. In terms of overall ranking, material requirements planning (0.1322) is the most critical factor among the 21 evaluation criteria, followed by technical innovation (0.0963) and innovation management and collaborative development (0.0942). Hence, in the overall value chain, planning of material requirements is essential for digitization and digital optimization.

We use the normalized weights to analyze the main dimensions. Of the seven dimensions, C and B are necessary for the motherboard industry. Based on the analysis, in the overall value chain of the motherboard industry, the front-end value chain is relatively crucial for the motherboard industry in the digitalization and digital optimization stages. However, the analysis results of 21 evaluation criteria show that C1, B3, and B1 are the most important criteria for the motherboard industry. In addition, because of the current small and diverse needs of customers, flexible material requirements’ planning is the most crucial factor.

This study used the digital transformation assessment of the motherboard industry constructed using FAHP to understand the causal relationship between evaluation criteria. It used the hierarchical connection weight values obtained through the FAHP method presented in the section titled, “Analysis of Fuzzy AHP” to assess the digital maturity of the motherboard industry (Table 8). It provides a review of the digital transformation maturity of the motherboard industry by experts of the industry. It also reviews the performance related to the implementation of digital transformation. The maturity assessment is based on the actual situation of the enterprise and the 21 assessment criteria defined in this study. Therefore, we need to examine whether the evaluation criteria are suitable for evaluating the enterprise.

In addition, because of different business types, the achievement rates of evaluation criteria differ by enterprise. Enterprises can also use this assessment to improve weak items. Therefore, it can effectively solve the pain points of business operations. In this study, the maturity assessment shown in Table 8 was distributed to the managers of company G, and the weighted scores were calculated and ranked. Managers assigned a score from 1 to 10 for each evaluation criterion. The discussion and analysis results are as follows.

Table 9 shows that the maturity assessment score of the motherboard industry assigned by Company G is 8.203. According to the classification of maturity level in this study, it can be classified as “very good maturity.” This means that although the industry has achieved good operational results in all aspects, there is room for improvement in certain aspects. Overall, the results show that Company G is in the “digital optimization” stage.