Digital transformation enablers in high-tech and low-tech companies: A comparative analysis

Abstract

Digital transformation (DT) is a complex, socio-technical, strategic, and often radical change influenced by various aspects, among which non-technological, organizational factors play significant roles. The literature on DT enablers suggests sensitivity to the organizational context, especially to the technology intensity of companies. This article analyzes the variations in non-technological DT enablers in high-tech and low-tech manufacturing companies. Through a systematic literature review, using coding techniques and cluster analysis, we created 20 aggregated enabler categories. With the Decision Making Trial and Evaluation Laboratory (DEMATEL) method, clustered DT enablers in high-tech and low-tech companies were compared. High-tech companies enable DT with the factors related primarily to organizational flexibility and dynamism. In contrast, low-tech firms propel DT with the enablers connected to more stable, traditional, managerial practices.

JEL Classification:

031 and 032

1. Introduction

The frequent and widespread introduction of new digital technologies in various areas of human life is accompanied by radical changes in organizations, industries, and societies (Kraus et al., 2021). The scale and pace of these technology-enacted changes force companies not just to implement new technological advancements into their processes and products, but also to drastically transform the very principles of doing business. As a result, digital transformation (DT), widely understood as strategic large-scale organizational change enabled by the integration of digital as well as non-digital technological innovations (Jones et al., 2021), has stimulated a large set of studies, with both theoretical (Loon and Quan, 2021) and practical implementations (Chawla and Goyal, 2022).

DT is becoming a strategic imperative for all industries and sectors of the economy, and particularly so where globally competitive forces are sharp, such as for companies in the manufacturing industries. The significance of DT for manufacturing companies is reflected in the nationwide programs initiated by many countries. The “Advanced Manufacturing Partnership” in the United States, “La Nouvelle France Industrielle” in France, “Future of Manufacturing” in the United Kingdom, and “Digital economy” in Russia are just a few notable examples of such programs (Liao et al., 2017).

The benefits that DT can potentially bring about to manufacturing businesses are multifaceted and include improved productivity and reduced costs (Hess et al., 2020), redesigned products and services (Nambisan et al., 2017), reinvented supply chains (Choudhury et al., 2021), enhanced innovativeness through better decision making (Carlsson, 2018), and new competitive advantages (Leão and Da Silva, 2021). At the same time, DT is associated with high risks and disruptions for companies and is characterized by meager success rates. According to Morakanyane et al. (2020), only 20% of companies succeeded in realizing full value from DT. These mixed results are conditioned by DT’s complex nature, dependent on the interplay of various socio-technical factors that are not fully within managers’ control (Nadkarni and Prügl, 2021; Wiggins et al., 2020). That is why there is an increasing emphasis on research investigating barriers, enablers, antecedents, drivers, success factors, and other ingredients of DT grouped into various sets and classifications (Ghobakhloo and Iranmanesh, 2021; Gölzer and Fritzsche, 2017; Kohnke, 2017; Moreira et al., 2018; Schallmo et al., 2017; Stentoft et al., 2021; Trenerry et al., 2021; Wolf et al., 2018; see Appendix 1 for a complete view). We, therefore, contend that there is a lack of coherence between different classifications of enablers, and there is a need to systematize them in a comprehensive structure.

One of the sources for different views on DT enablers is that they are analyzed from different angles and contexts. First, some authors focus more on particular technologies enabling DT in companies (Schallmo et al., 2017), whereas others also include a broader range of non-technological, primarily organizational, business-oriented factors (Moreira et al., 2018). DT can be considered to encompass both technical and non-technical enablers and success factors, drawing on the rich traditions of socio-technical systems (Boeing et al., 2020). Second, one of the dimensions that distinguish companies by their readiness for DT is the technological or R&D intensity, traditionally measured as the ratio of technological research and development expenditure to its revenues (Galindo-Rueda and Verger, 2016). Moreover, technology intensity influences many aspects of corporate innovativeness and related capability factors (Dong et al., 2021). Therefore, low-tech and high-tech manufacturing companies are expected to have different conditions for DT and different sets of DT enablers relevant to their contexts. Third, the variety of DT enablers observed in the current literature is accompanied by a multitude of different interpretations. For DT, as a corporate-wide and often radical change, the factors related to organizational flexibility and agility and “change readiness” are important. According to Evans and Bahrami (2020), corporations facing the challenges of necessary radical change should possess and use dynamic capabilities of flexibility encompassing all functions and subsystems of a business.

This study focuses on non-technological DT enablers and aims to test and evaluate whether these enablers are different in low-tech and high-tech manufacturing companies. We approach this question with a methodology including the extensive review of the studies related to DT enablers, qualitative coding technique and cluster analysis of the identified enablers, and the Decision Making Trial and Evaluation Laboratory (DEMATEL) method (Si et al., 2018). Our research is based on the expert knowledge of DT professionals from Russian low-tech and high-tech manufacturing industries. The study contributes to contemporary knowledge by systematizing DT enablers relevant to manufacturing companies, identifying DT enablers typical for low-tech and high-tech companies, and interpreting the identified differences from the perspectives of organizational flexibility theory and practice (Sharma et al., 2010).

The rest of this article is organized as follows. In the following section, a concise review of theoretical concepts of DT, its enablers, and their sensitivity to technology intensity is provided, followed by the research methodology. Subsequently, the findings are presented and then discussed. Finally, we conclude by presenting the theoretical contributions, managerial implications, limitations of the study, and suggestions for further research.

2. DT and its enablers

There is extensive research on DT identifying the essential features of this complex phenomenon. Compared with other concepts related to “digital” research discourse, such as digitization, digitalization, and Industry 4.0, the concept of DT is more oriented to business and organizational facets of digital technology implementation than purely technical considerations. DT is associated with major business changes (Fitzgerald et al., 2013), new business models, innovations, and performance of whole enterprises (Westerman et al., 2011), changing the rules of the game within organizations and industries (Hinings et al., 2018), and integration of technologies and business processes (Liu et al., 2011). It goes beyond trivial automation, referring to fundamentally new capabilities (Martin, 2008), change in a whole value creation process (Gölzer and Fritzsche, 2017), change in the business logic and model of a firm (Li et al., 2018), and the transformation of the entire company and its ways of doing business (Amit and Zott, 2001). DT goes beyond applying particular technologies in specific organizational processes. It is “a company-wide phenomenon with broad organizational implications in which, most notably, the firm’s core business model is subject to change through digital technology” (Verhoef et al., 2021: 892). Therefore, from a discipline perspective, DT fits more closely to strategic management than technology management. As a result, it is sensitive to many non-technological influences.

Such a multidimensional endeavor as DT cannot be implemented with only traditional management of organizational assets based on the formal strategies, plans, procedures, and commands. Managing DT is more like a choreography or orchestration of many internal and external factors than pure administration of a company’s resources (Chen and Tian, 2022). The current research focuses on identifying DT’s barriers, enablers, antecedents, contextual situations, and success factors. However, the current knowledge in this area is unsystematic. First, some researchers consider only digital technologies (Schallmo et al., 2017) as enablers, whereas others pay more attention to non-technological, organizational factors (Stentoft et al., 2021) or combine both technological and non-technological enablers (Moreira et al., 2018). Second, scholars approach DT enablers from various levels of analysis—national or regional economy (Alibekova et al., 2020), industry or a group of industries (Sanchis et al., 2019), or a focal company (Bouwman et al., 2019). Third, some authors construct complete classifications of enablers, sometimes overlapping (Ghobakhloo and Iranmanesh, 2021; Trenerry et al., 2021) and sometimes totally different (Kohnke, 2017; Wolf et al., 2018). Others identify only one or a couple of enablers (Dombrowski and Fochler, 2018). Finally, one group of researchers analyzes DT enablers relevant to a particular type of company or industry. For instance, Garzoni et al. (2020) investigated DT enablers in small- and medium-sized enterprises (SMEs). Büyüközkan et al. (2019) examined DT enablers in aviation companies. Some other authors analyzed manufacturing companies in different sectors (Liere-Netheler et al., 2018). The identified enablers are different, but the reasons for the observable similarities or dissimilarities are not clear.

In sum, the current understanding and state of knowledge of DT enablers comprises many studies of varying quality and degree of rigor, from surveys to reviews, often without controlling or measuring factors such as the degree of technological sophistication within firms. To partially deal with this problem, we focus our study on the following:

• The identification and structurization of:

• ▪ Enablers relevant to manufacturing companies, because they are highly affected by digital technologies;

• ▪ Enablers pertinent to a company’s level of analysis, because companies are the most active agents of DT (even in global, national, or industrial contexts);

• Testing the hypothesis that DT enablers in low-tech and high-tech companies are different.

The last aspect needs additional clarification. Although DT is a strategic, company-wide venture, technologies play a significant role. It is reasonable to expect that DT enablers in companies are conditioned by technology intensity (Galindo-Rueda and Verger, 2016). This conjecture is supported by the research on DT (not explicitly related to its enablers) in companies with different technological levels. Siachou et al. (2021) studied the specificities of DT in traditional companies and concluded that absorptive capacity and strategic interdependence determine the outcomes of DT. Bertello et al. (2021) discovered DT barriers in low- and medium-tech firms related to administration/finance, organization, strategy, and collaboration. Borovkov et al. (2021) identified DT barriers specific to high-tech manufacturing firms. These barriers included outdated manufacturing technologies, limited IT infrastructure capacity, special software infrastructure availability, lack of financial resources, and lack of qualified specialists. Baber et al. (2019) found that effectuation capability plays a significant role in DT, specifically in high-tech firms in Japan. Therefore, the current research suggests that DT and its enablers are sensitive to firms’ technology intensity. However, there are very few studies explicitly comparing DT enablers in companies of different levels of technology intensity.

Addressing the research and literature gaps as mentioned above, we hypothesize that DT enablers vary for companies with different technology intensity and formulate our research question as follows:

• RQ. Are non-technical DT enablers in high-tech and low-tech manufacturing companies different?

Being a corporate-wide, complex socio-technical change, DT implies the vital role of dynamic capabilities related to flexibility, agility, and change management (Ghosh et al., 2021). Therefore, the answers to the research question are interpreted with particular attention to the role of the enablers associated with organizational flexibility and other dynamic capabilities relevant to corporate change management, across firms with varying technological intensity.

3. Method

The study is based on a systematic literature review, qualitative coding technique, cluster analysis, and the DEMATEL method. The flow diagram of the research is presented in Figure 1.

Figure 1.

Research methodology flow diagram.

The left part of the diagram represents the literature review conducted to identify the comprehensive list of DT enablers in manufacturing companies. It includes two steps. First, the existing literature reviews of studies of DT in manufacturing companies were reviewed to identify the proper search terms needed to find DT enablers. Second, the papers related to non-technological DT enablers in manufacturing companies are selected and all mentions of the enablers were identified.

The right part of the diagrams shows the structurization of enablers and their comparison in high-tech and low-tech companies. The identified DT enablers were structured based on the authors’ judgment and coding technique in NVivo^® software (QSR International Pty Ltd, 2018). Then the enablers were grouped into a smaller number of categories using a cluster analysis approach, based on the experts’ assessment of semantic distance between DT enablers. After that, the DT enablers in high-tech and low-tech manufacturing companies were compared with the DEMATEL method based on experts’ evaluations of the pairwise influence of DT enablers on each other and calculated quantitative measures for the differences between the two groups of DT enablers. A more detailed methodology is provided below.

3.1. Literature review for DT enablers’ identification

As DT enablers may be clothed in different linguistic forms (antecedents, drivers, stimulators, imperatives, just to name a few) (Lokuge and Duan, 2021), the appropriate search terms were constructed. As the first step, the existing literature reviews on DT published in 2019–2021 were searched in Scopus; as Jones (et al., 2021) suggested, it comprises a representative collection of papers on DT. A total of 80 items with “digital transformation” and “review” in titles were identified. With the exclusion criteria mentioned in Figure 1 (top-left part, “1.1. review of reviews to construct search query”), 48 papers were eliminated and 32 articles were read in full text. It was found that such terms as “digitization,” “digitalization,” “digitalizing,” ‘digitizing,” and “Industry 4.0” were typical for the research related to the technological aspects of DT. Only “digital transformation” refers to the strategic business-oriented nature of DT. Based on the most frequently used synonyms of “enabler” (such as “driver,” “driving . . .” “antecedent,” and “success factor”), the search query was developed as TITLE-ABS-KEY (“digital transformation”) AND TITLE-ABS-KEY (enabler OR driver OR driving OR antecedent OR “success factor”).

In the second step, the non-technological DT enablers relevant to manufacturing companies were identified. With the constructed search term, 994 items indexed in Scopus till 2021 were found. Based on the selection process and criteria shown in the bottom-left part of Figure 1 (“1.2. identification of non-technological DT enablers”), 49 papers were included. The final list consisted of 63 papers which also included 12 articles from the first step and 2 papers recommended by the reviews at the final stage of publication. The descriptive information on the final list is present in Table 1.

Table 1.

Descriptive information on literature reviewed.

	2012	16	17	18	19	20	21	22	Sum
Type of paper
Case study				2		2	1	1	6
Conceptual	1	1	2	2	2	4	2		14
Interview-based					1		5		6
Quantitative data analysis					1	1	1		3
Review				1	3	3	16	1	24
Survey-based				1	1	4	3	1	10
Country/region of coverage
Not specified	1	1	2	2	3	7	15	1	32
Australia				1			1		2
Cross-country					2	2	6		10
Europe					2				2
Nordic countries							1		1
Germany, Austria					1				1
Germany, Austria, Switzerland							1		1
China								1	1
Czech Republic						1			1
Denmark				1		1			2
Germany				2					2
Hungary							1		1
Italy						1	1	1	3
Kazakhstan							1		1
Portugal							1		1
US						1			1
Serbia						1			1
Type of source
Journal	1	1		1	4	10	22	3	42
Conference			1	5	4	4	5		19
Book chapter			1				1		2
Focal industry/companies
Not specified	1	1	1	3	4	5	11		26
Mixed (manufacturing included)				1	2	5	12	2	22
Aerospace			1						1
Aviation 4.0					1				1
Industry 4.0				1	1	3			5
SME						1	2		3
Global							1		1
Manufacturing				1					1
Medium, family owned							1		1
Metal and steel							1		1
Mixed, SME								1	1
Total sum	1	1	2	6	8	14	28	3	63

SME: small- and medium-sized enterprises.

3.2. Structurization of DT enablers

The list of all enablers included full synonyms, different wordings with the same or close meaning, and very different terms. The number of enablers was reduced in two steps.

Within the first step, the semantic similarity of the identified enablers was analyzed using a coding technique widely known in qualitative research (Saldana, 2015). The aim was to group all identified mentions of enablers into a smaller number of enablers with close meanings. Using NVivo 11^® and following the recommendations of Bazeley and Jackson (2013), autocoding techniques were performed and, as a result, eight broad themes were found. Then, each author subcoded the initial items by more detailed codes. All discrepancies were identified, discussed, and resolved collectively. As a result, from 480 mentions in papers, 63 DT enablers were formulated.

Then the number of enablers was reduced even further by applying clustering analysis. Authors of similar research preferred to deal with 10–20 factors (in Asadi et al. (2022) – 12 items; Raj et al. (2020) – 15; Machado et al. (2021) – 16; Rajput and Singh (2019) – 20). To avoid arbitrariness in categorizing and making the enablers’ structure more practice-oriented, experts from manufacturing companies were approached (see section 3.4 of this article) and asked to evaluate the pairwise semantic similarity between DT enablers. A 0–9 Likert-type scale was used, with 0 as “complete similarity” and 9 as “maximum dissimilarity.” The aggregate semantic distance matrix was constructed by summing up the matrices from individual experts. The resulting matrix consisted of pairwise meaning distance assessments with possible values from 0 to 108 (maximum distance). The left-bottom triangle of the matrix was filled with transposed values. The resulting total distance matrix was subjected to cluster analysis, which was produced in RStudio with basic functions “factoextra,” and “cluster” packages.

To identify the best clustering algorithm and the optimal number of clusters (within the interval from 10 to 20), internal measures, such as connectivity, Dunn index, silhouette coefficient, as well as stability measures, such as average proportion of nonoverlap (APN), average distance (AD), average distance between means (ADM), figure of merit (FOM), were calculated (Brock et al., 2008) using clValid function from the eponymous package in R. Comparing k-means, k-medoids (PAM), and hierarchical clustering algorithms, we found in 4 measures out of 7 the hierarchical clustering method with 20 clusters showed the best results. In Table 2 the optimal scores and associate metrics and numbers of clusters are highlighted.

Table 2.

Connectivity (min is the best), Dunn index (max), silhouette coefficient (max), average proportion of nonoverlap (APN) (min), average distance (AD) (min), average distance between means (ADM) (min), figure of merit (FOM) for k-means, k-medoids (PAM), and hierarchical clustering algorithms.

Method	Metric	Number of clusters
		10	11	12	13	14	15	16	17	18	19	20
Hierarchical	Connectivity	40.965	44.360	51.799	56.071	59.067	60.938	63.464	69.607	71.607	74.007	76.132
	Dunn	.548	.548	.620	.671	.673	.673	.673	.698	.698	.729	.774
	Silhouette	.211	.218	.231	.256	.246	.268	.264	.255	.250	.249	.243
	APN	.015	.039	.009	.007	.004	.005	.016	.002	.008	.003	.003
	AD	119.516	115.004	106.511	100.671	96.506	92.196	89.613	85.451	83.597	80.378	77.948
	ADM	2.878	7.068	1.634	3.565	.896	1.887	3.205	.829	2.046	.753	.730
	FOM	12.913	12.789	11.974	11.745	11.302	11.192	11.175	10.800	10.792	10.528	10.412
Kmeans	Connectivity	47.379	48.614	56.193	57.873	59.067	60.938	63.464	69.607	71.607	74.007	76.132
	Dunn	.590	.590	.580	.665	.673	.673	.673	.698	.698	.729	.774
	Silhouette	.223	.216	.234	.255	.246	.268	.264	.255	.250	.249	.243
	APN	.022	.065	.032	.021	.011	.012	.023	.010	.015	.011	.012
	AD	118.849	115.867	106.922	101.197	96.923	92.580	89.998	85.821	83.967	80.783	78.415
	ADM	3.872	10.932	6.063	5.653	2.396	3.306	4.623	2.239	3.456	2.236	2.398
	FOM	12.846	12.765	12.076	11.822	11.444	11.313	11.309	10.908	10.901	10.640	10.606
Pam	Connectivity	55.318	57.120	57.148	57.466	61.989	70.165	74.015	76.110	81.938	84.063	86.063
	Dunn	.404	.404	.412	.644	.656	.564	.564	.577	.600	.600	.600
	Silhouette	.240	.253	.264	.285	.271	.254	.246	.246	.237	.229	.220
	APN	.054	.043	.048	.019	.020	.041	.028	.012	.018	.023	.033
	AD	117.246	111.733	106.936	99.651	95.272	94.259	90.874	85.951	83.909	81.984	79.824
	ADM	10.356	9.446	9.678	3.872	3.821	8.857	7.202	2.233	4.422	5.502	6.598
	FOM	12.748	12.496	12.320	11.552	11.285	11.277	11.145	10.695	10.675	10.648	10.633

APN: average proportion of non-overlap; AD: average distance; ADM: average distance between means; FOM: figure of merit.

3.3. Comparative analysis of DT enablers with DEMATEL

To compare 20 clustered enablers in high-tech and low-tech companies, the DEMATEL method was used. DEMATEL is an effective method for analyzing relations between factors within complex systems (Si et al., 2018). It has been successfully applied to rank elements of systems in terms of their importance and influence (Raj et al., 2020). DEMATEL is widely used in similar studies of enablers, antecedents, or barriers in different contexts (Asadi et al., 2022; Wang et al., 2021), including the fields related to DT and Industry 4.0 (Machado et al., 2021; Nimawat and Gidwani, 2021; Raj et al., 2020; Rajput and Singh, 2019). It has also been employed in the comparative studies of enablers and barriers (Jafari-Sadeghi et al., 2022). The algorithm of classical DEMATEL is well documented and the research followed the procedures described in Si et al. (2018). A brief explanation needed for the interpretation of DEMATEL results is provided in the next paragraph.

Besides DEMATEL, the contemporary research uses different methods to prioritize and reveal relationships among factors. To investigate enablers in the context of DT, scholars effectively use also MICMAC, ISM, TISM, AHP, and ANP separately or in different combinations (Adebanjo et al., 2022; Jain and Ajmera, 2021; Krishnan et al., 2021; Makki and Alqahtani, 2022). However, AHP does not consider the indirect effects and assumes that factors are independent (Gandhi et al., 2015), which is not relevant for this study. ANP may not be preferred because of its complex use and understanding (Makki and Alqahtani, 2022). TISM is very good in the qualitative and interpretative investigation of factors’ interaction (Sushil, 2012), which is not needed in the present research. A combination of ISM and MICMAC seems to be very close to DEMATEL because it identifies priorities and relations between factors. Although, ISM/MICMAC technique measures relations based on pairwise existence–absence evaluations and prioritizes factors in terms of discrete levels (Agrawal and Narain, 2021). DEMATEL can evaluate relations with the Likert-type-like scale and prioritize factors based on continuous scales, which makes this method more flexible and appropriate for the finer comparative analysis.

Within DEMATEL, experts evaluate the pairwise influence of factors on each other. Using a specific algorithm, these estimates are aggregated and converted into a square total relation matrix where each factor has in rows the assessments of its influence exerted on other factors and the influence received from other factors in columns. Then, each factor is measured with “prominence” and “relation” metrics based on the total relation matrix values. “Relation” (denoted as r—c, where “r” stands for “row” and “c”—“column”) is calculated by deducting the values in a column of a factor from the values in a row. “Relation” shows the net effect that the factor contributes to the system. Formally, it is the strength of influences exerted by a factor minus the strength of influences received. Positive “relation” means that a factor influences more than depends on others. Factors with positive “relation” are often called “causes” and negative “relation”—effects. “Prominence” (r + c) is calculated by summing each factor’s values in columns and rows. “Prominence” indicates the overall strength of the factor’s influences received and exerted. The higher the “prominence,” the more prominent the role of the factor can be considered.

Following the DEMATEL procedure, the evaluations of the pairwise influence of DT enablers (traditional scale from 0—no influence to 4—maximum influence) were collected from two expert panels (see section 3.4 of this article). The values from individual experts were averaged in two direct-influence matrices (for two panels). Then, two total relation matrices were calculated, “relation” and “prominence” for each enabler separately for high-tech and low-tech companies.

The differences in DT enablers were visually analyzed using influential relation maps with “prominence” and “relation” as axes. To measure the difference between DT enablers in low-tech and high-tech firms quantitatively, rank-correlation coefficients (Kendall’s tau-b and Spearman’s rho) were calculated. It was assumed that if the enablers have similar positions on the “relation” or “prominence” axes in both manufacturing companies, correlation coefficients will show a significant positive correlation. Otherwise, coefficients can be close to zero (enablers have different significance in terms of “prominence” and “relation” in two groups of firms) or negative (enablers have quite the opposite significance in two groups). The same coefficients were used to measure consistency of experts’ pairwise evaluations of factors mutual influences. The DEMATEL procedure was applied to experts’ individual matrices. Strong correlations of both prominence and relation were expected for in-group data.

Based on the visual and statistical comparison, the conclusions regarding the research question of whether DT enablers in low-tech and high-tech manufacturing firms are different were made. To validate the findings, they were presented at a panel discussion as part of practical conference dedicated to DT. The feedback did not contradict the findings and was incorporated into the discussion part of this article.

3.4. Expert panel

Scholars do not unequivocally define the size of expert panels appropriate for consistent and valid DEMATEL application. However, Adegoke et al. (2021) noted that “there is no limit (particularly the lower limit) to the number of experts to consider in the DEMATEL method.” Previous research suggests that 4–12 experts for a panel can be an appropriate number, provided that all experts have deep knowledge in the relevant areas and their selection follows applicable criteria (Jafari-Sadeghi et al., 2022; Raj et al., 2020; Shieh and Wu, 2016). The expert selection criteria in this research were based on previous studies (Jafari-Sadeghi et al., 2022; Peleckis, 2021; Raj et al., 2020) and included age (minimum 35 years old), education (minimum bachelor’s degree), work experience (minimum 5 years in a manufacturing company, either low-tech or high-tech), DT-related management experience (minimum 3 recent years of working as an IT-director, chief digital officer, or similar), recognition by other professionals (minimum two recommendations from professional bodies, such as Big Data Association, Digital Transformation Association of Russia, etc.) and a broad knowledge (knowledge of DT projects in a minimum of 5 companies, apart from their own, from the same industry). All experts worked for Russian manufacturing companies, but had the international experience 3 years and more (see Table 3). We selected 12 experts—6 with experience in DT in low-tech manufacturing companies and 6 in high-tech firms. The classification of companies as low-tech or high-tech manufacturing was based on Galindo-Rueda and Verger (2016).

Table 3.

Expert panel profile.

Age	Work experience in manufacturing	Industries (SIC code)	Work experience in DT	DT-related positions
48	12	Leather goods (3199)*	8	IT Director
45	8	Dairy products (202), beverages (208)*	4	Head of Data Management
38	9	Furniture (5021)*	9	IT Director
41	6	Furniture (5021)*	5	Chief Information Officer
48	15	Petroleum refining (291)*	4	Chief Digital Officer
51	18	Petroleum refining (291)*	5	Head of Data Office
50	8	Electronic components and accessories (367)	9	IT Director
41	10	Electrical goods (506)	8	IT Vice President
46	6	Pharmaceutical preparations (2834)	4	Chief Digital Officer
36	8	Pharmaceutical preparations (2834)	3	Head of Transformation Office
55	16	Aircraft parts and auxiliary equipment (3728)	5	Chief Information Officer
33	7	Aircraft (3721)	3	Chief Digital Transformation Office

*

Low-tech industries.

The experts were approached through video conferences in two stages. First, they provided their judgments for DT enablers’ clusterization (see section 3.2). Here, they acted as one group. Second, they participated in the DEMATEL analysis (see section 3.3), where they formed two panels. After each conference, the data received were processed (cluster analysis in section 3.2 and DEMATEL analysis in section 3.3) and the results submitted to the experts for their feedback. Finally, 8 out of 12 experts participated in the panel discussion organized to validate the overall findings.

The unique contributions of this study lie in its rigorous analysis of primary and secondary data, differentiation between high versus low technology firms, and hence the more fine-grained utility of the results that can be of practical use by those different types of businesses. For example, while Jones et al. (2021) provided a recent and useful foundational review of drivers and barriers, their study was a qualitatively constructed literature review, free of analytics and not comparing high- versus low-technology contexts, but rather a summary of past studies. Our research uses cutting edge analytics, builds on the large number of previous qualitative reviews, and provided new data from expert practitioners in both low- and high-technology companies in order to validate what was proposed by previous purely academic studies such as Jones et al. (2021). Furthermore, the present study used DEMATEL and related tools to ensure analytical rigor of the findings, in support of both academic and practitioner judgments. This has resulted in our study being able to identify both major and minor causes and effects of DT enablers, clearly showing for the first time the nature and extent of how these differ across technology sophistication levels. In sum, we build on the previous qualitative literature reviews (Jones et al., 2021) that aggregated and qualitatively identified drivers and barriers, to now provide fine-grained elaboration of these, segmented, informed by expert practitioners, quantified, and parsed in novel ways.

4. Results

In 58 papers, 480 mentions of non-technical enablers relevant to manufacturing companies were found. Purely technical enablers (such as artificial intelligence, robotics, the Internet of Things, etc.) were excluded. The enablers of the mixed socio-technological nature, such as technology management, and digital technology maturity, were included. The spelling was taken from the sources. Some authors provided the groupings of enablers. In these cases, broad categories were mentioned in round brackets after the enabler (for instance, “Collaboration (Human)”). Some authors used rather specific terms. In these instances, short comments in squared brackets were added (e.g., “Reducing transaction costs [expected benefits]” or “Brand permission [brand permission defines the limits of customers” willingness to accept a familiar brand name in new marketplace situations]’). In all, 480 mentions were reduced to 63 enablers, then grouped into 20 clustered enablers based on the experts’ opinions. The engagement of the experts at this stage helped confirm the correctness of 63 DT enablers and made the group of experts closely familiar with the meanings of the enablers, which was crucial for the next stage of research. The results of clusterization are shown in Figure 2. The detailed structure of DT mentions, enablers, and clusters is given in Appendix 1. With “E,” 63 enablers were marked, and “C”—20 clustered enablers.

Figure 2.

Cluster dendrogram for DT enablers.

Following DEMATEL procedures, experts evaluated 20 enablers in their pairwise mutual influence. The averaged direct-influence matrices for high-tech and low-tech companies are presented in Appendix 2. See Tables 5 through 14 for further details.

Based on the total relation matrices, “prominence” and “relation” for two groups of enablers were calculated and two influential relation maps shown for high-tech and low-tech firms were built. To make the visual analysis of DT enablers clearer, both the maps were merged into one with a horizontal axis representing “prominence” minus average “prominence” for a whole group of enablers as follows

$${(R+C)}_i^{*}={(R+C)}_i-\frac{{(R+C)}_i}{{\sum }_{i=1}^{20}{(R+C)}_i}$$

(1)

where (R + C)_i is the “prominence” of the ith enabler.

The merged influential relation map for enablers in both types of companies is shown in Figure 3.

Figure 3.

Merged influential relation map for DT enablers in high-tech and low-tech manufacturing companies.

According to the traditional interpretation of the DEMATEL influential relation map, the enablers above the horizontal zero line are called causes, and those below the line are effects. Besides, the enablers at the right of the vertical zero line can be interpreted as prominent or major and, to the left of the line—as minor ones. Gray circles show DT enablers for low-tech companies. Black dots depict enablers for high-tech companies.

A close look at the merged influential relation map helps to note that only 4 out of 20 DT enablers are in the same areas in both types of companies (see Figure 4), and 7 enablers have positions in the opposite quadrants (see Figure 5). Five enablers have different positions in terms of their “relation” (see Figure 6). Four enablers occupy different positions in terms of “prominence” (see Figure 7).

Figure 4.

Merged influential relation map for DT enablers with similar positions for low-tech and high-tech companies.

Figure 5.

Merged influential relation map for DT enablers with significantly different positions for low-tech and high-tech companies.

Figure 6.

Merged influential relation map for DT enablers with different positions with respect to “relation” (R − C).

Figure 7.

Merged influential relation map for DT enablers with different positions with respect to “prominence” (R + C)*.

To illustrate interaction between DT enablers, digraphs for two groups of companies were built according to the standard DEMATEL procedure (Figures 8 and 9).

Figure 8.

Digraph for high-tech firms’ DT enablers.

Figure 9.

Digraph for low-tech firms’ DT enablers.

Threshold values for high-tech and low-tech firms were .14 and .10, respectively. To make enablers and links more visible the enablers were slightly moved from their exact positions. Lines without arrows show mutual influence.

Digraphs are the additional tools to illustrate the prominence (r + c) and relation (r − c). Relation is visualized by the ratio of outbound links to inbound. Causes are characterized by the prevalence of outbound links, whereas effects—by dominance of inbound links. Prominence is characterized by the number of all links.

Based on visual inspection, the DT enablers in low-tech and high-tech companies are very different. Correlation coefficients support this impression. Table 4 shows the results of the calculations of coefficients, their low and upper bound values. p-values were found for the hypotheses that coefficients will be higher than the thresholds associated with a moderate positive correlation (.2 and .3 for Kendall’s tau-b and Spearman’s rho, respectively).

Table 4.

Rank-correlation coefficients of DT enablers in high-tech and low-tech companies.

	Lower bound (.95)	Point estimate	Upper bound (.95)	Threshold value	p-value
“Relation”
Kendall’s tau-b	–.3015	–.0316	.2384	.2	.0401
Spearman’s rho	–.5548	–.1489	.3144	.3	.0081
“Prominence”
Kendall’s tau-b	–.3253	–.0105	.3042	.2	.0769
Spearman’s rho	–.4652	–.0286	.4193	.3	.0679

For the “relation” and “prominence” points, estimates of correlation coefficients are well below the threshold associated with moderate positive correlation. p-values for “relation” are less than .05 and for “prominence”—less than .10. Hence, we found that the correlation between “relation” and “prominence” of DT enablers in high-tech and low-tech is negligible.

To prove the in-group consistency of experts’ judgments, DEMATEL was performed for individual expert direct-influence matrices. Correlations between experts of the same group were measured with the same coefficients of Spearman and Kendall. The results (see Appendix 3) showed positive and strong within-group correlation in terms of relation and prominence, which suggests good consistency of experts’ evaluations.

5. Discussion

Based on the above findings, it can be concluded that high-tech and low-tech manufacturing firms do have different sets of DT enablers. The most critical enablers (quadrant I) group includes only one enabler common for these two types of companies. The minor effects (quadrant IV) can be considered as non-important enablers or not enablers. Here, there are no common factors. In two other quadrants, there are only three enablers with similar importance in both groups of companies. The observed differences are worth noting for both theoreticians and practitioners and can best be interpreted by examining the enablers in all quadrants.

Starting from the most critical enablers (quadrant I), it can be seen that in high-tech companies, DT is promoted by the flexible and flat organizational environment (C13) characterized by a high level of empowerment and autonomy, outstanding collaboration, efficient teamwork, and by flexibility and change capabilities (C1) encompassing ability, innovative and digital culture, ambidexterity, and dynamic capabilities; a critical role of leadership capabilities and commitment from top managers (C10); the accent on learning and development (C11), innovativeness (C9), and to a lesser extent, customer orientation (C2) and development of partnerships and ecosystems (C14). Key DT enablers in high-techs look more connected with the “soft,” dynamic capabilities (communications, learning, leadership, change management, teamwork, customer orientation) and organizational flexibility (innovativeness, agility, flexibility). This finding is consistent, to some extent, with Acharya (2019), who found the notable impact of organizational flexibility on the performance of high-tech e-learning providers.

At the same time, low-tech firms rely more on more formal, tangible, “harder” elements—strategy (C19), corporate management (C17), and managerial control (C12). So it seems natural that for both types, the most critical DT enablers include mostly the internal resources and capabilities—these factors are under the control of companies and can be used as tools to influence other elements, even though they are quite different in nature between high- and low-technology companies.

From digraphs, it is visible that the most influential major causes for high-tech companies relate to dynamic capabilities in terms of innovativeness (C9), learning and development (C11), and flexible and flat organization (C13). They are characterized by the biggest number of all links among enablers for both high-tech and low-tech firms.

Interpreting the observed difference from the resource-based view (Khin and Hung Kee, 2022; Liu et al., 2011), it can be stated that low-tech companies possess lower organizational flexibility because of their more conservative resource base (Evers, 2011; Ghosh et al., 2021). Their structure and processes tend to be more stable. Their management systems are based on more formal administrative capabilities, procedures, and approaches. As a result, they cannot fully and properly use dynamic capabilities to propel DT. They may, however, enforce DT initiatives using more stable, control-oriented organizational competencies. To survive in unpredictable and highly competitive markets, high-tech companies develop their dynamic capabilities continuously (Yu-Yuan Hung et al., 2007). For high-tech companies, it is more typical to rely on flexible rather than hierarchical structures, apply leadership skills rather than administrative management procedures, and organize processes and projects based on agile teamwork rather than formal methodologies. That is why high-tech companies consider traditional managerial capabilities as less appropriate for DT. The enablers that are important for low-tech companies, such as “Strategy for DT” (C19), “Managerial control [expected benefits]” (C12), and “Finance resources availability” (C7), are among the unimportant factors (quadrant IV) for high-tech firms.

In quadrant II, the enablers can be perceived as the expected elements from DT results. Interestingly, low-tech companies have more enablers with (expected benefits) remark. They also have more enablers in this quadrant overall. In quadrant II, low-tech companies also have “Agility and change capability” (C1), “Innovativeness” (C9), and “Learning and development” (C11). These enablers are among the dynamic capabilities mentioned by high-tech companies as major causal enablers. It seems that low-tech companies consider DT as the vehicle for developing their dynamic capabilities, corporate agility, and innovativeness. The major effects of DT are connected with enhanced agility, change management, innovativeness, and organizational learning.

In high-techs, 3 out of 4 enablers with (expected benefits; (“Financial results” (C8), “Managerial control” (C12), “Process efficiency” (C15)) are among unimportant factors (quadrant IV). It does not mean that they are not interested in business results. Different expected benefits drive DT here. However, high-tech companies are interested in external business results, such as improved competitiveness (C18) and products suitable for digital trends (C16). High-tech companies, already being agile and flexible, use DT to increase the digitalization of their products and, as a result, improve competitiveness.

In quadrant III, there are causal but not major DT enablers. From the viewpoint of open systems theory (Roth, 2019), these results seem quite reasonable. The organizational change implied by DT is influenced by external drivers, which are usually out of the full control of companies. As a result, companies view these DT drivers as the causes they need to adapt. External enablers mentioned in this quadrant are more like environmental conditions that motivate DT initiatives to happen. It seems that low-tech companies are more dependent on external factors than high-tech companies. Low-techs have more external enablers and higher causal power (“relation”).

Quadrant IV consists of minor DT enablers with low influence. Here, we can indicate that, in high-techs, DT initiatives seem to be less sensitive to external factors (C4, C5) and financial factors (C7, C8), less reliant on formal management tools (C12, C19), and less focused on internal process efficiency (C15). The last observation seems counterintuitive and needs some consideration. However, it can be explained by the strategic focus of high-tech companies, which is often more connected with product differentiation than internal efficiencies (Treacy and Wiersema, 1993). That is why high-tech firms see product readiness for DT as a major expected effect and process efficiency as an unimportant factor.

Comparison of digraphs reveals that minor effects for high-tech firms have more inbound arrows, than those for low-techs. It means that for high-tech companies these factors are less influential, whereas minor effects for low-techs are slightly more balanced in terms of inbound and outbound links. Hence, for low-tech companies, these minor effects are not totally marginal and should also took into consideration as important effects.

During the panel discussion the experts and other participants paid their attention primarily on major causes and effects (quadrants I and II). They agreed with the findings and notes that DT in high-tech companies is often stimulated by the availability of dynamic capabilities related to innovativeness, flexibility, and organizational learning. Low-tech firms rely more on traditional management practices concerning strategy and control. Low-techs are more dependent on finance resources and technology management, because in comparison to high-techs they are at lower levels of digital maturity and technology change readiness. With the help of DT, high-tech companies are searching for long-term competitive advantages, whereas low-tech firms are more interested in short-term results (process efficiency and financial results) and frequently consider DT as a tool to develop their dynamic capabilities with respect to higher innovativeness, flexibility, and adaptiveness. Unexpectedly for the participants, many external enablers were found in minor causes or effects for both groups. According to their knowledge, DT is significantly driven by the pressure from competitors, clients, and government.

6. Conclusion

This research identified and analyzed the differences between non-technical DT enablers in low-tech and high-tech manufacturing companies. Based on a thorough literature review, the comprehensive structure of DT enablers with 20 clustered categories was built. The comparative analysis with DEMATEL found that high-tech and low-tech firms have significantly different sets of DT enablers. The interpretation of the differences concludes that both high-tech and low-tech companies consider the resources and capabilities already in place as the most critical enablers. High-tech companies rely more on the enablers related to organizational flexibility and dynamic capabilities, whereas low-techs rely on enablers connected with more stable capabilities, considering higher flexibility as the expected outcome effect. Dynamic capabilities make high-tech companies more flexible and adaptive to external drivers. The lower flexibility of low-tech companies motivates them to see DT as the opportunity, first, to develop dynamic capabilities and enhance its flexibility and agility. Organizational flexibility and dynamic capabilities and structures play a significant role in DT in both types of companies, but this role is notably different.

This research deepens our understanding of the DT complexity in manufacturing companies. The DT enablers were systematized and structured. It was found that technological context plays a significant role in DT initiatives in manufacturing companies. It was revealed that DT is primarily driven by organizational flexibility and dynamic capabilities in high-tech companies. In low-techs, DT is stimulated by more stable and traditional resources and capabilities, while dynamic capabilities and flexibility are considered the major effects expected from DT.

Practicing managers may take note of the recommendation to activate DT with the enablers relevant to their technology intensity. Management should also identify the enablers pertaining to their company’s context before initiating DT endeavors. It seems suboptimal to propel DT in low-tech companies by using dynamic capabilities which are not well advanced in these companies. On the contrary, it seems more appropriate to enable DT with the capabilities existing in a company, even if they are more based on the formal, procedural, and traditional management practices.

Practitioners can directly use our findings to assess their DT readiness and propensity using the identified and validated factors that might apply specifically to their level of technological sophistication, using the causes and effects identified in this study to anticipate driers and barriers, and hence be more effectively able to implement DT, or choose not to, or indeed to choose how best to do so, for example to proceed in an evolutionary manner as against more aggressively, to conduct pilot implementations versus proceeding on a more widespread basis within the firm. Our study provides for such thoughtful anticipation and DT management by practitioners based on bringing together and triangulating the previous knowledge in a vast literature with current expert practitioner judgments, from both high- and low-technology environments. These differences that are our novel findings, between low- and high-tech firms, should also lead to anticipation of challenges when mergers or acquisitions occur between firms of substantially varying technology levels.

The major limitation of this study is that the expert panel was represented by a select group of Russian manufacturing companies only and not by all high-tech and low-tech manufacturing industries, more globally based. The conclusion about the importance of DT enablers was derived from applying only one DEMATEL method, whereas, in some studies, authors prefer to use several methods. Perhaps, fuzzy or gray DEMATEL methods may be more appropriate for a few experts, leading to different results in future studies.

The findings suggest that the “technology intensity” conditions DT enablers. However, the causes leading to the differences in DT enablers in high-tech and low-tech companies are not clear. Therefore, further research should look deeper into the interactions between the technology base of companies and particular DT enablers. It will also be fruitful to perform the empirical analysis of relationships between DT enablers used by companies and the results of DT. Finally, we propose that follow-up studies should involve deep case studies using our framework of high- and low-technology firms, questioning just “how” drivers and barriers impact on DT effectiveness. In refining knowledge through such further research, empirical studies that follow this one, whether surveys, mathematical models, or qualitative studies, should structure elements such as industry concentration, competitive dynamics and degree of government support, and economic maturity of development, in order to create even more fine-grained insights than in the present study.

Research Data

sj-xlsx-1-aum-10.1177_03128962231157102 – Research Data for Digital transformation enablers in high-tech and low-tech companies: A comparative analysis

Research Data, sj-xlsx-1-aum-10.1177_03128962231157102 for Digital transformation enablers in high-tech and low-tech companies: A comparative analysis by Titov Sergei, Trachuk Arkady, Linder Natalya, RD Pathak, Danny Samson, Zafar Husain and S Sushil in Australian Journal of Management

This article is distributed under the terms of the Creative Commons Attribution 4.0 License (http://www.creativecommons.org/licenses/by/4.0/) which permits any use, reproduction and distribution of the work without further permission provided the original work is attributed as specified on the SAGE and Open Access pages (https://us.sagepub.com/en-us/nam/open-access-at-sage).