Applying the Delphi Method to Assess Critical Success Factors of Digitalization While Sustaining Lean at a Lean Automaker

Abstract

This paper describes the critical success factors for digitalization in manufacturing within the framework of Industry 4.0 and Industry 5.0 while sustaining Lean and Lean-reinforcement links during this transformation within the actual transformation journey of an enterprise (Toyota Motor Europe). In this regard, 11 critical success factors for “digital transformation”, 9 critical success factors for “Lean sustainability”, and 12 reinforcement links (between Lean and digitalization) were identified from the published literature and 56 one-on-one discussions with Toyota Production System experts. Both Lean and digitalization frameworks were developed using the ontology method. Then, a survey with these experts was conducted, in which the Delphi method was used as a survey tool for an analysis, which was performed in three steps: an analysis of psychometric properties was conducted, a stability analysis of the Delphi method was performed, and the significance of non-consensus hypotheses of the results were examined. The results show that “top management commitment” is the most critical factor for digital transformations, whereas for Lean sustainability, it is “keep Genchi Genbutsu (go and see)”. It was found that digitalization impacts Lean very positively (high/strong or high) for a majority of the links and lean on the positive side for the remaining links. These findings can help manufacturing firms make informed decisions regarding minimal waste, lead times, and the right strategy for digitalization.

1. Introduction

Today, manufacturing faces the biggest challenges in its history. The first, second, and third industrial revolutions took place one after another, whereas today, the fourth and fifth industrial revolutions are almost simultaneous. While embracing Industry 4.0 (I4.0) and Industry 5.0 (I5.0), manufacturing should also preserve its existing gains and philosophy, which are formed under the term “Lean”. “Lean” was coined in 1988 [1], derived from the Toyota Production System (TPS) [2]. I4.0 was coined in 2011, and shortly thereafter, I5.0 was coined. While I4.0 addresses technologies, I5.0 addresses humans, society, the environment, circular economy, resiliency, and the boundaries of the planet [3]. These emerging values are not directly addressed by the I4.0 paradigm. It is also important to understand the significant shift from a technology-driven I4.0 to the more human-centric-driven I5.0 [4].

As manufacturing has been implementing technology-driven I4.0, I5.0 began to be understood as a paradigm that recognizes the power of the industry to achieve societal goals beyond jobs and growth. It aims to become a resilient provider of prosperity, by making production respect the boundaries of our planet and placing the wellbeing of the worker at the center of the production process. In 2015, the United Nations General Assembly (UNGA) created “Sustainable Development Goals” (SDGs) [5]. In 2021, the European Union (EU) formally called for I5.0 [3]. Globally, many countries have introduced similar strategic initiatives: the National Strategy for Advanced Manufacturing (USA) [6], Society 5.0 (Japan) [7], and Made in China 2025 [8]. Regardless of the name, the I5.0 paradigm is a reality. Therefore, in digital transformation, manufacturing should also consider the I5.0 paradigm and its concepts, mandates, and expectations introduced by policymakers and society alongside the I4.0 technical framework.

Among these I5.0 references, the European Commission identifies key root concepts in a way that all possible elements and terms can be defined and grouped under them. In our study, we refer to these aspects of the I5.0 paradigm (Figure 1). It can be said that I5.0 is the key enabler and mandate of digitalization (Dx) next to I4.0.

There are numerous challenges introduced by I4.0 and I5.0 that also need to be considered by Lean philosophy and practices.

Manufacturing is being transformed by technologies and is becoming more connected and data-driven. With digitalization, cybersecurity has become a key concept in sustainability [9]. Innovation is key to survival in today’s rapidly changing market. The manufacturing need for new skills (both technical and transversal) introduces significant challenges to the education system and manufacturing including personalized education [10]. Just-in-time hiring strategies need to change radically to form closer collaborations between manufacturing and education (joint programs and hybrid education). Privacy and data protection regulations impose strict rules, and companies may face significant fines [11]. Human–robot co-working (cobot) is a challenging task for enterprises [12]. Moreover, human augmentation technologies such as wearable devices, robots, and eyeglasses may go beyond ethical dilemmas and be subject to regulations. With the COVID-19 pandemic, companies began to recognize the value of digitalization, knowledge integration in both the horizontal and vertical directions for agility, quick reactions to customer expectations, and adaptations to new business models [13].

Societal megatrends require workforce diversity. In the meantime, the aging workforce needs to be re-skilled and upskilled. Companies’ efforts to reach new talent introduce more diversity into the workforce, where company culture should be ready for it. Innovation and human centricity have been identified as key concepts of digitalization in many articles and are well-received by policymakers (e.g., the EU [3]). The rapid proliferation of new technologies will continue to transform roles within manufacturing, such as data scientists, compliance officers, collaborative robot experts, and virtual operators. Overall, these efforts imply more investments in the workforce and society [14]. The environment and circular economy paradigm are becoming increasingly relevant as companies realize the real value and profitability of this new, sustainable way of doing business. It relies on several strategies to extend the product life cycle through reusing, recycling, remanufacturing, and redesigning circular products and materials [15]. Sustainability, resiliency, and the circular economy are key challenges that manufacturing companies should manage and develop strategies for with respect to their products and business models.

All of these changes include regulations introduced by policymakers, societal realities, expectations that manufacturing should consider for sustainability, and resiliency.

The term “Lean” refers to a state-of-the art level in manufacturing. Lean principles, or TPS philosophy, as shown in the popular temple model in Figure 2 [16], cannot be addressed by the I4.0 paradigm alone. Key terms of the TPS, the Toyota Way philosophy, such as “Waste Reduction”, “Levelled Production”, and “Respect for People”, have definitions under the I5.0 paradigm. Toyota aims to undergo a Dx transformation while reinforcing Lean and considers Dx as a combination of I4.0 and I5.0. Digital transformation is inevitable, but Lean philosophy should be sustained.

For the TPS, or Lean visualization, the two-pillar temple model is a very popular form used in many studies. However, this model often leads people to understand Lean as a toolbox from which to choose the appropriate “tools”. However, as elaborated by Ruttimann and Stockli [16], Lean is not a toolbox; it is a philosophy for manufacturing. “Lean thinking” is another widely used term in the literature to emphasize the management framework of Lean that is made up of a philosophy, practices, and principles. For our research, we need a better representation of Lean that shows all its aspects and allows us to define relations to the digitalization paradigm. In his Ph.D. thesis, Masai [17] utilized an ontology method to develop a comprehensive representation of Lean. “Ontology” allows us to show concepts, principles, elements, methods, and tools with synergistic links and a philosophy. Having a holistic conceptualized view will allow us to identify the relations (multi-domain matrix) between Lean and Dx accurately for impact evaluation and to assess Lean-reinforcement points and will allow experts to evaluate critical success factors (CSFs) more meaningfully with a visual representation.

Studies have examined critical success factors for implementing I4.0 technologies from organizational and transformational aspects [18,19], or for specific areas such as project management [20], or with a focus on a particular technological implementation [21]. Some studies have also looked at CSFs for I4.0-technology implementation while considering some Lean principles [22,23]. Additionally, some studies have included CSFs from I5.0 [24,25]. However, none of these studies have examined a comprehensive list of CSFs for all concepts of I4.0 and I5.0. No study has explored the CSFs for a company to sustain a Lean framework while transitioning to Dx. Our study aimed to identify CSFs for both Dx transformations and Lean sustainability, as well as to validate relations between Lean and Dx within the context of I4.0 and I5.0 paradigms to identify Lean-reinforcement links. Alves [26] mapped I4.0 technologies to I5.0 principles and Lean thinking. Alves also emphasized why Lean thinking is an essential mindset for I5.0 transformation, stating that “Lean thinking is a fundamental mindset to promote societal transformations and prepare society to properly adopt and use Industry 4.0 technologies. These technologies continue to rapidly develop and have started to evolve to Society 5.0 and Industry 5.0”.

Considering our research topic, the level of perfect Lean understanding; digital transformation experience; and short-to-long-term business, technology, and environmental vision were key prerequisites for the panel members. The survey needed to be conducted with experts, and opinions needed to be aggregated. The Delphi method was selected, as it addresses needs and provides capabilities as a survey method.

We believe that identifying critical success factors (CSFs) and the connection between digitalization and Lean practices will help manufacturing management determine the right strategy and agile approach for their transformation journey while also sustaining and reinforcing Lean principles.

The four key research objectives of this article are shown in Figure 3.

Section 2 discusses Lean and Dx ontologies, reinforcement links, Dx transformation CSFs (CSFs–DT) and Lean sustainability CSFs (CSFs–LS). In Section 3, the research methodology is discussed. The results are presented in Section 4. Section 5 discusses the implications for manufacturing companies. Finally, Section 6 concludes this study.

2. Literature Review

2.1. Lean Framework (Ontology)

There are several lean ontologies. Khanna [27] developed a lean ontology for a specific domain of supply chain management. Masai [17] developed a comprehensive lean organization framework. What makes this ontology more meaningful and fit for our study is that he worked at Toyota Motor Europe during his study. Therefore, we decided to use this ontology as a starting point. From this aspect, Uschold and King’s [28] ontology-development methodology was selected for the Lean-framework development. For the building stage, Masai’s Lean ontology was selected and discussed with experts one on one, and 9 main concepts were identified (

Table 1. Lean framework’s main concepts [17].

No	Concept
1	Kaizen
2	Customer First
3	Stakeholder Satisfaction
4	Hitozukuri (making people)
5	Safety
6	Monozukuri (making things)
7	Just in Time
8	Mieruka (visual control)
9	Jidoka (stop in time)

). The final Lean ontology, with principles and terms, is presented in Section 4.

2.2. Dx Framework (Ontology)

“Ontology” is used more widely for I4.0 than Lean. Ontologies were developed for specific industrial domains, processes, services, or manufacturing concepts. Kumar et al. [29] reviewed and classified existing ontologies for I4.0. Currently, there is no universally accepted ontology-development methodology. Uschold and King and Fernàndez-López and Gómez-Pérez’s methodologies are believed to be most representative [30]. In order to properly reflect I5.0 paradigms alongside I4.0 from the view of the TPS, we decided to use the method proposed by Mizen et al. [31]. This method is summarized as follows:

Step 1: information collection (literature review);

Step 2: creating a glossary (root concepts and taxonomy);

Step 3: presenting the relationship among terms under main concepts.

Because Dx is defined as I4.0 + I5.0, our literature review covers both. Most of the works on I4.0 mainly focus on technologies and tools and include some characteristics and concepts that overlap with I5.0 concepts as well. This is quite normal, as I5.0 is not a chronological continuation of or an alternative to I4.0. This was the result of the forward-looking exercise of the Fourth Industrial Revolution. These studies also show links and relationships between these domains. A list of studies, including linkage aspects, is given in

Table 2. Studies on I4.0 and I5.0 elements, characteristics, concepts, and links to Lean.

Study	Elements (I4.0)	Concepts and Characteristics (I4.0)	Concepts and Characteristics (I5.0)	Links and Relations (Lean—I4.0)	Links and Relations (Lean—I5.0)
Breque et al. [3]			✓
Taisch et al. [9]			✓
Taisch et al. [14]			✓
Taisch et al. [15]			✓
Frank et al. [23]		✓	✓	✓	✓
Alves [26]	✓	✓	✓	✓	✓
Unal et al. [32]	✓
Marinelli et al. [33]	✓			✓
Gallo et al. [34]	✓			✓
Mayr et al. [35]	✓			✓
Pereira et al. [36]	✓			✓
Dillinger et al. [37]	✓			✓
Suleiman et al. [38]	✓	✓	✓
Rossini et al. [39]	✓			✓
Treviño-Elizondo et al. [40]	✓	✓		✓
Johansson et al. [41]		✓	✓	✓	✓

.

By utilizing the outputs of the literature review, a glossary of the root concepts and taxonomy was created, and a Dx ontology was proposed to experts and discussed with them one on one. The feedback was collected and reflected. The Dx ontology was formed under 9 key concepts (

Table 3. Dx framework’s main concepts.

No	Concept	Main Paradigm
1	Innovation	I4.0 and I5.0
2	Human Centricity	I5.0
3	Resiliency	I5.0
4	Security (cyber)	I4.0
5	Digital Manufacturing	I4.0
6	Digital Engineering	I4.0
7	Knowledge Management	I4.0
8	Vertical and Horizontal Integration	I4.0
9	Sustainability	I5.0

). The final Dx ontology is presented in Section 4.

2.3. Linkages between Lean and Dx

Many attempts have been made to demonstrate the relationships and connections between Lean and I4.0 (Table 2). The common approach to illustrating relationships was comparing Lean tools to I4.0 technologies. However, TPS, or Lean, is not a toolbox but rather a philosophy. While forming the Dx ontology and identifying key concepts, we aimed to create a framework that also encompasses the Dx’s focus and vision. This approach allowed us to identify linkages properly, evaluate CSFs accurately, and identify Lean-reinforcement links. Based on the findings of the studies (Table 2), 29 relationships between Lean and Dx were initially proposed to experts. The feedback was collected from experts one on one and was reflected. A total of 12 links were identified and are listed in

Table 4. Multi-domain matrix (Lean versus Dx).

			Digitalization Concepts
			1	2	3	4	5	6	7	8	9
		Link Number (1–12)	Innovation	Human Centricity	Resiliency	Security (Cyber)	Digital Manufacturing	Digital Engineering	Knowledge Management	Vertical and Horizontal Integration	Sustainability
Lean Concepts	1	Kaizen	1	3					9
	2	Customer First	2	4			7	7
	3	Stakeholder Satisfaction	2	4		6	7	7			12
	4	Hitozukuri (making people)	2	4			7	7		11
	5	Safety		4		6	7	7
	6	Monozukuri (making things)		3	5		8	8
	7	Just in Time			5				10	11	12
	8	Mieruka (visual control)							9
	9	Jidoka (stop in time)		3			8	8	10

.

2.4. CSFs–DT for “Dx Transformation”

The CSF approach was developed by a research group at the MIT Sloan School, introduced by Rockart [42] and developed over time. As a management term, a CSF should identify a “limited number” of areas in which satisfactory results will ensure targeted performance for an object [43]. With more factors, the process becomes complex, and decision making, focusing on core factors, and reaching targets will be difficult and may not be even possible.

Dx is not only an implementation of technologies. While early studies focused mostly on I4.0 technologies and identifying CSFs around it, recent studies have extended CSFs beyond technology to management aspects related to transformation; societal factors concerning employees [23,41]; and environmental awareness for sustainability, resiliency, and circular manufacturing [26]. Bhatai and Kumar [24] identified CSFs for implementing I4.0 technologies in manufacturing, considering factors of I5.0 as well. Nurbossynava et al. [25] identified CSFs for Dx transformation in manufacturing. Leyh et al. [44] identified CSFs for digitalization projects. The World Manufacturing Foundation (WMF) suggested success factors with a global view in line with many case studies; numerous academic essays; and data from different sources in its 2018, 2019, and 2021 annual reports [8,10,13]. Kumar et al. [45] identified CSFs for a circular and sustainable supply chain. A summary of the literature review, which helped us identify CSFs, is listed in

Table 5. Studies on critical success factors and digitalization.

Study	Number of Domains	Number of CSFs
Breque et al. [3]	3
Taisch et al. [9]	10
Taisch et al. [14]	10
Taisch et al. [15]	10
Bhatai and Kumar [24]	9	26
Nurbossynava et al. [25]	3	22
Suleiman et al. [38]	24
Leyh et al. [44]	6	25
Kumar et al. [45]		14

.

Experts in the enterprise, who are targeting the “mobility company”, are thoroughly experiencing Dx. CSFs from the literature review (Table 5) were grouped and reviewed with experts one on one. The experts’ feedback (ideas, suggestions, and votes) was combined, and 11 CSFs were identified and are listed in

Table 6. Critical success factors (CSFs–DT) of Dx transformation.

No	Factor
1	Commitment of top management and transformational leadership styles
2	Teamwork to achieve common objectives
3	Organizational transformations (breaking silos, talent management, and integrations)
4	Human centricity (diversity and wellbeing of employees with all aspects)
5	Availability of financial resources and investment strategies
6	Public education transformations and collaborations with universities/education
7	IT transformations and roadmaps (cybersecurity, IoT, cloud, legacy migration, data science, AI, etc.)
8	People transformations (skilling, upskilling, and re-skilling); empowerment; and citizen development
9	External support (incentives and regulations)
10	External support (consultations)
11	Manufacturing/R&D transformations (circular manufacturing/designs, virtualizations, and vertical and horizontal integrations)

.

2.5. CSFs–LS for “Lean Sustainability”

As stated, there is a common view that Lean and Dx are like twins and complement each other. There are studies that analyze Lean and Dx integrations and impacts of Dx in selected areas. Frank et al. [23] and Powel et al. [46] discussed Lean and the integration of digital technologies from a people point of view. Margherita et al. [47] and Johansson et al. [41] shared care and tension points (dialectical perspective and paradoxes) during this integration. Chavez et al. [48] discussed digitalization’s role on the social performance of Lean. There are no studies on “how to sustain Lean while transforming to Dx” with a holistic conceptual view.

To properly identify CSFs for Lean sustainability, we first checked if our Lean framework is a good representation. For this purpose, we referred to studies using the Lean maturity model. Considering the Lean status of the enterprise, we decided to apply a stratified sampling approach. We followed the following steps:

• Identify stratified sampling clusters using (1) institutional, well-accepted maturity models [49,50,51]; (2) a recent study using the Lean maturity model with an extensive literature review [52]; and (3) an academic thesis study proposed after ‘a’ and ‘b’ [53].
• Map maturity-model dimensions with our Lean framework model.
• Evaluate the Lean framework model.

The selected literature is provided in

Table 7. Lean maturity models and dimensions.

Model	Institute/ Author	Year	Number of Main Dimensions	Number of Sub Dimensions
LESAT-1	MIT [49]	2001	14	54
LESAT-2	MIT [50]	2012	15	68
Shingo Model	Miller [51]	2012	4	30
LAT	Pakdil [52]	2012	8	62
LMM	Maasouman [53]	2014	7	35

. For 112 out of 115 dimensions of 26 studies, one or more mappings were identified for the Lean framework. Based on this result, we concluded that our Lean framework addresses Lean at a very good level.

The mappings were discussed with experts one by one and prioritized with a digitalization transformation view point. A total of 9 CSFs were identified and are listed in

Table 8. Critical success factors (CSFs–LS) of Lean sustainability.

No	Factor
1	Keep Genchi Genbutsu (go and see)
2	Never compromise jidoka (stop in time) to keep problems on the surface
3	Keep and enforce JKK (build in quality with ownership), which will be more critical with digitalization
4	Keep/watch autonomation (automation with the human touch), with people and ownership being a core element of Lean
5	Transformation in nature is a challenge, and visual control will be more important than ever for JKK and jidoka
6	Keep standardization and strictly watch, which is the base for kaizen and employee empowerment
7	Keep an additional focus on supplier development, pace, resiliency, and sustainability, which need to transform as well
8	Trust the data, but do not lose nemawashi (consensus building), and utilize it meaningfully to be agile
9	Trust the data, but keep problem solving and the PDCA culture, as Dx needs a learning organization with innovation

.

2.6. Hypotheses

It is evident from the published literature that CSFs are specific elements or action areas that a business must focus on to reach its strategic objectives. A Dx transformation while sustaining Lean is a strategic objective of manufacturing today. Understanding Dx transformations and Lean sustainability CSFs and their significance helps managers in organizations to identify the right transformation strategy with agility and minimum waste. Defined with the I5.0 paradigm, Dx is a more complete framework with environmental and human aspects and is also more comparable with Lean as a framework. Having a holistic conceptualized framework of Dx and Lean and the identification and verification of linkages between them will also allow experts to evaluate Lean-reinforcement links and to evaluate CSFs more meaningfully. Therefore, as the first part of our hypotheses, we aimed to signify the relationship of Lean and Dx. According to the experience of experts in the enterprise and recent studies, it is a fact that Lean and Dx are being discussed together more and more as one framework (Lean + Dx).

The results of our study provide data on an enterprise’s Dx transformation journey, in which the enterprise is known as a pioneer of Lean. Understanding CSFs helps managers achieve their targets in an agile, efficient, and sustainable manner. Moreover, Lean and Dx frameworks can offer additional assessment opportunities for their Lean status and Dx strategies. The multi-domain matrix between concepts will provide deeper insights and will help to identify and reshape priorities and roadmaps for Lean reinforcement. In this respect, we proposed the research models shown in Figure 4, Figure 5 and Figure 6. In Figure 4, 12 hypotheses (H1.1–H1.12) are proposed, which signify the relationship between Dx and Lean, based on the links from Table 4.

In Figure 5, 132 hypotheses (H2.1.1–H2.12.11) are proposed, which signify the importance of CSFs–DT. In Figure 6, 108 hypotheses (H3.1.1–H3.12.9) are proposed, which signify the importance of CSFs–LS.

For each link, hypothesis statements were formed to support the experts for an evaluation of 12 links (Q 1), 11 CSFs–DTs (Q 2), and 9 CSFs–LSs (Q 3). The hypothesis-support statements are based on a literature review and experts’ feedback. The statements for the Link #1 group, {(H1,1), (H2.1.1–H2.1.11), and (H3.1.1–H3.1.9)}, are provided below:

(1) The core methods of Kaizen are ‘Problem Solving’ and ‘PDCA’. (2) Innovation not only focuses on ‘problem solving’ but extends it further to ‘problem finding’, which is not aware or stated by customers yet. (3) Like Kaizen, innovation also requires teamwork and new methods (e.g., design thinking and autodidacts) and also develops the ‘learning organization’ principle. (4) The extended types of problem handling of innovations will add new dimensions to the ‘learning organization’ culture. (5) Existing Kaizen types (teian, kaikaku, kakushin, and jishuken) and problem-solving types, such as QCC, will support innovation implementation, which has elements such as competition, coaching, curiosity, and creativity. (6) Digitalization requires agility. With 5C elements and education collaboration and new tools and methods, innovation will speed up this transformation and will reinforce the Kaizen concept.

The hypothesis support statements for links 1–12 are provided in Appendix A.

3. Methodology

An extensive literature review was conducted on Lean, Dx, and CSFs. It was observed that survey questionnaires are used in studies that involve a variety of professionals and firms. In some studies, where foresight in a specific area is the subject of study, a multidisciplinary structured group of experts is needed [54]. According to Gracht [55], “During the past 60 years, the Delphi multi-round survey procedure has been widely and successfully used to aggregate expert opinions on future developments and incidents”. The Delphi method was selected, because it is based on the principle that decisions from a structured group of experts are more accurate than those from unstructured groups. The Delphi method has also been evaluated as a method that provides opportunities for companies and organizations to be better prepared for future changes, with the advantages it provides in technology foresight studies. We determined that frameworks needed to be developed for both Lean and Dx, encompassing all concepts, principles, elements, methods, tools, and types to provide a holistic view. The impacts of Dx on Lean needed to be evaluated at the conceptual level. For this purpose, ontologies were developed for both domains. The linkages were represented as a multi-domain matrix, and CSFs were evaluated at the linkage level. The ontology approach primarily enabled an accurate evaluation and consensus building and supported a better identification of CSFs during the ideation sessions by our experts. The framework of this study is shown in Figure 7.

3.1. Delphi Method

The Delphi method was developed in the 1950s and was published by Dalkey and Helmer [56]. Over the past 60 years, this method has been widely used to aggregate expert opinions. The approach used to measure consensus and to determine when to stop rounds varies from study to study. Gracht [55] evaluated 15 different measures of consensus in the Delphi method. The Interquartile deviation (IQD) method was chosen to determine consensus, as “it is generally accepted as an objective and rigorous way of determining consensus”, according to Gracht [55]. The interquartile range (IQR) is the difference between the 75th and 25th percentiles of the data, and a smaller value indicates a higher degree of consensus. Raskin [57] identified an IQR of 1.00 or less as a consensus indicator. We used this metric for the measurement of consensus. The Delphi method typically includes four rounds of questionnaires, but three iterations are typically sufficient to identify points of consensus [55]. For our study, we defined our round criteria as follows: Run 4 rounds, including Round 1. For any non-consensus item, if the IQR value changes towards 1, run a 5th round with the agreement of experts.

3.2. Experts

For Delphi studies, it is essential to have a good sampling of experts to avoid biases. With this in mind, 14 diverse and multidisciplinary (minimum of two) experts were selected, all of whom are TPS leaders with a minimum of 15 years of company experience (

Table 9. Anonymized overview of experts.

Experts	Experience
	Quality	Engineering and R&D	IT and I4.0 Technologies	Manufacturing	Supply Chain and Logistics	People (HR)	Transformation and Reform	Sales and Value Chain	Other Regions (Japan and USA)
E-1	✓		✓				✓
E-2			✓					✓
E-3		✓		✓			✓		✓
E-4			✓	✓	✓	✓
E-5				✓	✓		✓
E-6				✓	✓		✓
E-7				✓	✓		✓
E-8		✓	✓						✓
E-9		✓	✓	✓	✓				✓
E-10		✓	✓				✓		✓
E-11	✓							✓
E-12			✓				✓
E-13						✓			✓
E-14	✓		✓	✓	✓		✓		✓
SUM	3	4	8	7	6	2	8	2	6

). Some of these experts also have experience in the enterprise’s Japanese and North American regions, in addition to Europe.

3.3. Questionnarie

From Round 2 onwards, three questions were asked for each linkage on the same page, along with linkage ontology drawings and hypothesis-support statements (Appendix A). The questions and scales are presented in Figure 8.

3.4. Analysis of Data

The psychometric properties (reliability) of constructs were analyzed using Cronbach’s α analysis. A stability analysis of the Delphi method was provided by applying the coefficient of variation (CV) method based on available data. The significance of non-consensus hypotheses on the results was examined using the CV method for hypothesis testing.

4. Results

4.1. Lean Ontology

Based on the literature review and discussions with experts, the number of terms in Masai’s ontology increased from 134 to 195 with updates. To reflect Lean philosophy, 11 principles are proposed and shown next to nine concepts. These principles can be at the concept level or attached to any specific term under the concept, in line with the term’s importance and impact on Lean. For the principles, links are defined as “develops”, which also implies that it “implements”. Figure 9 displays the Lean ontology, while Figure 10 shows each concept and its ontology.

4.2. Dx Ontology

The Dx ontology is formed using the 9 concepts and 194 terms (Figure 11). The literature review is the main input for it (Table 2). Figure 12 shows each concept with its elements, types, methods, tools, steps, and examples.

4.3. Dx Impacts on Lean

The survey results are presented in

Table 10. Dx impacts on Lean survey results.

			Results
Link	Dx	Lean	Q1	Q3	IQR	Likert	Mean/Cα	Likert Statement
L 1	Innovation	Kaizen	4	5	1	4.5		High/strong positive impact
L 2	Innovation	Stakeholder S/Hitozukuri/Customer First	4	5	1	4.5		High/strong positive impact
L-3	Human Centricity	Kaizen/Jidoka/Monozukuri	4	4.75	0.75	4.4		High positive impact
L 4	Human Centricity	Customer F/Stakeholder S/Hitozukuri/Safety	4	4.75	0.75	4.4		High positive impact
L 5	Resiliency	Just in Time/Monozukuri	4	4.75	0.75	4.4		High positive impact
L 6	(Cyber)Security	Stakeholder Satisfaction/Safety	4	5	1	4.5	4.3	High/strong positive impact
L 7	Dig.Man/Dig.Eng	Customer F/Stakeholder S/Hitozukuri/Safety	4	4	0	4	/	Medium positive impact
L 8	Dig.Man/Dig.Eng	Monozukuri/Jidoka	4	5	1	4.5	0.90	High/strong positive impact
L 9	Knowledge Mgmt	Kaizen/Mieruka	4	5	1	4.5		High/strong positive impact
L 10	Knowledge Mgmt	Jidoka/Just in Time	4	5	1	4.5		High/strong positive impact
L 11	Vert and Hori Integ.	Hitozukuri/Just in Time	3	4	1	3.5		Medium positive impact
L 12	Sustainability	Stakeholder Satisfaction/Just in Time	4	4	0	4		Medium positive impact

. A consensus was achieved for all links (100%). For six of the links, the experts evaluated the impact as “high/strong positive”, whereas for three of the links, the impact evaluation was “high positive”, and for the remaining three links, it was “medium positive”. The overall evaluation between Dx and Lean was a “high positive impact”.

4.4. CSFs–DT “Dx Transformation”

The survey results are presented as a heatmap in

Table 11. CSFs–DT: Dx transformation survey results.

	Likert Value Q1/Q3/IQR												Mm/Mc/MM Cα
No	Link 1	Link 2	Link 3	Link 4	Link 5	Link 6	Link 7	Link 8	Link 9	Link 10	Link 11	Link 12
1	4.6 4.25/5/0.75	4.5 4/5/1	4.5 4/5/1	nc 3.5/5/1.5	4.5 4/5/1	4.5 4/5/1	4.5 4/5/1	4.5 4/5/1	3.5 3/4/1	nc 3.25/5/1.75	4.5 4/5/1	4.5 4/5/1	4.2/4.4/4.5 0.93
2	4.5 4/5/1	4.0 4/4/0	4.5 4/5/1	4.5 4/5/1	4.5 4/5/1	3.5 3/4/1	4.5 4/5/1	4.0 4/4/0	4.0 4/4/0	4.0 4/4/0	4.5 4/5/1	4.0 4/4/0	4.2 0.86
3	4.0 4/4/0	4.5 4/5/1	4.5 4/5/1	4.5 4/5/1	4.5 4/5/1	3.5 3/4/1	4.5 4/5/1	4.5 4/5/1	4.5 4/5/1	4.4 4/4.75/0.75	4.5 4/5/1	nc 3/5/2	4.2/4.4/4.4 0.91
4	4.4 4/4.75/0.75	4.5 4/5/1	4.5 4/5/1	4.5 4/5/1	4.5 4/5/1	3.4 3/3.75/0.75	4.5 4/5/1	4.4 4/4.75/0.75	3.5 3/4/1	3.6 3.25/4/0.75	4.4 4/4.75/0.75	4.0 4/4/0	4.2 0.89
5	3.5 3/4/1	4.5 4/5/1	3.0 3/3/0	3.0 3/3/0	3.0 3/3/0	nc 3.25/4.75/1.5	4.6 4.25/5/0.75	nc 3/5/2	3.5 3/4/1	2.6 2.25/3/0.75	2.5 2/3/1	nc 3.25/5/1.75	3.3/3.4/3.8 0.90
6	3.5 3/4/1	nc 2/4/2	3.5 3/4/1	3.0 3/3/0	2.6 2.25/3/0.75	3.5 3/4/1	3.6 3.25/4/0.75	3.5 3/4/1	3.4 3/3.75/0.75	3.0 3/3/0	2.5 2/3/1	3.5 3/4/1	3.0/3.2/3.0 0.91
7	4.4 4/4.75/0.75	3.5 3/4/1	2.5 2/3/1	3.4 3/3.75/0.75	3.4 3/3.75/0.75	4.5 4/5/1	4.5 4/5/1	3.5 3/4/11	4.5 4/5/1	3.5 3/4/1	3.5 3/4/1	3.5 3/4/1	3.7 0.85
8	4.5 4/5/1	4.5 4/5/1	4.5 4/5/1	4.5 4/5/1	3.5 3/4/1	3.5 3/4/1	4.4 4/4.75/0.75	3.6 3.35/4/0.75	4.0 4/4/0	4.4 4/4.75/0.75	4.5 4/5/1	3.6 3.25/4/0.75	4.1 0.84
9	2.5 2/3/1	2.5 2/3/1	3.5 3/4/1	3.0 3/3/0	3.0 3/3/0	3.5 3/4/1	nc 2/4/2	2.5 2/3/1	2.5 2/3/1	2.5 2/3/1	2.5 2/3/1	3.5 3/4/1	2.8/2.9/3.0 0.90
10	2.6 2.25/3/0.75	2.5 2/3/1	3.0 3/3/0	2.5 2/3/1	2.5 2/3/1	3.6 3.25/4/0.75	3.5 3/4/1	3.4 3/3.75/0.75	3.6 3.25/4/0.75	2.5 2/3/1	2.5 2/3/1	3.5 3/4/1	3.0 0.89
11	3.5 3/4/1	3.4 3/3.75/0.75	2.6 2.25/3/0.75	2.6 2.25/3/0.75	3.5 3/4/1	3.4 3/3.75/0.75	nc 3/5/2	nc 3/5/2	3.5 3/4/1	3.5 3/4/1	3.5 3/4/1	3.5 1	3.3/3.3/3.6 0.90

. Among the 132 pairs, a consensus was achieved in 122 pairs (92.4%). No consensus was obtained for the 10 pairs. For non-consensus (nc) pairs, the experts’ tendencies are shown as a lower quartile (Q1) and an upper quartile (Q3) next to the IQR value.

4.5. CSFs–LS “Lean Sustainability”

The survey results are presented as a heatmap in

Table 12. CSFs–LS: lean sustainability survey results.

	Likert Value Q1/Q3/IQR												Mm/Mc/MM Cα
No	Link 1	Link 2	Link 3	Link 4	Link 5	Link 6	Link 7	Link 8	Link 9	Link 10	Link 11	Link 12
1	5.0 5/5/0	4.5 4/5/1	4.5 4/5/1	4.5 4/5/1	4.5 4/5/1	4.5 4/5/1	4.5 4/5/1	4.5 4/5/1	4.5 4/5/1	4.5 4/5/1	4.5 4/5/1	4.5 4/5/1	4.5 0.85
2	4.5 4/5/1	4.5 4/5/1	4.5 4/5/1	4.5 4/5/1	4.5 4/5/1	4.5 4/5/1	4.5 4/5/1	4.5 4/5/1	3.6 3.25/4/0.75	4.5 4/5/1	4.4 4/4.75/0.75	4.5 4/5/1	4.4 0.88
3	4.5 4/5/1	4.5 4/5/1	3.5 3/4/1	4.5 4/5/1	4.5 4/5/1	4.5 4/5/1	4.5 4/5/1	4.5 4/5/1	3.6 3.25/4/0.75	4.5 4/5/1	4.5 4/5/1	4.5 4/5/1	4.3 0.91
4	4.5 4/5/1	nc 3.25/5/1.75	4.5 4/5/1	3.5 3/4/1	4.5 4/5/1	3.5 3/4/1	4.5 4/5/1	4.5 4/5/1	4.5 4/5/1	4.4 4/4.75/0.75	4.5 4/5/1	3.5 3/4/1	4.1/4.2/4.3 0.90
5	3.6 3.25/4/0.75	3.5 3/4/1	3.5 3/4/1	3.5 3/4/1	4.4 4/4.75/0.75	4.0 4/4/0	3.6 3.25/4/0.75	4.5 4/5/1	3.5 3/4/1	3.6 3.25/4/0.75	4.5 4/5/1	3.5 3/4/1	3.8 0.90
6	4.4 4/4.75/0.75	3.5 3/4/1	4.5 4/5/1	3.5 3/4/1	3.6 3.25/4/0.75	4.0 4/4/0	4.4 4/4.75/0.75	4.4 4/4.75/0.75	3.5 3/4/1	4.5 4/5/1	4.5 4/5/1	3.5 3/4/1	4.0 0.91
7	3.5 3/4/1	3.5 3/4/1	3.5 3/4/1	3.5 3/4/1	nc 3.25/5/1.75	3.5 3/4/1	nc 3/5/2	3.5 3/4/1	3.5 3/4/1	3.5 3/4/1	3.5 3/4/1	3.6 3.25/4/0.75	3.4/3.5/3.8 0.93
8	4.0 4/4/0	3.5 3/4/1	3.5 3/4/1	3.5 3/4/1	3.5 3/4/1	3.6 3.25/4/0.75	nc 3/5/2	3.5 3/4/1	4.5 4/5/1	4.0 4/4/0	3.5 3/4/1	3.5 3/4/1	3.6/3.7/3.8 0.84
9	4.5 4/5/1	3.5 3/4/1	4.5 4/5/1	4.4 4/4.75/0.75	3.5 3/4/1	3.5 3/4/1	3.5 3/4/1	4.4 4/4.75/0.75	4.5 4/5/1	3.6 3.25/4/1	3.5 3/4/1	3.5 3/4/1	3.9 0.89

. Among the 108 pairs, a consensus was achieved in 104 pairs (96.3%). No consensus was obtained for the 4 pairs. For non-consensus (nc) pairs, the experts’ tendencies are shown as Q1 and Q3 next to the IQR value.

4.6. Psychometric Properties

The reliability of the constructs was examined by Cronbach’s alpha (α) values. A construct is considered reliable if its Cronbach’s α is more than 0.7 [58]. The Cronbach’s α of “Dx impact to Lean” was found to be 0.9, which is considered “excellent” (Table 10). The Cronbach’s α of “CSFs–DT” was found to be between 0.84 and 0.93, which is close to “excellent” (Table 11). The Cronbach’s α of “CSFs–LS” was found to be between 0.84 and 0.91, which is also close to “excellent”. These results confirm the reliability of the constructs.

4.7. Stability Test of Deplhi Method

According to Gracht [55], “it is important to distinguish between the two different concepts of consensus/agreement and stability” in Delphi studies. Traditionally, many Delphi studies stop the survey procedure for a certain projection when a pre-defined level of agreement, i.e., consensus, is achieved. However, Dajani et al. [59] remarked that a consensus is meaningless, if group stability is not reached. They define stability as “the consistency of responses between successive rounds of a study”. According to them, the coefficient of variation (CV) can be used as a measure of stability. They define stability as the absolute value of the difference in CV between successive rounds, which is less than a predetermined small value. CV is also a method used to measure consensus in Delphi studies. The coefficient of variation at or below 0.5 is found to be a cut-off point that is accepted as an indication of reasonable internal agreements [55]. Shah and Kalaian [60] used a CV at or below 0.2 to measure stability for successive rounds, referring to Dajani. We used the same value in our stability testing.

The Delphi method involves multiple rounds of questions and hypotheses (Hs). It is common for panel members to drop out during this process. Due to the high number of hypotheses in our study, we decided to eliminate questions where a consensus was reached to reduce attrition. As a result, a stability test could not be conducted for 183 hypotheses where a consensus was reached in Round 2. However, the minimum and maximum coefficient of variation (CV) values are provided as a reference, alongside the stability test results of the subsequent rounds, in

Table 13. Stability results of rounds.

	│CV(R2)│ (No of Hs = 183)	│CV(R3)–CV(R2)│ (No of Hs = 69)	│CV(R4)–CV(R3)│ (No of Hs = 31)
Min	0.11	0.08	0
Max	0.41	0.14	0.08

.

Based on the data in Table 13 and the high value of Cronbach’s α, it can be concluded that stability was achieved in the Delphi method across the different rounds.

4.8. H. Hypothesis Testing

The Delphi round summary for each hypothesis group is given in

Table 14. Summary of Delphi rounds.

No of Hs		Rounds						% of c
		Round 2		Round 3		Round 4
		c	nc	c	nc	c	ncfinal
H1	12	9	3	2	1	1	0	100
H2	132	88	44	23	21	11	10	92.4
H3	108	86	22	13	9	5	4	96.3

c, consensus; nc, non-consensus; nc_final, final result of non-consensus items for each hypothesis group.

. For the non-consensus hypotheses (10 of H2 (H2.4.1, H2.10.1, H2.12.3, H2.6.5, H2.8.5, H2.12.5, H2.2.6, H2.7.9, H2.7.11, and H2.8.11) and 4 of H3 (H3.2.4, H3.5.7, H3.7.7, and H3.7.8)), the IQR value did not change and stayed the same from Round 2 to Round 4. Therefore, we decided not to run Round 5.

It can be concluded, from the contents of Table 10, that Dx impacts Lean very positively (high or high/strong) for all links, except Links 7, 11, and 12, but these also lean on the positive side. The consensus rate is 100%. The overall average is 4.3 (close to high/positive impact). This can be interpreted as both frameworks being modelled adequately, the multi-domain matrix consistently showing relations, and hypothesis 1 (H1) being tested positively.

There were 10 non-consensus hypotheses for H2, according to the selected IQR consensus measurement method. The quality of survey results can be tested with another consensus measurement method to help reach a conclusion for the construct and the H2 group of hypotheses. The testing results of the survey data using the “coefficient of variation” consensus measurement method are provided in

Table 15. Alternative consensus analysis of 10 non-consensual hypotheses of the H2 group using the CV method.

H2 nc Hypo.	Likert Value Distribution					Mean	CV
	1	2	3	4	5
H2.4.1	0	5	1	7	1	3.3	0.31
H2.10.1	0	0	4	0	10	4.4	0.20
H2.12.3	0	3	1	6	4	3.8	0.29
H2.6.5	2	3	2	7	0	3.0	0.38
H2.8.5	0	1	7	0	6	3.8	0.29
H2.12.5	0	0	6	1	7	4.1	0.24
H2.2.6	0	0	6	0	8	4.1	0.24
H2.7.9	0	0	4	1	9	4.4	0.21
H2.7.11	0	0	5	0	9	4.3	0.22
H2.8.11	0	0	4	4	6	4.1	0.20

.

These results show that for the H2 group’s non-consensus items, the coefficient of variation (CV) measurement method indicates a consensus, according to the criteria proposed by Dajani [59] and summarized by Grach [55], which is significantly below the upper limit of 0.5 as well. This also demonstrates that the interquartile range (IQR) is a rigorous method for measuring consensus. Based on the alternative testing, the stability test results (Table 13), and an overall consensus rate of 92.4%, it can be interpreted that H2 was positively tested. However, the impacts of the non-consensus hypotheses need to be further analyzed with respect to the Q1 and Q3 values. If the Delphi method were to continue, a consensus could potentially be reached on the Q1 or Q3 value or somewhere in between. Based on this assumption, the minimum and maximum mean values (M_m and M_M) are presented in Table 13. Even in any possible scenario, the order of critical success factors for digital transformation (CSFs–DT) remained unchanged for the top five and bottom four CSFs. This indicates that the order of CSFs–DTs is consistent, with the possibility of only a change in the order for items 6 and 7.

There are 4 non-consensus hypotheses for the H3 group, according to the selected IQR method. The quality of survey results was tested using “the coefficient of variation” method, as conducted for the H2 group above, and the results are provided in

Table 16. Alternative consensus analysis of 4 non-consensual hypotheses of the H3 group using the CV method.

H3 nc Hypo	Likert Value Distribution					Mean	CV
	1	2	3	4	5
H3.2.4	0	0	4	1	9	4.4	0.21
H3.5.7	1	1	2	2	8	3.9	0.33
H3.7.7	0	0	6	1	7	3.9	0.24
H3.7.8	0	0	5	1	8	4.1	0.24

, as well as the experimental conclusions that can be drawn.

These results show that for the H3 group’s non-consensus items, the CV measurement method indicates a consensus and is also significantly below the top limit (0.5). Based on the alternative testing strategy, the stability test results (Table 13), and overall consensus rate (96.3%), it can be interpreted that H3 was positively tested. However, the impacts of the non-consensus hypotheses need to be further analyzed with respect to the Q1 and Q3 values, as performed for the H2 group. The border mean values (M_m and M_M) are shown in Table 12. In any possible scenario, the order of critical success factors for Lean sustainability (CSFs–LS) remained unchanged for the top five and bottom two CSFs. This suggests that the CSFs–LS orders are consistent with the possibility of a change only in the order for items 6 and 7.

5. Discussion

The survey results for CSFs–DTs “to have an agile, successful Dx transformation” are presented in

Table 17. CSFs–DT: Dx transformation summary.

Order	Factor	Value	Importance
1	Commitment of top management and transformational leadership style	4.4	Very important
2	Organizational transformation (breaking silos, talent management, and integrations)	4.4	Very important
3	Teamwork to achieve common objectives	4.2	Very important
4	Human Centricity (diversity and wellbeing of employees with all aspects)	4.2	Very important
5	People TF (skilling, upskilling, and re-skilling); empowerment; and citizen development	4.1	Very important
6	IT transformations and roadmaps	3.7	Moderately/very important
7	Availability of financial resources and investment strategies	3.3	Moderately important
8	Mfg./R&D TF (circular mfg./design virtualizations, and vertical and horizontal integrations)	3.3	Moderately important
9	Public education transformations, collaborations with universities/education	3.2	Moderately important
10	External support (consultations)	3.0	Moderately important
11	External support (incentives/regulations)	2.9	Slightly/moderately important

. The most important factors for Dx transformations were found to be a “Commitment of top management and transformational leadership style/program” and an “Organizational transformation”. In this regard, managers need to ensure they have companywide awareness, a top-down-direction approach, and the complete commitment of people making digital transformations to everyone’s job [61]. An organizational transformation should break silos, manage talents [62], and integrate with stakeholders [63].

“Teamwork”, “Human Centricity”, and “People Transformation” as a group follow the first two CSFs, indicating that the key enablers of Dx transformations are people working as a team, placing humans at the center to reduce tension in people transformations, as elaborated by Margherita et al. [47]. A transformational leadership style should promote ‘feedback-seeking’ behaviors [64] and prioritize the principles of “act for others, work with integrity” (Toyota Way 2020 [65]) to reinforce teamwork for agility. Moreover, transformational leadership and programs integrating digital technologies with existing Lean-based production systems create paradoxes, as elaborated by Johansson et al. [41]. The second group of CSFs also addresses both paradoxes and dialectical perspectives from a Lean thinking point of view. Managers should also see these three CSFs as key transformation enablers. Dx introduces a cultural shift, and “Human Centricity”, with its elements, is key concept to balance complexities. This result is in-line with the Lean philosophy and its key concepts and principles (Hitozukuri (making people) and learning organizations). These results are also in line with the findings that digitalization capabilities and integrations with Lean concepts and principles rely on humans and enhance human capabilities, stimulating a cumulative capability-development perspective, such as learning to learn capabilities and problem-solving skills, as recently elaborated by Galeazzo et al. [66] and Powell et al. [46]. In line with this discussion, the links, L 1, L 2, L 3, and L 4 show Lean-reinforcement paths related to these CSFs for managers.

The CSF “IT transformation technologies” follows, but contrary to common understanding, it is not at the top. The order of CSFs–DTs indicates that technology provides the expected benefit, if it is well understood and grasped as a need by the management and people and is utilized and developed by almost the entire workforce (citizen developments). On top of technologies, the key concept for digitalization is security (cyber) for the sustainability and resiliency of a company. Managers should prioritize the (cyber)security concept and consider link 6 as an essential reinforcement point for Lean.

The order of “Availability of financial resources” indicates that it provides the expected benefit once the company has a commitment and a program with people and skills one step ahead of the financial resources.

“Manufacturing and R&D transformation” implies a focus on integration with stakeholders (suppliers and customers) and the sustainability and resiliency concepts supported/enforced by policymakers. Together with “human centricity”, this CSF can be interpreted as a socio-technical reflection of Lean. This aspect has been modelled and confirmed by experts with the concept of “Stakeholder Satisfaction” in Lean, along with the principles of “Respect for people” and “waste elimination”. It is also emphasized and modelled in the Dx framework with concepts of “human centricity”, “sustainability”, and “resiliency”, which are linked to the Lean framework. The positive impact of Dx on Lean’s social performance was elaborated by Chavez et al. [48]. Managers should not only focus on the company but also on meeting customer expectations through mass personalization. Dx leads to mass personalization as an emerging manufacturing paradigm beyond customization focusing on uniquely made products and services for individuals at scale. Ahelerof et al. [67] elaborate on this point and how it can be enabled on a Lean base with digitalization.

The CSF “Public education transformation and collaboration with universities and education” follows. Managers should upskill and re-skill their existing resources. A collaboration with universities supports manufacturing tactically in this area. The just-in-time hiring strategy is no longer sustainable. Manufacturing needs a workforce with new skills, both technical and transversal. Managers need to support and collaborate with public education more through joint programs (e.g., STEAM) and longer internship programs, including at the high school level.

Both “External support” CSFs are considered important factors that will help and speed up Dx transformations. However, they are at the bottom of the list. This can be interpreted as if a company does not have a good level of understanding and engagement with the first 9 CSFs, external support will not have any significant impact on the Dx transformation. In this study, the values of the two external support CSFs are relatively low compared to the others. This may be due to the survey company being an enterprise with adequate financial and human resources compared to SMEs. If a similar survey were repeated with selected SMEs, these values would be higher. From a similar view, ‘collaboration with universities’ would be prioritized more. This can be considered a limitation of this study.

The survey results for CSFs–LS’s “to sustain/reinforce Lean while transforming to Dx” are ordered and presented in

Table 18. CSF-LS: Lean sustainability summary.

Order	Factor	Value	Importance
1	Keep Genchi Genbutsu (go and see)	4.5	Extremely important
2	Never compromise jidoka (stop in time) to keep problems on the surface	4.4	Very important
3	Keep and enforce JKK (build in quality w/ownership), which will be more critical with Dx	4.3	Very important
4	Keep/watch automation with the human touch, with people and ownership being core elements of Lean	4.2	Very important
5	Keep standardization and strictly watch, which is base for kaizen and employee empowerment	4.0	Very important
6	Trust the data, but keep problem solving and the PDCA culture, as Dx needs a learning organization with innovation	3.9	Moderately/very important
7	TF is a challenge, and visual control will be more important than ever for JKK and jidoka	3.8	Moderately/very important
8	Trust the data, generative AI, etc., but do not lose nemawashi, and utilize it meaningfully to be agile	3.7	Moderately/very important
9	Keep an additional focus on supplier development, pace, resiliency, and sustainability, which need to transform as well	3.5	Moderately/very important

. To sustain and reinforce Lean, the experts prioritized the “Keep Genchi Genbutsu (go and see)” principle, along with “Jidoka (stop in time)”, “JKK (build in quality w/ownership)”, “automation with the human touch”, and “Standardization”, which are key concepts and principles of Lean. The implementation of digital technologies, such as data collection and reporting, IoT, and AI technologies, would cause ignorance for these CSFs. The problem-solving culture follows them. A Dx transformation itself is a significant and continuous change (henkaten). The problem-solving culture and the PDCA cycle are essential tools for managing changes in this journey.

“Visual control” is another key concept of TPS. “Visual control” ensures the principles of right decision making (nemawashi), teamwork, and stopping in time to fix problems, as well as data science and generative AI work with data to generate meaningful and purposeful outputs. However, these outputs may not always be sufficient for decision making. Nemawashi is not only a consensus-building tool but also an information sharing and employee-engagement tool. This method verifies information accuracy and leads to the right decision. The importance of human factors for this phenomenon was elaborated by Lindner and Reiner [68]. “Supplier development” is also another recognized important CSF. Research suggests that knowledge diffusion occurs more quickly within Toyota’s production network than in competing automaker networks [69]. Toyota continuously supports its suppliers for Lean developments through training, crisis management, and mentorship. It can be interpreted that the CSF is low not because it is less important but because supplier development is a natural DNA thinking of TPS. Managers need to consider Lean practices and methods and implement them in their transformational program and hoshin (e.g., problem solving, PDCA training, and practices). The links L 5 and L 7 to L 12 identify reinforcement paths and help to define further practical actions (e.g., Lean practices extend to the whole company and close partners).

The non-consensus items and their impacts on the results and CSFs’ orders were analyzed and discussed in the hypothesis-testing section. The ambiguity that non-consensus items introduce is the potential change in the order of 6 and 7 in both Table 17 and Table 18. The impacts of these changes will have a limited effect on our discussion and conclusion. The overall quantitative results show that the Dx framework formed with the I4.0 and I5.0 paradigms reflects the Dx focus and vision at a very good level and makes it comparable with the Lean framework with the selected links. The ontologies and selected links show the transformation and impact paths and relations among concepts and underlying elements, types, tools, methods, and principles.

Although the focus of the Dx framework is manufacturing, the key concepts are general. Any company or business can utilize it as a template and can form a custom framework. The same approach also applies to Lean. The Lean framework has been proven to be applicable in any business area so far and can be applied with key principles and methods further (e.g., contextualization).

Digitalization is not a onetime tactical activity but a strategic, continuous process. For the careful planning and balancing of complexities and managing costs and cultural shifts, a roadmap needs to be defined and managed by the PDCA cycle. The CSFs’ order can help in the identification of priorities and steps. Another important input would be to assess the Lean and Dx maturity levels. These results would provide an input to the roadmap as well.

We believe that the Lean organization framework (Figure 10) derived from Masai’s study [17] with principles represents Lean thinking in a better way. The temple view shows some key terms directly and implies some others indirectly (e.g., under the Toyota Way). However, the concept of Monozukuri (making things) has not been mentioned directly so far. In some cases, it is thought to be synonymous with TPS, but it is the actual Gemba and arena of TPS, where it is put into practice. Monozukuri includes both product development (engineering) and production (manufacturing). This perspective of lean thinking was explained by Dombrowski et al. [70]. A Dx framework with links to the Lean framework provides a complete holistic view for LDx (Lean + Dx (I4.0 + I5.0)). The impact of Lean maturity on a Dx transformation was not a subject of this study. However, according to Tortorella et al. [71], “techno-logical adoption will not lead to distinguished results. LP practices help in the installation of organisational habits and mindsets that favour systemic process improvements, supporting the design and control of manufacturers’ operations management towards the fourth industrial revolution era”.

6. Conclusions

This article empirically examined the critical success factors for digitalization (I4.0 + I5.0) while sustaining Lean and Lean-reinforcement links during this transformation. In this regard, 11 CSFs for “Dx transformation”, 9 CSFs for “Lean sustainability”, and 12 reinforcement links (between Lean and Dx) were identified from the published literature and one-on-one discussions with TPS experts. To conduct our study, both Lean and Dx frameworks were developed using the ontology method. Then, a survey with experts was conducted at a company known as a Lean pioneer and that is experiencing a Dx transformation. The Delphi method was used as a survey tool. The analysis was performed in three steps. First, the psychometric properties were assessed. Second, a stability analysis of the Delphi method was conducted. Then, the significance of non-consensus hypotheses of the results was examined.

This article makes several key contributions to the literature. First, it validates CSFs and Lean-reinforcement links of digitalization using survey data. Second, the key finding of both CSFs is that the survey results do not point to one or two factors but rather to a group of factors that need to be considered. This means that both Lean and Dx should be viewed holistically and should not be considered as tools or technologies alone. These findings are useful for manufacturing firms and small-to-medium enterprises which are experiencing digital transformations.

The managers at these firms can put more focus on factors and links that have a significant effect on digital transformations while sustaining and reinforcing Lean.

The key findings of this article are as follows.

• Top management commitment and a transformational leadership style with a human centric view are prioritized enablers for digital transformations.
• “Keep Genchi Genbutsu (go and see)”, along with stopping in time, ownership of quality, and automation with the human touch are prioritized factors for Lean sustainability.
• An extension of kaizen with innovation, the implementation of technologies in manufacturing, and engineering with a cybersecurity concept are equally prioritized Lean-reinforcement links.

This study has a few limitations. CSFs were defined as a single layer. Future studies should add sub-factors and conduct research in specific areas of Dx technologies, human centricity, innovation, sustainability, and resiliency with depicted Lean concepts and principles. This study focuses on manufacturing. Future studies should also focus on other sectors and areas, such as supply chain and product logistics. As discussed in the previous section, this study has some weaknesses. A similar study should be performed with SMEs. Assessment tools and maturity models have been studied for both Lean and I4.0, and recently, including I5.0 concepts with the utilization of new methods and different approaches has been advocated for. Considering the common view that Lean and Dx are like twins and complement each other, the findings of this study can be used to develop a framework for Lean + Dx (I4.0 + I5.0), so-called LDx, and then propose an assessment tool or maturity model for it. Finally, the formalization and development of Dx and Lean organization frameworks using the ontology approach and the identification of reinforcement links would offer new insights and perspectives on the theoretical discussion of Lean and Dx.