Gumbi & Twinomurinzi (2020) conducted an extensive systematic review to assess the state of SMME readiness for SM adoption. The study found that SMMEs lack awareness (adequate knowledge) of SM, lack the know-how to perceive the relative advantage of SM, and are incompatible with SM due to a lack of expertise, skills and resources. Gumbi and Hossana (2022) expanded on these findings by conducting a conceptual research study to identify preconditions for SMME readiness for SM adoption and factors that influence these preconditions in emerging economies and low-income countries. The study identified SMKC, SMRA, and SMC as key preconditions for SMME readiness for SM adoption in emerging economies and low-income countries. SMKC refers to the development of SM awareness and knowledge competence in SM ADTs. SMRA refers to having a comprehensive understanding of the benefits of SM and the value that may be derived from it. SMC refers to making high-risk, proactive strategic investments to exploit market opportunities for compatibility with SM requirements. Gumbi and Hossana (2022) further identified that contextual preconditions (SMKC, SMRA, and SMC) may be influenced by SDDL, BMI, EO, and a CoI. SDDL is a method, process, or practice by which knowledge competence related to digital innovations and ADTs is developed. BMI refers to a firm's capability to assess the strategic objective and rationale for adopting technological innovations. EO is a measure of how a business operates strategically to discover and exploit market opportunities. A CoI refers to a set of organizational cultural values, norms, and artifacts which support a company's innovativeness.

The proposed SM frameworks as discussed in the previous section do not include or consider the context-specific (emerging economies and low-income countries) SMME preconditions (SMKC, SMRA, and SMC) or the factors influencing these preconditions, as uncovered by previous studies (Gumbi & Twinomurinzi, 2020; Gumbi & Hossana, 2022). Additionally, the proposed SM frameworks do not indicate the process or factors that influence SM readiness for adoption. The Conceptual Framework for IT Innovation Adoption, as shown in Fig. 2, explicitly identifies a process and factors under which innovation adoption is influenced. However, the process and factors are generic and broad, making them unsuitable for SMME readiness for SM adoption in emerging economies and low-income countries.

In this study, the Conceptual Framework for IT Innovation Adoption (Fig. 2) is adapted to formulate a conceptual SMME readiness framework based on Gumbi and Hossana (2022) approach, as shown in Fig. 3.

Fig. 3.

Conceptual SMME readiness framework for SM adoption.

A research model theoretically conceptualized by Gumbi and Hossana (2022), as shown in Fig. 4, was adopted in this study. The research model was conceptualized based on the IT innovation adoption process (Fig. 2) as discussed in the preceding sections. The primary objective of this study was to empirically develop an SMME readiness framework from the research model (Fig. 4) to answer the research study questions (RQ1, RQ2, RQ3 and RQ4) posed in the introduction.

Fig. 4.

Research model for SMME readiness for SM adoption.

In this study, critical realism (philosophy), multiple case studies (strategy), interviews (data collection method), retroduction and abduction (approach), and thematic analysis (data analysis technique) comprised the research design and methodology.

Actual Domain

RQ1

What is the level of SMME readiness for SM adoption in the manufacturing sector?

A critical realist research cycle using the emergent theory development approach was chosen for this study. Critical realist research is based on a scientific approach to the construction of explanatory models and theories (Mukumbang, 2023). The construction of these explanatory models and theories follows three stages of retroductive theorizing: the emergent phase, the construction phase, and the confirmatory phase (Mukumbang, 2023). The emergent phase, discussed in Section 3, has been completed, and its key results were published in Gumbi and Twinomurinzi (2020) and Gumbi and Hossana (2022). This paper focuses on presenting the key findings of the construction phase.

In the construction phase, the explanatory theory is built (constructed) using empirical data (Mukumbang, 2023) based on the conceptual theoretical framework (Fig. 4) discussed in Section 3. Explanatory theory-building generally includes case study research (single or multiple cases) involving formal data collection and analysis methods aimed at supplementing and confirming aspects of the developing theory in the emergent phase (Mukumbang, 2023). A multiple case study qualitative research design was used to answer the research questions and to facilitate the empirical development of the SMME readiness framework for SM adoption. Qualitative research designs and methods are utilized to understand a complex reality and the meaning of actions in a given context (Almeida et al., 2017). Multiple case study research provides an in-depth analysis of context-dependent knowledge and is a highly useful and powerful method for developing new theories (Mittal et al., 2020). SMMEs in the South African manufacturing sector were selected as the case studies for this research.

The manufacturing sector is the third largest sector in South Africa (Statistics South Africa, 2023), as shown in Fig. 5. The sector's GDP contribution is currently at 13% making manufacturing indispensable to South Africa as a means to achieving envisaged economic growth targets. However, despite this recognition, the sector's contribution to South Africa's economic growth prospects has been declining over the last two decades (Bhorat & Rooney, 2017; Kreuser & Newman, 2018). The sector's GDP has declined from approximately 21% in 1994 (Industrial Development Corporation, 2013) to 13% in 2022 (Statistics South Africa, 2023). Employment in the sector has continued to decline from 1.9 million jobs in 2008 (Statistics South Africa, 2009) to 1.1 million jobs in 2022 (Statistics South Africa, 2023b).

Fig. 5.

South African economic sector contribution.

SMMEs in the South African manufacturing sector

SMMEs in the South African manufacturing sector are defined and classified according to the National Small Business Act 102 of 1996 (Republic of South Africa, 1996, 2003, 2004), as shown in Table 1.

Table 1.

Manufacturing SMME classification

Size of class	The full-time equivalent of paid employees	Total turnover	Total gross asset value (fixed property excluded)
Medium	≤ 200	ZAR 51 million	ZAR 19 million
Small	≤ 50	ZAR 13 million	ZAR 5 million
Very Small	≤ 20	ZAR 5 million	ZAR 2 million
Micro	≤ 5	ZAR 200,000	ZAR 100,000

The manufacturing sector comprises a total of 198 274 SMMEs across the nine South African provinces (Small Enterprise Development Agency, 2022). Manufacturing SMMEs account for approximately 8% of the total SMME population in South Africa (Small Enterprise Development Agency, 2022). The provinces of Gauteng, Limpopo, Mpumalanga, KwaZulu-Natal, and the Western Cape account for more than 80% of all manufacturing SMMEs in the country, while the Northern Cape has the lowest number (less than 1%). The manufacturing sector is dominated by large enterprises when it comes to income contribution; however, in the past decade, the SMME sector has been steadily growing, as shown in Fig. 6 (Statistics South Africa, 2008a; Statistics South Africa, 2019).

Fig. 6.

Manufacturing industry Income by enterprise size, 2008-2017.

Semi-structured interviews were used for data collection. These interviews are well-suited to a case study research strategy as they enable the gathering of qualitative data that facilitates the investigation, exploration, deeper understanding, and explanation of contextual factors of a phenomenon within its specific context and setting (Saunders et al., 2009). Due to their distinctive flexibility, semi-structured interviews are regarded as the most important form of interviewing in case study research (Gillham, 2010). A questionnaire with seven SMME background (profile) questions and 25 open-ended interview questions, as shown in Appendix E, was used for data collection.

Retroduction is the logic of inference-making espoused by critical realism (Mukumbang, 2023). As a logic of inference, retroduction is typically associated with various forms of abductive reasoning, applied as the researcher moves back and forth between obtained data and an a priori theoretical framework (Mukumbang, 2023). Thus, the two logics of inference (retroduction and abduction) are considered highly complementary (Mukumbang, 2023). Retroduction and abduction are key analytical tools used in critical realism-based research studies (Meyer & Lunnay, 2013). When used together, these inference-making tools are crucial for formulating new conceptual frameworks or theories (Meyer & Lunnay, 2013). They enable a more rigorous form of analysis capable of differentiating the actual from the real by uncovering and understanding the complex processes of research participants (Meyer & Lunnay, 2013). Retroductive and abductive inferences are also complementary to deductive inference, particularly in advancing data analysis beyond the original research premise (Meyer & Lunnay, 2013), which is characteristic of critical realism-based research studies. In this study, retroduction, incorporating abduction, was used to empirically develop the SMME readiness framework for SM adoption. This process involved moving back and forth between qualitatively obtained data, the innovation adoption process (Fig. 3 in Section 3), and the a priori conceptual theoretical framework (Fig. 4) discussed in Section 3.

Abductive Thematic Network Analysis (ATNA) (Moutinho & Sokele, 2018; Rambaree, 2018) was used for data analysis, the development of the SMME readiness framework for SM adoption, and to answer the study's research questions. ATNA combines Abduction (abductive reasoning), Thematic Analysis (TA), and Thematic Network Analysis (TNA). Abduction is a process of enquiry for forming or explaining research propositions through the use of back-and-forth reasoning between theory and empirical evidence. TA is a process of identifying patterns to formulate themes from collected empirical data. Using TA researchers can organize segments of gathered data into themes, facilitated through coding (Moutinho & Sokele, 2018). TNA is a creative and systematic framework for linking themes in qualitative research. Through TNA, researchers can identify themes and then develop graphical representations of the linkages (networks) between them (Moutinho & Sokele, 2018). TNA can be data-driven, theory-driven, or a combination of both. This flexibility allows Abductive Theory to be combined with TNA, forming ATNA, for qualitative data analysis.

Content Analysis (CA) and TA are two of the most widely and frequently used approaches in qualitative data analysis (Humble & Mozelius, 2022). Both CA and TA identify themes, patterns, or codes inductively or deductively. However, CA is better-suited for broader applications, including both qualitative and quantitative research, larger data sets, and positivistic studies (Humble & Mozelius, 2022). In contrast, TA is more suited for deeper analysis, a more in-depth understanding of phenomena, and “pure qualitative” analysis, and is considered a “flexible method for qualitative analysis” applicable to various research designs (Humble & Mozelius, 2022). In this study, the objective of data analysis was to formulate and explain research propositions, identify patterns and themes, and link themes to empirically construct an SMME readiness framework for SM adoption and answer the research study questions (RQ1, RQ2, RQ3, and RQ4) as posed in the introduction. Thus, CA was deemed unsuitable, as it is constrained in formulating or explaining research propositions and linking themes to construct a framework. While TA is part of ATNA, on its own, it shares CA's limitations in these aspects.

Accordingly, ATNA was employed to formulate and explain research propositions, identify patterns, and link themes to empirically construct an SMME readiness framework for SM adoption. TA was used to open-code the data from transcribed text files, axial code the code groups, and develop the themes. TNA was applied to establish the linkages among the themes, using both obtained data and the adapted innovation adoption process. ATLAS.ti v.22.1.5.0 software was utilized for data analysis, following the steps proposed by Rambaree (Moutinho & Sokele, 2018), as shown in Fig. 7.

Fig. 7.

Abductive thematic network analysis.

Reliability and validity in qualitative research may be attained using the criteria of credibility, transferability, dependability, and confirmability (Zachariadis et al., 2013; Anney, 2014; Connelly, 2016; Cypress, 2017). Credibility was ensured through prolonged engagement with the participants, peer debriefing, and member checks. One of the researchers spent three months in the business environments of the participating SMMEs (prolonged engagement). Peer debriefing was conducted through discussions with manufacturing and ADT research experts within the CSIR. Member checks were carried out through iterative engagements with participants during the data coding process. Transferability was enhanced by employing purposive sampling and providing a thick description of the participants’ lived experiences, which were captured directly from the participants. Dependability was achieved by maintaining an audit trail which included the interview protocol, recordings, and transcribed documents. The coded data, memos, and themes in the transcribed documents were reviewed and validated by an expert and a senior researcher. Confirmability was established by utilizing the innovation adoption theory (Section 3), maintenance of an audit trail (interview protocol, recordings, and transcribed documents), and validating the coded data, memos, and themes through expert reviews. Additionally, iterative engagements with participants during and after the data analysis process further ensured confirmability.

To answer the research questions posed in the introduction section, a four-step ATNA process was followed, as discussed in the data analysis section (Fig. 7). This section begins with an overview of the selected case studies, followed by a detailed discussion of the key results obtained from the ATNA analysis.

Empirically developing an SMME readiness framework for smart manufacturing adoption

A four-step ATNA process was followed to empirically develop an SMME readiness framework based on the research model (Fig. 4) to answer the research study questions (RQ1, RQ2, RQ3 and RQ4) as posed in the introduction. TA was conducted to formulate themes (constructs) from the collected empirical data (step 1 and step 2). TNA was conducted to link (connect) the themes (step 3). Abductive reasoning was utilized in step 4 to empirically explain the research model's (Fig. 4) research propositions. Retroduction incorporating abduction was used to move back and forth between qualitatively obtained data and the innovation adoption process (Fig. 3 in Section 3) and the research model (Fig. 4) across all the steps (steps 1 to 4) of the ATNA process.

Step 1: Thematic analysis (Data Coding)

In step 1, audio data was transcribed into text files and subjected to open coding, as shown in Fig. 8. Following the process outlined in Fig. 8, 152 codes were generated from 239 quotations, as depicted in Fig. 9. Additional details can be found in Appendix A.

Fig. 8.

Open coded data from transcribed text file.

Fig. 9.

Codes (152) created from 239 quotations.

Step 2: Thematic analysis (Formulating themes - constructs)Development of code groups

In step 2, 32 code groups were formed from the 152 codes using axial coding, supported by 16 analytical reflexive group memos, as illustrated in Fig. 10. This process was guided by the a priori conceptual theoretical framework discussed in Section 3 (Fig. 4). Further details can be found in Appendix B – Table B.1.

Fig. 10.

Code Groups (32) created from 152 codes.

Emergent themes: constructs

Eight themes emerged from the aggregation and categorization of code groups: SMKC, SMRA, SMC, SDDL, BMI, EO, CoI, and SMMERA. The aggregation and categorization of code groups into themes were guided by the a priori conceptual theoretical framework discussed in Section 3 (Fig. 4). Through back-and-forth abductive reasoning between the empirical data and the SM innovation adoption process (Fig. 3), the study determined that the participating SMMEs can be classified as high-tech SMMEs and a non-high-tech SMME. A comparative analysis between the high-tech SMMEs and the non-high-tech SMME is shown in Appendix D – Table D.1.

Smart manufacturing knowledge competence (SMKC)

SMKC for the high-tech SMMEs emerged as a theme from the analysis of 24 codes categorized into four code groups, as shown in Appendix B – Table B.2. The theme (SMKC) revealed that high-tech SMMEs have high levels of SM awareness and knowledge competence of SM ADTs. For example, when asked about their understanding and knowledge of the SM concept, the CEO of Case A responded: “My understanding is that when you use smart manufacturing is that you utilize technology such as 3D printing and software tools to enable you to optimize the efficiencies of manufacturing.” The theme (SMKC) further revealed that the high-tech SMMEs are using ADTs such as smart sensors and see these ADTs as very important for their current and future business operations. The CEO of Case A elaborated: “We integrate sensors into aircraft as well, so it's not just on UAVs but it is also on aircraft, we have operational knowledge of how to use them, we've been using them for years.” The CEO further added: “Some of the tools that we are developing are dealing with the IoT, so in other words, the disaster management solution we are working on incorporates real things in the virtual world. That's the IoT using smart techniques to be able to optimize efficiencies… I think from where we stand today, we believe that is the future for the next 20-30 years.”

The SMKC theme for the non-high-tech SMME emerged from the analysis of 10 codes categorized into four code groups, as shown in Appendix B – Table B.3. In contrast to the high-tech SMMEs, the analysis of the SMKC theme for the non-high-tech SMME revealed that it has low levels of SM awareness and very limited knowledge competence of SM ADTs. The MD of Case D confirmed: “It's actually the first time to get really engaged with what this is when you sent me the questions and invited me.” The assessment of the theme (SMKC) further revealed that the non-high-tech SMME is not using any ADTs and considers even very basic digital technologies, such as barcodes, to be very expensive. The MD of Case D admitted that her knowledge of SM ADTs is “definitely low” and stated: “Right now we don't, we don't have that; we are not using any technologies as such because it's a hard labor kind of environment right now.” The MD added: “Another thing is that you have, I mean, you have other companies that still use this digital technology where they have barcodes, but barcoding is expensive for a small business.”

Smart manufacturing relative advantage (SMRA)

SMRA emerged as a theme from the analysis of 11 codes categorized into three code groups, as shown in Appendix B – Table B.4. The theme (SMRA) revealed that the high-tech SMMEs comprehensively understand the benefits (relative advantage) offered by SM from a value capture, value proposition, and value creation perspective. The CEO of Case A confirmed that SM offers cost reduction (value capture): “If you don't use smart manufacturing, which means optimizing costs and lowering costs for the consumer, you don't have a business.” The MD of Case C identified improved quality (value proposition) as a benefit: “…increased quality, meaning much more measurable quality”, while the MD of Case C identified improved production (value creation) as a benefit: “…increase productivity…And then also I would say even the throughput.”

The SMRA theme for the non-high-tech SMME emerged from the analysis of one code and one code group, as shown in Appendix B – Table B.5. In contrast to the high-tech SMMEs, the analysis of the SMRA theme for the non-high-tech SMME revealed that it found it difficult to explain the relative advantage of SM when asked to elaborate on the perceived value and benefits of SM from its own understanding. The MD of Case D could only manage to identify improved production as the relative advantage of SM: “I can say one of the benefits of adopting smart manufacturing would be consistent with our production. You see, the other thing is when you are working with people, sometimes they don't come to work, they're sick. But if a computer will not get sick as often, so once it's set up and it is serviced, the machines are serviced, production will continue.”

Smart manufacturing compatibility (SMC)

SMC emerged as a theme from the analysis of 10 codes categorized into three code groups, as shown in Appendix B – Table B.6. The analysis of this theme revealed that the high-tech SMMEs are making long-term capital investments in ADTs and associated skills, digitally enabled manufacturing infrastructure, and their own electricity from renewable energy sources despite barriers such as government policies, access to patient capital, corruption, and limited availability of ADT skills in the country. For example, the MD of Case B confirmed: “IoT plays a big role… we will have investment in terms of sensor development… [and] data analytics, machine learning will be massive investments … blockchain is going to be a combination of getting the right skills in terms of developers that are experienced and how one manages that, so that again there will be investment with technology partners.” The MD of Case C stated: “We want to implement the model ourselves where we decrease our dependence on the national grid… and start generating renewable energy… I'm looking for such a place where it's in a farm area where I can put the factory and have a massive solar facility, at least if I can generate two megawatts or one MW.”

The SMC theme further revealed that the high-tech SMMEs have embraced an entrepreneurial mindset (a growth mindset). The SMMEs strongly believe that they will not survive in the future manufacturing business environment unless they make the necessary long-term investments to transition from legacy manufacturing to SM. The CEO of Case A explained: “You cannot stay small, if you stay small you are dead… businesses work like that, you know, so you can't remain small forever. If not then you going to die and that's what I think is not well understood.” The MD of Case C added: “So, it is impossible to be in the manufacturing space and remain small, without automation, really, I don't see factories in the future becoming anything. I mean, it's either you do it or you're dead, you know.”

The SMC theme for the non-high-tech SMME emerged from the analysis of three codes categorized into two code groups, as shown in Appendix B – Table B.7. The analysis of the SMC theme revealed that, in contrast to the high-tech SMMEs, the non-high-tech SMME is not making any investments in elements or requirements of SM due to a lack of knowledge. The MD of Case D elaborated: “Well, I understand the importance of it, but it's not something that we have, we have really looked at like now…. one of the things is lack of knowledge about what smart manufacturing can do for your business, so a lot of research. I don't think small businesses really go out to understand what this is. Like I said, it's actually the first time that, like formally, I get to engage with someone about it ... But if you don't know, it doesn't become a priority.”

Self-directed digital learning (SDDL)

SDDL emerged as a theme from the analysis of 23 codes categorized into four code groups, as shown in Appendix B – Table B.8. It is observed from the theme (SDDL), that the high-tech SMMEs have a desire to learn and are intentional regarding learning about SM ADTs as well as other emergent 4IR digital technologies. This is confirmed by the CEO of Case A: “I am at least actively learning a lot…because I believe that blockchain will change how we do things in our business, and moving forward we need to understand it… I mean, right now, I'm learning about Web 3.0.” They display a high level of information acquisition competence, as evidenced by the MD of Case B: “Partners, so collaboration partners, in my instance, it would be universities. So, I engage the university quite a bit, and that is where the guys on both engineering and things like the data analytics, AI side of things.” Additionally, they demonstrate a high level of digital competence, as reflected in the statements of the MD of Case B: “I would say most structured kind of learning comes in, it's OK, a lot of research papers which are that I forgot to mention earlier, so there's a lot of kind of dissertations and PhDs, so the published documents that's available online… a vast array of online resources, so whether that be your MOOCS, the free, OK, free and sometimes there's a cost to it, online courses etcetera.”

The SDDL theme for the non-high-tech SMME emerged from the analysis of four codes categorized into four code groups, as shown in Appendix B – Table B.9. The analysis of the theme (SDDL) revealed that in contrast to high-tech SMMEs, the non-high-tech SMME has a very low interest in learning (desire to learn) about digital innovations, including SM and its related ADTs. They are not actively and deliberately seeking out information (learning with intention) about ADTs or any emergent 4IR technologies and are not using or accessing online platforms and resources to develop their knowledge competence of SM and its related ADTs, or acquire any information or knowledge related to digital innovation. The MD of Case D explained that, from their perspective, learning is ad hoc, reactive and happens once in a while: “So, once in a while, you try and find what is the best way of doing certain things. So, it's a developing thing every day, every day you learn, like today I'm learning smart manufacturing.” The MD also attested that they do not use any online platforms and resources as they largely rely on previous experiences for learning and developing knowledge competence: “No, I haven't done that. Most of what we are doing right now, I already like learned everything from my previous work. So, it was just to take whatever that was working and make it perfect.”. The MD further confirmed that they have limited digital competence when it comes to online platforms and accessing online resources: “I haven't went on Google to check it out like that. No, I don't know.”

Business model innovation (BMI)

BMI emerged as a theme from the analysis of 10 codes categorized into six code groups, as shown in Appendix B – Table B.10. It is extracted from the evaluation of the theme (SDDL) that the high-tech SMMEs are innovating their business models in multiple ways to exploit the relative advantages of SM. For example, the CEO of Case A is building a specific human resource capability to change the business model: “Cross qualification, so for instance, we try to look for people who can do multiple things, not just one thing… You have to have people who are specialists in multiple things.” Meanwhile, the MD for Case B is adopting AI/ML for data analytics and business intelligence applications to adapt the business model: “Okay, firstly capture as many data points as possible that's contextual and relevant to a process… start using things like machine learning and AI with that then actually start informing decisions… Then the second is adding business intelligence to that … add business logic and automate that business logic so that effectively, you have a platform that can inform you and understands your business model.”

The BMI theme for the non-high-tech SMME emerged from the analysis of one code and one code group, as shown in Appendix B – Table B.11. In contrast to the high-tech SMMEs, the analysis of the BMI theme for the non-high-tech SMME revealed that they are finding it extremely challenging to determine which components of their business model need adaptation to exploit SM. This is expected, as it has very limited knowledge competence of SM and how this innovation could be exploited to the benefit of the firm. When asked to elaborate on the changes that would be made to the current business model if SM was to be adopted, the MD responded: “Ohh, I don't know. I don't know how it will change now because, you see, it will not [change] because we started small.”

Entrepreneurial orientation (EO)

EO emerged as a theme from the analysis of five codes categorized into five code groups, as shown in Appendix B – Table B.12. It is evident from the appraisal of the theme (EO) that the high-tech SMMEs are proactively engaging in long-term, high-risk investments to strategically discover and exploit new market opportunities. These SMMEs are willing to commit significant investments and resources to market opportunities in uncertain environments, with highly unpredictable outcomes, to stimulate future market growth. For example, the CEO of Case A confirmed: “We take as much risk as we can tolerate. So, our view is that you have to continuously do that.” The MD of Case C added: “Basically, every opportunity has been a risk, every new product, it's basically being done at the risk. The factory is being done, I mean it's only now that we are gonna be getting people who want to manufacture in the factory, but it was all done without an order or without even a promise of an order… even if we lose, it doesn't stop the business.”

The EO theme for the non-high-tech SMME emerged from the analysis of five codes and five code groups, as shown in Appendix B – Table B.13. In contrast to the high-tech SMMEs, the analysis of the EO theme for the non-high-tech SMMEs, indicates that its focus is on short term objectives and goals (short-term orientation) and is not willing to proactively take risks to explore opportunities and growth in new markets. For example, when asked if the SMME would consider proactively pursuing and investing in long-term opportunities and projects, the MD of Case D responded: “…you know what, I would, I would really wait it out, I would not yeah. I would wait a bit and not really just jump to it, I would wait and see how it goes…”. When further asked if the SMME would consider pursuing high-risk, long-term opportunities and projects, the MD responded: “No, no, no, no, somebody must first go and get their hands burnt and then fingers burnt and then we'll come. It's a risk when you are still small, you know, don't try to take on an elephant at your first bite.”

Culture of innovation (CoI)

CoI emerged as a theme from the analysis of 10 codes categorized into three code groups, as shown in Appendix B – Table B.14. From the theme (CoI), it is determined that the high-tech SMMEs have a culture (values, norms, and behaviors) that facilitates organizational learning, for example, through self-management (Martínez-León & Martínez-García, 2011). This is evidenced by the CEO of Case A: “So, I spend a lot of time trying to recruit people, making sure that we recruit the right people so we don't have to manage anymore. So, we believe that people who are working for us self-manage. If they cannot self-manage then they should not be working here.” Additionally, they foster a culture that promotes innovation propensity, entrepreneurial spirit, creativity, and a risk-taking attitude (Bunyakiati & Surachaikulwattana, 2016). This is reflected in the CEO's statement: “We allow the environment where it is ideas that thrive not your position. That for me is important, that is part of our culture…It's not that kind of heavy management hierarchy, and that agile approach – it's how we just get on with doing things and getting work done. So, it's very much principles around agile working and also as a team.”

The CoI theme for the non-high-tech SMME emerged from the analysis of two codes and one code group, as shown in Appendix B – Table B.15. In contrast to the high-tech SMMEs, the analysis of the CoI theme indicates that the non-high-tech SMME appears to emphasize values, norms, and behaviors that focus on respect and restoration. This emphasis is highlighted by the MD of Case D: “I want the whole organization or company to be all about respect for our colleagues, respect that we give to each other, respect that we give in what we do… we are also all about restoration…I believe you get restored when your basic conditions are sort of somehow met at a certain level, so restoration is one of our culture elements that we are bringing in.”

SMME readiness for smart manufacturing (SMMERA)

SMMERA emerged as a theme from the analysis of two codes categorized into one code group, as shown in Appendix B – Table B.16. The analysis of the theme (SMMERA) reveals that the high-tech SMMEs are currently investing in SM and using SM ADTs such as cloud computing, the IoT, big data analytics, and smart sensors. When the high-tech SMMEs were asked whether they would be investing, adopting, and implementing SM within the next three years, the MD of Case B responded: ‘Yes, solidly yes… Umm, the next three years. I would say, a minimum of R10 million”, and the MD of Case C added: “Yeah, we will be, yeah.” According to Mittal et al. (2018b, 2018c), when an SMME has spent sufficient time, effort, and resources in practicing the SM paradigm and has developed new SM capabilities, the SMME is considered to be at an intermediate level of readiness for SM adoption. This indicates that the high-tech SMMEs are at an intermediate level of readiness and only one step away from being ready to implement SM (expert level).

The SMMERA theme for the non-high-tech SMME emerged from the analysis of two codes and one code group, as shown in Appendix B – Table B.17. In contrast to the high-tech SMMEs, the analysis of the SMMERA theme indicates that the non-high-tech SMME is not currently investing in SM, is not using SM ADTs, and is not aware of the concept of SM. When the non-high-tech SMME was asked whether it would be investing, adopting, and implementing SM within the next three years, the MD of Case D responded: “Ok, currently we are not ready, within the next three years, I have to make this business work.”Mittal et al. (2018b, 2018c) believe that when an SMME is largely unaware of the SM paradigm and SM relative advantage, the SMME is considered to be at a novice level of readiness for SM adoption. This means that the non-high-tech SMME is at the very lowest level of SM readiness for adoption.

Step 3: Thematic network analysis: linking themes (connecting constructs)

Innovation adoption is a phase-based process comprising the initiation phase, the adoption decision phase, and the readiness for the adoption phase (Fig. 4 in Section 3). In the initiation phase, knowledge competence associated with the innovation to be adopted is developed (Gumbi & Hossana, 2022). In the adoption decision phase, the assessment of the innovation's relative advantage from a practical, strategic, financial, and/or technological perspective for resource allocation, as well as whether the innovation supports the goals and objectives of the firm, is conducted (Gumbi & Hossana, 2022). In the readiness for adoption phase, the organization is prepared (made ready) for the innovation's adoption and general use (Gumbi & Hossana, 2022). Thus, for SMMEs to be ready for SM, SMKC needs to be developed in the initiation phase, the SMRA (value/benefits) needs to be assessed in the adoption decision phase, and the required SM investments for resource allocation need to be made in the readiness for adoption phase for SMC.

According to Gumbi and Hossana (2022), SDDL is used as a primary method and core competence in the development of knowledge competence related to digital innovations and ADTs. The authors further explain that for SMMEs to realize (perceive) SMRA, they need to innovate (change or adapt) their current business models (BMI). Gumbi and Hossana (2022) also identified EO as a key factor influencing SMC. According to the authors, a firm's SDDL, BMI, and EO are likely to emerge, develop, and mature in the presence of a CoI. Thus, a CoI is a key factor influencing SDDL, BMI, and EO. Based on the adapted innovation adoption process (Gumbi & Hossana, 2022), which is described in detail in Section 3, the linkages between the emergent themes were established through TNA (Appendix C – Fig. C.1 to Fig. C.11) and network diagrams, as shown in Appendix C – Fig.s C.12 and C.13.

Proposition 3

(P3): There is a positive relationship between BMI and SMRA.

A broad-scale, full implementation of any innovation requires a firm to test, use, and confirm the relative advantage of the innovation in alignment with the strategic objectives and goals of the organization. Long-term strategic investments, such as infrastructure, skills, and technologies need to be made to assimilate the innovation. These strategic investments must be made proactively, taking high risks in anticipation of future business growth. Thus, SMME SMC can only be achieved when SMMEs comprehensively understand the relative advantage of SM (SMRA), perceive it as outweighing the risks and costs of adoption, and proactively make long-term high-risk investments in anticipation of future growth (EO).

The high-tech SMMEs comprehensively understand SMRA and have high levels of EO, as shown in Appendix C – Fig. C.12. The non-high-tech SMME does not understand or know SMRA and has a very low level of EO, as shown in Appendix C – Fig. C.13.

It is therefore proposed that:

Proposition 4

(P4): There is a positive relationship between SMRA and SMC.

A CoI is a multi-dimensional organizational culture characterized by values, norms, and behaviors that promote risk-taking, market orientation, value orientation, innovation propensity, and organizational learning (Nadkarni & Narayanan, 2007; Dobni, 2008; Plambeck & Weber, 2009; Anderson & Harbi, 2012; Bock et al., 2012; Sharifirad & Ataei, 2012; Aksoy, 2017). The high-tech SMMEs are actively engaging in SDDL practices and processes to develop SMKC (organizational learning), as shown in Appendix C – Fig. C.12. These SMMEs are also innovating their business models (BMI) to realize SMRA (innovation propensity) and engaging in EO (risk taking and innovation propensity) decision-making processes and practices to achieve SM compatibility (SMC) (Appendix C – Fig. C.12). The SMMEs’ engagement in SDDL practices and processes is influenced and facilitated by an organizational culture fostering values, norms, and behaviors such as self-management, agile organization, collaboration, innovation propensity, knowledge and information sharing, trust and accountability, transparency, and a flat organizational structure (Appendix C – Fig. C.12). Similarly, their engagement in BMI is supported by values, norms, and behaviors that promote innovativeness, while engagement in EO processes and practices is influenced and facilitated by values, norms, and behaviors that encourage risk-taking, creativity, and proactiveness (Appendix C – Fig. C.12). High-tech SMMEs appear to have cultivated a CoI within their organizations, leveraging it to drive practices that orientate their enterprises toward SDDL, BMI, and EO processes. Conversely, the non-high-tech SMME emphasizes values, norms and behaviors focused on respect and restoration, as shown in Appendix C – Fig. C.13. This SMME has not cultivated any values, behaviors, and norms that foster a CoI, nor is it engaging in practices that promote SDDL, BMI, or EO.

It is, therefore, proposed that:

Proposition 7

(P7): There is a positive relationship between SDDL and a CoI.

RQ4 sought to understand what constitutes a suitable conceptual framework for SMMEs' readiness for SM adoption in the manufacturing sector. The resulting empirically developed conceptual framework for SMME readiness for SM adoption is illustrated in Fig. 11. The TA conducted in this study empirically confirmed all the constructs initially conceptualized in the research model (Fig. 4), namely, SMKC, SMRA, SMC, SMMERA, SDDL, BMI, EO, and a CoI, as shown in Fig. 11. Additionally, the TA identified 11 new subconstructs from the empirical data: Use of Advanced Digital Technologies (UADT), Key Advanced Digital Technologies for SMMEs (KADT), Barriers to Long-Term Investments (BLTI), Growth Mindset (GM), New Market Orientation (MO), Long-Term Orientation (LTO), Flat Organization Structure (OS), Knowledge and Information Sharing (KS), Collaboration (C), Trust (T), and Transparency (TR). These subconstructs are integrated into the framework, as shown in Fig. 11. The TA also led to the development of measurement items for all the subconstructs included in the framework. Measurement items were not identified or included in the emergent phase's conceptualized research model (Fig. 4). Using TNA and abductive reasoning, all nine research propositions were empirically confirmed, aligning with the constructs conceptualized in the research model (Fig. 4) and incorporated into the framework (Fig. 11).

Fig. 11.

Developed SMME readiness framework for SM adoption.

Critical realism is grounded in a tripartite, stratified ontology, divided into the domains of the real (structures, objects, and mechanisms), the actual (events), and the empirical (experiences) (Mcevoy & Richards, 2006; Modell, 2009; Zachariadis et al., 2013; Mukumbang, 2023), as discussed in the research and methodology section. The real domain refers to the social and political structures that possess causal powers, which activate mechanisms capable of generating events in the actual domain (Modell, 2009; Mukumbang, 2023). The actual domain represents the events or non-events that occur in the real domain (Mukumbang, 2023). These events are generated by causal mechanisms present in the real domain. The empirical domain encompasses observable events that are a subset of the actual, which are related to human perception, individual actions, and experiences (Mukumbang, 2023). This domain includes the information that becomes known through direct or indirect human experiences tied to the actual domain (Mukumbang, 2023). From a critical realist perspective, the SMME readiness framework for SM adoption (Fig. 11) was decomposed into these critical realist domains, as shown in Fig. 12.

Fig. 12.

Critical realist domains of manufacturing SMMEs.

RQ2 sought to identify the preconditions for SMME readiness for SM adoption in the manufacturing sector. SMKC, SMRA, and SMC are identified as the preconditions (actions that must be taken by SMMEs) within the empirical domain, as shown in Fig. 12.

RQ3 aimed to identify the factors influencing the preconditions for SMME readiness for SM adoption. SDDL, BMI, EO, and a CoI are identified as the prevailing structures and mechanisms (factors) within the real domain. These structures possess causal powers capable of influencing the preconditions (SMKC, SMRA, and SMC) for SMME readiness for SM adoption, as shown in Fig. 12.

RQ1 focused on identifying the level of SMME SM readiness for adoption in the manufacturing sector. The high-tech SMMEs are at an intermediate level of SM readiness and only one step away from expert-level implementation of SM, as uncovered by the TA in subsection H. These high-tech SMMEs are ready to implement SM within the next three years (an event in the actual domain) and are planning to invest, on average, a minimum of R10 million, as shown in Fig. 12.

When a CoI is present within the manufacturing SMME business environment, SDDL, BMI, and EO are activated. Conversely, when a CoI is absent, SDDL, BMI, and EO are deactivated. A CoI is an enduring structure with generative powers in the real domain (Fig. 13). SDDL, BMI, and EO are the causal mechanisms that prevail in the real domain when the generative powers from a CoI are triggered or activated (Fig. 13). SMKC (development of SM knowledge competence), SMRA (comprehensively understanding the benefits and value of SM), and SMC (proactively making high-risk investments in SM), are the actions that are triggered in the empirical domain when causal mechanisms are activated (Fig. 13).

Fig. 13.

A critical realist explanatory theory for SMME readiness for smart manufacturing adoption.

The effects of generative powers are contextual and dependent on the structures and conditions that hold these powers (Mukumbang, 2023). Thus, context has the potential to change the outcome by either activating or deactivating these generative powers (Mukumbang, 2023). This suggests that mechanisms do not always produce the same outcome in different contexts (Mukumbang, 2023). In the context of the non-high-tech SMME, the inhibiting environment prevents the SMME's actions (that is, SMKC – development of SM knowledge competence, SMRA—comprehensive understanding of the benefits and value of SM, and SMC – proactively making high-risk investments in SM) from contributing to the transformation of the organizational culture into a CoI by deactivating the generative powers. As a result, the CoI generative powers remain dormant, and the causal mechanisms (SDDL, BMI, and EO) remain inactivated. The inactivation of these causal mechanisms means that SMME readiness for SM adoption remains unobservable or does not occur.

The adoption of the SM paradigm is essential for SMMEs, as the rise of global pandemics like COVID-19 and the onset of the 4IR pose significant threats to the future sustainability, competitiveness, and viability of manufacturing firms in the business and operational environment. While the high-tech SMMEs are considered to be at an intermediate level of readiness for SM adoption and are already adopting and implementing the SM concept, the non-high-tech SMME is still at a novice level of readiness and is not yet prepared for SM adoption and implementation.

The study found that existing frameworks are unsuitable for supporting many manufacturing SMMEs in SM adoption, as they fail to address context-specific preconditions relevant to SMMEs. These frameworks were primarily developed for high-tech industry firms, corporations, and large multinational companies. Consequently, these frameworks are disconnected from the digitalization capabilities and SM maturity levels of many SMMEs, particularly non-high-tech ones. This study proposes a suitable framework for SMME readiness for SM adoption, incorporating SMKC, SMRA, and SMC as preconditions, with SDDL, BMI, EO, and a CoI as factors influencing these preconditions. In the framework, a CoI is empirically identified as an enduring structure with generative powers in the real domain, and SDDL, BMI, and EO as the causal mechanisms that prevail in the real domain when generative powers from a CoI are triggered or activated. When these causal mechanisms are activated, the development of SMKC, a comprehensive understanding of SMRA, and high-risk proactive investments for SMC occur in the empirical domain for SMME readiness for SM adoption (event in the actual domain).

Two primary limitations were identified in this study. The first is that the case studies used to construct the SMME readiness framework for SM adoption were limited to South Africa. The second limitation is that the sample size (four case studies), while adequate in terms of data saturation, comprised a higher proportion of high-tech SMMEs compared to non-high-tech SMMEs. This imbalance was influenced by factors such as COVID-19, new data privacy policies, and the reluctance of non-high-tech SMMEs to participate in the study due to limited knowledge or understanding of the concept of SM. To address these limitations, a critical realist quantitative research (confirmatory phase) study is currently underway to test, validate, and generalize the developed SMME readiness framework for SM adoption.

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Lucas Gumbi: Writing – review & editing, Writing – original draft, Visualization, Validation, Software, Resources, Project administration, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Hossana Twinomurinzi: Writing – review & editing, Supervision, Resources, Funding acquisition.