The effect of adopting artificial intelligence
technologies on intrapreneurial behaviour
in Small Medium Enterprises
Msc Business Administration
Small Business and Management
Tjarko Dijkstra
Master Thesis
June 2019
University of Groningen
Faculty of Economics and Business
2
Abstract
This thesis investigates the role of artificial intelligence (AI) technology on intrapreneurial behaviour in Small Medium Enterprises (SMEs). While past research mainly focussed on the antecedents of intrapreneurship and individual characteristics regarding intrapreneurship, this research focusses on the mechanisms which cause intrapreneurial behaviour to occur. In this study the critical realist perspective (Wynn and Williams, 2012) is used to examine the effect of AI technology adoption on intrapreneurial behaviour in SMEs. Using the retroduction process first affordances of the technology in its context are retroduced, after which the mechanisms are proposed. From the analysis three different mechanisms influencing intrapreneurial behaviour are proposed: embedded routine mechanism, structured data mechanisms, and artificial intelligence mechanism.
3
Inhoudsopgave
Abstract ... 2 1. Introduction ... 5 2. Literature review ... 7 2.1 Intrapreneurship in literature ... 72.2 Intrapreneurship and SMEs ... 9
2.3 Materiality, socio-materiality and technical subsystem ... 10
2.4 Theoretical perspective ... 12
3. Methodology ... 15
3.1. Case study research ... 17
3.2. Research design ... 17
3.3. Data collection methods ... 18
3.4. Case description ... 19
3.5. Data analysis ... 22
4.Retroduction of affordances ... 24
5. Proposed mechanisms ... 28
6. Theory building ... 31
7. Discussion and conclusion ... 32
7.1 Limitations ... 33
7.2 Conclusion ... 33
Bibliography ... 33
Appendix A: Semi-structured interview guide ... 37
Appendix B: Event description Company_A ... 38
Appendix C: Event description Company_B ... 38
Appendix D: Event description Company_C ... 39
Appendix E: Event description Company_D ... 39
Appendix F: Example coding ... 41
Appendix G: Codebook axial coding ... 42
Appendix H: Axial coding Company_A ... 44
Appendix I: Axial coding Company_B ... 45
Appendix J: Axial coding Company_C ... 46
Appendix K: Axial coding Company_D ... 47
Appendix K: Retroduction memo’s ... 48
Appendix L: Transcript interview Company_A ... 50
4
Appendix N: Interview transcript Company_C ... 58
Appendix O: Interview transcript of Company_D ... 61
Appendix P: Observations Company_A ... 64
Appendix Q: Observations Company_B ... 65
Appendix R: Observations Company_C ... 66
5
1. Introduction
In today’s dynamic economy, organisational change is seen as vital for organisations to survive. Especially in competitive markets organisational change is very important (Hirsch and Kirch 2012). There are many different drivers of- and ways for organisational change, among which the
introduction of IT (Walsham et al. 1996) and intrapreneurship (Kuratko et al., 2005). Recently the development of artificial intelligence (AI) technology and machine learning has come to the point of automating processes previously done by humans, even for smaller companies. With enough data AI technology and machine learning can make better predictions than human. When adopting AI technology in professional service firms (PSF) the current routines could be impacted. Professional service firms are defined by being knowledge intense, low capital intense and having a
professionalised workforce (Nordeflycht, 2010). Adopting AI technologies result in changes to the organisation’s work context and the nature of the routines performed within the organisation. Even though technologies possess the same aspects, the purpose for which it is used and how it is used is determined by its user. As work context is one of the enabling antecedents of intrapreneurship (Rigtering and Weitzel, 2013), changes in the intrapreneurial behaviour could be expected. In current literature little is known about the way AI technology adopted by organisations influence the
intrapreneurial behaviour of employees.
Intrapreneurship is known as entrepreneurial behaviour in existing organisations by its employees (Bal et al., 2007, Antoncic and Hisrich, 2003). In intrapreneurship an employee is making use of its creative skills in creating and introducing new products, services, processes or organisational systems (Covina and Slevin, 1991; Lumpkin and Dess, 1996; Zahra, 1996). Intrapreneurship shows through an employee identifying opportunities, subsequently shaping and developing these opportunities for the organisation he works in (Thornberry, 2001). The starting point in this process of
intrapreneurship is a new idea (Thornberry, 2001).
Intrapreneurship is important for organisations as it is a source for innovation which could improve its financial performance (Raug et al. 2009). Accompanying intrapreneurship is recognising
opportunities and exploiting them help to create new value for firms, failing to recognise new opportunities may harm a firm’s performance (Gertz and Baptista, 1996). Especially when competitors are innovating the risk of decreasing performances is high. Having intrapreneurship within a firm is a predictor of performance in small firms (Covin and Slevin, 1989).
6
have a professional workforce. As most online marketing consultancy companies are in the category small medium enterprises this study will focus on the online marketing consultancy companies of this category making the research practically relevant. Focussing on this group will also give practical relevancy as 99% of the organisations in the Netherlands are SMEs (Kamer van Koophandel, 2018). Research on intrapreneurship is often focussed on antecedents enabling intrapreneurship and what traits and characteristics of an employee strengthen the tendency for intrapreneurship. These researches have mainly been done in MNE’s (multinational enterprises), as intrapreneurship is also seen as a strategy for companies to withstand competition of start-ups threatening the existing organisation. In existing literature little research is done on intrapreneurship in SME’s which are passed the start-up phase. As intrapreneurship can help an organisation to be resistant to start-ups and help a company grow, the effect of intrapreneurship in existing SMEs past the start-up phase is very interesting.
Previous research has shown a positive supportive effect of technology on intrapreneurship (Antoncic and Hisrich, 2001). However, most researches on the effect of technology on
intrapreneurship is of quantitative nature. Subsequently the mechanisms through which technology in SMEs is influencing intrapreneurship is under exposed. This thesis will be an addition to literature by investigating the mechanisms through which AI technology in SMEs is influencing
intrapreneurship. Following the exploratory research question is central in this study:
Through what mechanisms is introducing artificial intelligence technology in SMEs influencing intrapreneurial behaviour?
7
2. Literature review
2.1 Intrapreneurship in literature
In previous research, intrapreneurship is being investigated using different definitions.
Intrapreneurship can be described as entrepreneurial behaviour within an existing organisation (Bal et al., 2007). Vesper (1990) has defined intrapreneurship as doing new things and depart from the customary to pursue opportunities. Intrapreneurship has been defined, by Stevenson and Jarillo (1990), as a process in which an individual inside an organisation is pursuing opportunities
independent of the resources within their control. For this research a broad definition will be used as ‘broad definitions better reflect the early stage of development of the field, avoid the need for
excessive retrenchment as new knowledge becomes available, and provide considerable latitude for a theoretical and empirical process to emerge that will eventually permit the unique parts of the whole to be classified, defined, and understood in relation to that whole’ (Sharma and Chrisman, 1999). In
this thesis the definition of intrapreneurship of Antoncic and Hisrich (2003) is used:
Intrapreneurship is defined as entrepreneurship within an existing organisation, referring to emergent behavioural intentions and behaviours of an organisation that are related to departures from the customary.
In this definition no difference is made regarding the size of the firm. Intrapreneurship is not only referring to new business creation within organisations as a devision of the current organisation, but also referring to other innovative activities and orientations within the organisation. These innovative activities and orientations include development of new business ventures, process innovation, product innovation, self-renewal, risk taking, pro-activeness, and competitive aggressiveness (Antoncic and Hisrich, 2003). Important to notice is that with intrapreneurship, the employee is operating at the boundaries of the core of the organisation. The intrapreneurial activities and orientations deviate from the existing routines, production- and support operations (Sandberg, 1992).
8
Dimension Definitions Theoretical grounds
New Ventures Creation of new autonomous or semi-autonomous units or firms Scholhammer(1981) Hisrich and Peters (1984 MacMillan et al. (1984) Vesper (1984)
Kantor and Richardson (1991) Stopford and Baden-Fuller (1994) Sharma and Chrisman (1999) New business
Persuit of and entering into new businesses related to current products
and markets Rule and Irvin(1988)
Zahra (1991)
Stopford and Baden Fuller (1994) Product/service
innovativeness Innovations in production procedures and techniques Scholhammer (1982) Covin and Slevin (1991) Zahra (1993)
Damanpour (1996) Burgelman and Rosenblom (1997)
Knight (1997)
Tushman and Anderson (1997) Process innovation Innovations in production procedures and techniques Scholhammer(1982)
Covin and Slevin (1991) Zahra(1993)
Damanpour (1996) Burgelman and Rosenblom (1997)
Knight (1997)
Tushman and Anderson (1997) Self-renewal Strategy reformulation, reorganisation and organisational change Vesper (1984)
Guth and Ginsberg (1990) Zahra (1991, 1993)
Stopford and Baden-Fuller (1994) Muyka et al. (1995)
Sharma and Chrisman (1999) Risk taking
Possibility of loss related to quickness in taking bold actions and
committing resources in the pursuit of new opportunities Mintzberg (1973) Khandwalla(1977) Miles and Snow(1978) Covin and Slevin (1986, 1989, 1991)
Stopford and Baden-Fuller (1994) Dess et al. (1996)
Lumpkin and Dess (1996, 1997) Lumpkin (1998)
Proactiveness Top management orientation for pioneering and initiative taking Covin and Slevin (1986, 1991) Venkatraman (1989)
9
Dess et al. (1997) Lumpkin (1998) Competitive
Aggressiveness Aggressive posturing towards competitors Covin and Slevin (1986, 1991) Miller (1987)
Covin and Covin (1990) Lumpkin and Dess (1996, 1997) Knight (1997)
Lumpkin (1998)
Table 1: Intrapreneurial dimensions including definition (Antoncic and Hisrich, 2003) Different circumstances have an influence on the employee showing intrapreneurial behaviour. Amongst these circumstances is the formal work context. Previous research has shown that the organisational structure is an important antecedent of intrapreneurship (Hayton 2005; Kanter 1985). When relating the structure of an organisation to an individual, Mintzberg (1993) has argued there are two parameters, being task specialisation and formalisation. Task specialisation can be divided into job task specialisation and the degree to which one holds responsibilities. Formalisation is the degree to which organisations control employees through formal job descriptions, procedures and rules (Mintzberg, 1993). The higher the formalisation of an organisation the more rigid this
organisation is. Contrary, lower formalisation in an organisation is related to flexible organisations. The degree of formalisation in organisations has an inhibiting effect on the degree of intrapreneurial behaviour (Zahra, 1991; MacMillan et al., 1986). The more formalised the organisation, the lower the degree of intrapreneurial behaviour.
Next to the formal work context is the organisational support and organisational values seen as important antecedents of intrapreneurial behaviour. With regard to intrapreneurship, support of superior managers is beneficial to intrapreneurial behaviour (Merrifield, 1993). Subsequently, if the values and believes in the organisation are open and supportive for intrapreneurial behaviour, it is more likely to show (Guth and Ginsberg, 1990).
Besides the formal work context and organisational support, resources available to the employee are important antecedents for intrapreneurship (Marvel et al., 2007; Day, 1994). Most important
resources identified for intrapreneurship are money (Menzel et al., 2007) and time (Knight, 1987). These resources give the employee the possibility of following up his novel ideas.
2.2 Intrapreneurship and SMEs
10
The focus of this thesis will be on the effect of adopting AI technologies online marketing consultancy agencies in the SME category. It is important to notice that besides size environmental aspects can influence the organisational structure of the environment (Meijaard et al., 2002). However, regardless the environment SMEs have certain unique characteristics regarding organisational structure compared to large companies. The following section will discuss the, most relevant, unique characteristics of SMEs’ organisational structure and compare it to large companies.
One of the commonalities between organisational structures of SMEs is it being flat and flexible. This characteristic shows in most SMEs by being centralised in decision making and little organisational levels. Flexibility is mostly recognised by the blurred departmental walls, employees hold different responsibilities and roles related to different departments (Huin, 2004; Hudson, Smart and Bourne, 2001). In large companies the structure tends to be more rigid and job responsibilities and roles are more formalised. This difference in organisational structure could result in different effects of implementing IT in SMEs. For example, if the implementation of IT automates certain routines. The automation of the routine could result in a shift of job role within the SME whereas in the large organisation the employee could become redundant while in SMEs the organisation is more flexible, offering the employee the opportunity to develop new ideas based on the new technology.
Another unique characteristic of SMEs structural organisation is the low level of institutionalisation of routines. Large companies tend to have a higher degree of institutionalisation of routines, which means that the way routines are performed is according to predetermined steps. Regarding SMEs the institutionalisation of routines is low, allowing for different ways of performing a routine. For large companies the implemented IT could embed certain institutionalised routines. Whereas the adoption of IT in SMEs could not only embed certain routines, they can also demand the employee to follow certain steps for the technology to work. The latter possible effect of the implementation of IT involves a higher level of institutionalisation of routines in SMEs, making the organisational less flexible and would hinder intrapreneurial behaviour.
Due to the low level of institutionalisation, in SMEs, the employees possess a higher degree of freedom and are allowed more initiative in how the routines are performed (Storey et al., 2010). This higher degree of freedom could also be reflected in the intrapreneurial behaviour shown by
employees. Where in large firms the performing of routines is institutionalised the employee in the SME decides himself how to work with the technology.
2.3 Materiality, socio-materiality and technical subsystem
11
technology has been studied thoroughly and different concepts of technology have been used in literature. In studying the effect of technology on organisations different terms have been used to define technology. The terms materiality, socio-materiality and socio-technical systems have been used in technology mediated change studies (Leonardi, 2012). All these terms can help in
understanding technology and the impact on organisations. To get a better understanding of the terms, its meaning and the relevance to this study is explained in the next sections.
Materiality is used to describe the aspects of technology which do not change regardless of time and place (Leonardi, 2012). As an example, a paperclip’s materiality is the iron it is made of, this does not change regardless the environment. However, in the office it is used to keep paper together and in the phone store employees can use it to open the simcard slot of a smartphone. The physical artefacts of the paperclip stay the same however its use differs over the different circumstances. For digital technology it is hard to observe the materiality since it is not tangible. However, the coding of the technology is the materiality of the technology. Regardless of the organisation in which the technology is implemented, SME or large organisation, the materiality will stay the same. Whether and in what manner the materiality is used determines the outcome of the technology in use. Knowing that technology could be used differently in distinct situations the social-material concept is introduced. To understand the effect of socio-materiality, a distinction is made between material agency and social agency (Leonardi, 2012). As discussed before, the materiality does not change over time and space, but the context in which it is used can change its purpose. It is the social agency, the intentional actions of the user, which enables the material agency, the way the technology acts. To use the example of the paperclip; the paperclip could only be used to open the simcard slot (material agency) because of the input of the employees of the phone store (social agency). The interaction of the social agency and material agency determine the effects of the technology in use. It is important to recognise the social agency interacting with the technology to understand the material agency which it evokes. In other words, it emphasises the role of the user in relation to the technology and the purposes it is used for. The space in which the technology’s user and the technology interact through social and material agency is known as the technical subsystem (Leonardi, 2012). When relating this concept to introducing technology in organisations and its effect on intrapreneurial behaviour. A technology can bring new material aspects which can be used in different ways. The social agency can use this technology in new ways to the organisation or in a manner which is completely different from what it originally was designed for resulting in
12
the paperclip for a purpose it was not designed for. Using the paperclip in this way replaces the need for a special tool to open the simcard slot. In the case of introducing new technology an employee could see possibilities, new ideas, for the organisation he works for. When he would develop these ideas, made possible by the new technology, intrapreneurial behaviour shows.
Figure 1. Conceptual model on AI technology enabled intrapreneurial behaviour
Figure 1 shows the concept of socio-materiality for this study. The intentionality is coming from the employee who is working with the technology, the materiality. During the interaction between the employee, enforcing his intentionality, with the materiality of the AI technology, a certain outcome is accomplished. In this study, the outcome of the interaction is being investigated, focussing on intrapreneurial behaviour.
2.4 Theoretical perspective
As both intrapreneurship and implementing technology in an organisation are sources of organisational change, this thesis will draw from organisational change perspectives. In
13
SMEs, a perspective is needed which treats the structure of the technology and agency conceptual different. This allows for an analysis of the effect of the two on each other. This section briefly describes the possible theoretical perspectives for this thesis, after which a thorough explanation on the theoretical perspective of this thesis is given.
In previous research different perspectives have been used to investigate the role of technology in organisational change. It is important to understand that the theoretical perspective chosen is colouring the eventual analysis. In technology mediated organisational change research different perspectives have been used with varying success. For example, the human agency perspective (behavioural approach) is focussing on the role of the people involved, arguing that an enterprise system can be reinvented or resisted by employees (Boudreau and Robey, 2005). Using this
theoretical perspective weakens the role of technology in organisational change as users redefine the technology through their actions. In case of AI technology, the human agency perspective
emphasises the role of the employee in technology mediated change. As technology is not only subject to its users but is also capable to change behaviour in organisations this perspective is not suitable for this thesis. A perspective assigning a more prominent role to the technology in
technological mediated change is institutional theory. This perspective argues that the technology is institutionalised during configuration, carrying the institutional logic. According to this perspective human agency is actively constrained by technology and leaves little room for the user to influence the effect of the technology. Other perspectives, like actor network theory (ANT), do treat agency and structure equal. However, they conflate agency and structure which hinders investigation on the effect of both on each other. As the model in figure 1 shows, structure exist independently of agency and therefor needs to be treated conceptually different.
As this thesis is built on the concept of socio-materiality, it is important to use a theoretical
perspective which keeps social agency and structure conceptually different. Keeping the social and structure conceptually different allows for discussing their effect on each other. In this research this concerns the effect of the adopted technology on the user of the technology and vice versa. Both the technology as well as the user of the technology could be the source of the intrapreneurial
14
The main focus of CR-based research is to provide clear and concise statements about causation, supported by empirical evidence (Wynn and Williams, 2012). The perspective is especially focussed on explaining how and why phenomenon occur (Wynn and Williams, 2012). Causation within CR does not imply the succession of events or direct correlations, but it is concerned with describing the process through which the events are caused (Wynn and Williams, 2012).
While CR is treating structure and agency as conceptually distinct, in reality agency and structure cannot exist without each other. In reality practice is influenced by the social structures in place and the other way around. However, social structure is influencing practice after which structure could be altered because of the practices (Bhaskar, 1979; Collier, 1994). The technology is designed with particular ostensive aspects in mind, it is the performative aspects which determine whether adaptions have to be made or opportunities arise for intrapreneurial behaviour.
CR perspective is a research paradigm that believes in a stratified reality (Wynn and Williams, 2012). In CR, reality is divided in three domains, being the real, actual and empirical (Bhaskar, 1979). For this research, the stratified reality is important to understand since the empirical data will not directly explain the role of the new technology.
‘The real consist out of the entities and structures of reality and the causal powers inherent to them as they independently exist ‘(Wynn and Williams, 2012).
‘The actual is a subset of the real, including the events that occur when causal powers of structures and entities are enacted regardless of whether or not these are observed by humans’ (Wynn and Williams, 2012).
‘The empirical is a subset of the actual and consist out of those events which are able to experience via perception or measurement’ (Wynn and Williams, 2012).
The stratified reality as explained above is structured in such a manner that mechanisms which are present in the real do not have to influence the actual, while they do exist. This also accounts for the actual, in which events could occur which are not experienced in the empirical. The domain of the empirical is concerned with the events which can be observed.
Using the CR-perspective makes it possible to retroduce the events observed in the empirical to mechanisms in the real. These mechanisms give us an understanding of how the adopted technology could affect intrapreneurial behaviour in organisation.
15
mechanism and its definition is quite abstract first the affordances of the technology are retroduced. Mechanisms are inherent to physical and social structures, constraining or enabling events to occur in a given context (Sayer, 2000; Smith, 2006). The definition of mechanisms is quite abstract which makes it hard to identify mechanisms. Therefore, the concept of affordance will be used to help identify the mechanisms in place. The concept of affordances consists of the same principles of mechanisms however it is more concrete in its description and therefore more actionable in research.
Affordances are defined as ‘the potential for behaviours associated with achieving an immediate concrete outcome and arising from the relation between an object and a goal-oriented actor or actors’ (Bygstad et al., 2016). Just as mechanisms, affordances can exist without being actualised. Whether the affordance is being actualised, or not, is dependent on the actor and its interaction with the object. In this thesis the affordance emerges from the relation between the implemented
technology and its user within an SME. Whether the affordance does occur in different but similar circumstances depends on whether the different users perceive the same affordances in the technology (Hutchby, 2001). For an affordance to be actualised the actor needs to have the
knowledge and competence. This entails that the affordance does not only relates to the technology but also to the actor and the circumstances. The affordance relates to the possibilities arising from the implemented technology interacting with its user resulting in an immediate concrete outcome. An affordance itself is an ever-present potential for action, the way it is actualised is dependent on the circumstances (Bygstad et al., 2016). Since the actualisation of affordances is dependent on the actor and circumstances the outcome of it is indeterminate. Next to this, the affordance is only being actualised when it would serve the objective of the actor. Due to the detailed definition of
affordances it can be seen as a subset of mechanisms. This does not mean that the affordances found will automatically be mechanisms. Affordances are the building blocks of more complex abstract mechanisms. These mechanisms can be derived from the interaction of several affordances. In the next part the methodology on how to retroduce the affordances and proposed mechanisms will be discussed.
3. Methodology
16
structures, mechanisms, events and experiences; an open system perspective; and emergent powers dependent upon but not reducible to lower level powers(Sayer, 2010; Collier, 1994; Bhaskar, 2013; Danermark et al., 2002). Important in critical realism is the stratification of reality into three nested domains (Bhaskar, 2013). Since it is assumed that mechanisms operate in an open system, the mechanisms activated may or may not be measurable or experienced. The mechanisms in the real could not be activated or counteracted by another mechanism which result in no events in the real. Mechanisms activated in the real do not necessarily be experienced in the empirical (Outhwaite, 1998; Bhaksar, 2013). Critical realist’s purpose is to uncover hypothesised mechanisms which could explain a given set of events to occur (Bhaskar, 1975; 1998). Besides this the purpose of using the critical realist perspective is to develop explanation for the way things act and how they are capable of doing so (Wynn and Williams, 2012).
The critical realism perspective is well suited for investigating the effect of adopting AI technologies on intrapreneurial behaviour because it adopts an open system perspective and treats the
technology’s materiality as a separate entity (Volkoff et al., 2007). This makes it possible to study the effect of the digital marketing technology’s materiality on the job role of the user.
As discussed, will this thesis use the concept of mechanisms and affordances to describe the effect of AI technology adoption on intrapreneurial behaviour. The affordance lens explicitly involves the technology and its user, while keeping them distinct. To identify the affordances and eventually propose the mechanisms influencing intrapreneurial behaviour within the SMEs the framework for CR data analysis of Bygstad et al. (2016) will be used. The framework identifies the critical steps involving CR data analysis and is based on extend literature of recent CR literature (Sayer, 1992; Danermark et al., 2002; Delanda, 2006; Volkoff and Strong, 2012; Wynn and Williams).
The first step is the description of events and issues, involving the clusters of observations made by the researcher or respondent of the interview (Sayer,1992; Volkoff et al., 2007). The second step involves the identification of key entities, which are the object of the case (Danermark et al., 2002). To identify the key entities the framework depicted in figure 1 will be used.
Thirdly the process of retroduction is proposed. Retroduction is a mode of inference which explains the events by the mechanisms which could have produced them (Sayer, 1992). During retroduction the unexplained phenomenon is explained through hypothetical mechanisms, which logically
explains the event under study (Minders, 2004). This step is set up in 4 phases. Firstly, the immediate concrete outcomes of implementing the technology is identified (Volkoff and Strong., 2013). The immediate concrete outcome is identified as a direct achievement through the use of the
17
user and the technology (Volkoff and Strong, 2013). After this phase the candidate affordances can be identified, which arise from the relationship between the user and the technology (Volkoff and Strong, 2013). These affordances reflect the potential for behaviour resulting in the concrete outcomes represented in the cases. The last phase is identifying the conditions which releases and stimulates the actualisation of the affordance (Volkoff and Strong, 2013).
During the fourth step the set of affordances are being analysed and associated mechanisms are being proposed (DeLanda, 2006). During this step the interaction of affordances is being analysed searching for dependencies between them. When the dependency between the affordances is established a shared mechanism can be proposed. The fifth step will assess the explanatory power of the mechanism by reviewing the empirical data in relation to the proposed mechanism.
3.1. Case study research
While multiple research methods can be used when doing a critical realist research, many
researchers argue case study research as the best method to investigate the interaction of structure, events, actions, and context to explicate and identify causal mechanisms(Miles and Huberman, 1994; Easton and Harrison, 2004; Ackroyd, 2010). As Yin (2003) describes, a case study investigates a contemporary phenomenon in its real-life context offering the possibility for an in-depth study. Furthermore, the case study approach will be used to investigate the phenomenon because of the rich data it provides (Yin, 2014), giving the possibility to formulate testable construct based on empirically valid results (Eisenhardt, 1989).
3.2. Research design
As this thesis is making use of case study research, an according research design is used during the writing of this thesis. To ensure the findings of this research are not coincidently related to one case, a multiple case study design is used. Using a multiple case study allows for cross case analysis finding similar results giving stronger support when empirically validating the mechanisms found. Next to ensuring the internal validity by using a multiple case study design, triangulation is used to ensure construct validity. Triangulation is concerned with the concurrent but separate collection of different but complementary data on the same topic (Morse, 1991). Using triangulation makes sure the subject under study is examined from different perspectives and sources. Using different sources to draw conclusions from is ensuring the validity of the research. To address the important aspect of triangulation this research will make use of semi-structured interviews in combination of
18
place when the user of the technology is using the technology. This allows for a good understanding of the different elements involved in the use of the technology and the process of using the
technology.
As proposed this thesis will make use of the concept of affordances to identify the mechanisms influencing intrapreneurship within SMEs when AI technology. To systematically assess every case in this study, the framework for critical realist data analysis of Bygstad et al. (2016) will be used. For this research semi-structured interviews and observations are used in data collection. Based on the theoretical framework of chapter 2 data was collected and subsequently an analysis was made per case. After each case was analysed a cross case analysis was done to compare resultant affordance. From this analysis the mechanisms are being proposed from which theory will be formed.
3.3. Data collection methods
The data focal in this research is gathered through a multiple case study. While a convenience sample is used, the cases do have to adhere to certain criteria to qualify as an appropriate case for this research. The sample is based on minimum variation sampling, which means that cases share similar characteristics and traits (Yin, 2014). Even though, due to the open systems perspective minimum sampling variation is difficult, all cases are selected based on certain criteria. To qualify as a case for this research an organisation must fit the criteria of an SME, as described in the introduction. Next to this the cases should be active in online marketing consultancy and adopted AI technology. The main AI technology subject to the study was the same for all cases. As for this study, a minimum sample variance is used, the decision is made to focus on cases using AI technology in their current routines. To gather the large amounts of data, semi-structured interviews are being conducted.
Semi-structured interviews offer the researcher the benefit of a guiding the interview touching upon the same subject for every case. Next to this, semi-structured interviews offer the researcher the possibility for follow-up questions for a better understanding of responses and to adapt to the different context of every case (Yin, 2014). To determine the interviewees for this research attention was given to the personal traits of the employee in the subject organisation. Literature shows that employees being open, extravert and conscientiousness are related to intrapreneurship (Obschonka and Stuetzer, 2017). Focussing on employees in the organisation showing higher levels of these traits increases the possibility for intrapreneurial behaviour to show.
19
entities and events involved, in the interviews, supporting the internal validity of this research (Thomas, 2004). Appendix A shows the interview guide. The aim of the interview is to identify the intrapreneurial behaviour before the technology implementation and identify the changes in
intrapreneurial behaviour due to the technology adoption. From the collected data, mechanisms will be derived which could affect intrapreneurial behaviour in SMEs.
Before the interviews, the interviewees were informed on the questions and concepts during an informal conversation prior to the interview, this to increase the transparency (Whetten, 1989). After the interview, a transcript of the interview was made. Interviewees were sent a copy of the transcript for confirmation of the accurateness of the data. For strengthening the results of this research observations were added to the data gathering process of the cases. Adding an extra data collection tool strengthens the validity of the research (Yin, 2014; Wynn and Williams, 2012). The case
observation was done while the technology was in use. This allowed the researcher to get a better understanding of the processes involving the technology. Next to this the observation helped the researcher to identify all entities involved using the digital marketing technology.
3.4. Case description
Since the case description and description of events and identification of key entities show great resemblance these will be merged in the following section. The subject cases of this study are all online marketing consultant agencies. For the all cases in this sample the AI technology provided by Google was the subject technology. The different AI technologies adopted aided in optimising the campaigns by setting the optimal bids and determine what search queries to advertise on. In all cases the technology was chosen because of its ostensive aspects. In the following part the cases are described, and the events are being presented per case.
Company_A: Event description
20
Before the AI technology was adopted the large part of the time was spend on manually setting bids for the campaigns of the clients. After adopting the technology time was freed which offered the opportunity for the consultant to investigate the viability of his ideas within the agency. The
technology offered time to explore opportunities outside of the consultant’s current job description. At the time of the interview the consultant was given time by his boss to pursue the new product for the agency. For this case the interviewee was one of the consultants of the firm.
Key entities
Key entities identified in the case of Company_A are the online marketing consultant as user of the adopted technology, the AI technology itself, and the online marketing campaigns. Next to this, the new technology following intrapreneurial behaviour can be seen as an entity.
Company_B: Event description
Company_B is an online marketing agency focussed on online retailers, mostly SMEs located in the North of The Netherlands. The agency has around 22 employees. The company is offering to set up and executing the online marketing strategy of retailers listed in the Emerce 100. The focus of the offered services is mostly on Google Ads, Microsoft Ads and Facebook Ads. However, the company also helps in managing comparison websites and affiliate marketing. The AI technology in use is provided by Google Ads and helps them set the right bids for search terms to achieve the goals of the client. For the technology to be able to adjust the bidding, it needs enough data to base its bids on. The AI technology determines the likelihood of a search query to be relevant for the client and adjust the bid, real time, to achieve the maximum results against the set goals.
Before the technology was adopted the agent had to constantly analyse the performance of the campaigns. When needed the bids had to be adjusted so the campaigns would perform conform the set goals. The bids would be the same until the consultant would adjust the bid again. This process of setting the right bids was labour intensive and time consuming. The technology is becoming better and better, taking more factors into account then a consultant could, therefor the performance of these campaigns is getting better compared to campaigns managed by consultants.
21
factor changes, a sales period for example, the technology cannot adjust quick enough. In these cases, the consultant should take over again.
Before the firm adopted the AI technology, campaigns were set-up using a certain strategy to assure the quality and right matching. However, for the technology to be able to perform well it needs a certain amount of data. With the previous campaign structure, data was spread over the different campaigns. The employee interviewed came up with a new strategy which makes it possible to bundle data in campaigns which made the technology suitable for more campaigns. The interviewee for this case was one of the consultants.
Key entities
Different entities are enacting in the case of Company_B. The first entity is the consultant who is adopting the technology for its campaigns. Second is the technology itself, which is interacting with the consultant. The third entity is the clients marketing campaigns, which are being managed through the technology.
Company_C Event description
This company is specialised in online marketing consulting. The organisation is employing 80
consultants spread over 2 locations. The consultants are advising and executing the online marketing strategy of major corporations. Without the technology the consultants had to manually set the bids for the campaigns occupying the time of the consultants. The company adopted the AI technology 2 years ago, gradually employing the technology for more campaigns. The ostensive aspects of the technology made it possible for the company to automate the bidding process of set-up campaigns. The adopted AI technology learns from patterns and in the data evolving itself without feedback from its users.
After the first implementation of the technology, the technology was tested with campaigns eligible for the use of the technology. Using large amounts of data, the technology makes profiles on which it predicts results and with that set bids for search queries. When set up right the AI technology
manages the campaigns previously managed by the consultant. With the technology adopted, the consultant gained extra time. This time was spent on executing the online marketing strategy of its clients. No real intrapreneurial behaviour following the technology adoption is shown.
22
Different entities play in important role in this case. Firstly, the technology was identified as an important entity in this case, as it is central to the automation of the bidding process. Secondly, the consultant was identified as an important entity as he is the one setting up the campaign and technology. Leading to the third entity, being the online marketing campaigns as these are being managed by the consultant or the technology.
Company_D Event description
Company_D is an online marketing consultancy. The company offers different services in online marketing amongst search engine advertising. For the search engine advertising the company adopted two different AI technologies. The first technology helps to advertise for search queries based on the content of the website, dynamic search advertisements. The second technology is using data to determine what bid is needed per search query to reach the set targets of the campaign. The different technologies have not been developed using feedback from the user, meaning the
ostensive aspects of the technology have been determined by the designers.
The dynamic search advertisements are set up by the consultant. When active dynamic search advertisements give input on new campaigns, which is not yet covered by the existing campaigns. The automated bidding technology is implemented through testing and gradually employed on eligible campaigns. When set-up correctly, the technology saves the consultant time spend on other activities.
The intrapreneurial activities following the adoption of the technologies have not been present in this case. However, using the dynamic search advertisements have not been the purpose of the technology in the beginning. Someone had come up with the idea to employ this technology as a tool to get insights in the blanc spot in the campaigns.
Key entities
Four main entities have been identified in this case. The first key entity is the consultant as the main user of the technologies. The second and third entities are the two different technologies. The fourth entity important in the use of the technologies is the online marketing campaign. The last entity is the website of the client.
3.5. Data analysis
23
AI technology. In order to produce theory out of the rich amount of data coding was used (Strauss and Corbin, 1998). The coding process organises large amount of data allowing for systematic analysis (Coffey and Atkinson, 1996). In this study, the coding process of Strauss and Corbin (1998) was used, involving three steps: open coding, axial coding and selective coding (Strauss and Corbin, 1998). The transcripts of the interviews and notes of the observation were coded in Atlas.ti 8. The first step in the process was open coding, during this step the transcribed interviews were broken down per question and closely examined and compared for similarities (Strauss and Corbin, 1998). In appendix F an example of the coding process is given. The codes assigned were derived from the responses of the participants, resulting in 81 codes. To avoid confusion, codes with similar meaning were combined, resulting in a list of 31 codes. The results of the open coding process are provided in a code book included in appendix G.
When all the data was coded according to the open coding process, axial coding was applied. During this step the relationship between the codes is explored. During this step the data was categorised for condition, context, actions, and consequences (Strauss and Corbin, 1998). During the axial coding process the data was categorised for the critical events related to the topic of this study as well as for the structural and contextual properties of the cases. In Appendices H-K the results of the axial coding process are displayed per case.
24 Figure 2 Data analysis process
Throughout the process of analysing the data, the focus was on the effect of adopting AI technology on the intrapreneurial behaviour.
4.Retroduction of affordances
In this section the retrieved data, from the cases, will be discussed and analysed according to the framework of Bygstad et al. (2016). The framework of Bygstad et al. allow for retroduction of the affordances on which the mechanisms influencing entrepreneurial behaviour will be built. The retroduction of the affordances is done using the axial coding resulting from the coding process. Per case the process of retroducing is described.
Company_A
Introducing the AI technology had two main motivations. The first reason for adopting the technology was because of external pressure. The second reason was to automate processes previously done by consultants. The technology automated the bidding process for online marketing advertisements. As a concrete outcome of the technology adoption, the consultant did not have to manually manage its campaigns. As a direct immediate outcome automation and efficiency can be identified. The automation and efficiency made it possible for the consultant to show intrapreneurial behaviour, discovering and testing the possibilities of his recognised opportunity.
25
the data of the online marketing campaign, the technology could set the bids for the campaign of the client. The automation of the bidding process was possible as the consultant had employed the technology for the online marketing campaign.
The proposed affordances arise from the relationship between the entities that may produce the functional outcome. When examining the interplay of the entities regarding the efficiency outcome, the outcome was possible due to the affordance of the technology being self-learning. Next to this the technology was able to perform routines previously done by the consultant. The consultant had to set-up the campaign aligning the goals of the client after which the technology could perform the routines. From this an embedded routine affordance is identified. Concluding in two affordances retroduced from the case of Company_A.
Different releasing conditions can be identified from the actualisation of the affordance. First is the amount of data available within the accounts of the clients is recognised. Without enough data, the technology would not be able to show the self-learning affordance. Having clear goals released the affordance for automation. The clear goals made it possible for the consultant to set-up the campaign for which the campaign should optimise.
Company_B
The adoption of the technology by Company_B was motivated for different reasons. The first reason was the higher rating Google gives for accounts adopting the technology. The second reason for adopting the technology was its performance, as the technology could perform the consultant his job equally or even better. Adopting the technology resulted in the consultant saving time previously spend on optimising the account of clients. As the consultant strived to adopt the technology for more campaigns, he saw opportunity for a new process of setting up campaigns. This new process makes it possible to combine data giving the technology extra ques to set bids.
The outcomes of the adoption of the technology could only occur because of the interaction of the consultant, technology, and the online marketing campaigns of the client. A connection between the technology and the data of the online marketing campaigns made it possible for the technology to perform according to the goals of the client. At the same time the user configures the technology and sets-up the campaign for the technology to perform. The data of the online marketing campaign interacting with the consultant offered the consultant the opportunity to detect the opportunity for a new process of setting up the campaigns.
26
of the user configuring the technology which in turn uses the data to set the bids, an automation affordance is proposed. Next to this an opportunity affordance is proposed. As an interplay of the consultant with the technology and campaigns with not enough data occurred. From this interplay, the consultant saw the opportunity to introduce a new process of setting-up the campaigns. For the affordance to be actualised, different releasing conditions can be identified. First, the technology is easy to set-up as it is pre-configured and self-learning. The account contained enough data which made it possible for the technology to perform. These conditions made the affordance of automation being able to occur. For the opportunity affordance to occur, the releasing condition is the support of his manager. Without his support, the actualisation of the opportunity would be possible harder.
Company_C
As Company_C deals with great amounts of data it decided to adopt the technology. The technology makes it possible to considerate more factors compared to a consultant. After adopting the
technology, the consultants saved time on managing campaigns as the technology performed this task. Next to this the technology gave the consultant insights in relevant target groups. The technology focussed on the most relevant search queries, but also on the most relevant target groups. These insights retrieved from the technology was used to accurately determine the target groups of clients and used in different campaigns.
The concrete outcome is the result of the interplay between the different entities. The outcome of time saving is the results of the consultant interacting with the technology. The consultant sets up the campaign, providing the technology with the different targets. Next the technology interacts with the data of the existing accounts to optimise the bids for the campaigns resulting. For the insights on the target group to retrieve, the consultant interacts again with the technology. All these interactions result in the outcome of the adoption of the technology in Company_C.
Following the analysis of the concrete outcome and the interplay of the entities, affordances can be proposed for Company_C. The first affordance proposed is automation as the technology took over tasks previously done by a consultant. The affordance of automation resulted in time saved for consultant while the technology is in use. The second affordance identified in the case of Company_C is one of data insights. Adopting the technology gave the consultant the opportunity to have insight in the data which could be used for determining the target group.
27
amounts of data in the online marketing campaigns of the clients make it possible to adopt the technology efficient. Without the amount of data, the concrete outcomes were less likely to occur. The second condition for the affordances to occur is the company culture. The company culture of Company_C is aimed at and open for constant change to be ahead of competition. For this the company encourage consultants to think creative and come up with new ideas. This culture stimulated consultants to come up with the idea to use the data for different purposes.
Company_D
The respondent of Company_D discussed two different technologies being adopted. The first was to fill gaps in the marketing campaigns not recently covered. The immediate concrete outcome was that the online marketing was expanded based on the content of the website. This was in line with the motivations of the company to adopt them. Next to this, blind spots were discovered in the current marketing campaigns. The other adopted technology had time saving as an immediate concrete outcome which was in line with the motivation of the company.
For these different outcomes to occur, different entities interacted. Firstly, the consultant is
interacting with the technologies, setting them up to perform. Next the technologies are interacting with the other entities. The dynamic search technology is interacting with the website of the client to identify different key words to advertise on. The automated bidding technology is interacting with the data of existing online marketing campaigns in the account. This interaction makes it possible to automate the bidding of the campaigns.
From the analysis of the concrete immediate outcomes and interaction of the key entities, the affordances will be retroduced. First proposed affordance is one of automation. Both technologies are offer automation of processes previously done manually. Dynamic search ad in the form of deciding on what search terms to advertise on. The automated bidding technology automate the bidding process by using the historical data. The second affordance is one of indexing. Because of the indexing affordance the dynamic search ad technology makes it possible to advertise on the content of the website of the client.
For the affordance to be actualised different conditions can be identified. First condition is the amount of data available in the online marketing account of the client. Using this data, the
technology is able to perform. Next releasing condition is the possibility for the technology to index the website of the client.
28
Case Proposed affordance
Company_A Embedded routine affordance Self learning affordance Company_B Automation affordance Opportunity affordance Company_C Automation affordance Data insight affordance Company_D Automation affordance
Indexing affordance
Table 2. The proposed affordances
5. Proposed mechanisms
In this chapter the proposed affordances, emerging from the case analysis, will be used to develop the mechanisms influencing intrapreneurial behaviour within SMEs. The affordances will be the building blocks of the mechanisms. To develop the mechanisms, the proposed affordances per case were grouped by comparing them for similarities. The cross-case comparison was made by
comparing the 4 different elements of the affordance retroduction process during the case analysis. Per mechanism empirical evidence from the cases will be provided. The cross-case analysis resulted in the affordances being abstracted into three mechanisms.
Embedded routine mechanism
The first mechanism identified is the embedded routine mechanism. This mechanism was identified by connecting the automation affordance of Company_B, Company_C and Company_D with the embedded routine affordance of Company_A. The argument for grouping these affordances in this mechanism is supported by the similar concrete outcomes of the affordances. In these cases, the affordance made it possible that the routine was substituted by the technology. The technology embedded routines, previously done manually, and so automating the routines in all the cases. With the routines performed by the technology time was saved, giving opportunity for intrapreneurial behaviour. The enabling conditions were an argument for grouping the affordances in this mechanism. For the technology to have this effect enough data is necessary.
Data structuring mechanism
29
opportunity affordance and Company_C data insights affordance. Consequently, data is the common releasing condition for the mechanism to be actualised in both cases.
Artificial intelligence mechanism
The third mechanism identified from the case analysis is artificial intelligence mechanism. This mechanism is derived from the cases of Company_A and Company_D. This mechanism is proposed based on the affordance of self-learning of Company_A and indexing affordance of Company_D. Because of the artificial intelligence mechanism, the technology is able to learn from historical data without interference from a user. This allows the technology to show trends which would not be brought to light without the use of the technology. Next to this, the technology adopted in Company_D interacts with the content of the website of the client indexing different terms. The technology consequently decides what content to advertise on. Through this affordance the relevant terms for the company are being brought to light. Important to notice is that the key entities
involved for the mechanism to occur, are the data and the technology.
Empirical evidence for mechanisms
As the proposed affordances have been associated with mechanisms, it is important to test whether there is empirical data supporting the mechanisms. In this part empirical data on the behaviour of structures from the cases will be explained using the mechanisms. The following examples portray the explanatory power of the mechanism.
Embedded routine mechanism
For the first mechanism, the embedded routine mechanism, empirical evidence can be found in all four cases. Company_A, Company_B, Company_C and Company_D have experienced automation of routines due to the implementation of the routines. Due to the interaction of the different entities the actualisation of the mechanism was possible. The interaction between the data and the technology allowed for automation of the routines:
Company_B: “With the adoption of the AI technology the process of setting the right bids is done by
the technology, based on the data in the account.”
30
It was when the technology, data and consultant interacted for the embedded routine mechanism to be activated. The consultant sets up the technology which interacts with the data to perform the routines previously done by the consultant. Resulting in time saved which can be used for intrapreneurial endeavours.
Company_A: “AI technology offered me extra time. Luckily this gave me the opportunity to explore
the possibilities with other tools which could benefit our clients.” Structured data mechanism
The data structuring mechanism finds empirical evidence in the cases of Company_B and
Company_C. For the structured data mechanism to be actualised, the technology interacts with the data available from the client. In the case of Company_B the technology structured the data so the opportunity for a new structure to set up the campaigns arises.
Company_B: “From the reporting of the campaigns I could analyse what campaigns could be bundled
so data could be combined for the technology.”
In the case of Company_C the technology structured the data so the consultant could analyse the specific target groups for the clients.
Company_C: “As the technology uses a lot of data to determine the best target group, the data from
the campaigns using the technology give insights in our target group.’
The empirical data shows the importance of data for the mechanism to be activated. In both cases an interplay between the technology and data is displayed for the affordances to show. The user
subsequently interpreted the outcome of the activated mechanism.
Company_C: “This information is used by this colleague to optimise other campaigns for our clients.”
Artificial intelligence mechanism
Empirical evidence for the artificial intelligence mechanism can be found in the cases of Company_A and Company_D. This mechanism is proposed based on the self-learning and indexing affordance arising from the cases. Both affordances are the result of the technology being able to act on its own within the boundaries set by its user. For Company_A the technology is adopted to set bids
automatically based on historical data.
Company_A: “Than the technology starts learning by accumulating enough data so it can predict
31
For Company_D the technology shows articifial intelligence by indexing the website of the client. From the index the technology identifies relevant terms to advertise on:
Company_D: “In this case the technology determines on what search queries to advertise, based on
your website.”
As the empirical data shows, the technology contains an artificial intelligence mechanism making it possible for the affordances to show. In both cases the entities involved are the technology and the provided data. As the technology differs per case, the required data necessary for the technology differs as well.
6. Theory building
In this part the retroduced mechanisms will be translated to build theory. Using the identified mechanisms, propositions will be formulated regarding the research question of this thesis. In Chapter 2 the theoretical foundation of intrapreneurial behaviour is discussed in relation to SMEs. The embedded routine mechanism is the basis for the first porposition. This mechanism, when activated, embed the routines which were previously done manually. This mechanism was stimulated to be activated through sufficient data. The immediate concrete outcome of adopting the technology resulted in extra time, one of the resources supporting intrapreneurial behaviour. Resulting in the following proposition:
Proposition 1: Intrapreneurial behaviour within SMEs is positively influenced through the embedded
routine mechanism when AI technology is adopted.
The second proposition is supported by the structured data mechanism. Because of the rich data the technology is processing, the structured data mechanism could be activated. The structured data mechanism reported the data, giving input for the consultant for novel ideas. Subsequently proposition 2 is proposed:
Proposition 2: Intrapreneurial behaviour within SMEs is influenced through the structured data
mechanism when AI technology is adopted.
32
Proposition 3: Intrapreneurial behaviour within SMEs is influenced through the artificial intelligence
mechanism when AI technology is adopted.
7. Discussion and conclusion
In the following part the managerial and scientific implications of introducing AI technology in SMEs is discussed.
Previous literature focussed on intrapreneurship focussed on the enabling factors for
intrapreneurship. These researches concern the antecedents supporting intrapreneurial behaviour (Rigtering and Weizel, 2013), or the characteristics of the employee strengthening the tendency for intrapreneurship (Obschonka and Stuetzer, 2017). Many researches neglected the mechanisms of how supporting antecedents are enabling intrapreneurship. This research is adding to the scientific field by explaining the mechanisms through which AI technology is influencing intrapreneurial behaviour.
Due to the exploratory focus of this research on the role of AI technologies on the intrapreneurial behaviour within SMEs, it is not suggested that the argued mechanisms are present in other contexts. However, the proposed mechanisms found empirical evidence in the cases under study. This could suggest the mechanisms could be activated in other SMEs when adopting AI technologies. Depending on the materiality with which the social is interacting, different outcomes can be
achieved.
The propositions of this research could be a good starting point for future research. As these
propositions cannot be generalised due to the explorative nature, an investigation on the effect of AI technology on intrapreneurial behaviour in SMEs amongst a larger sample would provide extra insights. Another subject for future research is the effect of the implementation of AI technology on an employee’s job role would be interesting. In the current research the focus was on intrapreneurial behaviour, activities which are outside the formal job role of the employee. In different cases the job roles of the respondent did change due to the adoption of the technology. As this was not the focus of this study, this was not thoroughly investigated.
For SMEs, adopting AI technology could result in more intrapreneurial behaviour. When one of the challenges of the SME is not having enough time resources for the employees to develop
33
7.1 Limitations
While this research provides relevant new insights into the effect of adopting Ai technology in SMEs, it does have its limitations. Using an exploratory research method, critical realism, makes it hard to generalise the results of this study. The mechanism retroduced in this research are the product of this research and its circumstances. With a limited amount of cases, the results are not statistically proven. It is therefore that the mechanism activated in these cases cannot be argued to occur in a different context.
Another limitation of this study is the time period in which it had to be conducted. The time constrain made it impossible to investigate every event at the exact moment of the event or just after. Instead, data was collected when the AI technology was already implemented. This could have resulted in respondent bias as they had to recall certain events.
Lastly this research could be limited because of researcher bias. This bias could have occurred during the semi-structured interviews and analysing the collected data. Even though different measures have been used to reduce this bias, one could not fully exterminate the influence of the researcher conducting the research.
7.2 Conclusion
While previous research on intrapreneurial behaviour have focussed on the enabling factors for intrapreneurship, this research has focussed on the effect of adopting AI technology on
intrapreneurial behaviour in SMEs. SMEs have unique structural characteristics which could cause them to respond different to the adoption of AI technology.
This thesis presented 3 different mechanisms which influence the intrapreneurial behaviour through adopting AI technologies in SMEs. Using the critical realist perspective, four SMEs were investigated using multiple data collection techniques. The data formed the basis for the retroduction of
mechanisms. The mechanisms found are embedded routine mechanism, structured data mechanism, and artificial intelligence mechanism. All of which can influence the intrapreneurial behaviour when AI technology is adopted in SMEs.
Bibliography
Antoncic, B., & Hisrich, R. D. (2001). Intrapreneurship: Construct refinement and cross-cultural validation. Journal of business venturing, 16(5), 495-527.
34
Bal, J., Bruins, A., de Jonge, J., Tan, S., Verheul, I., & Wennekers, S. (2007). Ondernemerschap in het primair en voortgezet onderwijs. EIM Report A, 200709.
Bhaskar, R. (2014). The possibility of naturalism: A philosophical critique of the contemporary human
sciences. Routledge.
Boudreau, M. C., & Robey, D. (2005). Enacting integrated information technology: A human agency perspective. Organization science, 16(1), 3-18.
Bygstad, B., Munkvold, B. E., & Volkoff, O. (2016). Identifying generative mechanisms through affordances: a framework for critical realist data analysis. Journal of Information Technology, 31(1), 83-96.
Ciborra, C. and G. F. Lanzara (1994). “Formative Contexts and Information Technology: Understanding the Dynamics of Innovation in Organizations.” Accounting, Management and Information Technology 4(2): 61-86.
Ciborra, C. U. (1991). From thinking to tinkering: the grassroots of strategic information systems. Proceedings of the 12th International Conference on Information Systems. New York: 283-292. Collier, A. (1994). Critical realism: an introduction to Roy Bhaskar's philosophy.
Covin, J. G., & Slevin, D. (1989). Strategic management in small firms in hostile and benign environments. Strategic Management Journal, 10(1), 75-87.
Danermark, B., Ekstrom, M., & Jakobsen, L. (2001). Explaining society: an introduction to critical
realism in the social sciences. Routledge.
DeLanda, M. (2006). A new philosophy of society: Assemblage theory and social complexity. A&C Black
Dobson, P. J. (2001). The philosophy of critical realism—an opportunity for information systems research. Information Systems Frontiers, 3(2), 199-210.
Earl, M. (1989). Management Strategies for Information Technology. Hemel Hemstead, Prentice Hall. Frese, M., Fay, D., Hilburger, T., Leng, K., & Tag, A. (1997). The concept of personal initiative:
Operationalization, reliability and validity in two German samples. Journal of Occupational and Organisational Psychology, 70(2), 139-161.
