Robotic Process Automation : an employees' perspective : an exploratory case study on the perception towards Robotic Process Automation among employees

(1)

An exploratory case study on the perception towards Robotic Process Automation among employees.

Master Thesis

Desley van der Zande – 11887869

MSc. in Business Administration – Digital Business University of Amsterdam – Faculty of Economics and Business

Supervisor prof. dr. P.J. van Baalen Second reader prof. dr. H.P. Borgman

(2)

Date of submission: 22-06-2018

Statement of Originality

This document is written by student Desley van der Zande who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

Acknowledgement

First of all, I would like to thank Peter van Baalen. His valuable and constructive feedback has been of great help in writing my thesis. I would also like to thank my thesis supervisors at KPMG: Nathalie Duijvesteijn and Gijsbert Sigmond. The assistance provided by them have been greatly appreciated. Lastly, since this is my graduate thesis, I would like to offer my special thanks to my parents who always supported me throughout my study.

(4)

Abstract

This study aims to explore the impact of the implementation of Robotic Process Automation (RPA) from the perspective of the workforce. Based on eight interviews with employees whose jobs are automated by RPA, this study explores how employees perceive this RPA implementation ex-ante and ex-post. Understanding the effects of this

implementation on the workforce is essential in providing insight into the long-term consequences RPA implementations. The findings show that employees perceive the RPA implementation as positive ex-ante and ex-post. Because of the simple nature of the to-be-automated process, employees expect that the implantation of RPA reduces their workload. Furthermore, they expect their jobs to become more interesting and diverse. When the

implementation is completed, they still feel positive about the implementation. However, they also mention concerns regarding the process errors and error handling, which they relate to a limited availability of expertise. This study contributes to the existing literature since it analysis the impact of automation on the workforce from an employees’ perspective. Furthermore, it provides recommendations for managers or change agents on the relative importance of the perceived usefulness, need for change and personal effects. It provides practical insights than can be used in change management practices related to the

implementation of RPA.

Keywords: Robotic Process Automation; Change Management; IT implementation;

Attitudes Towards Change; Employee Attitude; Human-Machine interaction; Need for Change; Usefulness; Valence.

(5)

1. Introduction

Over the past centuries, industrial revolutions resulted in major transformations in the way of working. (Ford, 2015). As a consequence of the revolutions, disrupting technologies required employees to adapt to a new working environment. The next revolution is currently taking place (Berg et al., 2018). This robotic revolution is characterised by the convergence of breakthrough technologies, such as artificial intelligence (AI), robotics and the Internet of Things (IoT). These technologies transform production processes and business models across different industries (World Economic Forum, 2017). Being in a transformational phase stirred the debate among journalists, economists and academics about the effect of automation on growth, employment and distribution of income (Berg et al., 2018; Roth & Kaivo-Oja, 2016).

Automation proponents agree that increased automation destroys some jobs but, argue at the same time that job automation creates new, more challenging jobs. Overall, the

automation proponents believe that the development of robotics will positively affect the workforce (Fung, 2013). Automation opponents, on the other hand, expect that 45-57% of the jobs in the United States will be threatened by robotics (Frey & Osborne, 2017). Nicholas Carr, a well-known opponent of the automation, warns for the negative impact of automation on the human ability to think (Carr, 2013). He uses the example of pilots to explain the automation paradox. Although technology reduces the cognitive burden on pilots, which has a positive effect on the overall safety, the pilots spend fewer hours on actually flying. This reduced experience results in less flying experience and a decline in flying skills over time. Carr (2013) worries about this paradox and argues the same effect is likely to occur across other jobs when the development of new types of automation continues.

One of the most recent types of automation is Robotic Process Automation (RPA). RPA is the next step within business process automation that is currently developing. Where

(9)

classical business process automation is built from scratch and requires alignment of IT-systems, RPA uses an outside-in approach meaning that the existing information systems remain unchanged (Van Der Aalst, Bichler, & Heinzl, 2018). As a result, RPA can adapt to changes of the underlying information systems in the same way as humans do. It can be used to reduce the workload of employees by replacing repetitive and simple tasks (Aguirre & Rodriguez, 2017).

Prior research widely addressed the effects of robotics on the labour market, and the consequences of business process automation on jobs (Autor, Levy, & Murnane, 2003; Lacity & Willcocks, 2015). However, little is known about the attitude of the workforce regarding a robotics implementation. Analysing this attitude is essential in providing insight into the long-term consequences of RPA since it influences employees overall job performance and affects the success of an implementation (Lines, 2004; Raineri, 2011; Venkatesh, Morris, Davis, & Davis, 2003)

Literature of change management and IT implementation will be combined to study the attitudes of the workforce regarding the implementation of RPA. This research tries to answer the question:

What are the expectations and evaluations regarding the implementation of Robotic Process Automation of employees whose job is affected by this implementation?

This research investigates how employees whose jobs are affected by RPA, respond to this implementation. This investigation includes a comparison of the employees’ ex-ante perception of the RPA implementation and the ex-post perception of the implementation. Furthermore, this comparison includes and exploration on which factors contribute to the employees’ overall attitude towards the entire RPA implementation.

(10)

Overall, the theoretical contribution of this research is threefold. First, it pays attention to the consequences of the expanding field of Robotic Process Automation on the workforce. Secondly, it provides guidelines for future research in the field of change

management. It investigates which factors contribute to employees’ attitude towards the RPA implementation and how these factors differ in relative importance. This relative importance has not been investigated before.

In addition to the theoretical contribution, this research provides recommendations for managers and change agents who are planning to implement RPA in the future. It provides insights into the employees' perspective of the RPA implementation, which can be useful in formulating communication strategies. Furthermore, the findings can be used as a basis for human resource management practices regarding human-machine interaction.

The second chapter of this thesis provides an overview of the existing literature in the field of automation. It explains the differences between existing and future types of business process automation and combines literature in the field of IT implementation and change management to address the issue of RPA implementation. The literature review ends by presenting a conceptual framework on which this research is based. The third chapter of this thesis explains the research method that is used to answer the research question. In the third section, the findings will be presented based on a within case-analysis and cross-case analysis. This thesis ends with the overall conclusion, the managerial implications and recommendations for future research.

2. Literature Review

2.1 Business Process Automation

The traditional type of business process automation is based on a fixed set of requirements and automates relatively predefined processes (Mohapatra, 2009). Some

(11)

examples are Enterprise Resource Planning, Customer Relationship Management and Supply Chain Management (Scheer, Abolhassan, Jost, & Kirchmer, 2004). These tools were mostly implemented to support enterprise knowledge workers to create higher output, increase productivity and improve quality and flexibility (Mohapatra, 2009).

Research in the field of business process automation emphasised positive as well as negative effects of business process automation on the workforce. Automation proponents argue that automation improves the overall productivity since it enables employees to complete more cases within the same time (Fung, 2013; Mohapatra, 2009). Additionally, Autor (2015), found that human workers conduct more non-routine-based tasks instead of routine-based tasks as a result of job automation. Non-routine-based tasks are less repetitive and requires more attention. Automation proponents therefore state that increased automation reduces the error-rates (Wickens, Li, Santamaria, Sebok, & Sarter, 2010).

Automation opponents, on the other hand, argue that automation threatens the labour market (Autor et al., 2003; Frey & Osborne, 2017; Jaimovich & Siu, 2012). Frey and

Osborne (2017), for example, found that 47% of the current jobs in the USA are at risk, meaning that they are potentially automatable. In addition, other automation opponents highlight the fact that also the employees who keep their job experience harmful effects of increased automation. Nicholas Carr, probably the most well-known automation opponent (Ford, 2015), addresses these downsides of automation for the remaining workforce. Carr, (2010), warns for the reliability of humans on technology since this reliability negatively affects our lives and our brains. Instead of performing tasks, workers will observe screens. This task shift inhibits their development of expertise and negatively affect their capabilities (Carr, 2013).

(12)

2.2 Robotics Process Automation

This study focuses on RPA, a new type of automation. According to the Institute for Robotic Process Automation and Artificial intelligence (2018), RPA is the application of technology that configures a robot that captures and interpret existing applications to process a transaction, to manipulate data or to communicate with other digital systems. RPA differs from traditional types of automation in three ways. First, it uses existing Information Systems to automate processes. Instead of replacing or aligning the existing systems, it interacts with these existing systems (Van Der Aalst et al., 2018). Secondly, RPA can adapt to changes within these underlying information systems, which makes it possible to handle exceptions (Aguirre & Rodriguez, 2017). Thirdly, traditional automation aims to enhance the workforce, while RPA focuses on virtualising the workforce (Fung, 2013). The RPA market is growing fast (Le Clair, Cullen, & King, 2017). While the market share of RPA was only $250 million in 2016, they estimate a market growth towards $2.9 billion in 2021.

2.3 Artificial intelligence (AI) & Machine Learning

In addition to the basic type of RPA, a more advanced type of RPA is currently in its early stages of development (Gartner, 2017). This advanced type of RPA involves the use of new developments within the field of Artificial Intelligence (AI), where human intelligence is exhibited by machines (Bini, 2018). These new developments belong to what is known as "machine learning" (Agrawal, Gans, & Goldfarb, 2017). Machine learning techniques enable the robot to learn human resolving capabilities through interaction with workers (Asada, Macdorman, Ishiguro, & Kuniyoshi, 2001; Van Der Aalst et al., 2018). In their paper Agrawal, et al. (2017), discuss the impact that improved machine learning techniques will have on the workplace. They argue that the role of humans within the workforce will become more crucial since virtual systems study the behaviour and the execution of tasks performed

(13)

by the human workforce. By observing the choices that these humans make, the robot learns how to execute different steps within a process and how to handle exceptions (Agrawal et al., 2017). As a result of this observational learning method, Agrawel et al. (2017) stress the importance of studying human actions and behaviour in the setting of job automation. This study is focused on the simple type of RPA without the addition of AI or Machine Learning since advanced RPA is still in its early stages of development. However, the contributions of this thesis apply to the more advanced types of RPA as well.

2.4 The Role of Individual Beliefs

According to the Thomas Theorem, the actual consequences of the situation itself are less important (Thomas, 1928). What matters is the individual perception of the situation. Their well-known quote “If men define situations as real, they are real in its consequences” (Thomas, 1928 p.572), forms the basis of a lot of theories in the field of sociology and psychology (Merton, 1995). More recent theories still rely on the Thomas Theorem. Baird & Williamson (2009), claim that uncertainty and fear among employees in dealing with new situations decrease organisational productivity. Similarly, Lin & Chen (2012), investigated the perception of cloud computing among IT professionals. Their findings suggest that the personal beliefs of IT managers prevent companies from using cloud computing, even when the benefits of this way of computing are widely addressed within the literature. Studies regarding IT adoption also emphasise the importance of individual perception. (Venkatesh et al., 2003), argues that the customers' acceptance of new technologies influences their use of this new technology. This view is supported by Abbas, Hassan, Asif, Ahmed, & Haider, (2018) who demonstrated that the attitudes of customers towards mobile banking apps influence their use of mobile banking apps.

(14)

Research in the field of job automation mainly consists of studies towards the

consequences of business process automation and future robotics developments on the nature of work and the labour market (Autor et al., 2003; Lacity & Willcocks, 2015). Little is known about individual perception towards the implementation of Robotic Process Automation. This research gap is remarkable since the evidence presented in this chapter suggests that the individual perception towards new implementations affect organisational change and

influence individual behaviour (Lin & Chen (2012); Thomas, 1928;Venkatesh et al., 2003). Soon, more advanced types of RPA will learn from human behaviour based on human-machine interaction (Gartner, 2017). It is therefore essential to gain a deeper insight into these human attitudes towards the RPA implementation. This study investigates the perceptions of RPA that exist among employees. It explores which factors influence the individual perception towards RPA ex-ante and ex-post. This study answers the research question:

By answering this research question, this study contributes to the understanding of the perception towards RPA among employees. It investigates which factors influence the

individual perception towards RPA and explores how these factors interact. This is the first study that explores this interaction and analyses the relative importance of these factors. Furthermore, this study originally contributes to the field of RPA since it is the first study within this research area that approaches this type of automation from the perspective of the workforce.

(15)

To answer the research question, this study is based on elements of the tripartite model. This model can be used to analyse individual perceptions towards change and is often used in the field of change management (Bouckenooghe, 2010; Rafferty, Jimmieson, & Armenakis, 2013; van Harreveld, Rutjens, Schneider, Nohlen, & Keskinis, 2014). The next section demonstrates this model and describes the key elements within this model. It links change management practices to the implementation of RPA and ends with a presentation of the conceptual framework

2.5 The Tripartite Model

The tripartite model explains the individual perception towards change (Ostrom, 1969). This model argues that the individual perception referred to as individual attitudes, consist of

cognitive, affective, and behavioural components. The components of this model are still

widely used within the literature in the field of change management (Bouckenooghe, 2010; Rafferty et al., 2013; van Harreveld et al., 2014). The cognitive component can be considered as the most critical component within the tripartite model because it influences emotions and behaviour (Bovey & Hede, 2001; Holt, Armenakis, Feild, & Harris, 2007). This effect is illustrated by Holt et al. (2007), who state that individuals use a cognitive scheme, to interpret a change. This interpretation influences their feelings and emotions towards change, which in turn forms the basis of employees' organisational behaviour. The cognitive component within the tripartite model is used as a basis for this research design and the interview protocol of this study.

2.6 The Cognitive Component

The cognitive component relates to individual beliefs, such as thoughts or

understanding of the object of change (Piderit, 2000). According to Rafferty et al. (2013), these beliefs are formed by the perceived need that change is needed, the perceived

(16)

usefulness of the change and the perceived personal effects of the change. The objectives of this study are based on these components. The following paragraphs explain these beliefs and present the objectives of this research.

2.6.1 Need For Change

The perceived need for change is in this study defined as The extent to which someone

believes that change is necessary. Kotter, (1995), mentioned this perceived need for change

as the first and most important step to successfully execute a change process (Kotter, 1995). Additionally, Bouckenooghe (2010), argues that establishing a sense of urgency is crucial in achieving a high need for change. Based on these studies, it is likely that the individuals' perceived need to automate the processes influences their overall attitude towards the RPA implementation. As a result, the first objective of this study is to explore how the perceived

need for change influence employees' expectations and evaluations related to RPA implementation.

2.6.2 Usefulness

According to the English dictionary, usefulness can be seen as: “The quality or state of being useful” (Cambridge Dictionary, 2018). Additionally, usefulness in the context of change implementation literature is often referred to as "the degree to which a person believes that using a particular system would be free of effort" (Davis, 1989, p. 320). This study combines both definitions and defines perceived usefulness in the context of RPA as the extent to which someone believes that RPA is useful to conduct the automated, or to-be automated task. In their research Brown, Massey, Montoya-Weiss, & Burkman (2002), investigated the effect of perceived usefulness on employees’ attitude towards a new IT implementation. Their research demonstrates that the perceived usefulness of new IT innovations among employees affects the acceptance of a new IT implementation among

(17)

these employees, especially when using the system was mandatory. Based on these studies, it is likely that the ex-ante and ex-post evolution of the usefulness of RPA influences the attitude of individuals towards the RPA implementation. As a result, the second objective of this study is to explore how perceived usefulness influence employees’ expectations and

evaluations related to RPA implementation.

2.6.3 Personal Effects

The last attribute that explains individual cognitive beliefs according to the tripartite model is personal effects. Self and Schrader (2009), argue that individual resistance to change depends on the perception of personal wins or gains. It is therefore likely that some

individuals resist changing if personal losses outperform personal wins (Armenakis & Bedeian, 1999; Lines, 2004; Self & Schraeder, 2009)

In the case of RPA, it is expected that employees experience both loss and gains. The promise to support workers by executing standardised tasks to improve efficiency indicates potential personal benefits. However, one can also argue that the RPA implementation threatens jobs by executing complete tasks. As a result of this threat, employees might experience personal risks. Pideritt (2000), further describes this contradiction in his

investigation towards personal effects. On the one hand, he mentions that employees who can adapt to a new implementation can take advantage of the opportunities that arise from the implementation. On the other hand, it is possible that they experience adverse effects such as losing authority, status, or economic losses (Autor, 2015). Little is known about the personal losses and wins that employees experience as a result of an RPA implementation. More information is needed to understand how employees evaluate the effect of the RPA implementation on them. As a result, the third objective of this study is to investigate how

(18)

perceived personal consequences influence employees’ expectations and evaluations related to RPA implementation.

2.6.4 Relative Importance

The last objective of this study is to explore the relative importance of these three attributes. According to (Finneman & Clark, 1996), the effect of the perceived personal effects should outperform the effect of the perceived need for change. They argue that individuals are likely to resist change if they believe that they will lose something of value. Following this argument, it is expected that employees focus on their self-interest rather than on the potential benefits of the organisation. This expectation is in contradiction with Kotter (1995), who argues that establishing a sense of urgency is the most critical focus point when leading change. Overall, it can be concluded that some predictions regarding the relative importance of these beliefs have been made. However, no prior literature exists that investigates this relative importance. It is therefore unknown if, for example, the need for change is more important than personal consequences, or if the usefulness is more influential on the overall perception than the need for change. To address this research gap, this study

investigates the relative importance of the need for change, usefulness and personal effects on employees' overall cognitive evaluation of the RPA implementation.

2.7 Conceptual Framework

To visually summarise the literature study and the four research objectives, a conceptual framework is constructed in Figure 1. This framework displays the cognitive beliefs that influence the individual attitude towards the implementation of RPA. Following the research objectives, this study explores how the RPA implementation influence these three elements. It investigates how these cognitive beliefs individually contribute to the overall attitude towards the RPA and in which hierarchical order. These investigations will be

(19)

conducted in the pre-implementation phase and the post-implementation phase to address potential differences between those phases.

Figure 1. Conceptual Framework on the attitudes towards RPA

3. Research Method

This research aims to explore the attitude towards the implementation of RPA among employees whose jobs are automated by RPA. It investigates which cognitive beliefs

influence the attitude towards the RPA implementation among these employees. Furthermore, it tries to explore how these beliefs influence the employees’ attitude towards the RPA

implementation and investigate the extent to which these factors contribute to their overall attitude towards the RPA implementation. This research covers both the pre-implementation and the post-implementation phase. The cognitive beliefs are examined according to the conceptual framework that is based on change management practices (Bouckenooghe, 2010; Ostrom, 1969; Rafferty et al., 2013). This research answers the question:

RPA

implementation Usefulness

Need for change Personal effects Attitude towards RPA implementation Pre-implementation phase Individual level RPA implementation Usefulness Personal effects Attitude towards RPA implementation Post-implementation phase Individual level Cognitive component Cognitive component

(20)

This chapter describes and discusses the methods used in this investigation. It describes the research philosophy, the research method and design, the respondent selection and the data collection method.

3.1 Research Philosophy

The research philosophy contains assumptions about the way in which the researcher views the world and is essential to consider since this view can influence the choices and strategies that are made regarding the research (Saunders, Lewis, & Thornhill, 2012). This study is interpretative in nature since it is based on the idea that humans are social actors who interpret the world around (Saunders et al., 2012).

3.2 Research Approach

This study uses an inductive approach. According to Saunders et al. (2012), an

inductive research approach is mostly used to explore phenomena to understand the nature of the problem. This study explores the attitudes of employees towards the implementation of RPA and provides a better understanding of the perspective of the workforce towards this implementation. This information can contribute to the development of new theory within the field of RPA as well as in the field of change management.

3.3 Method and Design

This study is based on a qualitative research method. This method is suitable for this study since it is used to answer questions about experience, meaning and perspective, mostly from the perception of the participant (Hammarberg, Johnson, Bourne, Fisher, & Kirkman,

(21)

2014). To analyse these individual attitudes, this study uses an embedded case study design. According to Yin (2003), this type of case study design applies to situations in which a single case is studied over multiple units of analysis. This study collects the qualitative data through interviews with multiple employees, working in different departments. The employees, in this case, relate to the multiple units of analysis. Each employee experienced job automation as a result of the RPA implementation. This implementation, therefore, relates to the single case on which this research is focused.

3.4 Respondents Selection

The respondents are selected based on a set of requirements. Firstly, this research is focused on the attitudes during the pre-implementation phase as well as the attitudes during the post-implementation phase. It is therefore required that the implementation phase is completed and that RPA executes the process. Secondly, it is required that the employee used to execute a task that is currently executed by RPA. This requirement guarantees a change in the daily work of the employees as a result of the RPA implementation, which enables the employee to reflect on this implementation.

To select the respondents, this study uses purposeful sampling, in combination with

criterion sampling techniques. In cooperation with KPMG, respondents are selected based on

the requirements as mentioned above and their willingness to participate.

3.5 Data Collection

Data for this study is collected through semi-structured interviews that are based on the conceptual framework. This way of data collection is appropriate when providing qualitative research (Gillham, 2005). It enables interviewees to express their thoughts and feelings and allows the researcher to ask questions based on issue related themes (Alvesson & Deetz, 1999). The interviews take approximately 40 minutes each and are recorded and

(22)

transcribed. After each transcription, the interviewees will check their transcripts. This avoids errors and strengthens the validity. Apart from the questions related to this research, the interview protocol also contains some questions that are related to a different research that is being executed by KPMG at the same time. These questions are related to the impact of RPA on job characteristics. It is beyond the scope of this research to analyse this issue. Therefore, the data drawn from these questions is not described. The interview protocol can be found in Appendix A.

3.6 Time Horizon

This research is conducted from January 2018 to June 2018 and is therefore limited in time. To explore the differences in cognitive beliefs between the pre-implementation state to the post-implementation state, a longitudinal research strategy would be preferable. This type of strategy allows the researchers to do a pre- and post-measurement. However, due to the limited timeframe of this study, a longitudinal study is practically impossible. RPA is still in its early stages, which is makes it hard to find cases on which a pre- and post-measurement can be conducted. To overcome this issue, the interview consists of two parts. In one part of the interview questions will be asked related to employees' evaluation of the RPA

implementation. The other part of the interview refers to their expectations of the RPA

implementation ex-ante, for example by asking ‘what was your response when you first heard of the implementation?’. Although it is expected that respondents are able to explain their earlier opinions, it is essential to take into account the likability of biased results when analysing the results related to earlier expectations. This will be further explained in the discussion section.

(23)

4. Findings

This chapter describes the findings of this research. Eight semi-structured interviews were conducted with employees whose job is automated as a result of the implementation of RPA. These interviews were conducted at four Dutch firms with over 250 employees within the following sectors: one banking firm, one within the fast-moving consumer goods sector and two energy utility companies. The RPA systems were up-and-running in all cases. An overview of the respondents, including the interview date, department and a fictional company name are displayed in Appendix B.

After the data was collected, literal transcriptions of the recordings have been created manually. To increase credibility, the employees reviewed their transcript to check for accuracy and resonance with their experiences. The approved transcripts were coded and analysed individually. A summary of the transcripts that include quotes and contextual information is displayed in Appendix C.

This chapter is divided into two parts. The first part displays how individual beliefs influence the individual attitude towards the RPA implementation, based on a within-case-analysis. The second part presents the findings related to the hierarchical order in which these beliefs influence the individual attitude towards the RPA implementation. By conducting a cross-case-analysis, this section aims to determine the individual contribution of the cognitive beliefs towards this attitude.

4.1 Within-Case-Analysis

The first part of this chapter presents the findings related to the three cognitive beliefs that have been discussed in the literature review. The cases are analysed individually to investigate how each cognitive belief affected the personal attitude of the employee towards the RPA implementation.

(24)

4.1.1 Findings on Need for Change

The results indicate that most employees saw the need to change the process before the implementation started. Some employees state that they perceived the initial task as too simple and repetitive to be conducted by the human workforce nowadays. Respondent 7 explains: “We always thought it was strange that this could not be automated earlier.

Automating these processes was already possible at the beginning of the computer age, right?” One of the employees, respondent 3, emphasises the need for change from an

organisational perspective: “Savings were needed because the company experienced tough

times.” Talking about the need for change, some employees argue that they saw the need for

change because they believe that RPA increases the customer satisfaction. Respondent 2 explains: “The robot cannot make mistakes, so in that case, the process will be executed

much quicker, which increases the customer satisfaction". In contrast, respondent 3 argues “I did not see the need for change at all. I always want to keep track of what is happening. If I do not see what is happing with all the trades for example, then I do not know the answers when customers ask me about their trade”.

The findings related to the perceived need for change in the post-implementation phase indicate that more employees see the need for change in the post-implementation phase in comparing to the pre-implementation phase. A commonly used explanation for this

perceived need for change in the post-implementation phase is related to the positive job effects that they experience as a result of the implementation. Three respondents experience that the task is conducted much faster now RPA automated it. Respondent 8 explains: "What

a human being could do in 15 minutes, the robot did in 3”. This respondent realised that

change was necessary when he received less customer complains. Only one respondent, respondent 3, did not perceive this change as necessary ex-post, which he relates to a lack of vision of the top management regarding the RPA implementation. He states: “They did not

(25)

have a clear goal in mind. They did not know what they wanted to achieve with this RPA implementation”.

The overall findings on the perceived need for change indicate that most employees see the need for change ex-ante. They perceive the initial task as too simple to be done by humans. Furthermore, they mention organisational benefits such as increased customer satisfaction and cost savings as important reasons to change the process. The expectations of most respondents regarding the positive effects of changing the process became true. Most respondents did see the need for change ex-post. From their perspective, RPA executes the process more efficiently compared to when they execute it manually. Furthermore, they experience a reduced amount of customer explains. Three respondents did not see the need to change the process before the implementation, because of the reduced control over the

process. Two of them changed their mind and agreed that changing the process was necessary when reflecting on the implementation.

4.1.2 Findings on Usefulness

The results regarding the perceived usefulness ex-ante indicate that most employees do not know enough to estimate the usefulness of the RPA implementation. Respondent 1 for example, mentioned that she did not have much faith in RPA, because she did not know what the robot could do. Likewise, respondent 3,5,6 and 7 mentioned that they did not have any expectations regarding the usefulness, because they did not understand how it works. Three respondents had negative expectations regarding the usefulness of RPA ex-ante, which they relate to the nature of the process. Respondent 2 explains: "We already estimated that this

process was too complex to automate because it uses multiple sources for its input".

Ex post, the opinions regarding the usefulness are divided. Four perceive RPA as useful. The respondents who perceive RPA as useful mention that it does what it has to do,

(26)

and that it is easy to work with RPA. In addition, respondent 4 mentions that RPA is useful because it can work outside office hours. However, most respondents also mention some concerns related to the usefulness of RPA. Two discrete reasons emerged for this. The first concern is comparable with the concerns during the pre-implementation phase. It relates to the suitability of the processes for RPA. Respondent 5 states: “When I see how it works it is

useful but, because of the many expectations it had a smaller impact than I expected”.

Respondent 6 agrees with this statement by saying that: “RPA can be useful but, only if

expectations can be executed as well. In addition, respondent 8 states that a large variety of

input sources made it hard to use RPA: “If one source changes the whole thing stops

working”. The second concern follows on from the first concern and refers to the time that is

needed to solve errors or handle exceptions. Respondent 2 states: “The robot got stuck, and

tasks were not completed which cost me even more time. It took long before it went well, think about months. We have suffered from this for a long time” In addition to that she

mentioned that the clients noticed this error as well; “We got emails from customers asking

why we did not open their account". She believes that it takes longer to repair the process

because this repair requires expertise in the RPA system that is not always available. She states: “none of us knew what went wrong”. Respondent 8 shares this view: “The company

takes a small risk by hiring people who do not understand the complete process. If an error occurs, they cannot rewrite the robot, because they do not know how. You lose knowledge especially when customers start asking you about what went wrong. That is tricky." This

reliability of expertise is also mentioned by respondent 6. “When the builder was not

available, and the owner was sick, the process stopped for a week, because we were not able to identify the error”.

Overall the findings indicate that most respondents find it hard to form an opinion about the usefulness of RPA ex-ante because they do not know how RPA works. This lack of

(27)

knowledge resulted in negative or positive as well as negative expectations. Some respondents doubted the usability of RPA, because of the complex nature of the process. Whereas none of the respondents expected RPA to be useful before it was implemented, half of the employees perceive RPA as useful when it was implemented. They experienced that it does what it has to do, think that it is easy to work with and appreciate the fact that it works outside office hours. However, a majority of the respondents pointed out some negative aspects of the usefulness of RPA. Some experience RPA as less useful because they do not think that the selected processes as suitable for this type of automation: they believe that RPA is not able to handle exceptions, or to execute processes that require an input of multiple sources. Others emphasise that solving the errors in RPA systems costs much time because it relies on individual expertise that is not always available.

4.1.3 Findings on Personal Effects

The findings regarding the personal effects indicate that most respondents expected positive personal consequences of the RPA implementation, ex-ante. Five employees expected that RPA would take over their tedious tasks, which enables them to do more interesting tasks. Respondent 8 explains: “We always had to do quite standardised tasks

manually. It is a psychological burden to do the same tasks every day. It makes your work sloppier. RPA, therefore, gives you the time to do other tasks which increases the enjoyment of work". In addition, three respondents expected personal benefits in a sense that they could

learn from the implementation process. Respondent 5 states: "I was asked to help with the

implementation. I liked that opportunity because it enabled me to improve myself and learn new things”. Three respondents mentioned job insecurity as a risk of RPA. However, only

one respondent, respondent 6, felt afraid to lose his job: “I was afraid that RPA would replace

(28)

two employees were only concerned about the jobs of others. Respondent 2 states: “If the

robot is being implemented, fewer employees are needed. In that sense, I thought about my colleagues but, not about myself.” He explained that he did not worry about his own job,

because he was more involved in the RPA implementation than his colleagues. Furthermore, he mentioned that he was mainly worried about the jobs of older employees. In contrast to this perception, respondent 7 states: “I am 62. I think that I am going to stop working at 64,

so for me the RPA implementation would not be that bad but, for the next generation…”.

Respondent 8 also refers to age as a factor that influences the perceived personal risks of RPA: “I noticed mainly at the beginning of the implementation some colleagues feared their

job future. It is a hot-button issue.” This respondent believes that most of these anxious

employees are "in the mid-50s”.

The findings related to the perceived personal effects in the post-implementation phase indicate that most employees experience positive personal effects. Like the

expectations ex-ante, most employees appreciate the perceived time gain as a result of the implementation. As a result of this time saved, respondent 4 mentions that he uses this time to help his colleagues, respondent 5 now spend his time executing piled up work. Furthermore, he mentions that it enables him to respond to client questions faster. Additionally, some respondents mentioned less work stress, which they link to the RPA implementation. Respondent 2, for example, states: “When I was working on the attendance cards, I was so

focused that I got irritated when a customer called. I thought, well... I am just right in the middle of the calculation, and now I have to answer the call... You got distracted every minute, and you knew you had a deadline”. Respondent 8 feels more motivated as a result of

this decreased workload: “You realise that you make progress instead of feeling that you will

never be able to finish work". Some respondents mentioned that the implementation provided

(29)

a rise in organisational status as a result of the implementation: “My name is mentioned more

within the company, and RPA is mentioned in presentations. Then I feel proud because I contributed to it”. The findings regarding the negative personal effects in the

post-implementation phase indicate that some employees are still worried about the jobs of others. Respondent 6 felt worried about the future of his job ex-ante. These concerns disappeared when the implementation was completed: “It only replaced one of my tasks, not even the

complete task, just a part of it”.

Overall these findings indicate that most respondents expect personal benefits as a result of the RPA implementation. They expect RPA to take over their annoying and simple tasks, which enables them to conduct more diverse tasks. Furthermore, they expect that they can learn from being involved with the implementation. When talking about the perceived personal risks, some mention the effects of RPA on job security. Most of the respondents are not worried about losing the jobs themselves. However, some are worried about the job of others. The findings related to perceived personal effects in the post-implementation phase indicate that seven employees perceived personal benefits related to the RPA implementation. These personal benefits mostly occur from a reduced workload and increased individual opportunities within the organisation.

5. Cross-Case-Analysis

Whereas the previous section is based on a within-case-analysis, this part consists of a cross-case-analysis. This analysis identifies similarities, patterns, and differences across the individual cases. It is used to explore the individual contribution of the cognitive components

need for change, usefulness and personal effects to the overall attitude towards the RPA

(30)

post-implementation phase. An overview of the table that is used for this analysis is displayed in Appendix E.

5.1.1 Analysis of Attitude Ex-Ante

The findings of the cross-case-analysis demonstrate that employees seem to have a positive attitude towards the RPA implementation ex-ante. Furthermore, the results indicate that respondents find it hard to assess the usefulness of RPA ex-ante. The findings further indicate that the attitude towards the RPA implementation is in line with their ex-ante perceived personal-expectations. Lastly, the findings demonstrate that respondents who reported a positive attitude towards RPA ex-ante, initially saw the urge to change.

In sum, the findings indicate that employees are positive towards the RPA

implementation ex-ante. Whereas the expected personal benefits and the perceived need for change seem to be an essential element of this attitude, usefulness is considered to be less important.

5.1.2 Analysis of Attitude Ex-Post

The findings regarding the ex-post attitude indicate that employees seem to have a positive attitude towards the RPA implementation ex-post. The findings further indicate that the attitude towards the RPA implementation is in line with their ex-post perceived personal-expectations and their ex-post evaluation on need for change. Whereas employees found it hard to assess the usefulness ex-ante, they could better reflect on this component ex-post. The findings regarding further indicate some concerns among employees related to the usefulness. However, is not possible to say if usefulness is in line with the overall attitude towards the implementation of RPA ex-post, based on the findings of this study. For some respondents, it is in line, for others not.

(31)

In sum, the findings indicate that employees have a positive attitude towards the RPA implementation ex-post. Whereas the expected personal benefits and the perceived need for change seem to be an essential element of this attitude, no indication can be provided regarding the importance of usefulness.

5.1.3 Analysis of Attitude Evolvement

Firstly, the findings demonstrate that most respondents have a positive ex-ante attitude towards RPA. This attitude remains unchanged in the post-implementation phase. Three respondents reported a change in attitude. The attitude of respondent 6 and 1 improved, while the attitude of respondent 2 declined. This observation applies to every respondent except for respondent 6, who experienced personal benefits after the implementation in contrast to what he expected before the implementation. Secondly, the findings indicate that the respondents who expected positive personal consequences as a result of the RPA

implementation, actually experienced personal benefits when RPA was implemented.

Thirdly, no patterns are visible for the evolvement of the beliefs that change is needed and the beliefs regarding the usefulness of RPA.

6. Discussion

This chapter discusses the overall findings of the research. It discusses how the factors

need for change, usefulness and personal effects seem to contribute to employees’ overall

attitudes on the RPA implementation ex-ante and ex-post. The last part of this chapter discusses the evolvement of the attitude during the implementation.

6.1 Need for Change

When talking about the need for change the employees mentioned the nature of the process when talking about the need for change. When processes are simple and repetitive,

(32)

employees mostly see the need for change. Furthermore, they seem to take into account the positive as well as negative effects of automation on the customer satisfaction. In reviewing the literature, no data is found on the association between the need for change and the customer satisfaction. It is therefore difficult to explain this result but, it might be related to the nature of the work. All the employees who were interviewed for this study used the RPA systems to process client data. Some respondents were also responsible for processing the client complains. As a result, it is possible that the respondents of this study are used to think about the customers’ perspective, which could explain their concerns about the customer satisfaction. Future studies, which take the customer-oriented nature of the job into account, need to be conducted to provide more clarity on this issue.

6.2 Usefulness

The perceived usefulness does not seem to be related to the overall attitude towards RPA ex-ante. This finding is explained by the respondents themselves, who stated that they did not know enough about RPA to form a view on usefulness ex-ante. The findings indicate that respondents mentioned concerns about the usefulness of RPA ex-post. These concerns are related to the suitability of the specific process to RPA. Respondents experienced that RPA is not able to handle exceptions or to execute processes that require an input of multiple sources. This finding is remarkable since RPA vendors explicitly promote the ability to handle exceptions and the possibility to use existing systems as a benefit of RPA in comparing to traditional types of automation (Blue prism Japan, 2018). However, the promises that have been made about the positive results of RPA can also explain these findings. Ford, Ford, & D’Amelio (2008) argue that resistance the change among employees can occur as a result of broken promises regarding the impact of the change. Creating too positive expectations can therefore negatively affect the success of the implementation. It is

(33)

therefore recommended to provide further research on the effect of ex-ante promises about the RPA implementation on the ex-post perceived usefulness.

Another point of concern that is mentioned regarding the usefulness is the reliability on individual expertise. In line with the arguments of Carr (2013), some respondents warn for the fact that the implementation of RPA results in a decrease of knowledge about the specific process. It can, therefore, be possible that the paradox of increased automation as mentioned by Carr (2013), also occurs on a small skill when RPA is implemented. However, more research on this topic needs to be undertaken before the association between this paradox, and the RPA implementation is more clearly understood.

When summarising the findings on the ex-post evaluation of usefulness, it can be indicated that employees might be concerned about the usefulness of RPA ex-post. They believe that it cannot handle the complexity of the initial processes and mentioned that the errors that occur cannot be solved correctly and on time. When interpreting these findings, it is important to realise that for all of the cases in this study, experts were hired to automate the process. As a result of this external help, information hand-overs are crucial. First, the expert needs to know how the process is being executed precisely, including all the exceptions, to build a robot that executes the tasks in a right way. Secondly, the companies’ process owner needs to know exactly how the robot is programmed. This information is needed, to make adjustments when errors occur, or when an input source changed. Based on the concerns that have been mentioned regarding the inability to deal with complex situations, it is possible that companies and RPA vendors do not pay enough to these hand-over moments. Further research could explore these hand-over moments in more depth to examine how these moments affect the success of the implementation and which aspects of these hand-over moments are considered to be important.

(34)

6.3 Personal Effects

There seems to be consensus among the employees that the RPA will positively impact them because it automates the tasks which they perceive as annoying and simple. This perception is not surprising, because automating simple tasks is the aim of RPA (Blue prism Japan, 2018). Additionally, it is a common view among the employees that the

implementation of RPA saved time, which they can now spend on other tasks. It is remarkable how the employees use this extra time saved. Within the findings, there is no indication of management guidance on how to use this extra time. As a result, employees seem to be a responsible for what kind of work they do in this time saved. Although it appears that some of the employees found tasks that seem to be valuable in the first place, such as answering client questions, it is doubtful if every employee is able to prioritise tasks in line with the organisational strategy and if he or she can cope with this responsibility. In organisations that are characterised by bottom-up management styles, for example,

employees are more used to work autonomously prioritise their tasks. However, in top-down steered organisations employees probably need more guidance to experience the

organisational benefits of RPA completely (Hornung, Rousseau, Glaser, Angerer, & Weigl, 2010). Future research is needed to investigate how the time that is saved as a result of an RPA implementation is used.

One unexpected finding was that only one respondent felt afraid to lose his job as a result of the RPA implementation ex-ante. Based on prior research regarding the effects of automation on the labour market it was expected that more respondents would mention this concern. Surprisingly, even the respondents who perceived the automation as a threat to the labour market did not feel afraid to lose their job. A possible explanation for the absence of this fear can be related to the employee involvement. It appeared that most respondents that have been interviewed for this study were involved with the implementation. Prior literature

(35)

regarding the involvement of employees in organisational change emphasise that employees who participate in the process of change, feel more in control over the complete process (Bordia, Hobman, Jones, & Gallois, 2004). It is possible that this feeling of control positively affects their perceived job security (Bordia et al., 2004). Talking about this issue, some respondents mention the effect of age on the perceived job threat. Employees in the age of 20-40 mentioned the fact that older colleagues feared for their jobs. However, this view was not supported by respondents in the age of 40-65. On the contrary, one respondent at the age of 64 mentioned his expected retirement as a reason not to fear his job. The one respondent that was afraid to lose his job ante is 38. However, he does not experience this fear ex-post, because he saw the relatively small impact on his job. Altogether, the findings of this study do report some concerns about job loss. Future research at least in the age category of 30-60 is recommended to investigate this matter further.

6.4 Ex-ante versus Ex-post Attitude

The findings indicate that employees are positive towards the RPA implementation ex-ante and ex-post. These attitudes towards the overall implementation of RPA seem to be in line with their belief on how the RPA implementation affects them personally and the extent to which they believe that changing the initial process is needed. Based on these findings, it is likely that self-interest contributes to the perception towards the RPA

implementation. These findings corroborate the Theory of Planned behaviour that states that people base their actions and intentions on perceived outcomes that are favourable for

themselves (Ajzen,1991). Additionally, the importance of the perceived need for change is in line with prior literature that stresses the importance of creating a sense of urgency

(Armenakis & Bedeian, 1999; Bagozzi et al., 2000; Bouckenooghe, 2010; Kotter, 1995; Rafferty et al., 2013). Over 50% of the companies that have been studied by Kotter (1995)

(36)

failed to create a sense of urgency. Based on the findings of this research, however, creating a sense of urgency does not seem to be an issue. Further work is required to establish this observation.

7. Conclusion

This thesis investigated the ex-post and ex-ante attitude of employees towards the implementation of RPA, by exploring their cognitive beliefs towards this implementation. To gain insights into these attitudes, this conclusion answers the research question: What are the

expectations and evaluations regarding the implementation of Robotic Process Automation of employees whose job is affected by this implementation?

Overall, the findings of this study indicate that employees have an overall positive attitude towards the implementation of RPA both ex-ante and ex-post. These findings suggest that the positive overall attitude ex-ante seems to be related to their expectation that the implementation will be beneficial for them personally, in combination with their perceived urge to change the process. They believe that the initial process is suitable for automation because it is simple and repetitive and expect a reduced workload and a more varied job.

Employees still have a positive attitude towards the implementation ex-post, which seems to be related to beneficial outcomes for them personally such as a reduced workload and more diverse job tasks. Some are less positive about the usefulness of RPA. They believe that it cannot handle the complexity of the initial processes and mentioned that the errors that occur cannot be solved correctly and on time.

7.1 Theoretical Contribution

This research extends our knowledge about the effect of RPA on the workforce. Therefore, this study contributes to the literature in the field of RPA two ways. First, it provides insight into the perception of the workforce towards RPA. Secondly, this work

(37)

contributes to existing knowledge about organisational change. It provides insight into the relative importance of the cognitive beliefs that employees have regarding an organisational change in the case of RPA. Lastly, this study assists in our understanding of human-machine interaction, by analysing the human workforce perception on the implementation of RPA.

7.2 Managerial implications

Based on the findings of this study, some managerial implications can be provided. Before the implementation, it seems to be essential for managers to select which processes are suitable to RPA carefully. Based on the results of this study, it is possible that the perceived need for change influence the overall perception of employees towards the

implementation. This observation indicates that a careful benefit-risk assessment, focused on the suitability of the initial process for RPA positively influences the overall attitude towards the RPA implementation. In addition, it is recommended to investigate how possible errors can be solved correctly.

Whereas the first recommendation focuses on the pre-implementation phase, the second recommendation applies to consequences of RPA. The results of this study indicate minimal managerial guidance on how to use the extra time that is saved as a result of the implementation. Although some employees might be able to prioritise their tasks, it might be more difficult for other employees. It is therefore recommended to take into account the possible job consequences of the RPA implementation for the human workforce before automating the process.

The last implication refers to the automation paradox (Carr, 2010). Most concerns that have been mentioned are related to the suitability of the process for RPA, the ability to solve errors correctly and the reliance on individual expertise. Based on these observations, it is recommended that managers take into account the effect of the implementation on the

(38)

organisational knowledge. As automation continues to advance, knowledge retention seems to become increasingly vital for organisations. Based on the findings of this study, it is possible that employees experience adverse effects related to knowledge retention, such as the inability to solve errors and difficulties in answering client questions, even in this early stage of RPA. It is therefore recommended for managers to take into account the possible risks of knowledge loss when implementing RPA.

7.3 Limitations

The findings of this research are subjected to a couple of limitations. Firstly, this research is focused on comparing the pre-implementation with the post-implementation phase. However, due to practical matters, it was not possible to interview respondents before and after the implementation. Consequently, the respondents were asked to express their expectations in the post-implementation phase. This could lead to biased results since the respondents' memories can be replaced by new experiences (Wichert, Wolf, & Schwabe, 2013). Secondly, the study uses a convenience sample that is limited to clients of KPMG. As a result, it is possible that other results occur for companies who did not cooperate with KPMG. Thirdly, prior research in the field of Psychology addresses a couple of individual differences such as personality factors that might influence the individual attitude. Due to the limited scope of this research and the complexity in measuring the personality traits in qualitative research, these factors have not been taken into account. Lastly, this study is limited in generalizability, because of the small sample size. Eight respondents were interviewed for this study. Due to the novelty of RPA, it was hard to find companies who recently implemented RPA. As a result of these limitations, the findings of this study need to be interpreted with caution.

(39)

7.4 Future research

This research has brought up a couple of questions that require further investigation. This research investigated the attitude of employees towards the implementation of RPA. More research is required to analyse the effects of these attitudes on employee behaviour. This is particularly interesting when human behaviour is used as learning input for more advanced types of RPA. In this study, some concerns were observed on to the knowledge retention as a result of the RPA implementation. Further research might investigate the links between the implementation of RPA and this knowledge retention. Additionally, it would be interesting to measure the effect of the RPA implementation on customer satisfaction.

Furthermore, as RPA advances continue, and more employees experience the effects of RPA, more quantitative studies should be conducted to find statistical evidence.

(40)

Appendices

Appendix A: interview questions

Introductie (verzoek tot opname, uitleg over het doel van het onderzoek, benoemen van anonimiteit), uitleg over het doel van het interview, voorstellen

Context

1. Kun je mij meer vertellen over de RPA-implementatie?

a. Indien niet besproken: Welke processen werden gerobotiseerd?

2. Wat was het doel van de implementatie?

3. Hoe is jouw werk veranderd sinds de implementatie van RPA? (Vraag uit ander onderzoek)

4. Hoe zag het veranderproces eruit?

5. Hoe keek je van tevoren aan tegen het implementatie proces van RPA? 6. Hoe ervaar je nu het implementatie proces van RPA?

7. Hoe keek je van tevoren aan tegen RPA an sich? 8. Hoe ervaar je RPA nu?

9. Hoe sta je over het algemeen tegenover de implementatie?

Cognitieve perceptie

10. Vond jij het voor de implementatie noodzakelijk voor jou persoonlijk dat dit proces geautomatiseerd werd? Waarom wel/niet?

11. Hoe kijk je daar inmiddels tegenaan?

Bruikbaarheid

12. Hoe gebruik jij RPA in je dagelijkse werkzaamheden?

(41)

14. Had je voorafgaand aan de implementatie gedacht dat RPA het behalen van je doelen makkelijker/moeilijker zou maken? Waarom?

Persoonlijke effecten

15. Welke persoonlijke voordelen ervaar je door de implementatie?

16. Welke voordelen dacht je dat RPA je te gaan ervaren toen je hoorde dat het geïmplementeerd zou worden?

17. Welke persoonlijke nadelen ervaar je door de implementatie?

18. Welke persoonlijke nadelen dacht je dat RPA je te gaan ervaren toen je hoorde dat het geïmplementeerd zou worden?

Afsluiten

15. Wat verwacht je verder nog van ontwikkelingen als RPA? 16. Zijn er nog dingen die je kwijt wil?

(42)

Appendix B: Selected Cases

Table 1:

Case Selection

Organisation Departement Interview date

Bank Fund agency 23-04-2018

Bank Data management 04-05-2018

Bank Settlement 01-05-2018

Bank Settlement 30-04-2018

Bank Tax 04-05-2018

Energy sector Procurement 04-05-2018

Energy sector CRM data management 24-04-2018

(43)

Appendix C: Case Information Context Bank

In 2006 Bank started with using RPA. They selected rule-based standardised processes to decide which process was suitable for automation. To overall goal of implementing RPA was threefold. Bank wanted to reduce FTE to improve their efficiency. Another reason was to reduce the error rate. Furthermore, by using RPA Bank could execute processes 24 hours a day instead of limiting the execution of processes to working hours. The change processes started in 2006 and Bank is still robotising processes, which results in incremental change implementation. Robotizing is part of the company's overall strategy and therefore affects multiple departments including the departments data management, tax operations, fund agency, and settlements. For the aim of this research, respondent of all three departments will be interviewed. To get a perspective and

overcome the bias of interviewing only remaining employees, one interview is conducted with an employee who recently left Bank as a result of the RPA implementation.

Respondent 1.

The first respondent works at the fund agency department. He is 50 years old, completed vocational education and, have been working at Bank for 24th_{years. His job is to}

set up an attendance ticket for customers who want to attend a shareholder meeting of

companies. When a request is received, he checks whether they do indeed have shares in that company and with how many votes they can enter the meeting. For some markets, mostly depending on the country, he must block the trading in shares. When checking these requirements, he makes an attendance card, or request the attendance card at another Bank. He explains this process as time-consuming, manually and not nice to do.

Furthermore, he explains that it involved many manual calculations and that he experienced a high time pressure. The process was executed by her and controlled by someone else. Altogether this whole process toke 2,5 hours a day. After the RPA implementation, this process is entirely executed by RPA