Implementation of Robotic

Process Automation -
A case study of issues, challenges and
success factors for RPA
implementation in banking and
financial services

Master’s Thesis 15 credits

Specialization: Management Control

Department of Business Studies

Uppsala University
Spring Semester of 2021
Date of Submission: 2021-06-02

Haris Camo
Niklas Grufman
Simon Harnesk
Supervisor: Jan Lindvall
 Acknowledgements

We would like to thank all respondents from the case-, and consulting company for your
valuable time and engagement in the study. We would also like to use the opportunity to thank
our supervisor Jan Lindvall for your valuable input and support throughout the process.

Haris Camo, Simon Harnesk & Niklas Grufman

Stockholm, 2nd of June 2021
Abstract
Robotic Process Automation (RPA) is an emerging lightweight IT solution that aims to
automate digital but manual business processes and has been particularly useful in the banking
and financial services industry. Whilst interest for RPA has been increasing along with the
increased pressure on organizations to leverage technology for efficiency to remain
competitive, studies show upwards of 50 percent of RPA implementations fail. Building on
problematizations of RPA implementation failures and scholarly urges for further research
within the field, this study aimed to investigate; issues and challenges to overcome, as well as
how RPA implementation may be facilitated. Based on a literature review including existing
implementation frameworks, a case study of semi-structured interviews with managers and
expert consultants was conducted, and project material was reviewed.

Several differences and similarities between existing theory and the respondents' experiences
were identified. Almost all issues and challenges emphasized by respondents were related to
the challenge of achieving adequate; change management, communication, education around
RPA, choosing the right processes to automate, and post-implementation and scale up. Several
of the common issues and challenges mentioned in theory, such as technical aspects, were not
experienced by the respondents and these mitigated issues and challenges may be dependent
on contextual preconditions. The most important facilitating actions were identified as; building
RPA knowledge, educating management and employees within the organization,
communicating new processes, RPA benefits and reasons for implementation, ensuring support
processes and establishing internal groups for RPA scale-up and stewardship, and carefully
choosing the right process to automate. Contextual preconditions likely facilitating the RPA
implementation for the case company were identified and interpreted as potential success
factors for RPA implementation. A structured approach with visualized project steps was
emphasized as important, and whilst the existing implementation framework compared to the
case was accurate and useful, the study suggest implementation frameworks may need to
include further attention to; change management, communication and education, and the
relative importance of the initialization- and post-implementation phases. The research is
concluded with a discussion on limitations and suggestions for research opportunities.

Keywords:
Robotic Process Automation, RPA Implementation, Implementation Framework, RPA in
Banking and Financial Services, Loan Administration.
Table of contents
1. Introduction 1
1.1 Problem background 1
1.2 Research gap 2
1.3 Research purpose 3
1.4 Research question 4
1.5 Study delimitations 4
2. Theoretical framework 5
2.1 IT and Business processes 5
2.2 Robotic process automation 5
2.3 RPA in Banking and Financial Services Industry 6
2.4 Benefits RPA 6
2.5 Issues and disadvantages with RPA 7
2.5.1 Limited use-cases and suitable processes 8
2.5.2 Issues with implementing RPA 8
2.6 Guidelines and models for implementation of RPA 9
2.6.1 Frameworks for RPA implementation 9
2.6.2 Implementation theory for technology fields outside RPA 14
2.7 Conclusions and analytical framework 15
2.8 Analytical model 16
2.8.1 Simplified analytical model 17
3. Methodology 18
3.1 Research approach 18
3.2 Research design 19
3.2.1 Qualitative case study design 19
3.2.2 Case company and participants 19
3.3 Empirical data collection process 21
3.4 Data analysis 22
3.5 Reliability, validity and replicability 23
3.6 Method discussion 24
4. Empirical findings 25
4.1 Structure 25
4.2 Fundamental aspects about the case and reasoning behind the RPA implementation
project 25
4.3 Phases and activities of the project 26
4.3.1 Initialization 26
4.3.2. Implementation 27
 4.3.3 Post-implementation 28
4.3.4 Support-processes 29
4.3.5 Model and summary of the implementation project phases 30
4.4 Issues and challenges 30
4.4.1 Human and organizational issues 31
4.4.2 Issues with choosing the right process to automate 32
4.4.3 Issues related to changes in systems 33
4.4.4 Issues for broader adoption and scale-up 33
4.5 Specific aspects the respondents would have done differently 34
4.6 Success factors and facilitating actions 35
4.6.1 Preconditions and contextual factors 35
4.6.2 Aspects to facilitate implementation 36
5. Analysis 39
5.1 Intentions and goals 39
5.2 Issues and challenges 40
5.3 Implementation approach and success factors 42
5.3.1 Success factors and aspects facilitating implementation 42
5.3.2 RPA implementation project approach 44
6. Conclusions 48
6.1 Issues and challenges for RPA implementation projects 48
6.2 Managing an RPA implementation project 49
6.3. Limitations and suggestions for future research 50
References 52
Appendices 57

List of figures
Figure 1. A typical RPA Process ................................................................................................ 6
Figure 2. Consolidated framework for RPA implementation projects ..................................... 11
Figure 3. Analytical model ....................................................................................................... 16
Figure 4. Simplified analytical model ...................................................................................... 17
Figure 5. Summary of the implementation project as visualized by the case company........... 30
List of tables
Table 1. Sample overview ........................................................................................................ 20
Table 2. Interview themes ........................................................................................................ 22
1. Introduction
This chapter starts by providing a descriptive background of Robotic Process Automation, and
associated challenges with the field. Further, we will cover the current streams of academic
research within the field in order to confirm a research gap and ensure the originality of the
thesis. Moreover, we will present our motivations for the research purpose, reflecting why the
study is practically needed and called upon in the given context. Thereafter, the research
questions are stated and followed by the discussion on expected theoretical, managerial and
societal contributions.

1.1 Problem background

The rise of emerging technologies and innovations has transformed our economic and business
world, presenting businesses with unprecedented opportunities to incorporate, exploit, and
leverage these modern digital solutions and innovations (Nwankpa and Merhout, 2020; Lin,
2018). As a consequence, businesses are under increased pressure to prioritize digital
investment and to turn it into creative business practices, new technologies, and business
models (Chae, 2019). Digital investment, which is defined as a firm's strategic technology
investment aimed at determining how ingenious digital technologies can potentially
differentiate its business, transactions, and processes (Nwankpa and Datta, 2017), is viewed as
a critical imperative for firms seeking to remain competitive and maintain market positions
(Syed et al., 2020; Yoo et al., 2012). Since many business processes are already performed by
computers, digitalization opens up a plethora of opportunities for optimization (Fischer et al.,
2020), such as the analysis and improvement of business processes.

Robotic Process Automation (RPA) is a recent emerging technology that aims to automate
digital but manually performed business processes (Lacity et al., 2016; van der Aalst et al.,
2018a). Unlike other process automation technologies, RPA ‘sits on top of’ the other IT systems
accessing these platforms through the presentation-layer, thus no underlying systems
programming logic is touched (Aguirre and Rodriguez, 2017; Lacity & Willcocks, 2015a). This
solution has been particularly useful in the banking and financial services industry (Madakam
et al., 2019), since the business processes are often managed through a combination of legacy
systems, new technologies, and manual processes (Vishnu, et al., 2017).

1
During recent years, interest for RPA in research and in terms of search-numbers in search
engines has been growing rapidly, indicating an increasing importance of RPA (Santos et al.,
2019). According to Gartner (2020), a leading IT research and advisory company, the RPA
market is expected to grow at double-digit rates annually through 2024. A successful RPA-
implementation can have major benefits for organizations, such as increased productivity
(Alberth and Mattern, 2017; Madakam et al., 2019; Vishnu et al., 2017; Syed, et al., 2020), full-
time employee (FTE)- and cost reductions (Lacity and Willcocks, 2015b; Santos et al., 2019)
and increased reliability and continuity of service (Santos et al., 2019). Moreover, RPA can
often be introduced more quickly and cheaper than IT solutions relying on APIs for system
integration, taking two to four weeks rather than months or years (Asatiani and Penttinen,
2016). Since this light-weight automation approach hence allows for return on investment
(ROI) to be achieved in a shorter time, RPA is often a strategically valuable alternative (van
der Aalst et al., 2018a).

Although the benefits from the implementation of RPA are well documented, it cannot be taken
for granted that an implementation of RPA will result in benefits for organizations (Syed et al.,
2020). In a study performed by EY, 30 to 50 percent of initial RPA projects fail (Lamberton,
2016), and only about 20 percent of organizations that implemented RPA in 2019 achieved a
business value that exceeded what they had expected (Willcocks et al., 2019). The field of RPA
has been described as an unexplored field, and academic research in implementation and
utilization has only recently begun to rise (Lamberton, 2016). Research on so-called RPA
"methodologies" take the form of lessons learned, recommendations, best practices, and
experience-reports from RPA implementations within organizations (Syed et al., 2020) and the
current access to established guidelines or best practices for maximizing the benefits of RPA
implementations (from adoption to delivery) are scarce (ibid.). As a result, finding a systematic
approach to optimizing the benefits from an RPA implementation becomes an open issue to
address (ibid.).

1.2 Research gap

Provided the many potential benefits of RPA described in literature, and the inherent difficulties
with implementation - the difficulties and eventual solutions is evidently relevant to investigate.
In terms of previous scholarly contributions to RPA, the field can be described as recent and
unexplored (Santos et al., 2019). This is also supported by several scholars including

2
Lamberton (2016) who suggests RPA remains an unexplored field and academic research in
implementation and utilization has only recently begun to rise, and Herm et al. (2020)
suggesting that “[...] from a research perspective RPA is poorly understood and only in the
early stage of scientific research. Hence, several areas have not yet been sufficiently
investigated and pose challenges [...]”. With regards to specific research on implementation
issues, solutions, and suggested approaches, scholars have also urged for extended research on
certain fields within the generally unexplored field of RPA. According to Syed et al. (2020),
guidelines and best practices for benefit realization from deploying RPA rarely exist and hence,
development of approaches to support benefit realization is an open issue to address.

A few scholars have presented general frameworks for implementation considerations, and
Gotthardt et al.´s (2020) suggest additional research is needed on implementation. In relation
to Santos et al.´s (2019) conceptual model for RPA implementation (presented in the theoretical
framework of this thesis) the authors suggest “future research should focus on applying the
proposed model on conducting a CS [case study], by using the steps identified and considering
RPA main topics and then refine the model based on the experience of conducting a CS, if
needed”. As will be seen, the conceptual model suggested to be adapted to a case has great
similarities with other models suggested by literature - pointing to the fact that adaptation and
evaluation of specific implementation models could be beneficial for the RPA implementation
field in general.

Conclusively, in addition to the obvious relevance of RPA knowledge described in the problem
background, the logic of this research will build on fundamental pillars of research gaps related
to; the field of RPA in general, best practices and RPA implementation approaches, as well as
application of previous RPA implementation frameworks on new cases.

1.3 Research purpose

Based on the practical relevance described in the problem background and the identified need
for further research within the field of RPA implementation, the aim of this study is to explore
common issues and challenges associated with RPA implementation. Furthermore, based on
existing frameworks and theory, the study aims to explore how projects may be managed in
order to facilitate implementation.

3
1.4 Research question
In order to address the research gap and contribute to the overall purpose of this research, the
following research questions were developed;
1) What are the issues and challenges for RPA implementation projects?
2) How can RPA projects be managed to facilitate implementation?

The study aims to investigate these questions in the context of the banking and financial
services industry, as will be described in the study delimitations.

1.5 Study delimitations

As the purpose of this study is to perform an overall mapping of problems, facilitating aspects
and suggested project approaches for RPA implementation, this study will focus on RPA
literature specifically and do not intend to make cross-disciplinary comparisons between other
fields of research due to the limited time and scope. Accordingly, this study will not include
BPM (Business Process Management) which is a common subject related to RPA in previous
literature. Provided the similarities and inherent connections between IT-implementations,
however, some references are made to ERP (Enterprise Resource Planning) implementation.

In this study, the term ‘implementation’ (of RPA) refers to the activities of automating new
processes and includes activities and long-term considerations of both initializing - before the
actual ‘implementation’ - and expanding the number of activities automated - after the
‘implementation’ of an automation. The same scope of ‘implementation’ is implicit in RPA
literature. The term ‘successful’ implementation is seemingly established in literature and
however remains undefined by scholars. Whilst this thesis will try to incorporate goals for the
case company´s implementation to put the conclusions in a context, implementation ‘success’
will be broadly defined as the foundational automation of a targeted process.

The fact that the case study only examines one company in the banking and financial services
industry, and with has certain pre-condition (to be described), may have a negative impact on
the generalizability and hence usefulness of this study. Moreover, provided the focus on
describing the entire process of an implementation and hence using a limited number of
centrally involved respondents along with triangulation through reviewing project
documentation - the study does not include e.g. end-users or other case company stakeholders

4
to describe challenges from different perspectives than the project team(s) and their focus on
project completion.

2. Theoretical framework
In this chapter, a theoretical foundation of previous research within the field of RPA and RPA
implementation will be presented. Based on the key topics identified, the theoretical framework
is concluded with an analytical model to be used for answering the research questions.

2.1 IT and Business processes

A fundamental part of business and management is that organizations seek to increase their
operational performance by redefining and managing their business processes in order to stay
competitive. In accordance with the definition provided by Pettigrew (1997, cited in Bryman,
2012) this research will define a process as; “a sequence of individual and collective events,
actions, and activities unfolding over time in context’. In order to achieve the objective of
staying competitive, information technology (IT) plays a major role (Syed et al., 2020). This
requires new work organizations that encourages agility and teamwork and concentrates more
on cognitive tasks that generates “know-how” and value. However, substantial parts of
organizations resources are often occupied by routines and repetitive tasks (van der Aalst et al.,
2018b). Despite the extensive use of process-aware information systems, most contemporary
businesses still rely on employees to initiate or stop procedures, to adapt them to different
situations, or to use their results for various purposes (van der Aalst et al., 2018b). Robot
process automation (RPA), an emerging trend in the area of business process automation, had
been proclaimed a possible solution to tackle this.

2.2 Robotic process automation

RPA is defined as a technology tool that uses software and algorithmic programmed robots
acting as human beings in system interaction to support efficient business processes (Lu et al.,
2017; Santos et al., 2019; Syed, 2020; van der Aalst et al., 2018). RPA is designed to relieve
employees of the burden of performing repetitive, simple tasks (Aguirre and Rodriguez, 2017).
It does not replace other systems. Instead, it is sitting on top of the other systems and performs
tasks which it has been programmed to complete (Lacity & Willcocks, 2015a). Hence, most
times, integration of RPA does not require any change in the existing systems (Lacity &
Willcocks, 2015b). In the same way as a human would interact with a software system by
clicking and typing, the RPA does it but with greater execution (Lu et al., 2017). However, it
5
would be no problem for a human to be reinstated and execute the task instead if needed
(Asatiani and Penttinen, 2016). RPA accesses systems in the same way a human does – with
its own username and password (Lacity & Willcocks, 2015a). One example where an RPA
would be suitable is when an employer logs in to one system, gathering data and processing
them using rules, then uploading this data into another system (Lacity & Willcocks, 2015a).
This process is illustrated in Figure 1.

Figure 1. A typical RPA Process (Adapted from Lacity & Willcocks, 2015a).

2.3 RPA in Banking and Financial Services Industry

With the widespread adoption of virtual banking, banks must invent new ways to provide the
best possible customer experience while reducing costs, maintaining security, and meeting
regulatory and compliance requirements (Vishnu, et al., 2017). To optimize operations and
improve efficiency, financial institutions have to focus on improving the speed and accuracy
of the core business processes – this objective can be achieved using RPA (ibid). The industry
is one of the most suitable for the technology (Madakam et al., 2019), as the organization's core
business processes generate a high volume of documents and are managed through a
combination of legacy systems, new technology, and manual processes (Vishnu, et al., 2017).

2.4 Benefits RPA

Several benefits with RPA is prominent in literature. One of the main benefits with RPA is that
the robots can operate around the clock, which provides reliability and continuity of services
(Syed, 2020), allowing for FTE reductions (Lacity and Willcocks, 2015; Santos et al., 2019),
and reduction of entry costs by upwards of 70 percent (Anagnoste, 2017). In fact, based on
FTEs replaced by robots, RPA technology has been shown to reduce the cost of human
resource-related spending by 20–50 percent and transaction processing costs by 30–60 percent
(Syed, 2020). By implementing robots for routine tasks, workers may concentrate on more
interesting value-adding activities that involve personal interaction, problem-solving and
decision making (Syed, 2019), also contributing to increased job satisfaction and employee
retention (Santos et al., 2019).

6
In comparison to humans, robots also make less mistakes and perform with higher efficiency,
resulting in increased productivity (Alberth and Mattern, 2017; Madakam et al., 2019; Vishnu
et al., 2017; Syed, 2020). RPA is also useful from a risk and compliance perspective as it can
maintain a log of work completed to ensure that the automated tasks and processes comply with
regulatory requirements (Syed, 2019). Furthermore, RPAs may be deployed more quickly than
other IT solutions that rely on APIs to integrate with systems, often taking two to four weeks
rather than months or years to introduce (Asatiani and Penttinen, 2016). As a result, RPA often
enable a fast ROI (Lacity and Willcocks, 2017; Santos et al., 2019) attractive for organizations.

2.5 Issues and disadvantages with RPA

In addition to scholarly attention to the many opportunities and potential benefits of RPA
provided in the previous section, several authors have highlighted limitations and potential
challenges with RPA. One of the main challenges with the use of robotics and RPA is
maintenance whereby complexity and costs of adapting the RPA solution to changing IT
infrastructure or organizational needs may be underestimated (Stolpe el al., 2017). When
systems change, robots often have to be updated which is time consuming and costly (Santos
et al., 2019). Furthermore, whilst RPA is an IT tool, the processes automated often belong on
the business side, whereby division of responsibilities between the IT and business side is
sometimes unclear during implementation and maintenance (Santos et al., 2019).

Lacking understanding of the term RPA and what it means, has also been described as an issue
whereby the technology may be interpreted as related to robotics, rather than software robots
(Santos et al., 2019). Furthermore, whilst providing efficient processing of tasks through
prompt actions, there is no human checking for errors before task execution meaning the robots
can make mistakes faster, not waiting for responses from applications (Santos et al., 2019).
Literature has also described that whilst some employees allocate working hours into new tasks
when RPA decreases the need for human processing, others simply replace their workers with
robots (Syed et al., 2020; Santos et al., 2019). Issues hence also appear with regards to impact
on employees who may see robots as opponents for job opportunities, which may have a
negative impact on the workplace (Asatiani & Penttinen, 2016). The buzz around RPA and its
potential impact for efficiency and reduction of workforce needed has provoked some public
opinion against automation software (Santos et al., 2019; Asatiani & Penttinen, 2016), as well

7
as created vast expectations and promises of impact which may be difficult to achieve
(Rutaganda et al., 2017).

2.5.1 Limited use-cases and suitable processes

The limitations in terms of suitable use-cases for RPA and that such limitations need to be
considered, is recurring in literature. Several criteria for when RPA is especially useful is
evident (Santos et al., 2019) including that the process to be automated preferably should;
include high volumes of transactions or information (Asatiani & Penttinen, 2016; Fung, 2014;
Lacity & Willcocks, 2015), be standardized and mature in terms of knowledge and experience
around it (Willcocks et al., 2017) and frequently interact with multiple systems (Asatiani &
Penttinen, 2016; Fung, 2014; Lacity & Willcocks, 2015). Santos et al. (2019) argue it is
important for companies to know whether their processes are suitable for automation, and
companies hence need to consider these disadvantages and limitations in terms of suitable
processes when adopting RPA to automate processes.

2.5.2 Issues with implementing RPA

Whilst the benefits that may be gained from implementing RPA are well documented, benefits
realization from RPA deployment cannot be taken for granted (Syed et al., 2020). The literature
is clear with regards to that RPA implementation is not always smooth sailing and that many
potential implementation issues may arise (Gex & Minor, 2019; Rutaganda, 2017; Santos et
al., 2019; Herm, 2020; Gotthardt at al., 2020). Accordingly, implementation challenges
represent important aspects of RPA which organizations need to consider (Santos et al., 2019).
Despite the existence of many RPA products, vendors and consultants in the market,
uncertainties about how to successfully implement and utilize the technology is common (Syed
et al., 2020). Although RPA is considered an easy to implement technology, in debt knowledge
within the field is necessary for successful implementation success (Herm et al., 2020).

Whilst the use of RPA has been steadily increasing, RPA solutions is to be considered as a
relatively new subject and the academic research within the field of RPA implementation and
utilization has only recently begun to rise, whereby systematic approaches for benefit
realizations of RPA implementation is still an open issue to address (ibid.).

8
In the ´Digital Directions´-report, produced with Microsoft, EY (2020) point to indications on
that upwards of 40 percent of RPA projects are considered as failed, and that EY´s consultants
experience common mistakes include; automating too much of a process, targeting the wrong
processes, disregarding IT infrastructure, underestimating skills needed, and underestimating
the need for continuous efforts post implementation. From a broader perspective, Willcocks et
al. (2018), however conclude that implementation issues are often a consequence of mistakes
from management rather than related to the tool. In an article for McKinsey, Leslie Willcocks
also mentioned change management and leadership as important aspects for RPA
implementation (Lhuer, 2016). Accordingly, Rutaganda et al. (2017) highlights incorrect
leadership at the top level of an implementation project as a common and critical mistake. In
their research they found that projects that were IT-led failed to a greater degree than those that
were business-led (ibid.). This is a common misunderstanding among organizations - that
emerging technologies such as RPA, are the territory of the IT-function (ibid).

Earlier in this chapter, the researchers (hereon after occasionally called ‘we’ in this study)
introduced potential benefits and purposes of RPA, however despite the many potential pitfalls
for implementation, and the seemingly high rate of failures and disappointments with
implementing RPA as mentioned above - Syed et al. (2020), claim that (then) existing literature
“lacks a clear framework on what the critical success (or failure) factors are”. The below
sections will present existing implementation guidelines for RPA systems identified, and
considerations of scholarly contributions on technology implementation outside the field of
RPA will be mentioned.

2.6 Guidelines and models for implementation of RPA

Whilst the overall literature on RPA implementation approaches is scarce, some contributions
regarding implementation frameworks exist. In this section, we will present the most developed
frameworks for RPA implementation along with some minor perspectives on implementation
outside the field of RPA.

2.6.1 Frameworks for RPA implementation

Some scholars have addressed RPA implementation approaches in recent publications. As will
be seen, most emphasis is made to general processes and ‘parts’ needed in implementation
projects, without much attention to specific actions. These frameworks may, however, provide

9
some helpful guidelines of the overall RPA implementation projects and the insights collected
for implementation improvement thus far.

Santos et al.´s (2019) conceptual model for RPA implementation success

In their review-article “Toward robotic process automation implementation: an end-to-end
perspective” aggregating insights from existing RPA implementation studies, Santos et al.
(2019) provides an overarching conceptual model of the main RPA topics including three broad
steps for successful RPA implementation, namely; strategic goals, process assessment, and
tactical evaluation. The first step “strategic goals”, suggest automation objectives need to be
established based on- and aligned with the company goals in order to motivate automation and
to provide a benchmark to evaluate and measure if objectives are achieved after eventual
implementation (ibid.). According to the model, these goals must consider the benefits,
limitations and future challenges of RPA in order to understand potential advantages
automation may bring, to avoid establishing unrealistic objectives which cannot be attained,
and to enable a long-term perspective for actions to address common challenges (ibid.). After
strategic goals have been established, Santos et al. (2019) suggests the process to be automated
needs to be assessed based on common RPA difficulties and avoiding processes requiring
adaptation to many changes in systems, a lack of rules, or other issues mentioned in literature.
Finally, after the most suitable processes have been chosen, the conceptual model suggests a
tactical evaluation of how to implement the RPA automation is conducted based on the same
factors identified in the literature (opportunities, difficulties and future challenges) to identify
e.g. integration needs.

Herm et al.´s (2020) ‘Consolidated Framework for Robotic Process Automation

Implementation Projects’
Based on a review of case studies and validation of the findings with RPA experts, Herm et al.
(2020) recently presented a consolidated and refined model for RPA implementation projects
in which the common stages of RPA implementation are segregated, and general guidelines are
suggested for each stage of the implementation process. The model is based on case studies
from a variety of heterogeneous contexts and the findings were generalized to capture general
RPA implementation projects (ibid.). The framework is divided into three phases for
implementing RPA projects, namely; initialization, implementation, and scaling, in which
some stages are performed once during each project, whilst others may be repeated (ibid.).

10
Many stages will likely be overlapping, as described in the below model adapted from the
authors (see Figure 2).

Figure 2. Consolidated framework for RPA implementation projects (adapted from Herm et
al., 2020).

Herm et al. (2020) describe their review suggest RPA implementation project initialization
phase starts with identification of automation needs through e.g., workshops, surveys,
interviews, documentation analysis, business discussions and through conversation with
employees of relevant department(s). The authors also emphasize that early alignment with
business strategy to ensure RPA implementation can support strategic goals is an important
part of the initialization phase. Here, companies need to consider potential usefulness,
importance, and added value of introducing RPA early on in the project, identifying success
factors and obstacles/ interests for their organization and stakeholders (ibid.). Furthermore,
Herm et al. (2020) suggest the initialization phase include screening of different (RPA)
technologies to determine whether RPA can be applied usefully and which technology and/ or

11
vendor is most suitable (not necessarily RPA but also e.g. AI or traditional BPM systems). The
authors also note such screening may be done proactively or exploratory, along with or after
the identification of automation needs (ibid.).

After verifying RPA as a suitable solution for the organization, Herm et al. (2020) suggest the
succeeding implementation phase include process selection focused on the prioritization and
selection of processes to be automated which require information from stakeholders and end-
users for better decision making. The authors explain previous case studies predominantly
consider low-complexity processes for testing and initial implementation, and suggest; level of
standardization, maturity (often synonym with level of existing documentation), execution
frequency and volume of processes should be considered to enable increased efficiency of the
automation (ibid.). The implementation phase also includes RPA software selection in which
the organization needs to consider aspects such as; eventual prior implementations, cost of the
software and skill requirements, in their decision-making (ibid.). Herm et al. (2020) also
emphasize that the increasingly maturing RPA market seems to allow for greater impact of
organizational factors rather than technical factors in software selection, and also that; vendor
reputation and support, data security of cloud solutions, as well as the available skills with
eventual external consultants often affect the decision.

Along with a longer period of evaluation of the business case with the eventual RPA solution
(which is commonly performed from early on in the initiation phase to late in the
implementation phase), the authors also suggest the implementation phase include a more
small-scaled verification of functionality, called proof of concept (PoC) (ibid.). Whilst the latter
PoC is aimed at determining whether the implementation of RPA is financially and technically
reasonable - preferably during several months to provide a detailed data-driven decision - the
business case evaluation should focus on ensuring organizational support based on indicators
such as improved processing times, human error rates and IT costs should be considered (ibid.).
After such activities has been completed to a sufficient degree, and decisions have been made,
Herm et al. (2020) describe RPA rollout takes place, comprising all activities with activating
the robots in the daily operations. The authors suggest RPA rollout strategies may not be RPA-
specific but apply to other software processes - the model lacked empirical data on how this
should be conducted.

12
After the defined business case and successful PoC has resulted in a rollout, and hence that the
implementation project has been completed, the authors suggest a phase of scaling starts
whereby companies can consider expansion of the RPA portfolio (ibid.). The authors suggest
such scaling and adoption is facilitated by templates and RPA libraries of techniques, that the
complexity of processes automated is gradually increased to allow for gradually improved
automation knowledge, that external service providers may be used as a support for increased
complexity of processes, and that affected employees must be involved in early stages of new
automations in order to ensure a positive (ibid.).

Herm et al. (2020) further suggest two additional categories of organizational considerations
for RPA implementation, included outside the model´s three phases of initialization,
implementation and scaling, namely that; a center of excellence (CoE) should be set up, and
that RPA support processes are key to ensure efficient use of robots. The CoE should be set up
to support monitoring and maintenance of software robots, including the definition of roles,
KPI:s, skills etc. (ibid.). The CoE is often anchored in the business side rather than the IT
department and provided the required resources, implementation of a CoE is more feasible for
large corporations, however even smaller companies should ideally have at least one FTE to
manage RPA knowledge and future projects (ibid.). Furthermore, the authors underline that
RPA support processes including top management support with regards to financing and
strategic awareness for RPA, governance guidelines and integration of IT and change
management is crucial for RPA implementation to ensure long-term success (ibid.).

Herm et al. (2020) conclude there is no generally valid procedure however that using a majority
of these steps will likely be a good fit for many implementations. The model, authors say;
“provides a clear methodological contribution on how to approach RPA implementation
projects comprehensively and it is of practical value for companies [...]” based on interviews
with RPA experts (Herm et al., 2020). The study was recently published and our literature
review identified no recognition-, criticism- nor application of the framework. The overall
description of RPA implementation considerations of Herm et al. (2020) also includes the main
points of Santos et al. (2019); strategic goals, process assessment, and tactical evaluation.
Overall, it is the most comprehensive model our literature review could identify - and it hence
seems well suited as a basis for our analysis.

13
2.6.2 Implementation theory for technology fields outside RPA
Technology Acceptance Model
Previous RPA implementation research has used external models such as Technology
acceptance model (TAM) to describe how users of technological tools experience technological
development (Legris et al., 2003). TAM is a model developed to describe an individual's
attitude towards technology. This model is based on “perceived ease of use” where the
experience towards technology is affected by how user-friendly and problem-free the
technology is, where higher ease of use has a positive impact on the individual's attitude to
technology. “Perceived usefulness” on the other hand is affected by how useful the technology
is and to what extent this technology will assist and boost the performance of the users (Davis,
1985). This study will only touch on the concept of TAM and will not go deeper into details as
the focus of this study concerns the implementation and not the experience around it provided
that we have an implementation method-oriented approach.

Learnings from ERP system implementation

Looking at implementation studies from business technology outside RPA, some literature can
be found investigating common issues and takeaways for system implementation. In their
Learning from adopters’ experiences with ERP: problems encountered and success achieved,
Markus et al. (2000) provide a list of common issues with ERP implementation and their
respondents conclude several actions could have been taken to avoid such implementation
issues. Markus et al. (2000) underline that companies experience problems in all phases of
system life cycles with business systems such as ERP, and note that many of the problems
companies may experience in later phases originated earlier however remained uncorrected or
unnoted. A greater, early attention to common issues should hence decrease risks for later
setbacks with system implementation (ibid.). The actions suggested to improve implementation
success comprise;
1. Doing a much better job of end-user training during the project phase.
2. Starting the project phase with plans for long- term maintenance and migration.
3. Documenting the reasons for configuration decisions, not just the parameters,
so that people not involved in the project phase can get up to speed quickly.
4. Not disbanding the project team when the project goes live, but instead staffing
a competence center for managing future evolution and learning.
Markus et al. (2000, p. 263)

14
Whilst these suggestions are based on implementation of comprehensive ERP systems with
different technology and scopes of RPA projects, the actions presented in their learnings
correspond well with the RPA specific recommendations for successful implementation
provided by Herm et al. (2020) and Santos et al. (2019).

2.7 Conclusions and analytical framework

In this theoretical framework we have briefly presented the fundamentals of RPA and its
commonly mentioned benefits. Identifying and describing foundational functionalities and
potential benefits is naturally relevant to gain an understanding of what the technology may be
used for, and potential benefits to realize with RPA implementation - to provide a basis for the
research on how such benefits may be realized through facilitated implementation.
Furthermore, the most common issues and disadvantages with RPA described by literature were
presented, including the commonly cited issue of maintenance efforts needed, as well as the
limited suitable use-cases for RPA automation. The latter, limited use-cases, may provide
criteria and guidelines to consider when evaluating processes to potentially automate. Previous
scholarly contributions describing suitable processes to be automated point to the fact that
banking and financial services may provide suitable use-cases for RPA. This industry
perspective supports the relevance of studying the this specific industry, and may also be used
for analysis to put the case company, it's characteristics and our findings in a broader context.

Based on the literature reviewed, many potential-, and common issues with RPA
implementation were presented. Although previous sources point to statistics displaying high
levels of RPA implementation failure, few sources discuss practical and detailed courses of
action to facilitate successful RPA implementation. Despite lacking research within the field,
especially two recent and (relatively) comprehensive models describing RPA implementation
success factors were identified and highlighted. Furthermore, the framework also included
results from literature searches outside the field of RPA implementation including TAM and
ERP technology implementation learnings. Similarities between the two RPA implementation
models were identified, and similarities could also be seen with implementation guidelines
outside the fields of RPA, providing a basis for analysis and comparison with empirical
findings.

15
Conclusively, whilst existing literature on RPA implementation methodology and best
practices for benefit realization is limited, the identified scholarly contributions and research
gaps provide starting points for a research design aimed at expanding empirical evidence on
best practices and implementation success factors.

2.8 Analytical model

Based on the findings presented in the theoretical framework, and the ambitions to describe
how RPA-implementation may be facilitated, the following analytical model (seen in Figure
3) was created with the intention to categorize important themes on the subject to be used in
our analysis.

Figure 3. The analytical model.

The first analytical component of ‘Potential benefits to realize and reason for RPA usage’ is
used to include fundamental drivers of RPA usage (to enable analysis of what companies may
want to achieve with the implementation and hence what the most important potential issues to
mitigate are). The central part of the analytical model is found under the horizontal divider,
namely ‘common issues and challenges potentially obstructing implementation’ and
‘implementation models and approaches suggested by literature’ focusing the analytical efforts
on identifying problems potentially causing implementation issues, as well as to identify best
practices and approaches to avoid issues and facilitate implementation. Through focusing on-
and applying the implementation model suggested by Herm et al. (2020) we intend to

16
benchmark previous implementation models with the approach used by the case company.
Inspired by Herm et al. (2020), large parts of the data-collection and analysis will be structured
around the models suggested phases of implementation projects, and the suggested actions for
each respective phase.

2.8.1 Simplified analytical model

Simplified further, the logic of the analytical model of this thesis and the above-described
approach can be visualized as seen in Figure 4 below, moving from understanding goals and
intentions, common issues and challenges to overcome to achieve these goals, and finally
looking at potential approaches to avoid/overcome issues and generally facilitating
implementation projects.

Figure 4. Simplified analytical model.

Based on this analytical model, both eventual similarities and discrepancies between theory
and empirical data within each theme should contribute to answering the research question. In
below sections, the methodological considerations made to ensure relevant data-collection
and practical analysis in line with the analytical model will be presented.

17
3. Methodology
The purpose of this chapter is to present the methodological choices made to meet the purpose
of this study and to create conditions to answer the research question, including argumentation
and reflections regarding the suitability and potential consequences of these respective choices.

3.1 Research approach

This study tried to capture the experiences of respondents having implemented RPA in their
organization concerning their perception of common challenges and best practices for
implementing RPA. The intention of this study was to capture experienced best practices and
compare these to previously developed frameworks regarding RPA implementations, with an
ambition to find possible areas of improvement. In order to investigate how, but also why, RPA
is implemented - to understand goals and eventual priorities during the implementation and
contextualize the findings - the researchers, hereon after called ‘we’ in this chapter, also
attempted to include considerations for the intended value to be created by implementation.

The epistemological stance of this study was reflected in different research approaches and
methodological considerations made. Three research approaches are common in research,
including; an deductive approach is centered around producing hypotheses to be confirmed or
rejected by testing already generated theories whilst an inductive approach instead aims to
generate new theory from our empirical experiences (Bryman, 2012). However, based on
Dubois and Gadde´s (2002) description of that an abductive approach can be beneficial if the
study intends to discover possible “areas of improvements” that perhaps can contribute to new
concepts and a development of existing models, we found the abductive approach to be an
appropriate choice for this study as it intended to examine best practices for RPA
implementation. By applying models and frameworks for successful RPA implementation
provided by previous scholars against empirical insights from the case company´s
implementation, the aim was to contribute to the development of existing theories and
frameworks.

18
3.2 Research design

3.2.1 Qualitative case study design

According to Bryman (2012), a qualitative study is suitable for studies which strive to create a
deeper understanding of various patterns in a specific event by capturing the perception of the
participants. Accordingly, to increase the understanding around what parts that may cause RPA
implementations to fail and which factors that may be decisive for the implementation – based
on actual experiences - a qualitative case study was considered to be the most suitable approach.
Supporting this logic, Yin (2009) highlighted that case studies are often considered suitable to
create an understanding of a company’s decision-making process and what impact these
decisions have on a company's approach. An additional factor that separates a case study from
other studies is that it may generate an in-depth understanding of a real event or phenomenon
(Yin, 2009). Accordingly, a case study may contribute to a more specialized illustration of how
to best implement RPA based on the case company´s perceptions and experiences from
practice. Criticism has however been directed at case studies related to their tendency to be
influenced by preconceived notions (Yin, 2009), which the researchers of this study reflected
on and constantly tried to avoid. According to Bryman (2012), researchers have to decide on
what level the empirical data is to be collected on. This research is based on a single case design
based on a convenience sample. This approach was deemed suitable provided the intentions to
capture a detailed picture of the implementation process and it´s different steps, which would
have been difficult to achieve if several companies were to be analyzed during a short period
of time. Moreover, provided the inclusion of expert consultants, the single case study design
would still allow for data collection based on vast experiences from several previous
implementations and cases – applied to and explained in the context of the case studied.

3.2.2 Case company and participants

According to Bryman (2012), theoretical sampling is an ongoing process where data is
collected and analyzed in order to know what data to collect next, and where to find that data.
In accordance with Bryman (2012) description that convenience sampling is based on what is
available to the person(s) conducting the study, the researchers had an established relationship
with the case company. The case, however, also matched the purpose of the study in many
regards. The relevant case company is a Swedish venture capital- and advisory group providing
loans, financing and business development advisory to small and medium sized companies.

19
The group has roughly five hundred employees, which compared to the major players in a
Swedish banking and financial services context is relatively small - however the company's key
business activities revolve around credit which is a key business of the industry. Furthermore,
the case company's communication regarding the RPA project was positive and indicated the
project was ‘successful’, theoretically increasing their opportunities for investigating
facilitating aspects. As mentioned in previous chapters, banking and financial services business
offer suitable opportunities for RPA (Syed et al., 2020), motivating our choice of sector - and
provided that the company's processes were typical for the industry and seemingly considered
their project as successful, we deemed the case to be suitable for the study.

Through conversations with the company, it was revealed that they hired consultants in their
implementation process and a snowball sampling (as described by Bryman, 2012) began where
new respondents emerged over time from various recommendations. This selection process
resulted in contact with three respondents (presented in Table 1) with key roles- and vast insight
into the project and the steps taken during the implementation. Respondents has been given
abbreviated titles in order to;
1) ensure anonymity and to disclose neither the participants’ identities, nor the names of
their respective firm, and to;
2) facilitate the reader's ability to see statements and quotes from the respective respondent
in relation to whether he or she has a consultant or client perspective.
The coded names are based on whether the respondent belongs to the client organization (the
case company implementing RPA), abbreviated as “CL”, or the consultant organization (the
firm assisting the case company with the RPA implementation), abbreviated as “CO”. The
respondents from the same organization are distinguished by numbering (1 or 2).

Table 1 – Sample overview

Participant* Profile Interview Interview
duration setting

CL1 Internal project manager for the RPA project, heavily involved in initiating 65 min Video call
RPA usage. Has started and managed the customer service department
within the case organization.

CL2 Head of digitalization with the case company. Three plus years of experience 63 min Video call
within the case company and broad background within innovation, tech and
startups.

CO1 CEO of the consulting group's subsidiary focused on Automation. Has 15 58 min Video call
years of experience in management with a focus on IT. Extensive experience

20
 with automation and four plus years of experience with RPA. Ultimately
responsible for the delivery.

CO2 Business analyst and project manager from the consulting firm. N/A E-mail
Three plus years of RPA implementation before the project, focused on areas correspondence
such as identification and evaluation of process candidates, team lead of
development teams, testing, deployment, maintenance, CoE and strategy.

*Participants' designated names are based on if the respective participant is part of the client-
(case company) or consultant organization.

3.3 Empirical data collection process

In order for the research question to be answered following the logic of the analytical model,
semi-structured interviews were considered to be the best choice as this technique makes it
possible to create a more in-depth understanding of the company’s implementation. Semi-
structured interviews have been described as appropriate to create a broader understanding of
a specific case (Kvale & Brinkmann, 2015). Accordingly, we found semi-structured interviews
suitable to facilitate empirical data collection from practical experiences and increased
understanding of how respondents experienced an implementation, the different phases and
actions, difficulties and success factors - to answer our research question. Semi-structured
interviews thus became the obvious choice as we were able to create an in-depth understanding
of the case.

Kvale and Brinkmann (2015) describe semi-structured interviews can create an exchange of
knowledge between the interviewee and the respondent in which the questions should be
adapted based on the person being interviewed, to achieve the best possible result. Accordingly,
the focus and follow-up questions was somewhat adjusted between the interviews of CL- and
CO respondents to best capture the case company´s and the ‘expert’ perspectives respectively.
Respondent CO2 could only participate in the study via email correspondence due to time
constraints, and hence - since the project phases had already been described by the other
respondents - the interview questions sent to respondent CO2 excluded theme C (described in
Table 2).

In order to ensure relevant knowledge regarding the implementation process all respondents
included in the study had key positions during the implementation project. Including only the
knowledgeable, key actors from the project - meant a restricted number of participating
respondents. In order to counteract eventual downsides of the smaller sample-size, the data

21
collection also comprised reviewing project material and most importantly the case company´s
own visualization of the project and its phases. As described by Bryman (2012), such a multi-
method and reviews of additional material enables and triangulation and hence increased
reliability. This approach enabled comparisons of the respondents' answers with the additional
material, increasing the reliability through triangulation and ability to double check the answers
provided by the respondents.

To answer the research question, the empirical data collection followed the logic of the
analytical model (presented in Figure 3), in turn based on the purpose, theoretical framework,
previous research and research gaps presented in this study. The intention was to capture RPA
implementation intentions, issues and challenges, success factors and approaches used - in
order to facilitate greater understanding of how RPA implementation may be facilitated. In the
below table (table 2), the main themes of the interview guide and the logic behind their
inclusion is presented. The full interview guide can be found in Appendix 1.

Table 2 – Interview themes

Theme Objective

A Interviewee profile and role To understand the background and experience of the interviewee,
contextualizing the collected data and ensuring relevance of participation

B The project - To capture the overall approach and actions during initialization,
Phases and actions implementation, post-roll-out and support processes (based on Herm et al.´s
2020 categories to provide structure and enable comparison)

C Issues and challenges To capture eventual issues and challenges obstructing RPA implementation
and how these were solved

D Success factors To capture keys to facilitate RPA implementation

3.4 Data analysis

A common approach to qualitative data analysis is thematic analysis (Bryman, 2012), which
allow a deeper understanding of text than words or keyword-in-context (KWIC), via an analysis
that permits researchers to identify thematic subjects with both explicit and implicit expressions
(Guest et al., 2011). To identify themes that were true to our data set, we followed Ryan and
Bernard's (2003) guidelines which emphasize attention to thematic- or linguistic clues
contained in respondents' responses, and as suggested by Guest et al. (2011), a hierarchical
structure with themes and sub-themes was used. Provided that the analytical model and the
interview guide revolve around distinct, separate themes, the thematic analysis naturally tended

22
to follow the same structure. We found these methods for data analysis favorable for
distinguishing and presenting the meaning of our data in a logical order.

All interview data files were temporarily saved, transcribed, and then deleted. The transcription
began after the first interview and were conducted simultaneously as the remaining interviews,
which enabled us to recall the key findings and to identify suggestions of useful questions and
questions for the remaining interviewees. During the transcription we tried to reflect the
statements of the respondents in literal and also non-verbal statements, such as chuckling or
emphasizing certain words, so that we could reflect and analyze their responses fairly and
appropriately.

3.5 Reliability, validity and replicability

According to Bryman (2012), a study's credibility is reflected in three underlying criteria which
are validity, reliability, and replicability, all of which form the basis of the study's credibility
within the field of business economics research. Bryman (2012) describe that a study's
credibility can be measured based on "external reliability", which shows the study's
replicability. With this in mind, we have chosen to ask questions regarding the implementation
of RPA to a company that has recently worked with this in the financial sector. It was assumed
respondents ability to recall important aspects of the implementation would be dependent on
their memory and hence time passed since implementation. We hence assumed a study of a
recently completed project should provide more accurate answers, which affected the choice of
case. We have also worked with the internal reliability of the study, which implies that the
content of the study is perceived in a similar way by several observers (Bryman, 2012). Since
we are three authors, we were able to discuss the interviews afterwards and reflect upon the
answers, increasing the likelihood that the answers have been interpreted correctly.

Furthermore, Bryman (2012) highlights that internal validity concern whether the respondents'
answers reflect upon the theory development generated by the study. As the study sought to
investigate existence of eventual areas of improvement in the suggestions made by previous
studies, the ambition and focus was to further develop existing frameworks by finding and
highlighting eventual differences in the case company's implementation approach and success
factors, compared to theory. Finally, Bryman (2012) also presents the external validity, which
demonstrates the extent to which the outcome of the study is generalizable. As this is a case

23
study, we are aware that the generalizability may decrease as the answers are based on a specific
company and adjust interpretations of our results accordingly. As previously described
however, the case company was chosen based on our initial perceptions of that the case
company and its products were somewhat typical for the industry, in an attempt to increase
generalizability.

3.6 Method discussion

As described in previous sections, the methodological approaches are based on careful
considerations and attempts to answer the research question in the most accurate way possible.
Several weaknesses are however inevitable provided the methodological choices made. As
emphasized by Bryman (2012), data collection in a qualitative study can easily be affected by
subjectivity, and researchers interpretations in the analysis. Furthermore, it is argued that the
outcome of qualitative studies can rarely be generalized (ibid). To reduce the risk of
subjectivity, the interview guide was developed from a neutral position based on our analytical
model (based on previous literature) and especially on previous implementation frameworks.
Furthermore, in accordance with the recommendations related to ethical research described by
Diener and Crandall (1978), we have been careful not to disclose any information that may
reveal the companies nor respondents' identities, carefully informed respondents on the
research purpose and how the data will be used, and made sure to make fair representations of
previous literature and our findings.

Conclusively, the methodological choices made in this study is built on wide and conscious
considerations of pros and cons with different alternative approaches. Irrespective of the
potentially negative aspects mentioned herein, we deem the methodological choices made as
suitable for the purpose of the research provided the contextual conditions.

24
4. Empirical findings
In this chapter, the findings of the qualitative data collection will be presented in clusters of
themes and headings, following the structure of our interview guide and analytical model.

4.1 Structure
The empirical findings in this chapter will be presented in accordance with a thematic analysis
conducted based on the main themes of the analytical model and interview guide (in turn based
on the literature review, research gaps identified, consequential research question and purpose
of this research). First, a brief description of the relevant automation and the case company´s
fundamental intentions with implementing RPA will be presented, followed by a breakdown of
the different actions taken and considerations made during the main phases of the project,
namely initialization, implementation and post-implementation, as well as overarching support
processes. Subsequently, issues and challenges- as well as aspects facilitating the RPA
implementation will be presented. References to respondents will be made through the assigned
abbreviations (previously described in Table 2).

4.2 Fundamental aspects about the case and reasoning behind the RPA
implementation project
In order to understand the project and the characteristics of the automation to facilitate analysis,
an initial objective of the data collection process was to capture; what processes the project
intended to automate and the main intentions for the case company to implement RPA.

The RPA implementation was described as part of long-term strategic digitalization- and
efficiency initiatives, in relation to which the group had considered RPA during several years.
The primary, specific drivers described by respondents CL1 and CL2, was an increased need
for processing capacity and efficiency, evident opportunities for ROI - and high alternative
costs for meeting the increased need for capacity - as well as ambitions to decrease processing
mistakes and to minimize the amount of administrative and “boring” activities performed by
employees. The targeted process to automate was related to loan-application with the ambition
to remove time consuming and manual processing performed by advisors through moving
information from the client’s loan applications into- and between the different systems needed
to document, process and make a decision regarding the application.

25
4.3 Phases and activities of the project
A key objective of the empirical data collection was to capture what activities the case
company performed during the implementation of the RPA solution. In below sections, these
phases and the respective activities will be presented.

4.3.1 Initialization
With regards to the initial phases of considering automation and the early discussions leading
up to the RPA implementation project, it was evident throughout the interviews that
automation-initiatives and analysis of characteristics and functioning of processes had been
considered long before the RPA solution was launched. Following the major strategy work that
was carried out in 2018 where the key word was digitization, the company's digitization
manager, respondent CL2 carried out a survey and interviews regarding the processes and
activities for the two main products, in order to investigate possible processes digitization and
streamline, creating a clear process map of each product. What was seen according to
respondent CL2 was "[...]one product had an unstructured process, while the other was
structured". Respondent CL2 concluded that “…the product-process was time consuming,
inefficient and boring to handle for the employees”. Respondent CL1 explained that "[...]
during this work we talked about RPA and robotics [...] I was given the mandate to spend time
diving into RPA and reading reports and going to seminars and creating an image for myself".
Moreover, respondent CL1 got in touch with different RPA-vendors in the industry to gain
knowledge about how RPA works and how to implement the technology. Realizing they did
not have the knowledge or resources to build their own robot internally, the case company,
according to respondent CL1 and CL2, turned to their own IT supplier for insights.

The business case

A recurring theme regarding the initialization phase of RPA projects was the focus of
establishing a business case. According to Respondent CL2, the pandemic was "[...]what got
us started". Respondent CL1 described they realized that they would get more resources from
the management and owners to meet the expected increased demand for their structured
product. Respondent CL2 recalled that "[...]march showed an incredible demand for the
product. We went up by 300% compared to the same period last year. To be able to handle this
there were two options. Either we automate, or we hire an army of people to handle that."
Respondent CL1 and CL2 presented RPA as a solution to this problem for the top management.

26
The top management approved the initiative to start the RPA project as a solution and according
to respondent CL2 time efficiency and ROI were the deciding factors.

Selecting process and vendor

An evident part of the implementation project was selecting the process to automate and the
vendor to provide the services. Selecting process was rather easy according to the case
company. The goal was to automate processes related to the product which saw a big increase
in demand. For this goal, the case company turns to the consulting company for help with
automation. Based on respondent CO1’s experience, this was "[...]a classic example of where
you choose a process based on subjective assessment", which respondent CO1 suggests can
increase the risk of a failed implementation. Ideally, in this phase, according to respondent
CO1, processes with high business value and low complexity should be chosen. The respondent
described this evaluation can be visualized in a matrix, which has been attached in Appendix
2. They also need to meet the criterias of being "[...]high volume, repetitive and rule-based with
few exceptions”. According to respondent CL2 which functions and units would be involved
in the large interview work that took place whose purpose was to map every click end users
made step by step of the particular aimed process. According to respondent CO1, an important
part of this phase is to work iteratively and having weekly meetings for identifying faults
increases at the same time as questions regarding the robot or project can be answered.

According to respondent CL2, UiPath was chosen because they could deliver a cloud-based
solution in Sweden, and respondent CO1 believes that the choice of UiPath was appropriate as
the organization works with big data. Respondents concluded great similarities between
vendors and that the consulting firm's relation with the provider was important. Respondent
CO1 adds that "It can be devastating if you choose the wrong platform", and explains the choice
of platform should be based on organization´s data-processes, systems and applications.

4.3.2. Implementation
As seen in above sections on the initiation phase, the respondents emphasize much of the work
related to RPA is conducted as preparations before the more technological aspects of the
implementation. A quote from respondent CO1´s illustrates this attitude, saying; “I do not think
it [the phase] is that interesting, honestly”. Instead, the respondent said, it is more of a sanity
requirement to have a sound way of working and a systematic approach. The steps needed for

27
implementation, however, are described by respondents throughout the interview and included
in the material and models received by the case company; Acceptance Tests (AT), Request For
Change, and Deployment, as described in Figure 5, were categories of activities included in
the implementation phase.

Developing and testing the robot

The consulting company was responsible for the testing of the robot according to respondents
CL1 and CL2, whilst the general project team interacted with workers, to validate the
functionality (following the recommended approach from the consulting firm). Respondent
CO1 believes that this is not a difficult phase in the project as long as it is done in a systematic
way, describing that; "From a development perspective, it is not so revolutionary. You work on
validating requirements, you develop a proposal for a solution, you validate the solution, you
build on the solution and acceptance test it before it is put into production ".

Robot roll-out
Once everything had fallen into place, the last phase of the implementation began, to deploy
the robot. Before making this happen, a request for change was applied to a specific group
within the organizations which decides for every new IT-implementation. In this case, RPA
was approved and thus the robot was deployed. An important aspect in this phase is according
to respondent CO1 "[...] involve the business" and work iteratively. In this regard, respondent
CO1 considers that the case company did not succeed and explains that "[...] once the case
company had built the robot and would put it into production, they were not clear enough in
the communication with the business". Respondent CO1 continues "It had not been ensured
that the end users received the information and gained sufficient understanding about the robot.
It may be small things, but it creates a friction that can be negative in the long run if you want
to scale up the automation".

4.3.3 Post-implementation
Immediately after the roll-out was made, its functionality was monitored by the consulting
company around the clock for two weeks according to respondent CL2, remediating bugs and
deviations. It was decided that the consulting company would continue to assist with
maintenance, delivery of the platform and support, but without round-the-clock surveillance.
Respondent CO1 explains that the most common errors that occurred and still are occurring are

28
not due to the robot, but due to "[...] people who pushed or changed where they should not,
which can ruin the robot." Respondent CL1 emphasizes that it has not been an unproblematic
post-implementation. Changes in the underlying system have been made since the deployment
without informing the consulting company affecting the functionality of the robot. To
counteract these problems, respondent CL1 emphasizes that organizations should view the
robot as a living being that needs to support and be evaluated. Respondent CO1 has a similar
view “"A robot needs a supervisor who evaluates its work on an ongoing basis". Respondent
CL1 also emphasized that stewardship and maintenance of the knowledge and further
development of RPA (as mentioned in the maintenance and management in Figure 5) is
important, however that the case company may not have fully succeeded with these activities
due to other priorities.

4.3.4 Support-processes
In the following sections, the findings related to support functions such as center of excellence
and management support are presented.

Center of excellence and internal RPA knowledge

When talking about support processes and when asked about allocating resources or
establishing a new part of the organization to work with RPA in and manage knowledge,
respondent CL1 answered that; "my straight answer is no” [...] the intention has been that it
should be someone or some people who work with it within the organization but have not got
there”. Respondent CL2 also points out the importance of their own competence about the
robot within the organization "I think in the long run that it is best if the case company builds
up its own department for RPA''. Respondent CO1 agrees in this matter and emphasizes the
importance of having abilities within the organization, post-implementation. This can be
managed by either hiring or establishing knowledge within the organization. Both the case
company and consulting firm feels that the management has not prioritized or allocated
resources for this post-implementation.

Management support
During the phase of identifying technologies and ways of digitizing within the case company,
both respondents CL1 and CL2 were given a mandate by the management to set aside time for
investigating RPA. When asked directly if they experienced that they received management

29
support, respondent CL2 answered the following "We received very strong support from above,
our CEO bought into RPA directly and saw that this was the future". Throughout, the
management support seems to have been high during the implementation phase. However, the
respondents from the case company suggest that the support was not forthcoming when asked
for resources to scale up RPA within the organization.

4.3.5 Model and summary of the implementation project phases

As described in the project documentation material received by the case company and
respondent CL1, the different specific phases of the project were described and visualized in a
process the consulting company tends to use for the same purpose. The documentation received
in our data collection concerns especially the part of the process after involving the consulting
firm and not the extensive preparations and initial phases described in previous sections of this
chapter. In summary, this breakdown of the project´s main parts were described as consisting
of first; establishing contact with the relevant department of the consulting firm, identifying
roles internally and creating a project organization before developing a working-process for
RPA. The latter project phases were visualized through a number of fundamental sections in a
simplified flow-chart. An adapted version of this flow-chart can be found in Figure 5. The
activities of each respective phase are described in Appendix 3.

Figure 5. Summary of the implementation project as visualized by the case company

(adapted from material received from the case company).

Whilst this process description is brief and simply provides an overview of the main phases,
respondent CL1 was ascertained transparency and visualization of the process and ‘the steps to
be taken’ was crucial in order to make sure the organization was onboard with the process.

4.4 Issues and challenges

Throughout the data collection and interviews, a large number of potential issues and
challenges with RPA implementation have been identified. In general, we have seen that the
30
respondents' emphasized the human and organizational issues to a larger degree than the
technical aspects. This relationship and general predominance of human and organizational
challenges is exemplified with a quote from respondent CL1, emphasizing that; “[...] what I
thought was going to be problematic was not problematic… and that was the technological
aspects. [...] The challenge is the human behind [the technology] and the organization”. In
accordance, most of respondent CO2´s issues (and solutions) revolved around human aspects.
In below sections, the most prominent issues and challenges as described by the respondents
will be presented.

4.4.1 Human and organizational issues

During the conversation with the respondents of the study, it emerged that change management
was one of the problematic factors that was shown to influence the implementation of RPA.
Respondent CL2 believes that there was opposition from certain parts of the organization
regarding robotization, where the robot was associated with a cutback in the workforce. The
implementation initially gave the impression of "here comes a robot" which made employees
think about whether their job was at risk of being automated, which would have an impact on
their employment. With regards to the previous emphasis to that respondent CL1 did not
consider the technological aspects as problematic, but instead that the human- and
organizational aspects were the most challenging - the same respondent provided a breakdown
of how these problems were evident in respective parts of the project. Before the project was
started, respondent CL1 explained, it was challenging to make people feel comfortable working
with a robot. During the process, when the robot was configured, it was challenging to get the
organization to stick to implementing the robot without evaluating why things are happening
the way they are at the moment. It was not the robot’s task to rationalize elements, it was to
map elements as they already did in the organization. CL1 says that this has been a challenge
in their PDD but also for their acceptance testers to stick to the focus issues where they should
not develop the process but rather take it as it is. Developing the process is another workshop
and not the main purpose of their implementation. However, according to CL1, the consultants
were pretty good at putting an end to this when they started to slip away from the main subject.
Moreover, as will be described in relation to the below issue of choosing the right process to
automate, respondents CL1 and CL2 also described that it was challenging to ensure that
employees recalled and mapped the processes correctly, including all necessary steps.

31
4.4.2 Issues with choosing the right process to automate
A common theme of potential issues in the RPA implementation process, recurring throughout
our interviews, was the potential issues with choosing the right process to automate.
Respondent CL1 believes that if they would have an organization that realized the benefits with
RPA and observed the needs for more robots, they would probably have ended up in a situation
where several processes could have been automated. Then the organization could have
processed which robots to focus on first. This is something they have not come to use, but if it
had been possible to present about 10, 15 or 20 processes to choose from and to get a good
decision support for, a scale up of the RPA implementation could more easily be carried out.
This was something they did not find available at the start of the implementation process.
Furthermore, respondent CL2 believes that the first time you perform an RPA, you take very
large scoops. It is a lengthy process that must be automated that goes over a number of units,
departments and systems. Respondent CL2 believes that this process is problematic and even
though they implemented the robot, not all parts were automated, and a few parts had to
continue as they did before. As described by respondent CL2, one of the biggest challenges
when choosing a process was to choose a process that is achievable to automate and not to take
too big "steps" directly "and divide the implementation into smaller sub-processes.

According to respondent CO1, this is a classic mistake as it often leads to choosing a process
of subjective judgments or political decisions. The challenge when rolling out is therefore
whether you have captured the requirements regarding the process that is to be automated, i.e.,
to have the people in the business to truly understand the importance of having to set
requirements at a detailed level or explain at a detailed level how the process works and how
this can be automated. Furthermore, respondent CO1 highlights that the harder part of change
when automating processes is when people start to become redundant. Then it is important to
get help from those with expertise in the area. Change management is therefore important,
which is about managing expectations among people within the organization through
information and education. It is also important to emphasize that automating without
maintenance is not optimal as processes no longer work after changes. Therefore, respondent
CO1 believes that the choice of process to be automated should also be based on the process
remaining as it is and not a process that undergoes constant development. When a development
in a process takes place, it will affect the robot where it would have to be reprogrammed at each
such event.

32
4.4.3 Issues related to changes in systems
The robot is programmed to copy a work routine and perform a task in a specific way.
Therefore, changes in the underlying system will have a direct impact on the robot's
performance. Respondent CL2 states that if there was an update where a specific bottom
changed color or was moved to another location, the robot would no longer understand how to
move forward as it was programmed to follow a specific pattern. Getting the organization and
colleagues in key positions to think about the robot was therefore challenging in this case as a
simpler change could have a major impact on the robot's work process. Respondent CL1 also
shares this view and says that no one wants to oppose the automatization as it would be
considered a direct misconduct towards the organization. On the other hand, it can be
considered difficult in situations where everyone works very hard in a pandemic, to remember
to think about the possible outcome in the event of a change. It is the human factor where the
organization must think about how a change in some way can affect the robot's working
process. This applies not only in the first stage but also in the 2nd and 3rd stage where the robot
works. It was thus clear from the respondents who worked with the implementation of RPA
that changes made by people who did not understand how this will affect the robot were
considered problematic, as each change entails the need for reprogramming so the robot can
relate to the new way of working.

4.4.4 Issues for broader adoption and scale-up

When it comes to scale-up of the RPA implementation, it is something that has currently been
deprioritized. CL1 believes that if it was up to the IT-department to decide, they would have
had significantly more robots than they currently have. The problem with scale-up is that the
organization does not see the benefit of further automating processes in today’s situation. The
decision model in the company is based on the fact that the organization needs to demand a
further development of robotization, something that may be due to the lack of knowledge and
insight into the usefulness of the robot and what it can achieve. This is something that could
possibly be due to fear and uncertainty, which has resulted in IT not being able to continue
working with their ideas, since no formal request has been made regarding further development.
Furthermore, respondent CL1 claims that this may be due to the increased workload on
managers that has arisen in connection with the pandemic, which may be the main reason why
the scale-up process has suffered. Furthermore, respondent CL2 highlights that if the
organization decided to implement additional robots, they can reveal that such a process is

33
already available and put on “standby”. They already know how the teams should be composed.
Since the first robot has been implemented with the help of respondent CO1, a more realistic
schedule and cost calculation can also be presented, something that respondent CL2 sees as
something positive where they are ready for more.

Furthermore, respondent CO1 explains that the problem of scale-up can be explained by the
lack of understanding regarding the automation process where the end users did not gain a
sufficient understanding during the implementation of the robot. All of a sudden, there was a
robot doing parts of their jobs. According to respondent CO1, this can create friction within the
organization and have a negative impact on the long run and harm the scale-up process.

4.5 Specific aspects the respondents would have done differently

Whilst the respondents’ descriptions of the issues and challenges mentioned in this chapter
included indications of lessons learnt and hence potential actions to counteract negative effects
of these challenges, the interviewees also mentioned a small number of specific lessons learnt
and aspects they would have handled differently throughout the RPA implementation process.
Three of these will be highlighted below as especially evident.

A recurring and specific insight was that the respondents (CL1, CL2 and CO1) emphasized that
a small supporting organization or resources dedicated to RPA and maintaining, developing
and scaling the usage within the organization would have been beneficial. Furthermore,
respondent CL2 specifically advocated that, a posteriori, “having all the facts in hand, one
should probably take a larger amount of small processes which you automate with human
control-points in between them. That would have facilitated the work and I think it would have
been quicker”. By automating smaller parts of processes at a time with short sprints and many
smaller ‘victories’ along the way, a more positive attitude could have been created, meaning
that selling RPA implementation to the organization would have been easier. This, the
respondent said, is something they take with them for forthcoming implementations.
Furthermore, the theme of skills and education of employees was a recurring theme the
respondents emphasized as important. Accordingly, respondent CL1 highlighted that he would
have focused increased RPA education in the organization and with the board of directors
especially, if the process were to be repeated. If I were to re-do the journey [of the RPA
implementation] again, he said; “[...] We could perhaps have started with a trainee-program

34
for group management within the field of RPA. Then, this program [and the knowledge] would
have seeped down the decision-lines in order to make sure everyone is aboard the train and to
say that ‘this is not dangerous’. That is something we should have wanted to do, if we were to
do it again. Some sort of educational module”. The same respondent also expressed that he
underestimated the overall RPA skill-level needed to implement such a solution, and that they
had to revise their ambitions of managing the implementation themselves, serving as a learning
on that it is relevant to seek professional assistance.

4.6 Success factors and facilitating actions

A key objective of the data collection and interviews was to capture what aspects the
respondents considered as important to facilitate the RPA implementation. Whilst the many
issues and challenges presented in earlier sections of the empirical findings represent clear and
many times straight forward aspects to carefully consider in order to enable implementation -
the respondents provided many opinions on what is important to succeed. The most prominent
of these actions and considerations mentioned will be presented below.

4.6.1 Preconditions and contextual factors

To begin with, in addition to actions actively taken to facilitate RPA implementation, we found
some aspects of the case company's situation as potentially affecting their abilities to implement
a RPA solution, based on the respondents' descriptions. A central aspect of this case company's
situation is that they had outsourced their IT to the company presented as ‘the consulting
company’ in this report. As respondent CL1 described it, the consulting company already had
an established relationship with the case company, already “[...] managing our digital
infrastructure with servers and knew how we worked etcetera. That meant we did not have to
start all over again [with a new relationship and setup]”. Similarly, respondent CL2 described
that IT-infrastructure was not an issue since the consulting firm already managed their
infrastructure and that a infrastructure-project had recently been conducted. Should the
consulting company have had another existing IT provider and hired the consulting company,
the respondent (CL2) said, “[...] it would likely have been a lot messier. Now [with the existing
relationship and insights into the IT environment] they [the consulting firm] could talk- and
solve a lot internally”.

35
Moreover, the fact that the case company had already conducted work to understand and map
their processes - was described as a facilitating factor by especially respondent CL, naturally
facilitating- and shortening the process selection and mapping. In addition, the respondents
emphasized that the circumstance of the Covid-19 pandemic and the drastic effects it had on
the demand for their loans, created a clear business case and facilitated internal marketing of
the RPA solution.

4.6.2 Aspects to facilitate implementation

Selecting the right process
A recurring theme throughout the data collection process was the question of selecting the right
process and the work that should go into such decisions. In a conclusive manner, respondent
CO1 summarized what he perceived as important aspects for a successful RPA implementation,
in three key parts, two of which relate to the process to be automated, namely; (1) selecting the
correct process, (2) making sure you have really understood the process and why you want to
automate it. The respondent also explained that companies should avoid the common mistake
of choosing a process based on subjective judgement or politics. Regarding (1), both
respondents CO1 and CO2 recommended targeting the least complex process for automation -
for the reason that these are easiest to robotize while it can be used as a first example when
promoting RPA within an organization. As described in previous sections of the empirical
findings, respondent CL1 pointed to that they developed a retrospective list of evaluation
criteria for future decisions between processes to automate (attached in Appendix 2). A proper
foundation for such a decision, he said, is something he recommends companies focus on in
case several processes are potential targets. According to the descriptions of respondent CL2,
it is furthermore easy to choose a too broad process scope and instead indicate it is better to
automate small parts of the process at a time. When the process is selected, respondent CO2
emphasizes the importance of “[...] involve the business and the process experts throughout the
development”. This leads us to the importance of change management in RPA implementation.

Change management, communication and education

The importance of change management and communicating with employees with regards to
the RPA project, it´s impact on (and benefits for) the organization, was one of the most
prominent points made by the respondents throughout the interviews and collected material. As
mentioned under previous sections on issues and challenges for RPA implementation, the

36
respondents often highlighted that the organizational and human aspects are perceived as more
problematic than the technological parts. Accordingly, the descriptions of keys to facilitate
RPA implementation and ‘success factors’ were to a large degree focused on how to manage
people and the organization. In respondent CO1´s summary of aspects to manage in order to
succeed with RPA an implementation project, the third and final aspect was described as
“[...]making sure the end user understands how it affects them when the RPA is in production”.
Communication with the end-user to make sure how their work is affected and may have to be
adopted, he described, is crucial. On the same theme, respondent CL2 described that an
important part of their approach was to perform seminars and weekly demos of the RPA within
the organization (for those interested) in order to spread the word about the implementation and
make sure employees feel involved and prepared for the robot.

Similarly, the word ‘education’ of advisors (employees) was recurring, and CL2 described
education as crucial for those whose chores were to be automated in order to make sure they
are involved. According to CL2, their approach was to involve a smaller number of managers
and ‘champions’ first in order to facilitate buy-in and let them spread the word about the project
and education to the rest of the teams. When showing the ‘advisors’ the boring things they no
longer had to perform, he said, the RPA solution was easily accepted. Furthermore, related to
the importance of change management, are also the questions of expectations according to
respondent CO1 who described a common issue to avoid is that many ‘people’ have too high
expectations, thinking “[...]it's just a click and drag and then you have 3000% ROI”. In regard
to how to succeed with RPA implementation, respondent CO1 also summarized these points
and the general need for change management and communication by saying change
management in RPA implementation projects is important and should revolve around
“[...]managing expectations among people within the organization through information and
education”.

RPA knowledge
Another evident theme throughout the data collection process was RPA knowledge.
Respondent CL1 highlighted that the RPA initiatives must be derived from and driven by ‘the
business side’ rather than from the IT department and if RPA knowledge is limited, decision
making regarding RPA will be harder. The skill requirements for RPA implementation projects
is, according to respondent CO1, underestimated provided that RPA is “low code” and hence
easier to get started with, however that the processing of fully developing and using an RPA
37
solution is no less complex than any other software solution. Hence, he emphasized, it is also
important to have experienced developers involved.

Structure and approach is important

References to that structure and RPA implementation approaches are important, and were
recurring during the interviews. According to respondent CL2, having intermediate goals
during the project would have been a better approach to have. That way, it would have gone
faster, quicker and more qualitative, while not creating too high expectations within the
organization that can not be met. Moreover, respondent CO1 emphasizes having a systematic
way of working with the implementation phase by structured approaches to testing, validating
and launching the RPA solution represent no icing on the cake but is rather simply required.
The model for this approach streamlined the ongoing work by documenting in a structured way
all the steps the robot would do. When approval was needed from managers to proceed in
different phases of the project, this documentation facilitated the decision-making of managers
according to respondent CL1.

The importance of a clear business case

Throughout, the respondents emphasized the importance of knowing why to automate before
one automates. Without a business case, they agree that the risk of failing with RPA
implementation increases. According to respondent CO1, the importance of having a business
case before choosing to automate a process is one of the most important success factors.
Respondent CL1 agrees and believes in particular that time and money are aspects that must be
evaluated before choosing to invest in a new solution.

4.7 Final remarks regarding the empirical findings

In this chapter we have presented the results of a thematic analysis based on the analytical
model to investigate issues, challenges and facilitating aspects of RPA implementation. The
themes are evidently interconnected and many of the respondents' answers and the data
collected from project material reconnect to intentions, challenges and potential facilitating
aspects in the same points made. In the subsequent analysis, these findings will be condensed
and presented in accordance with the logic of the analytical model.

38
5. Analysis
In this chapter, we will analyze the empirical findings with the ambition to gain a deeper
understanding of the respondents' perceptions of; issues and challenges of RPA
implementation, how implementation projects should be managed, and how these empirical
findings relate to theory.

5.1 Intentions and goals

The logic of the analytical model of this thesis was to capture the case company's goals and
intentions with the RPA implementation and hence facilitate understanding of priorities for
issue-mitigation during implementation. Many aspects of the company's reasoning are
interesting in relation to theory. The main driver of the implementation was the need for
increased efficiency and processing capacity in order to meet higher demand, faster and with
greater cost efficiency than what could have been achieved through hiring high numbers of
employees. Based on the common scholarly descriptions of that robots allow for increased
efficiency and FTE reduction (Lacity and Willcocks, 2015; Santos et al., 2019), may reduce
entry costs (Anagnoste, 2017) and may be developed much quicker than IT solutions relying
on APIs (Asatiani and Penttinen, 2016) allowing for faster ROI (Lacity and Willcocks, 2017;
Santos et al., 2019); the case company´s main drivers are much in line with the benefits and
drivers mentioned in theory. Furthermore, the secondary goals of RPA implementation
identified in the empirical data also match scholarly descriptions of RPA benefits, including;
allowing for more value-adding activities (Syed, 2020) and increased job satisfaction (Santos
et al., 2019). Security and compliance aspects (as described by Syed, 2019; Santos, 2019) was
not emphasized as equally important. Moreover, the respondents descriptions of why the
processes was automated, is much in line with both descriptions of the suitable use-cases for
RPA, including the processes high volume of information, interaction with several systems and
that they are standardized in terms of knowledge and experiences (as described by Asatiani &
Penttinen, 2016; Fung, 2014; Lacity & Willcocks, 2015, 2017) and also fit Vishnu et al.´s
(2017) descriptions of typical banking and financial services companies processes.

Altogether, the case company's goals with- and drivers for RPA implementation are; (1)
pronounced and thoughtful, (2) in line with theoretical descriptions of RPA benefits (with some
exceptions), (3) typical for the industry and match theoretical descriptions of suitable use-cases

39
(indicating generalizability related to similar companies and processes), and; (4) manifold,
hence requiring diverse considerations for benefit realization in implementation.

No specific and pronounced effects could be isolated and derived from these individual
intentions, although some consequences were found to be likely. The case company´s
characteristics, clear business case- and urgent need for the implementation, evident ROI and
hence consequential; tolerance of costs, usage of consultants and generally systematic approach
for implementation - may have created a suitable context for RPA implementation. These
contextual aspects are relevant for the analysis of success factors and eventual mitigation of
issues and challenges, in accordance with theory highlighted in below sections.

5.2 Issues and challenges

A key objective of this research was to investigate issues and challenges of RPA
implementation. When comparing the empirical findings with theory, it was evident that several
of the common issues and challenges mentioned in the literature were not relevant in the case,
and directly dismissed by respondents, including; a lack of understanding of what RPA means
and that its definition (Santos et al., 2019), IT infrastructure (Stolpe et al., 2017), nor
uncertainties whether IT or business was leading responsible for the implementation (Santos et
al., 2019). As mentioned in previous sections, preconditions and previous actions may have
mitigated issues, in this case related to; hiring the existing IT supplier who were in control of
the IT infrastructure as the consulting firm and clearly defining roles. Furthermore, costs related
to RPA implementation, suggested to be a common issue by Stolpe et al. (2017) was not
described as an issue even though the costs became higher than expected. This latter mitigation
of cost-issues could likely be explained by high expectations of ROI, strong financial resources
and a clear business case. Furthermore, related to mitigation of common issues, the challenge
of choosing a suitable process to automate (as described by Santos et al., 2019) was emphasized
by respondents, although their previous work of mapping and understanding their processes,
and evident need to automate the specific part of their business - facilitated the case company's
decision making.

Whilst the scholars included in our literature review has emphasized that a series of more or
less technical issues are common during RPA implementation (e.g. Stolpe et al., 2017; Santos
et al., 2019), almost all issues and challenges emphasized by respondents relate to the challenge

40
of adequate; (1) change management, (2) spreading of information, and (3) education- and
engagement of management and employees around RPA and the solution. One example with
additional theoretical connections, was making sure employees were fully informed and
involved in the changes before and during roll-out, and this was described to have led to friction
when handled somewhat deficiently by the case company. As described in theory (by e.g.
Santos et al., 2019; Asatiani & Penttinen, 2016), employees may be afraid to lose their job when
tasks are automated, and the same challenge was mentioned by the respondents who
emphasized the importance of vast change management efforts to counteract such feelings of
uncertainties. Similar attitudes and fears of the effects of technology are also described in the
Technology Acceptance Model (Legris, 2004) which, we conclude, may be a useful theory to
describe these attitudes the respondents said may occur during RPA implementation projects.
Moreover, the challenges of achieving adequate communication and change management, as
described by the respondents, led to undesirable actions amongst employees which altered the
conditions in the system. Such issues with changing IT environment/ information for the robot
to work with, has also been highlighted by Santos et al. (2019).

Although considered as successful, the case company´s RPA implementation was not free from
issues and challenges, which is consistent with the prevailing literature which has concluded
that issues and challenges are common (e.g. Gex & Minor, 2019; Rutaganda, 2017; Santos et
al., 2019; Herm et al., 2020; Gotthardt at al., 2020). Especially the empirical findings related to
that respondents described; (1) post-implementation issues and scaling of RPA was some of
the most challenging parts of the project, and (2) that technical aspects are considered as less
problematic than human/ change management questions - was surprising in relation to the
common issues emphasized in literature. Along with other current priorities during Covid-19,
the scale-up- and post-implementation issues were described as mostly related to lacking
managerial RPA knowledge and hence ambitions to further automate and utilize existing RPA
know-how, and generally keeping the organization attentive towards the robot regarding
decision making and system changes.

Based on the case-specific pre-conditions, including; long preparatory work on process

mapping and RPA functionalities, and an existing IT supplier who had expertise in RPA
implementation, it is reasonable to speculate that many of the common technical issues
regarding RPA implementation may have been mitigated. In accordance with the descriptions
of Lacity and Willcock's (2018) on that issues and challenges are often related to management
41
mistakes than with the RPA tools themselves - we conclude many of the main issues and
challenges described by respondents are related to human aspects, management and
communication. Accordingly, since these issues and challenges were described as related to
lacking or insufficient actions - these prevailing issues were also described as seemingly
possible to counteract. The most common examples of such suggested actions are described in
the section below.

5.3 Implementation approach and success factors

Whereas comprehension of issues and challenges are necessary to identify opportunities for
improvement, the main objective was to investigate eventual suggested solutions and aspects
facilitating RPA implementation, and how these insights relate to previous scholarly models.
The content of these findings has been divided into two main topics and sections below.

5.3.1 Success factors and aspects facilitating implementation

A fundamental objective of this study has been to identify aspects facilitating RPA
implementation and solutions to common issues. As described in previous sections, we found
that the contextual preconditions, as described by respondents, likely facilitated the RPA
implementation with regards to; having clearly defined IT roles (outsourced), established
contact with a consulting firm knowledgeable within RPA, recently had reviewed their IT
infrastructure and mapped their processes, and that an urgent need for increased capacity
created an evident business case for RPA. Whilst the IT structure and knowledge likely
mitigated potential technical issues, as concluded in previous sections, the careful mapping of
processes is in line with Santos et al.´s (2019) emphasis on the importance of understanding
and choosing the right processes for automation in order to facilitate implementation.
Moreover, the precondition of having an urgent need for RPA and clear business cases is in
line with Herm et al.´s (2020) urge for creation of- and recurring reviews of a business case for
RPA, to facilitate buy-in and implementation. These pronounced preconditions are hence to be
seen as facilitating aspects according to both respondents and theory.

In accordance with the Microsoft-produced Digital Directions´- report´s emphasis on the level
of RPA knowledge is often underestimated, the empirical data collection showed the
respondents considered RPA knowledge as an important facilitating aspect and that the
complexity of RPA implementation should not be underestimated (EY, 2020). Although this

42
insight may seem obvious, it can also be connected to the respondent’s emphasis on the
importance of educating management and employees within the organization - as well as
communicating both aspects related to; RPA knowledge to increase learning, benefits and
logics behind the implementation for change management purposes, and generally avoiding
misinformed employees potentially disturbing the RPA work. These issues should be avoided
for several reasons as described in the previous section. In general, and as also mentioned in
the above section on issues and challenges, many of the important activities to facilitate RPA
implementation mentioned in the empirical findings - revolved around the human aspects,
leadership and change management. Whilst change management is described in Herm et al.
(2020), and the importance of leadership was mentioned by (Rutaganda et al., 2017), the
attention to both RPA education, communication and change management was not highlighted
as crucial success factors by the literature reviewed in this research. These empirical findings
indicate a potential need for additional attention in RPA implementation theory, or at least
further research to validate or discard such need.

Moreover, support processes were also recurring in the empirical findings regarding facilitating
aspects. Whilst the case company had not established an internal RPA-team to conserve
knowledge and expedite future projects - most respondents claimed such internal constellations
would be highly desirable to facilitate implementations. This is in line with Herm et al.´s (2020)
suggested usage of a CoE and emphasis to support processes. Moreover, many of the
facilitating aspects identified seemingly match Markus et al.´s (2000) “actions suggested to
improve implementation success” for ERP systems, including; “doing a much better job of end-
user training”, having “[...] plans for long- term maintenance and migration”, and; “not
disbanding the project team when the project goes live, but instead staffing a competence center
for managing future evolution and learning” (Markus et al., 2000). Although no efforts were
made to further investigate guidelines for ERP implementation and eventual coherence to our
findings, these similarities may indicate that connections between different areas of IT
implementation- and RPA implementation literature may be useful.

Conclusively, as mentioned in previous sections - the empirical and theoretical findings on

issues, challenges are in many ways interconnected with the facilitating aspects mentioned
herein. As will be described below, the connections between the identified issues, challenges
and facilitating aspects, can very much be reconnected to the overall approach and structure of

43
how an implementation model may be constructed to provide guidelines on how to facilitate
RPA implementation.

5.3.2 RPA implementation project approach

In an attempt to investigate how RPA implementation should be managed according to theory
and empirical experiences, the structure of the data collection and analysis of this research has
revolved around scholarly suggested phases and models for RPA implementation. A
fundamental conclusion is that the perceived, empirical success factors (facilitating aspects) are
interconnected with the structure and activities under the different phases of the implementation
project. Both literature (e.g. Herm et al., 2020; Santos et al., 2019) and respondents highlighted
that a structured approach for RPA implementation outlining phases and activities is valuable
for implementation projects. This insight is also in line with the scholarly suggested need for
additional research within the field of implementation approaches and models.

Furthermore, a key finding of this research is that the phases and key activities described by
implementation literature, and our main theoretical implementation framework by Herm et al.
(2020), very much correspond with the activities mentioned by respondents both in terms of
actions, considerations and respective order of each type of action.

Additional takeaways also exist in relation to the respondents' descriptions of the extent,
relative length and importance of what Herm et al. (2020) call the initialization phase - which
was seemingly much greater than Herm et al. (2020) indicate, compared to the other activities
included in the model. We also conclude this focus and emphasis corresponds with the success-
factors identified (described in previous section) related to; carefully understanding-, choosing-
and mapping processes to automate; creating and evaluating a business case, and; focusing on
communication and creating knowledge around the benefits and requirements of RPA from
early on. The importance of the initialization and preparatory work with implementation
projects should hence be recognized.

Compared to what is indicated in Herm et al.´s (2020) visualization at first glance, the empirical
data collected suggest ‘process selection’ is a comprehensive activity including discussing-,
understanding-, mapping-, and finally deciding on a suitable process. Whilst the same activities
are reflected in Herm et al.´s (2020) description, we suggest more emphasis is made to the

44
importance of these activities, and that the visualization of ‘process selection’ is increased in
Herm et al.´s (2020) model. Furthermore, in contrast to the respondents suggestions on that
(what Herm et al., 2020 would describe as) ‘software selection’ is not to be seen as difficult,
and that one respondent suggested this selection can take place later on in the project - we
actually found the case company used the ‘software selection’ process as part of their, what
Herm et al. (2020) would call, ‘screening’ of RPA in the first place, talking to vendors to
understand the technology and what it may contribute with. Consequently, no unambiguous
suggestion on how existing models could be adopted with regards to software selection can be
derived from our research.

Furthermore, based on the empirical data collected, we found that the Herm et al.´s (2020)
‘proof of concept’- and ‘roll-out’ phases (described as relatively un-problematic by the
respondents) was performed in an efficient and structured way based on previous knowledge.
Especially the ‘request for change’-process included a wide variety of smaller controls,
managed in a structured manner based on routines and existing governance. Whilst Herm et
al.´s (2020) model also visualizes the ‘roll-out’ phase as concise, our interpretation is that the
outsourced IT and consultant assistance may have facilitated roll-out with the help of know-
how and consultancy processes.

An important conclusion is also that the post-roll-out and scaling activities, described as the
last phase of Herm et al.´s (2020) implementation model, received relatively much attention in
the empirical data collected - compared to what the model indicated. The case company´s own
visualization of the project included a management and maintenance phase, and respondents
described that; continuous monitoring and reconfigurations of the robot, management of RPA
knowledge and initiatives within the organization (closely related to support processes) and
especially scaling the usage of RPA in the organization, as also mentioned by Herm et al.
(2020), was one of the most challenging aspects.

Provided the focus on phases and timeline-oriented structure of Herm et al.´s (2020)
implementation model, it is not difficult to see why ‘support-processes’ received restricted
amounts of attention. What is interesting, however, is the vast attention and focus the
respondents made to questions regarding humans and change management. Although briefly
mentioned by scholars including Leslie Willcocks (cited in McKinsey & Company, 2016) and
also included but not explained nor commented in Herm et al.´s (2020) model, the respondents'
45
great emphasis on the importance of human questions and change management (based on the
issues and challenges and success-factors described in this report) is not equivalently
represented in RPA implementation literature and guidelines. Establishing a COE, however, as
mentioned in Herm et al.´s (2020) model, was also suggested by respondents, especially to
facilitate future projects and scale-up. As suggested by one of the consultants and as can be
concluded based on the recurring suggestions on ensuring- and percerving RPA knowledge -
the mere actions of collecting and using RPA knowledge may be the central actions needed,
rather than creating an CoE.

Although the activities suggested by Herm et al.´s (2020) implementation framework was
similarly described by respondents, and that the chronological order of these activities to a large
extent matched the order of the model - an important finding was that the project included
iterations and simultaneous activities, moving back and forth between activities. Whilst Herm
et al. (2020) primarily pronounce the importance of iterations, in describing a ‘continous cycle’
is needed for CoE and support processes, the empirical findings suggest that especially the early
activities and ‘proof of concept’ likely need to involve iterations. Implementation models would
hence potentially reflect the true project process if such iterations and complexity were
illustrated.

Conclusively, the empirical findings show extensive similarities between Herm et al.´s (2020)
‘Consolidated framework for RPA implementation projects’ and the project approach used and
suggested by the respondents, but also some evident discrepancies. Whilst Herm et al. (2020)
framework seemingly reflect the overall approach and suggested actions to be taken during
RPA implementation, the empirical findings also indicate that development of the framework
to further emphasize the importance of; change management, communication and education,
and the relative importance-, and demands of both the initialization- and post-implementation
phase. The differences in emphasis throughout the project approach is seemingly in line with
the empirical discrepancies compared to theory identified in the previous issues and challenges
section. The conclusive recommendations made in relation to the project approach were hence
built on what was perceived as challenging as well as facilitating for the RPA implementation.

The research approach of this case study did not allow for isolation of eventual impacts from
preconditions and contextual aspects, however as described in previous sections, several
aspects of the case company's situation and actions previous to the project may have facilitated
46
implementation in relation to especially the technical implementation questions. Other banking
and financial services companies hence need to take these eventual effects into consideration.

The overall highlighted importance of a structured approach, and the usefulness of visualized
project steps, also supports the relevance of developing implementation models and reviewing
the included activities to ensure relevant questions are considered. This finding also seems to
support the relevance of this study and the suggested need to further develop guidelines for
implementation.

47
6. Conclusions
Many practitioners seem to struggle with RPA implementation and researchers have urged for
further research within the RPA implementation field in general, as well as for application of
existing implementation framework to new empirical cases, in particular. The purpose of this
study was to investigate issues and challenges associated with RPA implementation.
Furthermore, based on existing implementation frameworks and theory, the aim was to explore
how projects may be managed in order to facilitate implementation. Semi-structured interviews
were conducted with employees and consultants having recently implemented a RPA solution
in the banking and financial services industry, documents describing the implementation
approach was reviewed, and the empirical findings were substantiated through theoretical
comparisons to answer the research questions:
1. What are the issues and challenges for RPA implementation projects?
2. How should an RPA project be managed to facilitate implementation?

6.1 Issues and challenges for RPA implementation projects

Whilst many potential issues and challenges were identified in the data collection and analysis,
some were especially evident and in line with theory, including; challenges with choosing the
right processes to automate, post-implementation and scale up issues - although the latter was
not equivalently emphasized in literature. Furthermore, the analysis concluded that almost all
issues and challenges emphasized by respondents relate to the challenge of achieving adequate;
(1) change management, (2) spreading of information, and (3) education- and engagement of
management and employees around RPA and the solution. The case company generally did not
experience several of the common issues and challenges mentioned in theory, such as technical
issues and cost-acceptance. Based on commonly described success-factors, the analysis
concluded much of the mitigated issues and challenges may be dependent on contextual
preconditions and actions taken by the case company before the implementation project. The
relevance of the common issues and challenges described in theory may hence not be ruled for
other contexts.

The analysis also concluded that most of the key issues and challenges identified in the case
study were directly linked to suggestions on how to facilitate implementation - indicating that
issues and challenges may be mitigated, as described in the section below.

48
6.2 Managing an RPA implementation project
To answer the question on how to manage RPA implementation projects, both (1) the
implementation structure and approach, and (2) particular success factors and facilitating
actions suggested by the respondents were considered and compared with existing theory. The
analysis concluded that the success-factors and implementation approach suggested by the
collected data, is profoundly interconnected. The most important facilitating aspects and
success-factors were identified as; building RPA knowledge and focusing on educating
management and employees within the organization, communicating (RPA knowledge,
benefits and logics behind the implementation for change management purposes), and carefully
choosing the right process to automate. The value of support processes including management
support, change management and establishing internal groups for RPA scale-up and
stewardship, was also emphasized.

The analysis also concluded that the key facilitating aspects emphasized by respondents
seemingly match scholarly contributions related to how one may improve implementation
success for ERP systems. Although no efforts were made to further investigate guidelines for
ERP implementation and eventual coherence to our findings, these similarities may indicate
that connections between different areas of IT implementation literature and RPA
implementation may be useful.

Furthermore, the analysis concluded that certain contextual preconditions and actions taken
before the project may have facilitated the RPA implementation for the case company, through
having; clearly defined IT roles (outsourced), an established contact with a consulting firm
knowledgeable within RPA, recently had reviewed their IT infrastructure and mapped their
processes, and that an urgent need for increased capacity created an evident business case for
RPA. Moreover, a structured approach with visualized project steps was also emphasized as
important to facilitate implementation. This conclusion also supports the relevance of
reviewing implementation models and the included activities to ensure relevant questions are
considered. This finding also seems to support the relevance of this study and the suggested
need to further develop guidelines for implementation.

One of the most important conclusions of this study is that Herm et al.´s (2020) ‘Consolidated
framework for RPA implementation projects’ provided a seemingly accurate breakdown of

49
most necessary actions and considerations also suggested by respondents. The empirical
findings however also indicate that implementation frameworks such as Herm et al. (2020) may
need increased attention to; change management, communication and education, and the
relative importance-, and demands of both the initialization- and post-implementation phase.

6.3. Limitations and suggestions for future research

This study has concluded several interesting differences and similarities between theory and
the empirical findings of the case company. An analysis of why differences may be evident was
presented, and new potential considerations for implementation frameworks was discussed.
The authors would like to emphasize several limitations of the study.

The fact that the study only examined one company, in the banking and financial industry, and
with certain pre-condition, may have had a negative impact on the generalizability and hence
usefulness of this study. Moreover, provided the focus on describing the entire process of an
implementation and hence using a limited number of centrally involved respondents along with
triangulation through reviewing project documentation - the study did not include e.g. end-
users or other case company stakeholders to describe challenges from different perspectives
than the project team(s).

Whilst the research questions and overall attempts to describe how projects should be
conducted are normative, the researchers have tried to highlight that contextual individual
differences likely have vast effects on how implementation should be conducted.
Generalization of the results presented in this study is hence limited to especially the banking
and financial services industry and processes surrounding lending. Provided the extensive
harmony between the universal and context independent implementation framework analyzed
and the empirical findings - many of the conclusions may however be interesting for
practitioners outside the targeted industry of this study.

In order to increase reliability and generalizability, the researchers suggest future studies within
the field of RPA implementation should include multiple companies, investigate and compare
additional contexts-, industries-, and types of companies with and without assistance from
consulting firms. Provided the assumed effects of contextual preconditions described in this
study, attempts to isolate and investigate issues and facilitating aspects in different contexts

50
would likely increase the practicality of RPA implementation theory. In addition, the
researchers also suggest future research to be cross-disciplinary and include increased focus
on; implementation theory related to other software solutions, as well as on organizational-,
and change management questions, to develop RPA implementation models further.

51
References

Aguirre, S. and Rodriguez, A. (2017). ‘Automation of a business process using robotic

process automation (RPA): a case study’. Appl Comput Sci Eng Commun Comput Inf Sci.
Conference paper

Alberth, M. and Mattern, M. (2017). ‘Understanding robotic process automation (RPA)’, The
CAPCO Institute Journal of Financial Transformation, November, Automation No. 46, pp. 1-
8.

Anagnoste, S. (2017). ‘Robotic automation process – the next major revolution in terms of
back office operations improvement’, Proceedings of the International Conference on
Business Excellence, Bucharest, Vol. 11, available at:https://doi.org/10.1515/picbe-2017-
0072

Asatiani, A. and Penttinen, E. (2016). ‘Turning robotic process automation into commercial
success’, Journal of Information Technology Cases. Vol. 6, pp. 67–74.

Bryman, A., (2012). Social Research Methods 4e. Oxford, UK: Oxford University Press.

Chae, B., (2019). ‘A General framework for studying the evolution of the digital innovation
ecosystem: The case of big data’. International Journal of Information Management, Vol. 45,
pp.83-94.

Crandall, R. and Diener, E., (1978). ‘Determining Authorships of Scientific Papers’. Drug
Intelligence & Clinical Pharmacy. Vol. 12(6), pp.375-375.
Davis, F. (1985). ‘A Technology Acceptance Model for Empirically Testing New End-User
Information Systems: Theory and Results.’ Massachusetts Institute of Technology, pp. 1-291.
Dubois, A. and Gadde, L., 2002. ‘Systematic combining: an abductive approach to case
research’. Journal of Business Research, Vol. 55(7), pp.553-560.

52
EY (2020). Digital Directions: A perspective on the impact of digital technologies. Available
via: https://www.ey.com/en_gl/alliances/digital-directions-a-perspective-on-the-impact-of-
digital-technologies

Fischer, M., Imgrund, F., Janiesch, C. and Winkelmann, A., (2020). ‘Strategy archetypes for
digital transformation: Defining meta objectives using business process management’.
Information & Management. Vol. 57(5), p.103-262.

Fung, H. P. (2014). ‘Criteria, use cases and effects of information technology process
automation (ITPA)’. SSRN Scholarly Paper No. ID 2540023, Social Science Research
Network, Rochester, New York. Available via: https://papers.ssrn.com/abstract=2540023

Gartner - Gartner Says Worldwide Robotic Process Automation Software Revenue to Reach
Nearly $2 Billion in 2021. Available via: https://www.gartner.com/en/newsroom/press-
releases/2020-09-21-gartner-says-worldwide-robotic-process-automation-software-revenue-
to-reach-nearly-2-billion-in-2021.

Gex, C., and Minor, M. (2019). ‘Make Your Robotic Process Automation (RPA)
Implementation Successful’, Armed Forces Comptroller, Vol. 64(1), pp. 18–22.

Gotthardt, M., et al. (2020). Current State and Challenges in the Implementation of Smart
Robotic Process. Automation in Accounting and Auditing. ACRN Journal of Finance and
Risk Perspectives, Vol. 9(1), pp. 90-102.

Guest, G., MacQueen, K. and Namey, E., (2011). Applied thematic analysis. 1st Ed.
Thousand Oaks, CA: Sage Publications.

Herm, V., et al. (2020). A Consolidated Framework for Implementing Robotic Process
Automation Projects. in D. Fahland et al. (Eds.): BPM 2020, LNCS 12168, Springer
International Publishing. pp. 471–488

Kvale, S., and Brinkmann, S. (2015). Interview. Learning the Craft of Qualitative Research
Interviewing, 3rd Ed. Thousand Oaks, CA: Sage Publications.

53
Lacity, M., Willcocks, L. and Craig, A. (2016) ‘Robotizing global financial shared services at
royal DSM’. Paper Ser. Finan. Serv. Vol. 46(1), pp. 62–76

Lamberton, C. (2016). Get ready for robots: Why planning makes the difference between
success and disappointment. EY FS Insights. Available via:
https://www.ey.com/en_gl/financial-services-emeia/get-ready-for-robots

Legris, P., Ingham J. and Collerette, P. (2004). ‘Why do people use information technology?
A critical review of the technology acceptance model’. Information & management.
Vol. 40(3), pp. 191–204.

Lhuer, X. (2016) ‘The next acronym you need to know about: RPA (robotic process
automation)’. McKinsey & Company. Available via: https://www.mckinsey.com/business-
functions/mckinsey-digital/our-insights/the-next-acronym-you-need-to-know-about-rpa

Lu, H., Li, Y., Chen, M., Kim, H. and Serikawa, S. (2017). ‘Brain Intelligence: Go beyond
Artificial Intelligence’. Mobile Networks and Applications. Vol. 23(2), pp. 368-375.

Lin, P. (2018). ‘Adapting to the New Business Environment’. CPA Journal 88. Vol. 12,
pp.60–63.

Madakam, S., Holmukhe, R. and Kumar Jaiswal, D. (2019). ‘The Future Digital Work Force:
Robotic Process Automation (RPA)’. Journal of Information Systems and Technology
Management. Vol. 16, pp. 1-17.

Markus, M., Axline, S., Petrie, D. and Tanis, C. (2000). ‘Learning from adopters' experiences
with ERP: problems encountered and success achieved’. Journal of Information Technology.
Vol. 15(4), pp. 245-265.

Nwankpa, J. and Datta, P., (2017). ‘Balancing exploration and exploitation of IT resources:
the influence of Digital Business Intensity on perceived organizational performance’,
European Journal of Information Systems. Vol. 26(5), pp.469-488.

54
Nwankpa, J. and Merhout, J., (2020) ‘Exploring the Effect of Digital Investment on IT
Innovation’. Sustainability. Vol. 12(18), p.73-74.

Rutaganda, L. (2017). “Avoiding pitfalls and unlocking real business value with RPA”. The
CAPCO Institute Journal of Financial Transformation. Vol. 46, pp. 104–115.

Ryan, G. and Bernard, H., 2003. ‘Techniques to Identify Themes’. Field Methods. Vol. 15(1),
pp.85-109.

Santos, F., Pereira, R. and Vasconcelos, J.B. (2019). Toward robotic process automation
implementation: an end-to-end perspective. Business process management journal. Vol.
26(2), pp. 405-420

Stolpe, A., Steinsund, H., Iden, J., and Bygstad, B. (2017). ’Lightweight IT and The IT
Function: Experiences from Robotic Process Automation in a Norwegian Bank’. Proceeding
from the annual NOKOBIT conference. Vol. 25(1).

Syed, R. et al. (2020). ‘Robotic Process Automation: Contemporary themes and challenges’,
Computers in Industry. Vol. 115, pp. 103-162

Tonnquist, B. (2014). Projektledning. 5 uppl., Stockholm: Bo Tonnquist och Sanoma

Utbildning AB

van der Aalst, W., Becker, J., Bichler, M., Buhl, H., Dibbern, J., Frank, U., Hasenkamp, U.,
Heinzl, A., Hinz, O., Hui, K., Jarke, M., Karagiannis, D., Kliewer, N., König, W., Mendling,
J., Mertens, P., Rossi, M., Voss, S., Weinhardt, C., Winter, R. and Zdravkovic, J. (2018).
’Views on the Past, Present, and Future of Business and Information Systems Engineering’.
Business & Information Systems Engineering. Vol. 60(6), pp. 443–477.

van der Aalst, W. M. P., Bichler, M.,and Heinzl, A. (2018b) ’Robotic Process Automation’
Business Information Systems Engineering. Vol. 60(4), pp. 269-272.

55
Vishnu, S., Agochiya, V. and Palkar, R. (2017), “Data-centered Dependencies and
Opportunities for Robotics Process Automation in Banking”. Journal of Financial
Transformation. Vol. 45, pp. 68–76

Willcocks, L. Lacity, M. and Craig, A. (2015a) ‘The IT Function and Robotic Process
Automation’, The Outsourcing Unit Working Research Paper Series, Paper 15/05. The
London School of Economics and Political Science.

Willcocks, L., and Lacity, M. (2015b). ‘Robotic Process Automation: The Next
Transformation Lever for Shared Services’. The Outsourcing Unit Working Research Paper
Series, Paper 15/07. The London School of Economics and Political Science.

Willcocks, L. Lacity, M. (2016) ‘The Next Transformation Lever for Shared Services’, The
Outsourcing Unit Working Research Paper Series, Paper 16/01. The London School of
Economics and Political Science.

Willcocks, L., Lacity, M. and Sauer, C. (2017). Outsourcing and Offshoring Business
Services. Cham: Springer International Publishing.

Willcocks, L. Lacity, M. and Hindle, J. (2018) ‘KEY TO RPA SUCCESS, Change

Management & Capability Development - People, Process & Technology’. Executive
research report. Knowledge Capital Partners.

Willcocks, L. Lacity, M. and Hindle, J. (2019). KEY TO RPA SUCCESS, Part 5: The Path to
Maturity. Executive research report. Knowledge Capital Partners.

Yin, R. K. (2009). Case study research: Design and methods, 4th Ed. Thousand Oaks, CA:
Sage Publications

Yoo, Y., Boland, R., Lyytinen, K. and Majchrzak, A., (2012). Organizing for Innovation in
the Digitized World. Organization Science. Vol. 23(5), pp.1398-1408.

56
Appendices
Appendix 1
Interview guide

Checklist before the interview:

- Describe our research, the purpose and the themes we will cover
- Define the case and limitations of scope (describe the specific implementation project
we are referring to and that the respondent may answer the questions based on
previous experiences if that is stated)
- Describe the practical aspects with the interview
- Describe ethical considerations (including their anonymity)
- Ask for permission to record and describe how the recordings will be used

1. Interviewee profile and role in project

1.1 Experience and role definition
- Can you tell us about your professional role? (e.g. tasks, special field of knowledge,
years of experience etc.)
- Have you worked with RPA technology or implementation before this project? If yes:
for how long/ approximately with how many projects?
1.2. Role and responsibility in this specific project
- What was your role in the RPA implementation project
- What responsibilities did you have? What tasks did you perform?
- During what timespan were you involved?

2. The project
2.1 Initialization
- Tell us about the initiation of the project? What was the background of the project?
- Was there a business case? If yes;
- What was the reasoning and;
- When was it formulated?
- Was there a connection between the RPA implementation and your business
strategy? If yes, how?
- What were the first steps/ actions your organization took? Preparations?
- Tell us about the process of choosing an RPA vendor/ software?
- Tell us about the choice of which process to automate?
- Could you conclude certain criteria for processes suitable? If yes; which
criteria?

2.2 Implementation
- Did you work with testing and verifying the functionality of the robot? If so,
how?
- By this time of the project - had the business case changed/ had your
perceptions of usefulness changed?
- Tell us about the roll-out phase.
- Did you have any specific strategy?
- Did you use some sort of pilot-roll-out or certain processes?

57
2.3 Completion of the project and post roll-out activities
- After the implementation of the software was completed and the process(es) were
automated… Can you describe any actions taken “post-implementation”?
- Were there any activities aimed at scaling up the robot-technology usage within the
organization?

Eventual support processes

[Over the entire project and afterwards]
- Can you describe any support processes, functions, activities etc. that
has been practiced during/ after the project?
- E.g. change management, governance, IT integration etc.
- Did the organization establish a “center of excellence” or some sort of
group to support monitoring and maintenance of the robots?
- Has any measures been taken to capture and maintain RPA
knowledge within the organization?
- How would you describe the level of “management's support” for the
project? [From the board, executives, and downward]

3. Issues and general approaches

- What were the most challenging/ problematic aspects of the RPA
implementation project?
- How were these challenges/ problems solved?

- It appears that upwards of 50% fail with their RPA implementation. What do you
think may be the reason for this, based on your own experience?

Specific examples of common issues [to be asked if the respondent does not
mention the specific problem(s) during the interview]
- Was there any issues with targeting the right process/ selecting the right
scope for automation?
- If yes, how did you overcome these issues?
- Was there any issues regarding IT infrastructure?
- If yes, how did you overcome these issues?
- Was there any issues regarding the skills needed for the implementation
project?
- If yes, how did you overcome these issues?

4. Keys to success
[Reconnect back to the different phases and activities mentioned by the respondent and ask
what they did right. Do they consider any actions as particularly important?]
- What aspects of the RPA implementation project and process do you consider
as especially important for the success of the project?
- Which part of the implementation process did you consider to be most
decisive for the final result?
- In addition to this part, were there other “decisive” factors that
contributed to the outcome of the implementation?
- Was there any part in the implementation process you should have done
differently
- If this is the case, how?

58
 5. Other questions
- With regards to RPA implementation success… is there anything you feel we have
failed to ask about?
- Any preconceptions you had before the project and things you felt you
learned?

[Finish the interview, thank the respondent for their participation, provide information on
publication and ask if they would like to take part of the thesis when published]

59
Appendix 2
Suggested evaluation criteria for process selection

Business impact vs. complexity-matrix for process evaluation

(Adapted based on descriptions from respondent CO1)

Evaluation factors for choosing what process to automate

Adapted from the case company´s retrospective lessons learnt PowerPoint presentation.

60
Appendix 3
Activities included in the respective phases of the case company´s project overview

Phase Activities

Evaluation As described by respondent CL1 the evaluation phase is orientating in

nature, aimed at learning about RPA and finding arguments for and
against using RPA or instead e.g. hiring additional advisors and
generally evaluating the process to automate and whether RPA is
suitable.

Configuration The consequent configuration phase consisted of mapping the process to

be automated in a Process Definition Document (PDD), developing
scenarios for the robot to process and mapping all the detailed steps the
robot need to perform through workshops with different internal parties
and interviewing employees normally performing the tasks. The PDD
was then reviewed and accepted by relevant decision makers before the
Solution Design Document (SDD) was created (mostly by the consulting
company), followed by acceptance tests (AT) within the organization,
observing the robots work, ensuring the robot is conducting the processes
correctly throughout all steps.

Implementation In the implementation phase, a Request For Change has to be approved

from an IT perspective (following the same decision-process as all other
IT changes in the organization, e.g. ensuring that the robot does not
negatively affect other systems, deciding what server should be used, and
safeguarding GDPR compliance - before finally deploying the robot.

Management The final phase of the visualized project plan overview is management
and and maintenance of the RPA involving reconfiguring and developing the
Maintenance robot in accordance with changes in e.g. processes (as described in
previous sections of this chapter).

(The summary is based on reviewing visualizations received from the case company and on
descriptions from respondents)

61