Dynamic roadmap for successful implementation

RoboticProcessAutomation:
DynamicRoadmapforSuccessfulImplementation
Guðrún Lilja Sigurðardóttir
Thesis of 30 ECTS credits
Master of Science (M.Sc.) in Engineering Management
June 2018

1
Thesis of 30 ECTS credits submitted to the School of Science and Engineering
at Reykjavík University in partial fulfillment of
the requirements for the degree of
Master of Science (M.Sc.) in Engineering
Management
June 2018
Supervisors:
Dr. Páll Jensson, Supervisor
Professor, Reykjavík University, Iceland
Dr. Þórður Víkingur Friðgeirsson, Supervisor
Professor, Reykjavík University, Iceland
Examiner:
Páll Kr. Pálsson, Examiner
Assistant Professor, Reykjavik University, Iceland

iii
Copyright
June 2018

5
June 2018
Abstract
The purpose of this MSc thesis seeks to uncover the efficient way of implementing
Robotic Process Automation (RPA) successfully. This study aims to give
organizations insight into how to successfully implement RPA and what factors to
beware of to avoid failure. The main literature that is supporting this research is
previous research of RPA along with a comparison to IT development via Business
Process Management. Experts in the field that have experience of implementing
RPA were interviewed for this research. The results are present as a dynamic
roadmap for RPA implementation with a description of what risk factors it is
necessary to beware of to avoid failure.
The results indicate that the value is not only in the robot. When a number of
automated processes increases then RPA gets more complicated. Organizations
need to be able to understand RPA and therefore build a structure for RPA with
appropriate RPA team. Including change management right from the beginning is
fundamental; when the changes are applied, employees have to be prepared for
them. Companies need to continuously work on the process assessment and
therefore identify, optimize and prioritize processes for the RPA lifecycle.
Companies that have RPA team with lack of knowledge could experience that
some part of the RPA lifecycle is not carried out successfully, which can have an
adverse effect later on in RPA lifecycle and result as project failure. RPA project
evaluation depends mainly on the objectives of the project and therefore the type
of processes that are automated.
Keywords: RPA, software robot, BPM, roadmap, implementation.

vii
júní 2018
Útdráttur
Í þessari ritgerð er að leitast við að finna hagkvæma og skilvirka leið við
innleiðingu á Robotic Process Automation (RPA). Markmið rannsóknarinnar er að
fyrirtæki og stofnanir fái yfirsýn yfir helstu þætti í innleiðingarferli á RPA með
árangursríkum hætti og þá einkum það sem þarf að hafa í huga til að forðast mistök.
Rannsóknir sérfræðinga í innleiðingu RPA eru bornar saman við almenna
tækniþróun á sambærilegu sviði og loks eru viðtöl tekin við sérfræðinga sem starfa
við innleiðingu á RPA víðs vegar um heimin.
Niðurstöður rannsóknarinnar er svo teknar saman með uppsetningu á sérstökum
ferli þar sem leitast er við að útlista þau atriði sem eiga að koma í veg fyrir mistök
í ferlinu. Af niðurstöðum rannsóknarinnar má draga þá ályktun að innnleiðing á
RPA verði flóknari eftir því sem sjálfvirkum ferlum fjölgar. Þá er mikilvægt að
fyrirtæki tileinki sér þá aðferðafræði sem nauðsynleg er fyrir innleiðingarferlið. Í
því felst m.a að sérfræðiþekkingar er þörf á öllum hlutum lífsferils RPA ef vel á að
takast. Stöðug forgangsröðun, hagræðingar og greiningar á öllum ferlum ásamt
innleiðing á breytingarstjórnun, strax í upphafi er nauðsynlegur grunnur þegar
sjálfvirkum ferlum fjölgar. Mat á árangri RPA verkefnis fer eftir upphaflegu
markmiði hverju sinni og þar með tegund þeirra ferla sem á að sjávirknivæða

9
Thesis of 30 ECTS credits submitted to the School of Science and Engineering
at Reykjavík University in partial fulfillment of
the requirements for the degree of
Master of Science (M.Sc.) in Engineering Management
June 2018
Student:
Supervisors:
Dr. Páll Jensson
Dr. Þórður Víkingur Friðgeirsson
Examiner:
Páll Kr. Pálsson

xi
The undersigned hereby grants permission to the Reykjavík University Library to reproduce
single copies of this Thesis entitled Robotic Process Automation: dynamic Roadmap for
Implementation and to lend or sell such copies for private, scholarly or scientific research
purposes only.
The author reserves all other publication and other rights in association with the copyright
in the Thesis, and except as herein before provided, neither the Thesis nor any substantial
portion thereof may be printed or otherwise reproduced in any material form whatsoever
without the author’s prior written permission.
date
Master of Science

xiii
“It has become appallingly obvious that our technology has exceeded our humanity.”
- Albert Einstein

xv
Acknowledgments
This master’s thesis marks the end of master’s program in engineering management at Reykjavik
University. Several people have provided their support during the work with this thesis. Without you,
this thesis would not have reached this level.
To my supervisors, Páll Jensson and Þórður Víkingur Friðgeirsson, I would like to thank you for your
guidance and support through the creation of this thesis.
Then I would like to thank the interviewees for taking their time off for the interviews and support me
with their knowledge.
Finally, I would like to thank my family for patience and endless support they have shown me during
this semester.

xvii
Contents
Acknowledgments ........................................................................................................... xv
Contents......................................................................................................................... xvii
List of Figures................................................................................................................. xix
List of Tables .................................................................................................................. xxi
List of Abbreviations...................................................................................................xxiii
1 Introduction................................................................................................................. 1
1.1 Background........................................................................................................... 2
1.2 Research questions................................................................................................ 3
1.3 Assumptions and Limitations................................................................................ 3
1.4 Structure of the thesis............................................................................................ 4
2 Research methodology................................................................................................. 5
2.1 Literature Review Methodology............................................................................ 6
2.2 Data Collection..................................................................................................... 6
2.2.1 Interviews .................................................................................................... 6
2.3 Data analysis......................................................................................................... 8
3 Robotic Process Automation and a comparison to IT development.......................... 9
3.1 Robotic Process Automation ................................................................................. 9
3.1.1 RPA classified as lightweight IT............................................................. 11
3.1.2 RPA tools ............................................................................................... 12
3.2 Comparison of RPA and IT development via BPM ............................................. 14
3.2.1 BPM implementation methodology......................................................... 15
3.3 Summary ............................................................................................................ 16
4 Previous research of RPA Implementation .............................................................. 17
4.1 Company 1: RPA consultant service provider ..................................................... 17
4.2 Company 2: Financial Shared Services................................................................ 17
4.3 Company 3: Mobile telecommunication provider................................................ 18
4.4 Company 4: Energy Sector.................................................................................. 19
4.5 Company 5: BPO service provider ...................................................................... 20
4.6 Summary ............................................................................................................ 21
5 Dynamic Roadmap for RPA Implementation .......................................................... 22
5.1 Phase I ................................................................................................................ 22
5.1.1 Process Assessment ................................................................................ 22
5.1.1.1 Choose wrong processes for RPA ............................................. 23
5.1.1.2 Process assessment not carried out correctly ............................. 24

xviii
5.1.2 Proof of Concept..................................................................................... 26
5.1.2.1 Focusing only on technical PoC ................................................ 27
5.1.2.2 Lack of understanding of RPA requirements............................. 27
5.1.2.3 Failure to build a successful Business Case ............................... 28
5.2 Phase II............................................................................................................... 30
5.2.1 Project design and build .......................................................................... 30
5.2.1.1 RPA without an Operating Model............................................. 30
5.2.1.2 RPA without Change Management ........................................... 31
5.2.2 RPA Lifecycle ........................................................................................ 32
5.2.2.1 RPA team with lack of knowledge ............................................ 32
5.2.2.2 Underestimate testing in RPA Lifecycle.................................... 34
5.2.2.3 RPA projects evaluation............................................................ 35
5.3 Summary ............................................................................................................ 38
6 Discussion................................................................................................................... 42
6.1 Findings Reliability............................................................................................. 42
6.2 Building Center of Excellence............................................................................. 43
6.3 Re-engineering processes for RPA...................................................................... 44
6.4 Further research .................................................................................................. 45
7 Conclusion.................................................................................................................. 46
7.1 Summary of research .......................................................................................... 46
8 Bibliography .............................................................................................................. 49

xix
List of Figures
Figure 1: Research process (Kothari, 2004)......................................................................... 5
Figure 2: The circular process of coding (DeCuir-Gunby et al., 2011)................................. 8
Figure 3: Example of a simple process for RPA (Willcocks & Lacity, 2016)....................... 9
Figure 4: Feature of RPA processes (Sutherland, 2013), (Willcocks & Lacity, 2016). ....... 10
Figure 5: Example of the process that was automated using Blue Prism (Screenshot from
Blue Prism software) ........................................................................................................ 13
Figure 6: Implementation methodology for BPM (Chang, 2006)....................................... 15
Figure 7: Company 1 main stages of RPA implementation................................................ 17
Figure 10: RPA lifecycle for Company 4 .......................................................................... 20
Figure 11: Company 5 main stages of RPA implementation.............................................. 20
Figure 12: Dynamic Roadmap for RPA implementation ................................................... 39

xxi
List of Tables
Table 1: Interviewees identification and their roles in RPA projects......................................7
Table 2: Heavyweight vs. Lightweight IT (Bygstad, 2017)..................................................11
Table 3: Factors that can cause failure when implementing RPA and appropriate references
...........................................................................................................................................41

xxiii
List of Abbreviations
RPA: Robotic Process Automation
IoT: Internet of Things
IT: Information Technology
AI: Artificial Intelligent
FTE: Full Time Equivalent
PoC: Proof of Concept
CofE: Center of Excellent
BPR: Business Process Reengineering
BPM: Business Process Management
ROI: Return on investment
SME: Subject Matter Expert

1
1 Introduction
Technology is quickly advancing; computers have started listening and speaking, and we have
robots that can handle imprecise tasks that are currently done by humans (Brynjolfsson &
McAfee, 2014). Organizations have to respond to increased challenges from customers’ needs
and remain competitive by reducing cost. Robotic Process Automation (RPA) enables
organizations to automate repetitive and tedious tasks and therefore provides a way of reducing
operating costs (Forrester Research, 2014). Organizations that have made use of the RPA
technology have not only noticed massive savings on full-time equivalent (FTE), they have
also experienced other benefits including improved service, quality and speed, lower error rate
and staff satisfaction (Willcocks & Lacity, 2016).
According to Lamberton (2016), 30-50% of the first RPA projects within an organization fail.
The term RPA is relatively new, however, the market is growing fast. To be able to avoid the
common mistakes in RPA project failure, it is crucial to define a roadmap for the
implementation.
In this research, it is investigated how to successfully implement RPA from the beginning of
the roadmap, where the business problems are defined, to the end of the entire process when
RPA is part of the organization’s strategy. A dynamic roadmap to successfully implement RPA
is presented as well as a description of what risk factors organizations need to be aware of to
avoid project failure. The roadmap is adapted from a combination of literature and interviews
with experts in the field.
This thesis is structured in the following way: chapter 2 is an overview of research
methodology. In chapter 3, the main features of RPA and its tools are defined, it is also
explained why RPA is classified as lightweight IT while examining the main difference
between lightweight and heavyweight IT. To get a better understanding of RPA, a comparison
of RPA and IT development via BPM and the methodology for BPM implementation is
described. In chapter 4, an overview of previous research of RPA implementation is provided,
to give a better understanding of the RPA implementation roadmap. In chapter 5, the dynamic
roadmap for RPA implementation is presented. Results and further research are discussed in
chapter 6. Finally, the conclusion of this thesis can be found in chapter 7.

2
1.1 Background
Automation is a system that functions without direct human interaction. Many automated
systems have this in common: taking out the most unreliable factor (human error) thus
improving precision, quality, and accuracy. The first idea of how to automate processes using
software came in 1935, when the computer scientist Alan Turing described how a systematical
algorithm could work processes more effectively. His ideas on algorithms and automation had
a lasting impact (Middelburg, 2017).
The beginning of the technological development started in 1964 when the first laboratories
researching artificial intelligence were opened at MIT, and following, in 1965 the first Robotics
Institute was opened (Willcocks & Lacity, 2016). Service Automation is believed to be the next
wave of development in automation (Middelburg, 2017). Middleburg (2017) described service
automation as a technology that is used to automate services and deliver optimal user
experience, where the primary objective is to automate redundant manual labor. Willcocks &
Lacity (2016) define RPA as service automation, but other terms apply to service automation.
For example, a scripting tool, artificial intelligence, cognitive computing, BPM, etc. RPA
addresses the part of service automation that automates structured processes (Willcocks &
Lacity, 2016). Robotic Process Automation is defined by the IEEE Standards Association as:
“A preconfigured software instance that uses business rules and predefined activity
choreography to complete the autonomous execution of a combination of processes,
activities,transactions, and tasks in one or more unrelated software systems to deliver a result
or service with human exception management” (IEEE Std 2755-2017, 2017).
The term Robotic Process Automation (RPA) was first used in 2012 and was created by
marketing director Patric Geary, who worked for the RPA software company, Blue Prism
(Hindle, Lacity, Willcocks, & Khan, 2018). RPA is a newly developed technology and there is
no proper research from the early stages of use. RPA began to gain popularity in 2014 and
2015 when companies started to announce considerable savings due to automation. The market
for RPA back-office automation was becoming more significant by early 2016, but it was still
relatively small-scale during this time (L. Willcocks & Lacity, 2016). According to Horses for
Sources Research (2017) and Everest Group research (2017), the global RPA market which
includes both RPA services and RPA software increased by about 64% from 2016 to 2017
(from $271 million to $443 million) (Fersht & Snowdan, 2017) (Everest Global, 2017). HfS

3
Research (2017) reported a 42% increase in the market from 2017 to 2018 and an expected
increase around 94% from 2018 to 2021 (Fersht & Snowdan, 2017).
1.2 Research questions
To investigate the dynamic roadmap for successful implementation of RPA and what factors
can cause failure when implementing RPA, the research questions are the following:
How to successfully implement a Robotic Process Automation?
What is Robotic Process Automation?
What factors can cause failure when implementing RPA?
What factors are common for RPA project evaluation?
The researcher found it interesting to investigate how to implement RPA successfully for many
reasons. First, RPA is relatively new and there is little prior research on RPA implementation.
Second, the researcher believes that RPA will have a high impact on the Icelandic labor market
in the next few years and finally, the researcher thinks RPA is an interesting tool to help
companies to increase efficiency.
Objectives of this research
This research aims to answer the questions mentioned above, with a primary focus on the first
question. To answer these questions, a previous research of RPA implementation for different
companies was analyzed, theory of implementation workflow was studied, and interviews with
experts working in the RPA field were studied. An RPA roadmap with the dynamic structure
for the most suitable implementation methods will be created and described in detail, as well
as a table with all factors that are considered to be essential to be aware of to avoid failure.
1.3 Assumptions and Limitations
The field of RPA has little prior research and there are not many organizations that have been
working with RPA for a considerable amount of time, thus, this study is largely limited. The
researcher found it challenging to find candidates for an interview due to the fact that RPA is
a relatively new method, and the number of RPA experts located here in Iceland is limited. The
interviews are based on four experts’ perspectives, so a small number of interviewees are a
limitation. The interviewees have all worked for the same consultant company which can limit

4
the results. They have experiences of implementing RPA for different types of business, but
the types of businesses are not exhaustive so the researcher cannot conclude that the findings
conform to all type of businesses. The roadmap presented in this research is a dynamic roadmap
so it can differ between companies. This research focuses mainly on experience from RPA
developers that have used software robots from Blue Prism.
Experts’ degree of knowledge is also a limitation. Expert opinion toward RPA implementation
could change in the next few years as popularity increases.
1.4 Structure of the thesis
The study begins with an introduction where the RPA background and the subject itself is
discussed. Following this, a detailed description of RPA and a comparison of RPA with IT
development via BPM is covered to get a better understanding of what distinguishes RPA from
other automation tools. Furthermore, a brief overview of previous research of RPA is presented
to assess the current state of RPA implementation. In chapter 5, the interview findings have
been compiled, and the research questions will be answered. A dynamic roadmap for successful
RPA implementation is presented and described as well as the factors that can cause failure.
The discussion in chapter 6 provides a comparison of literature and interviews as well as the
main difference between them and, finally, a conclusion summarizing the main results and key
concepts closes the thesis. The structure of the thesis is as follows:
i. Introduction
ii. Research methodology
iii. Description of RPA
iv. Comparison of RPA and IT development
v. Previous research of RPA
vi. Dynamic Roadmap for RPA and risk factors
vii. Discussion
viii. Conclusion

5
2 Research methodology
The research method is qualitative. Qualitative research provides findings that are not based
on statistical procedures or other quantitative methods. The results of qualitative research are
based on analyzes of different types of knowledge (Hoepfl, 1997). Qualitative research can
refer to research where only a small sample of interviews are used for the research, both
structured and unstructured (Bryman & Burgess, 1994). The process for this research is based
on a research process described by Kothari (2004), which is provided in Figure 1. The process
starts by describing the research problem, which is done by discussing with experts in the
subject field, and then examining available literature review. When research questions have
been designed, the researcher executes a literature review in order to prepare the research
design. Collection of data was done through interviews and analyzing of the data was done by
“coding”. Coding is the process of segmenting sentences into categories and labeling these
categories with terms (Creswell, 2014). Finally, the researcher finished writing the report and
started preparing for a presentation.
Design research
Analysis of data
Report & Present results
Define research problem
Review concepts and theories
Collect data
Execution of the project
Review previous research findings
Figure 1: Research process (Kothari, 2004)

6
2.1 Literature Review Methodology
According to Kothari (2004), the researcher may review two types of literature, conceptual and
empirical literature. Conceptual literature consists of concepts or theories, while empirical
consists of previous research (Kothari, 2004). The literature review for this thesis consists of
both literature types.
A literature review was conducted throughout the research. The literature review covers RPA´s
main features and background, implementation methods, and comparison to IT development
via BPM. The previous research analysis is based on case studies. These case studies include
a description of RPA project implementation for different companies. The researcher focuses
on five previous research case studies and the key themes of RPA implementation that arise
within them.
2.2 Data Collection
According to Kothari (2004), the data collection can take place in two ways; through an
experiment or a survey. If data collection is done through a survey, the researcher can choose
any of the following methods (Kothari, 2004):
 By observation
 Through personal interview
 Through telephone interview
 By mailing of questionnaires
 Through schedules
The data collection for this research was done through personal interviews. In this chapter, it
will be discussed in detail how the data was collected.
2.2.1 Interviews
The researcher decided to collect data through personal interviews that were all anonymous.
Four people were interviewed: two from Norway and two from Iceland. The researcher chose
these four people as they have all participated in RPA projects with different roles and
therefore, may have different opinions toward RPA implementation in general. All of them
have gained Blue Prism certifications as developers by passing an accreditation exam. The
interviewee’s identification and roles can be found in Table 1.

7
INTERVIEWEES ROLES
INTERVIEWEE 1 Process analyst and developer
INTERVIEWEE 2 Developer
INTERVIEWEE 3 Lead developer, process analyst, RPA leader and IT
infrastructure
INTERVIEWEE 4 Project manager, developer and process analyst
Table 1: Interviewees identification and their roles in RPA projects
All interviews took place face to face with a semi-structured method. In a semi-structured
interview, the interviewer should have prepared a checklist to be sure that the interview covers
all areas of research questions. A semi-structured interview is more flexible than a structured
interview, which allows the experts to expand their response (Alshenqeeti, 2014). The
researcher decided to use a semi-structured method in order to ensure that facts the experts
thought were necessary to mention regarding the interview questions would also be discussed.
All the questions used in the interview were prepared prior to the interview and all interviews
were tape-recorded. Each interview took approximately one hour. The primary objectives of
the interviews were to gain input from experts on how to implement RPA successfully, what
risk factors are important to prevent failure, and what factors are common for RPA project
evaluation.
A lot of time was spent on designing the structure of the interview. The preparation consisted
of reading the literature of RPA and implementation method for other tools. It also included
training in robotics software and participation in a successful implementation of RPA for an
Icelandic company. After the preparation, five main stages were considered to be integral to
RPA implementation. The interview consisted of questions regarding these stages, a brief
description of what happens at each stage, what is important at each stage to succeed and what
can cause failure. The stages divide into two phases; three stages for phase I and two for phase
II. The stages are as follows: Process Assessment – Business Case - Proof of concept - Project
Design and Build - RPA Lifecycle. In all interviews, the researcher asked if the experts wanted
to add any additional stages that are an important part of an RPA roadmap or skip any stages.
All information from the interview was transcribed unless it had a company or interviewee
identifier.

8
2.3 Data analysis
The researcher conducted a qualitative data analysis based on six steps provided by Creswell
(2014). First, data was organized and prepared for data analysis which involved transcribing
interviews; interviews that were in Icelandic were translated into English. The next step
required reading the transcribed data to provide a general sense of information. The data was
then coded as described previously in chapter 2. The researcher created codes with a
combination of emerging and predetermined codes. Emerging codes are those that are
developed during the data analysis (Creswell, 2014). Coding was then used to emphasize the
circular process where the researcher reviewed raw data based on theoretical findings and
current literature (DeCuir-Gunby, Marshall, & McCulloch, 2011). The circular process of
coding, as described by DeCuir-Gunby et al. (2011), can be seen in Figure 2.
Research literature
Theory
Raw data
Code development Coding
Figure 2: The circular process of coding (DeCuir-Gunby et al., 2011)
The next step involved categorizing the codes to generate themes for the research study, which
was done by gathering relevant codes, sorting them into groups and shaping the themes to the
general description. When the code had been categorized, the findings were summed up. All
post-processing of the data was done in the commercial spreadsheet software Excel (Microsoft
Excel for Windows, version 15.0 year 2013) and Word (Microsoft Word for Windows, version
15.0 year 2013).
The dynamic roadmap for RPA implementation was created as well as a table of factors that
can cause failure. The last step of the data analysis included interpretation of qualitative
research.

9
3 Robotic Process Automation and a
comparison to IT development
In this section, the main features of the concepts and classification of RPA are described
including what features characterize RPA as a tool by providing an example of how the
framework of an automated process looks in a robotic software. For a better understanding of
the RPA concept, a comparison of RPA and IT development, using Business Process
Management (BPM), was carried out as well as a description of BPM implementation
methodology.
3.1 Robotic Process Automation
Robotic Process Automation (RPA) is a methodology where a computer software is used to
complete a specific process that was previously done by a human. Robotic automation software
does not replace systems. Instead, it works with the system and performs a particular task in
the same way as it has been asked to complete (Sutherland, 2013). RPA interacts with a
computer system the same way a human would, but much faster and at a lower cost. Instead of
using a salary-paid employee to do a repetitive task on the computer, RPA can be used to do
the processes that includes the typing and clicking the same way as a human (Lu, Li, Chen,
Kim, & Serikawa, 2017). RPA does not require changing old systems. RPA can be integrated
with any software used by humans and it can be implemented in a short period of time for the
purpose of carrying out operational procedures (Asatiani & Penttinen, 2016).
Some processes are better suited for robotic automation than others. One example of a simple
process that is well suited for RPA is a process that includes the tasks of signing in to a
computer system, getting data from the system and transforming it to another digital output
eventually passing it on to another computer system (Willcocks & Lacity, 2016). This process
can be seen in Figure 3.
Figure 3: Example of a simple process for RPA (Willcocks & Lacity, 2016).
Sign in Get data
Transform
data
Move data

10
RPA automation potential increases with more routine and more manual processes. Non-
routine tasks with little recurring patterns are not well-suited for automation (Asatiani &
Penttinen, 2016). Sutherland (2013) and Willcocks & Lacity (2016) defined features that a
process has to have to increase the opportunity of being suited for RPA. Figure 4 shows an
example of these features.
RPA
Rules -
driven
Access
multiple
systems
Limited
exception
handling
Human
likely to
cause error
Operated
without
human
judgement
Important
service level
24/7
High
transaction
volume
High level
of
standardiza-
tion
Routine and
manual
High
manual cost
Figure 4: Feature of RPA processes (Sutherland, 2013), (Willcocks & Lacity, 2016).
RPA robots exist as a software that is installed on a computer. It is called a software robot due
to its operating principle (Asatiani & Penttinen, 2016). One robot is equal to one software
license (Lacity & Willcocks, 2016). The RPA software robot communicates with other IT
systems on the front-end, while other traditional software are integrated via the back-end
(Asatiani & Penttinen, 2016). This is one reason why RPA is classified as lightweight IT, but
other conventional software are classified as heavyweight IT. A more detailed explanation of

11
the RPA classification and the main differences of lightweight IT and heavyweight IT can be
found in chapter 3.1.1.
3.1.1 RPA classified as lightweight IT
According to Bygstad (2017), RPA is classified as a lightweight IT where it does not disturb
underlying computer systems while software that is integrated via back-end is classified as
heavyweight IT. The differences between heavyweight and lightweight IT are, for example,
heavyweight IT contains the knowledge to have control over large systems and developing a
solution over advanced integration while lightweight IT is a new term that controls for example
apps, sensors, and Internet-of-things, where users or vendors frequently do the deployment
(Bygstad, 2017). Lightweight IT is used to describe front-end software that is typically adopted
outside of the IT department (Willcocks & Lacity, 2016). Heavyweight IT is used to describe
back-end software where the owner is the IT department (Bygstad, 2017). Table 2 summarizes
the main difference between heavyweight and lightweight IT according to Bygstad (2017).
HEAVYWEIGHT IT LIGHTWEIGHT IT
A knowledge regime, driven by IT
professionals, enabled by
systematic specification and proven
digital technology and realized
through software engineering
A knowledge regime, driven by
competent users need for solutions,
enabled by consumerisation of digital
technology and realized through
innovation processes.
PROFILE Back-end: Supporting
documentation of work
Front-end: supporting work processes
OWNER IT department User and vendors
SYSTEM Transaction systems Process support, apps, BI
TECHNOLOGY PCs, servers, database, integration
technology
Tablets, electronic whiteboards,
mobile phones
IT
ARCHITECTURE
Fully integrated solutions,
centralized or distributed
Non-invasive solutions, frequently
meshworks
DEVELOPMENT
CULTURE
Systematics, quality, security Innovation, experimentation
PROBLEMS Increasing complexity, rising cost Isolated gadgets, security
DISCOURSE Software engineering Business and practice innovation
Table 2: Heavyweight vs. Lightweight IT (Bygstad, 2017)
Heavyweight and lightweight IT require different controls, but they are also mutually
dependent on each other (Bygstad, 2017). An RPA team always has to integrate heavyweight
IT at some part of the RPA project, for example, when it comes to setting up the computer and
getting access to heavyweight IT systems then the IT function is most often responsible for this
(Stople, Steinsund, Iden, & Bygstad, 2017). Stople et al. (2017) investigated the relationship

12
between RPA development and IT function and came to the conclusion that the connection
between heavyweight IT and lightweight should not be firmly tight, but there is a challenge
that requires more integration to heavyweight IT that needs to be considered. These challenges
are for example the maintenance of the robot and further development of RPA (Stople et al.,
2017).
RPA projects are mainly business-driven projects, but IT always has to be part of the project
to set up the underlying infrastructure for RPA (Forrester Research, 2014). RPA is designed as
a tool for subject matter experts (SMEs) to do their own automation with small IT components
rather than explaining to an IT expert how to perform all tasks that will then automate SME
processes (Lacity, Willcocks, & Craign, 2016). RPA tools can also be modified through a
simple logical statement, so it does not require a user with excellent programming skills
(Asatiani & Penttinen, 2016). Further characteristics of RPA tools are described in chapter
3.1.2.
3.1.2 RPA tools
The key to making a quick transition from a process which is done by a human to an efficient
automated process is to choose the right automation software (Mohapatra, 2013). When
organizations have decided to implement RPA they have to find out what RPA software is best
suited for the business. Around mid-2017, there were over 45 tools that were marketed as RPA
(Hindle et al., 2018). According to Hindle et al. (2018), Blue Prism is the market leader in
enterprise RPA. Other vendors are, for example, UiPath, Automation Anywhere, Workfusion,
and Advanced System concept (Lu et al., 2017)
Distinguishing “real” RPA applications is important. RPA software does exactly what it is
trained to do so there are no terms of “intelligence” in RPA software (Burgess, 2016). Machine
Learning is a subdivision of Artificial Intelligence (AI) that provides learning capabilities. It
allows computer systems to learn from data or experience, rather than by following pre-
programmed rules (The Royal Society, 2017). With AI it is possible to provide structured
outputs from unstructured inputs. Even though RPA does not have self-learning capabilities, it
can be used for the further process of the structured output from AI (Burgess, 2016). This is
one of many examples of how RPA can be used as an extension to other tools.
RPA software is sometimes confused with screen scraping tools. Screen scraping automation
tools are easier to learn and cheaper than other process automation tools, but it can sometimes

13
be hard to distinguish between screen automation and process automation tools (Lacity et al.,
2016). Screen scrapers only understand a window located in a specific location, so it relies on
X and Y coordinates. Recognizing a window defined by a location will no longer work if the
window is moved to another screen (Willcocks & Lacity, 2016). Blue Prism is an RPA tool,
not a screen scraping tool; it interacts with data through Java, Html, Access Bridge and Surface
Automation (Willcocks & Lacity, 2016). Figure 5 gives an example of a process that was
automated using robotic software from Blue Prism. The process has the five main tasks, which
are the following:
1. Log in to System
2. Open Excel sheet
3. Get data from Excel sheet
4. Put data information from Excel to Order system
5. Take note of Order reference
Figure 5: Example of the process that was automated using Blue Prism (Screenshot from Blue
Prism software)

14
Figure 5 shows the main page on process level in Blue Prism. The software includes drop and
drag icons that resemble a process (Lacity & Willcocks, 2016). The process is divided into
different steps. These steps refer to another page where more detailed information for the
specific steps can be found, which is called function level. Function level includes actions that
are displayed in object level (Stople et al., 2017). For the process in figure 5, an example of
one page in object level is Login to System. Once the RPA developer has instructed the robot
in what to do, the controller can monitor the process in the so-called Control Room in Blue
Prism (Stople et al., 2017).
3.2 Comparison of RPA and IT development via BPM
One of the main differences between RPA and IT development via BPM is that RPA is
distinguished as lightweight IT whereas IT development is distinguished as heavyweight IT
(Fersht & Slaby R., 2012). BPM requires excellent programming skills, and it uses a new
application and back-end system while RPA reuses the existing application (Forrester
Research, 2014).
It is more expensive and time-consuming to use IT development than RPA. RPA technology
is best suited for the requirement that is too short-lived to justify lengthy IT development. IT
development needs a person with excellent programming skills while RPA can be done by a
person that has only a few months of training in RPA (Fersht & Slaby R., 2012). IT automation
costs more than RPA and it is better suited for re-coding business logic or the data access layer
(Willcocks & Lacity, 2016). Primary goals for BPM is re-engineering process from as-is to to-
be while the main goal for RPA is automating as-is processes (Forrester Research, 2014). Re-
engineering processes and automating processes is not the same. Automation is about taking
the as-is process and automating it, but re-engineering is more about re-thinking and re-
designing processes. Organizations have spent a large amount of money on automating existing
process without thinking about whether the process task is really necessary. Automation should
not be implemented for a process unless the process has been re-thought and re-designed
(Mohapatra, 2013).
RPA does not have to be a competitor to other IT developments; it is more about being able to
use these tools together and to know what tools suit an organization’s needs. For example,
BPM often requires steps that have to be made by a human, but RPA can be those steps so RPA
can then be more like an extension than a replacement (Forrester Research, 2014).

15
3.2.1 BPM implementation methodology
Chang (2006) described implementation methodology for BPM that includes six phases:
commit, research, analyze, design, implement and support. The implementation methodology
can be seen in Figure 6.
Commit
Research
Analyze
Design
Implement
Support
Figure 6: Implementation methodology for BPM (Chang, 2006)
When the first phase is implemented, the company is committed to the decision to adopt BPM
from top-level executives. The changes that are implemented in the organization are cultural,
and changes must have designed and implement business improvement. The research phase is
about analyzing the existing business process and selecting a business process management
tools. The next four phases go in a cycle with the objective of implementing the project by
focusing on the small amount of the processes that were analyzed in the research phase. Once
a project is finished a new cycle starts again from the beginning. The cycle starts by analyzing.
The analysis consists of project planning and analysis of the processes that are supposed to be
implemented for the specific project. The design phase is about building a prototype and
building the best process management solution. The implementation phase is about developing
the program that is needed for process solution, testing it and then going live with the process

16
solution. The last phase involves monitoring the performance of the new process to determine
whether goals have been reached, finally preparing for the next project (Chang, 2006).
3.3 Summary
RPA is a software robot that can perform routine, standardized and rule-based processes the
same way as a human can. RPA automation opportunity increases with processes that are well-
suited for RPA. RPA is classified as lightweight IT while other traditional IT software is
classified as heavyweight IT, which also explains the main differences between RPA and IT
development via BPM. IT development is often more expensive and time-consuming than
RPA. IT development also requires a person with extensive programming skills while
automating processes with robotic software can be done by a person with only a few months
of training. The re-engineering process is the primary goal for IT development via BPM, which
is also essential to keep in mind before processes are automated to make sure that unneeded
tasks are not automated. RPA does not have to be a competitor to IT development as RPA and
other automation tools can co-exist, making an organization more efficient.

17
4 Previous research of RPA
Implementation
In this chapter, there will be a brief overview of case studies that are related to the
aforementioned research questions. These case studies are from different companies that have
successfully implemented RPA as well as companies that have implemented RPA in a client
company.
4.1 Company 1: RPA consultant service provider
Company 1 offers financial processes and outsourcing services to companies. They provide
RPA for employees to get rid of repetitive tasks of financial processing. Company 1 had a
successful pilot project and then implemented RPA internally. They have also been offering
consulting on RPA implementation to companies. The client’s implementation process starts
with a workshop, so the employee understands the potential of RPA and then reviews and maps
out the process with experts to identify potential uses for RPA. The second stage is about
creating a business case that includes information on how automation tools can increase
performance. The third stage, which is the last stage, is configuration and deployment of robots
to perform tasks that have been described step-by-step for all processes (Asatiani & Penttinen,
2016). The main stages for Company 1 can be seen in Figure 7.
Process
assessment
Business
Case
RPA
lifecycle
Figure 7: Company 1 main stages of RPA implementation
4.2 Company 2: Financial Shared Services
Company 2 implemented RPA with help from a consulting company. The first stage was a
Proof of Concept (PoC). At this stage, the company’s technical and financial ability to
implement RPA was analyzed. After the PoC they started to break down processes and redesign
processes to make it worthwhile for automation. Redesigning steps in processes can be done

18
by eliminating steps that are not needed. The next stage was to launch RPA where 19 processes
were automated in 5 months, which is about 3.8 processes per month. The RPA lifecycle
continued by automating 25 processes. The main benefits of the automation were 45 FTE
savings, so employees were redeployed to other, more important tasks, and services were
delivered faster. There were also other benefits, like increased compliance and accuracy and 9
months return on investment (ROI) (Lacity et al., 2016). Figure 8 provides the implementation
stages for Company 2.
Proof of
Concept
Process
assessment
Redesign
processes
RPA
lifecycle
4.3 Company 3: Mobile telecommunication provider
Company 3 implemented RPA with help from Blue Prism consultants. The first stage was to
eliminate the nonvalue processes, optimizing and simplifying core processes. This stage took
the company two years. In company 3 RPA will not replace BPM, as it is a tool that can be
used along with other BPM tools. After the first stage, processes that were ready for automation
were identified. The next stage is called PoC, which was implemented to check whether RPA
is financially and technically possible. Two low-complexity processes were tested for RPA at
PoC stage (Willcocks & Lacity, 2016).
Some conflicts came up regarding IT and RPA. The IT team had incorrect and negative ideas
about RPA. They assumed that Blue Prism was a screen scraper tool and they didn’t see the
reason for implementing RPA where they already had a mature BPM system. A comparison
between IT automation and RPA was performed. The comparison showed that the payback
period for IT automation was much longer than for RPA. At this point, they had already
answered that RPA had both financial and technical ability in the company (Willcocks &
Lacity, 2016).
Blue Prism helped to train two back office members, which took about 12 weeks. At stage
three, RPA was executed and 15 processes were automated in three months with a payback
period of 12 months and hundreds of FTE savings and redeployment (Willcocks & Lacity,

19
2016). The main implementation stages describe here above for Company 3 can be seen in
Figure 9.
BPM
Proof of
Concept
RPA
lifecycle
Figure 9: Company 3 main stages of RPA implementation.
4.4 Company 4: Energy Sector
The RPA journey for company 4 started by automating one process. The project failed, but
they picked a process that was not well-suited for RPA, thus, not finding the real business
value. Eight years later the actual business value was revealed. They automated about 25
processes which returned ROI of 200% within 12 months. A Center of Excellence (CofE) was
used to govern service automation capabilities; they work across the organization and
priorities, identify and develop automated processes before they go live. The first stage was
identifying processes suited for RPA. (Willcocks & Lacity, 2016).
Company 4 had firmly established Robotic Operating Model. Their structure for the
automation project started by getting a request for automation. CofE accepted or denied
processes for RPA. If they accepted process, they started defining, designing the process for
robotic software, configuring the process in robotic software and verifying the automated
process. When the automated process had been tested, it was executed and then took over
quality assurance and continuous improvement (Willcocks & Lacity, 2016).
There are a few lessons learned from this case, one of which is very important; it is integral to
send the right message to staff. If employees are scared of losing their jobs and feel like RPA
is a threat to their job security they will not welcome the benefits of fewer repetitive tasks.
There were other factors like the importance of bringing IT in early and considering RPA as a
complementary tool to other systems that are considered important for RPA (Willcocks &
Lacity, 2016).
From Figure 10, the RPA lifecycle for Company 4 after CofE had accepted the process for
automation can be seen.

20
Figure 10: RPA lifecycle for Company 4
4.5 Company 5: BPO service provider
The first stage of implementing RPA for company 5 was selecting a process to automate and
learning about Blue Prism. For the proof of concept, they took a process that was not too
complex, good for testing, and easy to measure cost before and after automation. They designed
the process and configured the robot, which took about three weeks. The next stage took two
weeks. It included quality assurance and testing the process before they went live. After stage
2, optimization of the processes was performed and more processes were automated. The main
benefits were FTE cost saving and decreasing service time. The payback period for the first
process was six months (Fersht & Slaby R., 2012). Figure 11 shows the main stages of
implementation of RPA in Company 5.
Process
Assessment
Proof of
Conecpt
RPA
lifecycle
Define
Design
ConfigureTest
Execute
Verify

21
4.6 Summary
From the case studies, the researcher concluded that the most important stages of RPA project
implementation are: Process Assessment - Business Case - Proof of Concept - Project Design
and Build - RPA Lifecycle. Proof of concept involves testing the financial and technical ability
for RPA in a specific company. If proof of concept turns out to be successful for the company,
they have proved that RPA can be a good business case. The project has to be designed and
built as company 4 had done; by building a structure for RPA with the robotic operating model.
The next phase would then be RPA lifecycle where other processes from the process
assessment are automated. For each process that is automated, it has to be defined, designed,
and developed in robotic software, tested, executed and then verified. The benefits of
automating processes were similar for all discussed case studies. Below are examples of
potential benefits:
 FTE saving
 Increased customer service – lower error rate – faster delivery
 Staff satisfaction – redeployed to more important task
 ROI
 Increased number of automated processes
 More accuracy

22
5 Dynamic Roadmap for RPA
Implementation
In this chapter, main findings and how to successfully implement RPA are presented. The
results are divided into Phase I and Phase II and in five stages, three for Phase I and two for
Phase II. These stages represent the questions the interview was built on. At the end of the
chapter, a dynamic roadmap is introduced along with a table that summarizes all factors that
were considered to be important to prevent failure. The dynamic roadmap for RPA provides a
process that offers organizations a starting point of the implementation plan for RPA.
5.1 Phase I
Here, the first phase of a dynamic roadmap for RPA implementation will be reviewed.
Interviewee 4 emphasized the importance of the first step of the roadmap, which often seems
to be overlooked. The first step involves identifying the business problem, challenges,
obstacles or tasks that make working days more strenuous. Once identified, the company can
begin looking into how to solve the problem, making RPA a potential solution:
“You should not mainly focus on RPA as a solution. It is not always the correct
solution. You have to look into what the business problem is before implementing
RPA.” – Interviewee 4
The second step is about figuring out whether RPA is the right solution for the business
problem. Once a company has decided to use RPA it has to determine what RPA software
provider should be used to automate their processes.
5.1.1 Process Assessment
Process assessment is an assessment of process opportunity for RPA and is one of the most
important factors to be able to build a good business case. Interviewee 3 described the purpose
of process assessment and what it is mainly based upon:
“First of all, process assessment is performed to identify whether there is a
potential or whether the benefits are bet to pay off for automating it. So, it is

23
based on the process assessment meeting and also the prioritized list of
potential” – Interviewee 3
5.1.1.1 Choose wrong processes for RPA
Process selection plays a crucial role in successfully implementing RPA in organizations. The
right process needs to be selected, and they must be prioritized. The choosing of processes for
RPA projects is an assessment that must be carried out by process analyst that know RPA very
well. Both interviewee 1 and 3 talk about the importance that experts in RPA process analysis
offer organizations help in choosing the processes for PoC. Otherwise, a task too difficult that
is not suited for RPA could be mistakenly selected:
“Letting the customer pick the processes by them self is likely to cause failure
(1).” –Interviewee 3
Interviewee 4 talked about what characterizes proper process analyst. Adequate process analyst
has to make sure that people understand RPA and what criteria is needed. It is also important
to look into RPA with a larger perspective. For example, compiling RPA with other tools and,
therefore, automating the entire process instead of only automating part of it. Interviewee 4
described this as an example:
“You could have one process that retrieves information from one system and then
put it into another one. Then there could be another process where another
person goes to a webpage and extract numbers, which is the information that you
receive. That’s why you need to see the bigger picture when you are working with
RPA, and that’s important in the process assessment part. (2)” – Interviewee 4
All interviewees talked about the importance of not choosing too complex processes for PoC.
Otherwise, the company risks failure. Interviewee 3 mentions that it is good to keep in mind
the success factor for successful RPA process and not to overlook process complexity. All the
interviewees recommended starting with an easy process for PoC projects because then, in
Phase II, it is possible to reuse the objects that have already been created for easy processes
and build on them. Interviewee 3 explained why easy processes should be chosen for PoC:
“We are building everything from scratch, but next time we will have the general
object, so you save a lot of time by doing that way. That’s why you should always

24
start with easy processes and build objects because then you can reuse it. (3)” –
Interviewee 3
Interviewee 3 discussed that processes which are automated in Phase II should depend on how
much experience the developer has. Processes that have the highest beneficial value should be
chosen for Phase II, but if developers have little experience, Phase II should start with processes
which are easier and then work up to a more complicated process later in the RPA lifecycle.
All the interviewees agreed that the most critical part of process assessment is to choose a
process that is suited for RPA. Interviewee 2 and 3 identify examples of processes that should
not be chosen for RPA to prevent failure (4):
 “Processes that have too many exceptions” - Interviewee 2
 “Computer applications, used by the process, that are too slow or do not
work well with RPA” - Interviewee 2
 “Processes that include human judgment” - Interviewee 2
 “Inputs that come from many different sources” – Interviewee 3
 “Standardized is very, very important. Processes that are not
standardized - for instance, if there is a free text field or the inputs come
from many different sources.” – Interviewee 3
5.1.1.2 Process assessment not carried out correctly
All interviewees agreed that process assessment should start with a presentation to the
employees in order to ensure that they understand RPA and what criteria’s are essential for
RPA processes. A survey is sent to process experts, and there must be few days between from
when the survey is sent out until the requested feedback, so the experts get enough time to think
about the processes. Interviewee 4 also talked about the importance of giving people time to
think about the process opportunity in the company by describing his/her own experience.
Interviewee 4 participated in a meeting where the RPA process criteria were introduced and,
at the same meeting, employees were given only 15 minutes to think about the processes as to
automate, which did not result in a good opportunity assessment:

25
“They need more time to think about the processes. Because they do the same
task every day and it is hard for them to know what processes they should look
into or not (5).”- Interviewee 4
Interviewee 1 talked about his own experience regarding the process assessment meetings and
noted that they could give better results in some cases if the interviews are in the employees'
native language. Interviewee 2 mentioned that an important part of process assessment is to
consider who evaluates process difficulty. It can cause incorrect process assessment when
different process expert evaluates difficulty for processes. Interviewee 2 recommends that the
same person estimates how difficult the processes are. Otherwise, the process assessment could
be overestimated or underestimated:
“A survey is a good way to select processes to look further into for automation,
but then it is important to have the same person that estimates the difficulty level
for all the processes so there will be consistency between the values. For
example, if I ask a process expert to give a grade from one to three for how
complex his process is, one process expert could find his process to be very
complicated while another process expert, which could be more used to
performing difficult tasks thinks the same process has a difficulty level of one.
(6)” – Interviewee 2
In the survey, process experts are asked to list processes that they think are suited for RPA,
how much time it takes for them to perform the processes themselves, the difficulty level, etc.
From this list, the process analyst knows more about what process expert it should interview
to get further information about RPA processes:
“From the initial meeting, we get the feel of what is extremely easy. Because then
they show us on a very high level the processes, and then we get a good feel for
what processes should be for the next phase, Phase II”. – Interviewee 3
The results from the meetings and the surveys are that processes which are considered to have
highly beneficial value for the company, therefore, recommended to be automated in Phase II.
Interviewee 1 and 3 said that the processes chosen for Phase II are processes that intersect
between high expected benefits and short development time from the process assessment.

26
Interviewee 1, 2 and 3 talked about the importance of having a very accurate assessment of
processes which are considered to be highly beneficial for the company by sitting side by side
to process experts and watch the process being executed manually. Interviewee 1 described his
own experience of having two meetings for process assessments:
“Normally, the first meeting for process assessment is just a high-level
assessment, and sometimes process owners are hearing about RPA for the first
or second time and therefore do not realize what is ideal for RPA. The advantage
of having at least two meetings is not only about getting detailed information
about the process but also increasing the number of processes that present an
opportunity for RPA. For example, when we meet the process experts for the
second time, the process experts have often thought of additional processes that
could also be ideal for RPA (7)” – Interviewee 1
Interviewee 1, 3 and 4 also described the advantage of sitting side by side to process experts,
watch them perform the process, and ask enough questions to see if there is some part of the
process that is out of the scope for RPA. Interviewee 3 talked about the importance to not just
looking for processes that take a lot of time to perform but also to look for processes that are
recurrent or have a heavy workload:
“Look for high value instead of high time. The potential for like how many FTE’s
you will automate is the time it takes to do the process times number of cases.
Better if it is a small process that probably takes 2 to 3 minutes to do it manually
but the workload cases are really high (8)” - Interviewee 3
Interviewee 1 said that the time it takes to perform a process assessment in a company depends
mostly on the size of the company and how many departments will be part of the assessment.
5.1.2 Proof of Concept
Proof of concept (PoC) enables identification of good and bad practices (Hindle et al., 2018).
Every automation project needs a PoC to test whether the tool is suitable for the project and if
it is the wrong tool or the wrong time, it may affect the project cost and cause failure in testing
automation. The test case has to be prepared for the automation test (Kumar, 2017).

27
5.1.2.1 Focusing only on technical PoC
Interviewee 4 suggested that PoC should be more about business than technology. PoC for
RPA is not only about proving that the technology works it is also about proving that RPA is
a solution to the business problem. Interviewee 3 emphasized that it is crucial to not only focus
on the development part in PoC projects, it is also important to focus on the operating model
and to link RPA to the company’s strategy. Interviewee 4 recommended that the primary focus
in a PoC should be on preparing the organization for all requirements needed for RPA based
on the process assessment:
“It has been run thousands of pilots now, and we know that the technology works
and then why spend another for this one, for technical proof of concepts we know
it is working. (8)” – Interviewee 4
All interviewees recommended 6 to 8 weeks for a PoC project, depending on whether the
processes that are automated are to be used in a production environment or only piloted in a
test environment. All interviewees recommended that moving the process from a test to a
production environment should not be part of the PoC and interviewee 3 and 4 agreed on that
only one process should be automated. However, interviewee 1 and 2 recommended
automating at least 2 or 3 processes in PoC projects. Interviewee 1 and 2 suggest this in order
to allow employees and management to build faith that the technology works. Interviewee 3
and 4 think it is more important to make sure the organization understands RPA and what is
needed to succeed with RPA than focusing on that the technology works:
“The value is not in the robot. If you are not going to proceed with this then why
should you spend time on developing two processes? We can create one to verify
whether it will work and that should be good enough before you decide whether
you want to move on or not (9).” – Interviewee 3
5.1.2.2 Lack of understanding of RPA requirements
Interviewee 4 discussed the importance of making sure the organization understands what RPA
is and what is needed to understand the requirements of RPA. Interviewee 2 talked about what
team is needed for PoC project as it depends on the goal of the project. For example, if the goal
is to train people in RPA, then companies will need someone who has good experience in RPA
and, therefore, should be able to train people. Interviewee 4 mentioned that if moving a process

28
from testing to production is part of the PoC project then it is also important to have a controller
to monitor the robot. Interviewee 3 described the most important roles for PoC project:
“[It is] Most important to have someone dedicated to the operating model,
strategy, and process assessment, then you should have someone dedicated to the
development. (10)” – Interviewee 3
Interviewee 2 talked about what happens when PoC finishes, which depends on the result from
PoC and whether the company will continue to Phase II, where more processes are automated,
and employees are trained further on how to handle the robot in production by themselves.
Interviewee 3 mentioned that a common mistake made by companies after PoC, is that they
think that they are ready to move on their own, often leading to failure. It depends on the goal
for a PoC, and what the company wants to do upon completion. It is common that employees
are trained in PoC projects, then, in Phase II, they can participate in development under
guidance. Interviewee 1 recommended, based on his own experience that Phase II should start
right after PoC:
“Ideally Phase II should start right after proof of concept, so the employees have
not forgotten the training from proof of concept (11)” – Interviewee 1
5.1.2.3 Failure to build a successful Business Case
The business case provides all necessary information that has to be known to understand
whether the project is worth the required investment. Two things are almost always a part of a
business case: a business need and a cost-benefit analysis (Project Management Institute,
2008).
Interviewee 1 described the business case for RPA as a beneficial opportunity for the company
to automate. Interviewee 2 and 4 mention that the business case is estimated on how time-
consuming the processes are, which mainly depends on the complexity of the process.
Interviewee 3 described that the complexity of the process is mainly based on which system it
uses. Interviewee 4 stated that the business case, which is estimated in PoC is a big risk
primarily due to change management. Employees can be scared of losing their jobs, and,
therefore they estimate time for their process as being lower than it actually is:
“They sometimes estimate a lower time because they are scared of losing their
jobs. We don’t know how time-consuming the processes are, we trust them, and

29
that is a big risk for us. Because we are providing a business case to them which
is based on their estimates. (12)” – Interviewee 4
Interviewee 4 continued to say that the estimate for a process’ time is normally not based on
disturbance, it is more like a perfect scenario, which is rarely the case, and thus, the process
normally takes more time than estimated:
“There is no such thing as a perfect scenario in every day work so when they say
that a process takes like 5 minutes it usually takes 15 due to questions and so on.
It is usually way more than they estimate (12).” – Interviewee 4
Interviewee 4 also pointed out that during a pilot project the business case and process
assessment is at a very high level. When PoC has been completed, it is important to continue
working on the process assessment in order to optimize processes. The process analyst needs
to look at the processes with a bigger perspective, not only look at the as-is processes that are
ready for automation but also at the to-be processes. In Phase II it is possible to get more
information about the complexity of processes and therefore more into the business case. Then
an estimation of development cost can be done. How many developers are needed is based on
RPA processes opportunity, time and whether external or internal developers are needed. The
focus needs to be on building capabilities while also being mindful of benefits realization.
Companies need to make sure they get further funding’s, especially if they need external
resources:
“In Phase II you should get more detailed about process complexity, and then
you can start thinking about how we should schedule that, should we use both
external and internal resources. Should we try to rush and make all the
development this year and start on the benefits realization or should we build up
internal capabilities and take it slower (13).” – Interviewee 4
All interviewees agreed that a good business case is crucial for moving forward with RPA.
Otherwise, the organization doesn’t find a reason to continue with it. Interviewee 4 described
what needs to be included in a successful business case:
“To have a successful business case you need to include the operating model.
You need to include the total cost, not only development, and you need the
organization structure around RPA (14)” – Interviewee 4

30
Interviewee 3 mentioned that a good rule of thumb is to have at least 3 FTE’s for a business
case to find the advantage of going forward with the RPA project. RPA requires a team, so
business cases under 3 FTE’s should never go forward:
“Blue Prism is a simple technology, but it adds complexity to the IT
infrastructure. Someone has to maintain it, and someone (a controller) has to
monitor the robots if they go down because of changes in the systems. There must
be a developer that can maintain, so if you can’t find more than 3 FTE’s then
why on earth are you going to do it. (15)” – Interviewee 3
5.2 Phase II
Phase II only takes place if the PoC meets the required business value and if there is an
opportunity for other processes to be automated with a robotic software.
5.2.1 Project design and build
5.2.1.1 RPA without an Operating Model
All interviewees discussed the importance of building a structure for RPA in a company. RPA
with no organization structure can cause failure. The importance of building an organizational
structure for RPA increases when the number of automated processes increase:
“How are you going to let RPA succeed? RPA gets more complex when the
number of automated processes that are running in production environment
increases. There are so many things that can go wrong (16)” – Interviewee 2
Interviewee 2 talked about the importance of having someone at the company that takes care
of the robot’s maintenance. This individual must know how the robotic software works and has
to have some IT background. The project manager must have participated in RPA projects and,
therefore, can help companies build an operating model for RPA. Interviewee 3 described the
operating model as everything that is not directly related to development. It is about linking
RPA to the company’s strategy and finding where to put RPA:
“So, it is how this links to the company strategy, where in the organization do we
put RPA, do we have it in multiple places or do we only have it in one place? If
a company, for example, is in many countries, it has to have several RPA teams

31
at different places, and then there is a Center of Excellence that is supporting all
of these teams. (17)” – Interviewee 3
All interviewees discussed the importance of having a project manager that has experience
participating in RPA projects. Interviewee 4 described the most critical roles that are needed
for the RPA lifecycle which are the roles that are included in the operating model. It is
important that the company has a controller before a process is put into production. Then, it
will need a manager, process analyst, developers, and the many roles needed to support the
function:
 Process owner
 Process experts
 Application owner
 IT infrastructure
5.2.1.2 RPA without Change Management
Interviewee 1 mentioned the importance of employees not being concerned about the robot
replacing their jobs. There must be a willingness to improve processes, if employees are afraid
of losing their jobs they will not willingly participate in the project. When more processes are
automated then change management becomes more necessary. Interviewee 3 discussed the
importance of having someone that takes care of change management:
“You have the management level, and then you have the operating level.
However, nothing is linking those levels, so the operation level is centralized in
building roots but is not in centralized of doing any change management, so there
has to be someone that is responsible for that (18).” – Interviewee 3
Interviewee 4 emphasized the importance of including change management in PoC. The main
reason for including change management is to prepare people for expected changes so they will
be more inclined to accept changes early on and contribute. Interviewee 4 described his
experience of a project that failed mainly due to the RPA being kept as a secret to employees
so, in that way, people got scared. Interviewee 4 described that there was no change
management from the beginning in that project, so the project failed and now, one year later,
they only have one process in production and no operating model:

32
“Change management should be included during the pilot not only after. You
should start on it during the pilot. You need it because after the pilot you will
look at benefits realization and people will be reallocated, and in that way you
need to have them on board during the pilot to make sure when the change hits
that they are ready (19)” – Interviewee 4
5.2.2 RPA Lifecycle
According to the interviews, the RPA lifecycle consists mainly of these steps:
 Design the process for development (process breakdown)
 Develop in robotic software
 Test the process
 Execute the process from testing to production
 Verify in production
 Evaluate the benefits
5.2.2.1 RPA team with lack of knowledge
Designing the process for development consists of breaking down the process into small units.
Interviewee 1 and 2 discussed the importance of having a process expert who knows the process
very well and can describe every step for all variation of the process flow. The process analyst
will then look through all cases, subdivide them into smaller parts for the development, and
gain a better understanding of how the overall picture for the process looks like:
“Important to break down the process into as small units as possible and take
enough time in process breakdown otherwise it is going to have a negative effect
on development (20)” – Interviewee 2
When process breakdown has finished, the development phase will start. Interviewee 4
mentioned that one of the main factors that leads to failure for the development is when the
process breakdown is not good enough, eventually affecting the development part and leading
to failure. Interviewee 4 also talked about the importance of the RPA developers agreeing on
the best development guidelines in order to make sure all developers follow the guidelines,
thus making it is easier for outsiders to look at the code. Interviewee 1, 2 and 3 all emphasized

33
that it is important to block all changes regarding the process flow when development has
started. They stated that the process needs to be ready for automation when the development
begins. Constant change requests from the business affects the time that has been planned for
the development part of the process. All those interviewed agreed on that the process re-
engineering or process improvement should be part of the RPA roadmap to make sure that the
process is ready for automation:
“It has to be clear that the process is ready for automation after process
breakdown and that has to be done in accordance with the process experts (21)”
– Interviewee 2
Interviewee 3 mentioned that the developer must have excellent knowledge or experience with
the RPA tool that the company has chosen to use to automate the processes. Interviewee 3
continued to describe what is needed to have excellence knowledge for development in Blue
Prism; developers need to have experience with working with Blue Prism and be certified by
Blue Prism. Developers receive a certification once they have passed the accreditation exam.
Interviewee 2 gave some examples of what can go wrong in development if developers do not
have enough knowledge to be a good developer, by describing factors that can lead to a bad
development:
“Fails to make an error messages if there are errors. Fails to have enough
decision making, for example, checks weather robot has written right text in all
text fields. Fails to take care of data missing problems and the robot writes
nothing (22).” – Interviewee 2
Interviewee 4 talked about essential part of having an objective in RPA lifecycle which is very
important to keep in mind. When automating a lot of process without having overall objectives,
it will often lead to failure long-term. To be able to get more funding and defend having a good
RPA team, there must be some value for automating processes:
“You need the management’s attention, and you need guidelines from
management on which processes to fulfill otherwise you will fail in the long-term
(23)” – Interviewee 4

34
5.2.2.2 Underestimate testing in RPA Lifecycle
All those interviewed stated that an important part of the development is to have good exception
handling. Exception handling is something that is supposed to take care of the process if
something goes wrong. For example, if the system freezes or if it is not possible to open the
system that the process uses, etc. Interviewee 1 and 3 also mentioned that it is imperative that
the system, process use, and the robotic software work well together to succeed in
development. Interviewee 3 mentioned the complexity of the process is based on the system,
whether it is java-based, windows-based, browser-based, etc. After the development in robotic
software, the testing with prepared test data will take place. Interviewee 1 stated that when
testing the process development begins, it will soon be revealed how well the process was
developed. In testing it is necessary to have good test data:
“It is very important to have good test data with plenty of side events. Then we
will see how well the exception handling has been built for the process. A best-
case scenario is to have a weekly clone of test data (24)” – Interviewee 1
Interviewee 1 also described that in testing, it is good to have someone who has not been
involved in process development to look at the code and try to run it with different test data.
This person often tests for something that the developers may not have thought about testing.
Interviewee 2 stated that testing phase is often an underestimated part of RPA lifecycle. It is
important to give enough time to test, test with different cases, and try all edge cases as well.
Interviewee 3 talked about what would most likely cause failure in the development part, which
is when the test data is not good enough:
“You don’t have representative data for what shows up in production which
means that you are not testing for everything (25)” – Interviewee 3
Interviewee 4 suggested that testing should be part of everyday work in RPA lifecycle and
should be like a formal goal checklist. As described by interviewees 2 and 3, the so-called user
acceptance test (UAT) is when the robot is tested on training systems and is then assured that
the robot meets all exception criteria. In UAT the process is tested with real-time data:
“In the UAT it is tested the exception test criteria, so you know that the robot will
do everything correctly (26)” – Interviewee 3

35
Interviewee 3 continued to describe what happens after UAT. This situation is called “soft
launch,” (when moving from testing to production). It is then important to verify that the robot
is doing exactly what it had been doing in the test environment. Interviewee then talked about
the importance of having same testing and production environment. If it is not the same, it can
cause a lot of work for the developers by changing the code and then testing it again. In
production, the process is no longer run on training systems, so it is important that it has been
tested very well:
“Soft launch is like taking it softly and smoothly in production, so you start
running cases one by one and while monitoring everything, and then you start
three cases at a time, still monitoring and then at some point we say okay it
behaves exactly the same as in the testing environment, we have tested everything
in the testing environment that’s when we are sure it is going to run smoothly in
production (27)” – Interviewee 3
5.2.2.3 RPA projects evaluation
Interviewee 4 later stated that the company implementing RPA must choose key performance
indicators (KPI) based on their overall strategy. When management knows the business
problem, they can provide guidelines on how to work with RPA to underpin the business
problem. How RPA projects are evaluated depends mainly on the objectives of the project and,
therefore, what type of processes are automated. It depends on the company’s KPI and what
factors are essential for evaluating RPA projects. Interviewee 4 concluded that evaluation
should be part of the RPA lifecycle.
Five main factors were considered to be the most common for evaluating RPA projects. These
factors are full-time equivalent (FTE), error rate, staff satisfaction, ROI and automation
potential.
Full-time equivalent (FTE)
Full-time equivalent (FTE) is used to convert hours that are worked by several employees into
hours that are worked by full-time employees. FTE is normally 8 hours per day, 5 days of
week, 52 weeks per year, but it can differ between companies (Bragg, 2017).

36
Interviewee 3 described that the potential for the number of FTE would be automated by being
calculated as the time it takes to do the process multiplied by the number of cases. Interviewee
1 talked about the necessity to take into account human efficiency when calculating FTE
savings. The robot does not get disturbed, and it does not get tired, but these two factors can
both lower a human’s ability.
Error – rate
Error-rate is supposed to decrease when processes are no longer done manually, but it also
depends on processes. Interviewee 1 described how error-rate decreases when the robot
performs the task exactly as it is supposed to without human error.
Staff and customer satisfaction
Staff and customer satisfaction are both factors that can change but are hard to measure.
Interviewee 2 stated that staff satisfaction could change by implementing RPA when
employees get rid of boring tasks and get new tasks. Satisfaction among staff should increase
in most cases, but it can also decrease, i.e., when the goal of the project is to get rid of
employees. Interviewee 1 added that staff satisfaction could increase in the long run both with
civil servant and management. Continuing, interviewee 1 mentioned that customer satisfaction
could change if the objective is to automate processes that fulfill better services, i.e., when
making sure customers are answered as soon as possible or ensuring better service quality,
without error.
Return on Investment (ROI)
According to Rohs (2004) the return on investment can be calculated using the following
formula:
𝑅𝑂𝐼 (%) =
𝑇𝑜𝑡𝑎𝑙 𝐵𝑒𝑛𝑒𝑓𝑖𝑡𝑠 − 𝐶𝑜𝑠𝑡
𝐶𝑜𝑠𝑡
∗ 100 (Rohs, 2004)
All those interviewed agreed that the typical benefit factor used to measure ROI for RPA
projects is FTE savings. Interviewee 3 also added that it is hard to measure other potential
benefits. Interviewees also agreed that the cost for RPA projects depends on how many
employees are involved in the project and whether the people participating are experts in RPA,

37
or if they need to be trained. It also depends on how many processes are going to be automated
and, therefore, how many software robots are recommended for the project. Typical cost factors
for RPA projects that are using Blue Prism for automating processes are the following:
 Software Robot (license cost) * number of robots
 Training people in RPA (FTE cost)
 RPA team (FTE cost)
o Project manager
o Process analyst
o Developers
o Process owner
o Process experts
o Application owner
 IT infrastructure (SQL server, application server, FTE cost, etc.)
Automation potential
Interviewee 2 and 3 both described the automation potential as being a common factor for
evaluating RPA projects. Interviewee 3 also added that the automation potential is a percentage
of the process that handles automation and then the rest of it will go to exception handling. One
example of this is a process that reads a standardized email from a client. If the robot can read
80% of the email, then 20% goes to exception handling and would be sent to business for
manual handling. This can happen because for some reason the robot cannot read it, suggesting
that the software is down or data is missing.

38
5.3 Summary
The main findings are summed up in the dynamic roadmap that shows the process for
successful implementation of RPA and a table that includes risk factors that can cause failure
when implementing RPA. Figure 12 shows the dynamic roadmap for RPA implementation and
Table 3 shows nine risk factors that are considered to be the most common factors that can
cause failure when implementing RPA. Table 3 also provides references for these factors.

39
Figure 12: Dynamic Roadmap for RPA implementation
Define
Design
Develop
Test
Execute
Verify
Evaluate

40
RISK FACTORS REFERENCES
CHOOSE WRONG PROCESSES FOR
RPA
- Don’t let the customer pick the processes (1)
- Need to look at bigger picture of RPA (2)
- Need to start with an easy process and then build on it (3)
- Choose processes not suited for RPA (4)
PROCESS ASSESSMENT NOT
CARRIED OUT CORRECTLY
- People need time to think about processes (5)
- It is a risk when process difficulties are estimated by different persons (6)
- Look more into high beneficial processes (High workload vs. low development time) (7) (8)
FOCUSING ONLY ON TECHNICAL
POC
- Focus on preparing the organization for all requirements needed for RPA (8)
- Organizations need to understand RPA and what is needed to succeed (9)
LACK OF UNDERSTANDING OF RPA
REQUIREMENTS
- Need to have the most important roles for RPA (10)
- There cannot be too long time that passes between PoC and Phase II (11)
FAILURE TO BUILD A SUCCESSFUL
BUSINESS CASE
- Time estimation can be a big risk (12)
- Continue working on process assessment and business case in Phase II (13)
- Need a business case with all the structure around RPA (14)
- A business case under 3 FTE should never go forward (15)

41
RPA WITHOUT AN OPERATING
MODEL
- RPA gets more complicated when the number of automated processes increase (16)
- Link RPA to organization strategy (17)
- Build a Center of Excellence that supports RPA teams (17)
RPA WITHOUT CHANGE
MANAGEMENT
- Has to be someone responsible for change management (18)
- Must make sure, when changes hit, that employees are ready (19)
RPA TEAM WITH LACK OF
KNOWLEDGE
- Take time for process breakdown (20)
- The process has to be ready for automation after process breakdown (21)
- Need developer with good knowledge otherwise it leads to bad development (22)
- Need management guidelines for automation objectives (23)
UNDERESTIMATE TESTING IN RPA
LIFECYCLE
- Need to have test data with plenty of side cases (24)
- Need to have representative data of what shows up in production (25)
- UAT – make sure the robot meets all exception test criteria (26)
- Soft launch – important to have same testing and production environment (27)
Table 3: Factors that can cause failure when implementing RPA and appropriate references

42
6 Discussion
This study aimed to answer the question of how to implement RPA successfully and what risk
factors one should beware of to avoid failure. Because RPA is relatively new term, the
researcher thought it was also important to define the RPA concept, compare it to IT
development via BPM and, finally, describe the methodology of BPM implementation for the
reader to have a better understanding of the subject.
In this chapter, the researcher will discuss the findings reliability and how it aligns with the
literature, discuss the importance of building a Center of Excellence for many RPA teams as
well as having re-engineering as a part of the roadmap, and finally provide suggestions of what
can be done for further research.
6.1 Findings Reliability
The findings are based on both literature and interviews. The researcher studied previous
research of RPA implementation and interviewed experts that have implemented RPA with
various roles. The researcher thought it was important to gain experience from various roles,
thus being able to get answers for what is important to succeed and what to beware of to avoid
failure for all steps in the roadmap. The researcher sought to hear different experiences of both
good and bad implementations and define the positive and negative parts that are involved in
the roadmap, based on expert experience. Even though the interviewees have experience with
various roles and in different companies, the main findings for the roadmap were quite similar,
thus strengthening results and findings.
If BPM implementation methodology is compared to the RPA implementation roadmap, it is
obvious that BPM focuses more on building software to automate processes, while RPA
focuses more on using a robotic software to automate processes. Despite this, the roadmap for
RPA implementation and BPM implementation methodology have much in common, also
strengthening the results.
RPA can be used in conjunction with other IT development tools just as was mentioned in
Chapter 4, by company 3. An RPA roadmap (discussed in Chapter 4) is based on previous

43
research of RPA implementation from different companies and was then defined and confirmed
by those interviewed. Then, interviewed experts suggested some minor changes. The risk
factors brought forth by the experts did not conflict with previous research of RPA
implementation. In Chapter 4 it is mentioned that company 4 experienced project failure when
they picked the wrong process for RPA upon implementation, resulting in the company not
finding value in the technology. They had also experienced the importance of change
management by sending the right messages to staff. Company 4 had built a structure for RPA
with Center of Excellence that took care of all parts in the RPA lifecycle. In chapter 4 it is
mention that companies need to be able to prove financial ability for RPA, which requires
companies to have a successful business case and they will also have to prove technical ability
for RPA. The financial and technical ability can both be proved by executing a PoC project.
The lessons learned in chapter 4 align well with the research findings in this report.
For a more in-depth analysis of the roadmap, risk factors and the common RPA projects
evaluation factors, the researcher confirms that these findings are in-line with previous
literature. Most of the findings can be recognized through the literature, and the researcher
finds that none of the findings contradict it. The researcher concluded that the results are
valuable for companies to use as a starting point for RPA implementation.
6.2 Building Center of Excellence
When the number of automated processes increase, RPA becomes more complicated. RPA
requires an operating model. The structure for RPA can consist of many RPA teams across the
organization, but then should include a Center of Excellence (CoE) that supports those teams.
According to Willcocks & Lacity (2016), the CoE works across the organization units with the
primary objects of identifying processes, priorities processes for automation and then
developing processes in robotic software before they go into production. They also take care
of controlling and monitoring the robot in production (Willcocks & Lacity, 2016).
The RPA team must then include individuals who have been trained or have experience in the
needed roles for RPA in order to make sure all parts of the RPA lifecycle will be successful.
All stages of RPA lifecycle that were previously mentioned affect each other. If the process
breakdown is not acceptable, it will affect the development stage, and bad development will
interfere with testing. If processes are not tested well enough, it will affect the process in the
production stage, often leading to failure. CoE takes care of identifying and prioritizing

44
processes for automation. Processes that are identified need to be well-suited for RPA, and
there must be some objective for automating these processes for management to be able to
defend those roles needed for RPA.
6.3 Re-engineering processes for RPA
Changes often must be made for processes before they are ready for automation. RPA is not
merely an automation tool, it also leverages new action for the future, i.e., how processes can
be further re-engineered (Lacity & Willcocks, 2016). The advantage of having process re-
engineering as part of the roadmap for RPA is quite straightforward; process re-engineering
can improve opportunities for processes suitable for RPA in the company, and ensuring the
process is ready for automation. If processes are not ready for automation, this could have a
negative effect on the development. The number of opportunities increases if processes that are
not considered suitable for RPA are considered suitable or at least part of it after re-engineering.
Automation should not be implemented for a process unless the process has been reconsidered
and redesigned (Mohapatra, 2013). When a company uses automation to mechanize old ways
of the operating processes, they leave the current process and then use technology to speed
them up. Automation may then have a negative impact on the existing process, but it can be
positive if it is deployed correctly by re-thinking and re-designing the process before
automation (Hammer, 1990).
The main idea of BPR is to look at the process from a “clean slate” perspective, then deciding
how it is possible and whether it is possible to construct the process based on improvement and
efficiency (Mohapatra, 2013). Processes that are suited for RPA must be standardized and rule-
based, not including human judgment. Re-engineering as part of the roadmap for RPA consist
of re-designing and re-thinking the process so it will be suited for the robot, thus increasing
efficiency. Estimating process re-engineering for RPA projects from as-is to to-be processes,
where to-be process is the process that is ready for automation, will require an answer to
questions like the following:
Can we make the process more standardized?
Can we eliminate human judgment in the process?
Can we eliminate steps in the process?
Can we operate the process in another way?

45
When questions about the process re-engineering possibility have been answered, and the
process has been improved, part of or the entirety of it is ready for RPA. The process changes
for RPA do not always have to be radical and take a lot of time, it can be, for example, more
convenient to only make the process standardized or simple rather than to re-design the entire
process.
6.4 Further research
Highlighting the important limitations of this study also indicated where there is a need for
further research. First, research investigating the implementation of robotic software other than
Blue Prism in greater detail is needed to compare implementation methods. Two different case
studies could be conducted that compare different implementation methods for different robotic
software’s. Second, RPA is a relatively new term, which indicates that expert experience and
RPA implementation inspection is constantly changing. To echo expert opinion, RPA
implementation requires continuous research to ensure successful implementation.
In order to research how RPA works with other tools like BPM, BPR, and Lean, the researcher
thinks it would be interesting to compare different cases where RPA is implemented in
conjunction with other tools.

46
7 Conclusion
RPA has become widespread abroad, especially in the Nordic countries. RPA is a relatively
new subject here, in Iceland, which is why the researcher found a need to study it further. The
aims of this thesis included: investigating RPA, defining RPA, assessing how to successfully
implement RPA in companies, and finally, determining the factors that may cause failure. The
chosen research questions were the following:
How to successfully implement Robotic Process Automation?
What is Robotic Process Automation?
What factors can cause failure when implementing RPA?
What factors are common for RPA project evaluation?
To answer these questions, the researcher conducted a qualitative research methodology by
surveying four experts of varying expertise. The interviewees all have experience of
implementing RPA in various roles. The researcher finished a training course in Blue Prism
and participated in a successful RPA implementation project to get more understanding of
RPA.
The conclusion of what the researcher considered crucial to achieve the objective of this
research is summarized in this chapter.
7.1 Summary of research
In Chapter 3, the research question about what is RPA was answered. For organizations, it is
critical to have some knowledge of RPA and its distinguishable qualities in order to achieve
successful implementation. The analysis of RPA revealed that it is a software robot that can be
integrated with other software in order to perform tasks the same way as a human. Processes
that are automated with RPA need to be standardized, rules-driven and routine. RPA is a tool
that is classified as lightweight IT in comparison with traditional IT development via BPM
which is classified as heavyweight IT. RPA tool is designed for SME to do their own
automation, outside the IT department but with a small IT component. Organizations do not
have to choose between RPA and traditional IT development as they can both work in
conjunction for processes that require steps made by humans.

47
In Chapter 4, previous research was summarized on RPA implementation to approach the
answers for the remaining research questions. Five different cases of RPA implementation were
studied. The most important stages of RPA implementation roadmap and the foundation for
how the roadmap was built were conducted from these cases. These stages are Process
Assessment, Business Case, Proof of Concept, Project Design and Build, and RPA Lifecycle.
In Chapter 5, the rest of the research questions were answered. Nine risk factors were
summarized and identified as key risk factors from expert interviews. A dynamic roadmap was
presented illustrating steps needed for successful RPA implementation along with a description
of what is crucial at each stage in order to prevent project failure. The researcher recommends
starting with a proof of concept project, where it is proved that RPA has the financial and
technical ability for the organization. Building a successful business case is crucial for the
management to be able to see that there is a business value for moving forward. The researcher
recommends that the proof of concept project should focus on both technical and business
sectors of RPA, primarily focusing on the business aspect to make sure companies understand
RPA and are prepared for what is needed to succeed with RPA.
After proof of concept project, companies are rarely ready to go further on its own. It will
depend on the process opportunity assessment and the business case whether management
agrees to move forward. For companies to be able to move forward with RPA, it requires
building a structure for RPA and for RPA to be linked to the organization strategy. The RPA
team is composed of a project manager, developers, process analyst, and a controller all with
knowledge and experience with the RPA process aimed to be automated. The project manager
must have experience leading RPA projects, developers are to have experience in the robotic
software, process analyst must understand the processes that are well suited for RPA, and the
controller takes care of the production aspect. For an organization with many RPA teams, it
requires a Center of Excellence that support those teams.
Companies must continuously work on the process assessment, therefore identifying,
optimizing, and prioritizing processes for the RPA lifecycle. Before processes are automated,
it needs to be ensured that the process is ready for automation. All improvement for the
processes should be made before the process is defined for development. Test cases have to be
prepared for testing with all possible side cases to make sure that the process will be ready for
production. If companies do not have the right RPA team, it can have a negative effect on the
RPA lifecycle. All parts of the RPA lifecycle are somehow linked to each other. If some part

48
of the lifecycle is not carried out successfully, it will have an adverse effect later on in the RPA
lifecycle.
Because employees may be hesitant to participate in the project out of fear of losing their jobs,
it is critical to ensure a willingness among them in order for successful RPA implementation.
When more robots are built, then change management becomes necessary. Including change
management right from the beginning is fundamental; when the changes are applied,
employees have to be prepared for them.
RPA project evaluation depends mainly on the objectives of the project and therefore the type
of processes that are automated. RPA evaluation complies with key performance indicators that
an organization has chosen, which should be based on their strategy. There were five-factors
that were considered to be most common when evaluating RPA projects. These factors are FTE
savings, error-rate, staff and customer satisfaction, ROI, and automation potential. FTE savings
is the most common factor that is used to evaluate an RPA project.

49
8 Bibliography
Alshenqeeti, H. (2014). Interviewing as a Data Collection Method: A Critical Review. Sciedu
Press, 3(1), 39–45. https://doi.org/10.5430/elr.v3n1p39
Asatiani, A., & Penttinen, E. (2016). Turning robotic process automation into commercial
success – Case OpusCapita. Journal of Information Technology Teaching Cases, 6(2),
67–74. https://doi.org/doi:10.1057/jittc.2016.5
Bragg, S. (2017, December 24). How to calculate FTEs. Retrieved March 8, 2018, from
https://www.accountingtools.com/articles/how-to-calculate-ftes.html
Bryman, A., & Burgess, R. G. (1994). Analyzing Qualitative Data. London, United
Kingdom: Routledge.
Brynjolfsson, E., & McAfee, A. (2014). The Second Machine Age: Work, Progress, and
Prosperity in a Time of Brilliant Technologies. New York, United States: W. W.
Norton & Company.
Burgess, A. (2016). Artificial intelligence and robotics: Separating reality from the hype.
Retrieved February 25, 2018, from http://www.sourcingfocus.com/site/opinions
comments/artificial_intelligence_and_robotics_separating_reality_from_the_hype/
Bygstad, B. (2017). Generative innovation: a comparison of lightweight and heavyweight IT.
Journal of Information Technology, 32(2), 180–193.
https://doi.org/10.1057/jit.2016.15
Chang, J. F. (2006). BUSINESS PROCESS MANAGEMENT SYSTEMS - Strategy and
Implementation. New York, United States: Auerbach Publications.
Creswell, J. W. (2014). Research Design: Qualitative Quantitative, and Mixed Methods
Approaches (4th ed.). California, United States: Saga Publication Inc.
DeCuir-Gunby, J. T., Marshall, P. L., & McCulloch, A. W. (2011). Developing and Using a

50
Codebook for the Analysis of Interview Data: An Example from a Professional
Development Research Project. SAGE Publishing, 23(2), 136–155.
https://doi.org/DOI: 10.1177/1525822X10388468
Everest Global. (2017). Robotic Process Automation (RPA): Technology Vendor State of the
Market Report.
Fersht, P., & Slaby R., J. (2012). ROBOTIC AUTOMATION EMERGES AS A THREAT TO
TRADITIONAL LOW-COST OUTSOURCING. Bengaluru, India: HfS Research, Ltd.
Fersht, P., & Snowdan, J. (2017, June 10). The Robotic Process Automation market will
reach $443 million this year. Retrieved January 21, 2018, from
https://www.horsesforsources.com/RPA-marketsize-HfS_061017
Forrester Research. (2014). Building A Center Of Expertise To Support Robotic Automation -
Preparing For The Life Cycle Of Business Change. Massachusetts, United States.
Hammer, M. (1990). Reengineering Work: Don’t Automate, Obliterate. Harvard Business
Review, 68(4), 104–112.
Hindle, J., Lacity, M., Willcocks, L., & Khan, S. (2018). ROBOTIC PROCESS
AUTOMATION: Benchmarking the Client Experience. Knowledge Capital Partners.
Hoepfl, M. C. (1997). Choosing Qualitative Research: A Primer for Technology Education
Researchers. Journal of Technology Education, 9(1), 47–63.
IEEE Std 2755-2017, IEEE Guide for Terms and Concepts in Intelligent Process Automation.
(2017). United States of America: IEEE Standards Association.
Kothari, C. R. (2004). Research Methodology - Methods and Technique (2nd ed.). New
Delhi, India: New Age International Publisher. Retrieved from
http://www2.hcmuaf.edu.vn/data/quoctuan/Research%20Methodology%20-
%20Methods%20and%20Techniques%202004.pdf
Kumar, B. A. (2017). Proof of concept (Poc) selenium web driver based Automation

51
framework. International Journal of Advanced Engineering and Research
Development, 4(7), 279–290.
Lacity, M., & Willcocks, L. (2016). Robotic Process Automation: The Next Transformation
Lever for Shared Services. London, United Kingdom: The Outsourcing Unit.
Lacity, M., Willcocks, L., & Craign, A. (2016). Robotizing Global Financial Shared
Services. London, United Kingdom: The Outsourcing Unit.
Lamberton, C. (2016). Get ready for robots -Why planning makes the difference between
success and disappointment. United Kingdom: EYGM Limited.
Lu, H., Li, Y., Chen, M., Kim, H., & Serikawa, S. (2017). Brain Intelligence: Go beyond
Artificial Intelligence. Mobile Netw Appl, 1–8. https://doi.org/10.1007/s11036-017-
0932-8
Middelburg, J.-W. (2017). Service Automation Framework (1st ed.). Zaltbommel,
Netherlands: Van Haren Publishing.
Mohapatra, S. (2013). Business Process Reengineering - Automation Decision Points in
Process Reengineering. New York, United States: Springer Science Business.
Project Management Institute. (2008). A Guide To the Project Management Body of
Knowledge (PMBOK GUIDE) (4th ed.). Newton Square, Pennsylvania: Project
Management Institute, Inc. Retrieved from
https://www.works.gov.bh/English/ourstrategy/Project%20Management/Documents/
Other%20PM%20Resources/PMBOKGuideFourthEdition_protected.pdf
Rohs, F. R. (2004). Return on Investment (ROI): Calculating the Monetary Return of a
Leadership Development Program. Journal of Leadership Education, 3(1), 27–38.
Stople, A., Steinsund, H., Iden, J., & Bygstad, B. (2017). Lightweight IT and the IT Function:
Experience from Robotic Process Automation in a Norwegian Bank (Vol. 25).
Presented at the NOKOBIT 2017, Oslo: Bibsys Open Journal System.

52
Sutherland, C. (2013). FRAMING A CONSTITUTION FOR ROBOTISTAN. Bengaluru: HfS
Research, Ltd.
The Royal Society. (2017). Machine Learning: the power and promise of computers that
learn by example. London, United Kingdom: The Royal Society. Retrieved from
https://royalsociety.org/~/media/policy/projects/machine-
learning/publications/machine-learning-report.pdf
Willcocks, P. L., & Lacity, M. C. (2016). Service Automation - Robots and The Future of
Work (1st ed.). Ashford, United Kingdom: Steve Brookes Publishing.

School of Science and Engineering Reykjavík
University
Menntavegur 1
101 Reykjavík, Iceland
Tel. +354 599 6200
Fax +354 599 6201
www.ru.is

Dynamic roadmap for successful implementation

More Related Content

Similar to Dynamic roadmap for successful implementation (20)

More from Srinivas Kannan (8)

Recently uploaded (20)

Dynamic roadmap for successful implementation