You are on page 1of 51

Master’s Thesis

Is Self-Service Business
Intelligence a hoax?
A descriptive study of casual users’ independence
using SSBI in the data mining process

Author: Sandra Hansson


Supervisor: Zenun Kastrati
Examiner: Anita Mirijamdotter /Päivi Jokela
Date: 2021-05-24
Course Code: 4IK50E, 15 credits
Subject: Information Systems
Level: Graduate
Department of Informatics
Table of Content
Abstract 4
Acknowledgements 4
1 Introduction 5
1.1 Background 5
1.2 Research problem 6
1.3 Gaps in the knowledge and purpose of the research 6
1.4 Delimitations of the study 7
2 Literature review 8
2.1 BI activities 8
2.2 Definition of SSBI 10
2.2.1 Make it easy to collect and improve BI results 10
2.2.2 Easy to use BI tools 11
2.2.3 Easy managed and fast deployed data warehouse solutions 11
2.2.4 Easy accessed data sources 11
2.3 SSBI tools 12
2.4 Users 13
2.4.1 Casual users 13
2.4.2 Power users 14
2.4.3 IT experts 14
2.5 User-related challenges of SSBI 14
2.5.1 Access and data usage 14
2.5.2 Low user skills 15
2.5.3 Difficulties when using tools 15
2.5.4 Difficulties creating and changing content in reports 15
2.5.5 Difficulties to assure quality of reports 15
2.5.6 No formal education 15
2.5.7 Low interest in the SSBI tools 15
3 Method 16
3.1 Research design 16
3.2 Sampling methodology 16
3.3 Data collection 17
3.4 Data analysis 18
3.5 Validity, reliability and generalizability 19
3.6 Ethical considerations 20
4 Result 21
4.1 Motivation and ambition 21
4.2 Ease of use 21
4.2.1 Easy to use SSBI tools 21
4.2.2 Creating dashboards 23
4.2.3 Data analysis and creating own reports 23
4.2.4 Efficient analysis process 24
4.2.5 Labor-intensive and time-consuming 25
4.2.6 Creating own solutions 25
4.2.7 Adapted to work tasks 26
4.2.8 Requiring new skills 26

2
4.3 Data availability 27
4.3.1 Knowledge of available data 27
4.3.2 Locating available data 28
4.3.3 Data definition 29
4.3.4 Adding missing data 30
4.3.5 Managing and filtering data 31
4.4 Data quality 32
4.4.1 Report quality 32
4.4.2 Governance of reports 33
4.4.3 Redundant reports 34
4.5 Identified user-related challenges 34
5 Discussion 36
5.1 Method 36
5.2 Result 37
5.2.1 SSBI tools are not sufficiently user-friendly 38
5.2.2 ”DIY” data preparation and collection 38
5.2.3 Whose data is correct? 39
5.2.4 Data access permissions limiting data exploration 40
5.2.5 No governance leads to poor report quality 41
5.2.6 Lack of good education and required knowledge 41
5.2.7 Dependency requirements switched focus 42
5.2.8 Summary 43
6 Conclusion 44
6.1 Conclusions 44
6.2 Contribution 45
6.3 Future Research 45
References 46
Appendices 48
Appendix A: Interview questions for casual users 48
Appendix B: Interview questions for power users/IT-experts 49
Appendix C: Informed consent 50

3
Abstract
When using Business Intelligence (BI), organizations can improve decision-making by
compiling, understanding and utilizing the data held by the organization. Self-service
Business Intelligence (SSBI) has emerged as a new focus within BI and aims to make BI tools
available to business users, and relieve IT-experts involvement of in ad hoc reporting and
analysis. The aim of this research is to examine challenges of casual users being self-reliant
when using SSBI tools, and the way they are using them in the data mining process. This was
done through a qualitative study, interviewing seven individuals from different user groups:
casual users, power user and IT-experts. From the results it appears that most casual users are
not sufficiently self-reliant in the data mining process using SSBI. The visualization is the
prime area for SSBI, which most casual users manage themselves if the data and dashboards
are pre-defined for them. SSBI is becoming increasingly more common, which leads to more
and more casual users with increased experience who need to be able to dig out their own data
for interpretation and analysis. Yet, without additional knowledge, such as data knowledge or
SQL skill, casual users are in need of support when it comes to more complex operations than
ad hoc analysis and reporting, still creating a request-response relationship to power users and
IT-experts. The major challenges, limiting casual users from being self-reliant, are: not
sufficiently user-friendly tools, poor data definition and lack of data knowledge, limited data
access, indigent validation of reports and lastly, inadequate education.
Keywords SSBI, Casual user, self-reliant, challenges, data mining process.

Acknowledgements
The author would like to thank her supervisor Zenun Kastrati, whose support has been
significant and contributed valuable insights regarding the content of the study. I also want
thank the respondents for their time and commitment in this study.

May 2021,
Sandra Hansson

4
1 Introduction
1.1 Background
Traditionally, managers have used computers directly to assist and facilitate decision making,
which is a crucial activity in a competitive market, but it has mostly been applied to well-
structured problems (Poonnawat & Lehmann, 2014). Computerized systems have entered
diverse areas of management, and most managers know that information technology is
important to their organizations and use information technology extensively. With Big Data,
the volume and variety of data have exploded and the data management has been forced to
make adjustments. Although, today the challenges don’t lay in managing Big Data, but rather
making sense out of it (Behringer, Hirmer & Mitschang, 2017).
The cornerstones of today's modern management are analytic and Business Intelligence (BI)
technologies. BI is an umbrella term that includes methods, systems, architectures, tools and
best practices that enable access to and analysis of information to enhance and optimize
decisions and performance (Llave, 2018), and is used in the whole decision-making process.
Hence, BI technologies are used to translate raw data into useful decision-making
information, such as data warehousing, data mining, visual dashboards etc. for decision
support (Behringer, Hirmer & Mitschang, 2017). Most importantly, it is necessary for
enterprises to implement BI tools to manage the data mining process, which are able to
capture, store and organize big data as well as create and present valuable insight. BI tools
together with Data Visualization software can help them monitor and manage business
activities, and also increase speed and response time (Cremona, Ravarini & Vom Brocke,
2019).
Usually, decision-makers, data clients, researchers and experts interested in the use of BI for
decision support are divided into two types: power and casual users (Behringer, Hirmer &
Mitschang, 2017; Lennerholt, van Laere & Söderström, 2020). In order to develop and
operate BI effectively, power users have the requisite technical skills. They can pick and use
data, create reports for instance, and evaluate content on their own or present it to casual users
so that they can make choices. Hence, casual users are usually organizational personnel,
managers and users with business understanding but who lack technical BI expertise and
therefore rely on pre-defined reports or dashboards produced and supplied by power users to
support their decision-making (Lennerholt, van Laere & Söderström, 2020).
Today the supply of power users, such as data analysts and data scientists, doesn’t meet the
demand, which in turn creates a bottleneck in the request–response relationship between them
and casual users (Lennerholt, van Laere & Söderström, 2020). To decrease the dependency of
power users and IT experts, a lot of organizations are looking to technology as a solution:
Self-service BI (SSBI). These tools allow organizations to still be able to compete even if
their staff does not consist of big data specialists. SSBI tools are designed to help
organizations reach their big data goals, and the emphasis on encouraging casual users to
become more self-reliant and less dependent on power users but also the IT departments, is a
key element of SSBI (ibid.). Accordingly, business change, IT backlog, and the need to be an
analytics-driven enterprise are the primary drivers of self-service BI (Imhoff & White, 2011).
Moreover, SSBI is marketed to allow customers to process data at the pace that the company
needs, and because of that self-service analytics has been a huge success. The rapid sales

5
growth of these vendors in recent years attests to the growing popularity of self-service
analytics (van der Lans, 2015). Accordingly, vendors of SSBI implies these tools allow casual
users to generate their own reports and dashboards, without the dreaded bottleneck, in a very
simple and timesaving manner. And technologies for SSBI are great, as they allow casual
users to get the reports they want by themselves, using clever drag-and-drop interfaces. They
can create dashboards that tell the story of their data, and use their own data for interpretation
and reports. Although, all this can only be done if it is prepared, extracted, stored, processed,
put at their disposal, secured, etc. by someone else. And the data are often only as good as the
team that does the preparation. So the question is: are causal users self-reliant when
performing BI activities using SSBI tools?

1.2 Research problem


Even though SSBI tools are pushed and marketed as the holy grail of the BI and analytics
market, these tools still do not fulfill the demands to make casual users fully self-sufficient, as
both power users and IT experts are still required for support in the data mining process
(Lennerholt, van Laere & Söderström, 2020). Like aforementioned, SSBI is pushed and
marketed as a tool that enterprises can conduct by themselves without the assistance of
experts. Although, these tools aren’t adequate for casual users to be fully self-reliant in the
data mining process, yet they’re becoming more commonly used in enterprises in a self-
sufficient manner (van der Lans, 2015).
Although the improved tools aim to solve the challenges of user-related problems, the
challenges are still there but might have switched focus. Accordingly, even though SSBI
offers many advantages, implementing SSBI is not as easy as expected (Lennerholt, van
Laere & Söderström, 2020). State-of-the-art SSBI studies mention problems only briefly
when implementing and using SSBI. And, much of the current reports are focusing on
different solutions to the challenges of this problem (e.g. Behringer, Hirmer & Mitschang,
2017; Poonnawat & Lehmann, 2014; Schuff et al., 2018). Moreover, it was tough finding
research focusing on mainly the user-related challenges of SSBI, yet, there are a few within
the information systems domain (Alpar & Schulz, 2016; Lennerholt & van Laere, 2019;
Lennerholt, van Laere & Söderström, 2020).

1.3 Gaps in the knowledge and purpose of the research


As there isn’t a vast amount of research conducted on the specific topic, the current studies
urge other researchers to do more reports to expand the knowledge in the area (Lennerholt &
van Laere, 2019). Literature on SSBI problems is still immature and further research is
required on the challenges of SSBI implementation (Lennerholt, van Laere & Söderström,
2020). Bani-Hani, Tona & Carlsson (2018) states that due to lack of empirical research on
SSBI, it is still unclear how SSBI tools assist casual users in participating in the conversion of
data into information.
With the immature literature, this research aims to dive deeper in the user-related challenges
that appear when using the SSBI tools, focusing on the non-technical casual users. User-
related challenges mentioned by scholars focus on casual users being limited in their use of
SSBI tools. Inadequate data access, data quality issues and the insufficient technical skill of
users, to mention a few challenges, which all have a negative effect on the casual user and
their ability to be self-reliant when using SSBI.

6
Accordingly, there are lots of opportunities to unravel the uncertainty of SSBI and how it
supports casual users in the data mining process and to find new/other challenges or simply to
verify the existing research. Descriptive studies accept the significance of context and setting,
and try to consider the participants' lived perceptions of the phenomenon under investigation.
Also, the strengths are to create detailed explanations of dynamic situations that have yet to be
or with limited attention in literature (Marshall & Rossman, 2011). Thus, the purpose is to
create a descriptive study that describes the challenges of casual users’ self-reliance when
using SSBI, as well as investigating how SSBI tools assist casual users in participating in the
conversion of data into knowledge.
RQ’s: ”In what ways are casual users self-reliant when it comes to performing BI activities
using SSBI?” and ”What are the occurring challenges for casual users when using SSBI tools
in the data mining process?”

1.4 Delimitations of the study


Since this study is focusing on the casual user’s experience and usage of the SSBI tools, the
ability of other users using SSBI tools is not interesting for this research. The research objects
are SSBI tools and how they support casual users in the data mining process. Accordingly,
there are a vast amount of different SSBI tools, but for this particular study the number has
been narrowed down to two, in regards to the scope of the research. The tools chosen for this
particular study, Tableau and Qlik, have been proven to be very popular and widely
incorporated among organizations, therefore, information about other tools will not be
brought up or discussed. The aim is to explore and describe the ways in which casual users
are self-reliant when using SSBI, hence the individuals participating are working in
organizations where these tools are already implemented, and are not in the middle of/
planning a implementation process.

7
2 Literature review
The following section presents the theory that underlies being able to answer the study
research questions. The literature review begins with a brief account of what Business
Intelligence activities is, consequently Self-Service Business Intelligence tools is presented, its
different user types, SSBI objectives and user-related challenges.

2.1 BI activities
All enterprises and organizations have visions, questions and goals. They collect the
necessary data, analyze it, and determine which actions to take to achieve their objectives in
order to answer these questions and track performance against these goals (Sharda, Delen &
Turban, 2014). Data mining has become a commonly used method for dealing with a variety
of complex business issues and opportunities. It has been shown to be extremely effective and
beneficial in a variety of situations. Many of these business data mining applications aim to
solve a pressing issue or investigate a new business opportunity in order to gain a sustainable
competitive advantage (ibid.).
Accordingly, BI includes a range of tools, applications and methodologies that allow
organizations to acquire data from internal systems and external sources, prepare the data for
analysis, and also to develop and run queries against the data (Sharda, Delen & Turban,
2014). In almost every enterprise, the journey of data from the source to user reports takes the
following steps: extraction, transformation, storage, securing, modeling, querying,
visualizing, dashboarding (ibid.). This journey is the activities carried out in the big data
chain, which is a modern data value chain that has arisen to face new challenges related to
data. Hence, in organizations that rely on big data to obtain useful insights and hidden
information, the big data chain involves four stages of data flow, which are: data collection,
data processing, data analysis and decision-making (Janssen et al., 2017).
The most widely used model is Simon's decision-making process model (Barnard & Simon,
1947) to explain and better understand how big data can be incorporated into the decision-
making process, and also how it has an effect on the organizational structure. The model has
four phases: Intelligence, Design, Choice and Implementation, and decision-makers in various
fields regularly adopt this model (Elgendy & Elragal, 2016). Thus, the big data chain
collection and processing steps are embodied in the Intelligence phase of the decision-making
process, and the main people involved in these first two steps are IT experts (Abbasi et al.,
2016). The next stage is the data analysis which is incorporated in the Design phase, where
possible courses of action are developed and analyzed through a conceptualization or a
representative model of the problem (Elgendy & Elragal, 2016). Next is the Choice phase,
which includes the decision-making, where alternatives and evaluation criteria are identified,
methods are used to evaluate the impact of the alternatives/solutions proposed (ibid.). And
lastly, the implementation phase is the phase where the chosen alternatives and solutions are
implemented.
These mentioned phases and steps are the main activities carried out in the big data data
mining process for decision-making. To increase the chances of success in data mining
projects, data mining researchers and practitioners have proposed many processes based on
best practices. These efforts have resulted in a number of standardized processes, such as
Cross-Industry Standard Process for Data Mining (CRISP-DM), which is argued to be the

8
most common one. CRISP-DM is a six-step process that begins with a thorough
understanding of the business objects and why data mining projects are required, and
ends with the deployment of the solution that met the relevant business need (Sharda, Delen
& Turban, 2014). Since data mining is motivated by experience and experimentation, the
whole process is an iterative cycle and can be very time-consuming, depending on the
problem situation and the analyst's knowledge/experience. Sharda, Delen & Turban (2014)
explains the six steps of CRISP-DM accordingly:
1. Business understanding, it is essential to create an understanding of the managerial need
for new knowledge, as well as a clear statement of the study's business goals.
2. Data understanding, the data mining process' key task is to locate appropriate data from a
variety of databases, while following the business understanding.
3. Data Preparation, or data preprocessing, is the process of preparing the data defined in the
previous phase for analysis, using data mining methods. The fact that real-world data is
typically incomplete, noisy, and inconsistent, require a large investment in this phase. The
different data processing steps of the data preparation are: Data Consolidation, Data
Cleaning, Data Transformation and Data Reduction.
4. Model building, where various modeling techniques are chosen and applied to a prepared
dataset in order to address the particular business need in this phase. The evaluation and
comparison of the different models constructed is also part of the model-building process.
The data mining task may be prediction (either classification or regression), association, or
clustering, depending on the business need.
5. Testing and evaluation, the accuracy and generality of the built models are assessed and
evaluated. This phase determines whether the chosen model(s) meet the business goals and,
if so, to what extent.
6. Employment, even if the model's goal is to perform a simple data exploration, the
knowledge obtained from that exploration must be structured and presented in a way that
the end-user can understand and profit from. Also, maintenance activities for the deployed
models may be included in the deployment phase. Since everything about a business
changes constantly, the data that reflects its operations must also change. Models based on
old data become outdated, irrelevant, or misleading over time.
Scholars state that big data require cutting-edge technology for processing, storage, access
and data analysis to make better and smarter decisions (Cremona, Ravarini & Vom Brocke,
2019). Thus, to manage the data mining process it is necessary for enterprises to implement
BI tools, which are able to capture, store and organize big data as well as create and present
valuable insight. BI tools together with Data Visualization software can help them monitor
and manage business activities, and also increase speed and response time (ibid.).
Accordingly, advanced new technologies, such as visualization tools, support the extraction of
knowledge hidden in corporate archives or archival public records (Sharda, Delen & Turban,
2014).
In addition, contemporary research suggests that communication and cooperation between
casual users and power users has been shown to be very important in improving decision-
making, as it improves the transformation of information into knowledge (Sidani & Sayegh,

9
2019). And BI tools promote collaboration and knowledge sharing during the decision-
making process (Cremona, Ravarini & Vom Brocke, 2019). Although, in this predicament the
focus is on the promises of dependence of casual users with help from technology.
Accordingly, casual users typically ingest information that is made available to them by IT
experts and power users through a request or a routine agreement between departments. As a
result, they use BI to ingest data, which they then turn into knowledge using their intuition,
prior experience etc. to make decisions and take actions. Surprisingly, BI only assists casual
users during information use during this process (Bani-Hani, Tona & Carlsson, 2018).

2.2 Definition of SSBI


Today, more businesses are adopting a new business intelligence model, which is
characterized by a self-service data approach. IT is in charge of data management (security,
accuracy, and access), allowing users to interact directly with their data. Accordingly, SSBI,
which has emerged as a new approach to BI, has the ability to increase casual users
participation by enabling them to not only consume but also author information on BI tools.
As a result, they can access data and generate information in the form of reports and basic
analytical queries without relying on power users or IT experts (Bani-Hani, Tona & Carlsson,
2018). This is the result of the boundaries between BI and the Data Science (DS) areas
blending more and more, and tomorrow's solutions will consist of a combination of BI and
DS to an increasing degree. Accordingly, DS components are embedded in the BI without the
user being aware of what functionalities are on the back of the solutions (Svahn & Ax, 2020).
Bani-Hani, Tona & Carlsson (2018) define SSBI as ”a new approach to BI that increases the
level of co-production and decreases the level of individual’s dependency during user’s
engagement with a broad range of applications and tools comprehensively embedded
throughout the process of solving an analytical task. This duality of high levels of co-
production and low level of dependency are key elements at the core of SSBI.” Although, as
aforementioned, current scholars state that due to lack of empirical research on SSBI, it is still
unclear how SSBI tools assist casual users in participating in the conversion of data into
information (Bani-Hani, Tona & Carlsson, 2018). Yet, as stated by Imhoff and White (2011)
these tools concentrate on four key objectives, which is elaborated in these following sections.

2.2.1 Make it easy to collect and improve BI results


From a business perspective, it's of utmost importance that the information is presented in an
SSBI environment in such a way that it is easy to understand. In addition to understanding the
information, SSBI must provide a BI environment that facilitates the discovery, access, and
exchange of reports and analyses in addition to understanding the data (Imhoff & White,
2011). By sharing reports and forums that allow for interactions, such as feedback on
analytical results, organizations can identify best practices of information processing. Users
must be able to customize their dashboards for data presentation and visualization so that the
information is displayed in a consistent manner for each specific situation (ibid.). Moreover,
Imhoff and White (2011) suggest without business context, SSBI is unlikely to succeed.
Specific business concepts that are easily accessible, as well as data lineage that is recorded
and registered, are required for easy to consume and enhance BI.

10
2.2.2 Easy to use BI tools
Analysis, reporting and ad hoc tools allow users to access, analyze and model business
problems, create synthetic reports and share information stored in the Data Warehouse. Not
only must BI results be simple to collect and refine, but the tools that produce them must also
be easy-to-use. When organizations transition from traditional BI reports to SSBI, the tools
need to become more user-friendly, as users do not possess the same competence as the IT
experts or even power users (Imhoff & White, 2011). Bani-Hani, Tona & Carlsson (2018)
agrees and conclude in their paper that SSBI, like other technologies, particularly SST, relies
heavily on ease of use. One of the SSBI premises is to reduce operational complexity so that a
business consumer with limited analytical or technological expertise can still use a range of
resources and participate in data analytics. Users should be able to quickly access and
translate raw data into information. To some extent, users can interact with tools and software
that are reasonably simple to use. For example, the drag-and-drop functionalities may allow
users to perform some calculations, by effectively hiding a complex operation at the back end
that the end user is not necessarily aware of (Bani-Hani, Tona & Carlsson, 2018).
Accordingly, a self-service tool's most important feature is its user-friendly drag-and-drop or
click-based interface. These tools don't require any programming language or technical skills
to use, which makes them ideal for business users.

2.2.3 Easy managed and fast deployed data warehouse solutions


An ETL process reads data from one or more databases, transforms extracted data, and loads
data into data warehouses when transferring data. The Data Warehouse gathers structured,
relational, and purified data in order to provide data in a consistent format to the organization
(Sharda, Delen & Turban, 2014). To cut costs, improve time to value, and help increasingly
extreme data processing, SSBI needs to imply examining alternative ways to distribute data.
Agile methods, software-as-a-service (SaaS) and cloud offerings, as well as analytic database
management systems, all help to achieve these objectives. When users must collect data on
their own, the SSBI tools must provide adequate efficiency and scalability for analytical
workloads and broad data sets. By utilizing SSBI, users can be directly involved in the
production of data which means that business units can distribute their own applications, to
meet their specific requirements (Imhoff & White, 2011).

2.2.4 Easy accessed data sources


Imhoff and White (2011) have uncovered that if data sources are not easy to access, SSBI
won’t take place. Unlike BI, in SSBI users get access to external data, outside of the Data
Warehouse, to be able to build analyzes and create reports. External data, such as
environmental, regional, and demographic data, can be accessed using SSBI software. SSBI
manages external and unstructured data in addition to the conventional data contained in a
Data Warehouse (Imhoff & White, 2011). All data, including email and social media, must be
completely available to all users. When a consumer gathers data for analysis, he or she has the
ability to understand the meaning, which in turn results in more accurate data that satisfies the
requirement. Also, the capacity of information workers to comprehend the full picture is
becoming increasingly important. Before SSBI can use this contextual data, it may not be
feasible or even appropriate to carry it into the data warehouse (Imhoff & White, 2011).

11
To summarize, with more easy-to-use SSBI tools, Visual Analytics/dashboards, better
performance of Data Warehouse solutions, easier access to data sources and more DS
components embedded in the BI, the causal users dependence of power users and IT experts
should decrease. However, current studies have proven that causal users of these tools still
aren’t fully independent and very often in need of support to perform BI activities.

2.3 SSBI tools


SSBI applies to all activities in the BI supply chain, from accessing source data to using BI
findings to make business decisions (Imhoff & White, 2011). Accordingly, instead of using
intuition, precedents, and conventional mindsets, self-service BI helps to create a new culture
around the use of data on a daily basis. Through visual interfaces and representations, these
tools help to to explore, make sense of, and communicate data, instead of using SQL queries
and custom scripts (Behringer, Hirmer & Mitschang, 2017; Sharda, Delen & Turban, 2014).
Technical advancements have emphasized self-service options and capabilities to increase
service delivery. Software vendors are attempting to keep up with this trend by incorporating
a self-service component into their BI offerings (Bani-Hani, Tona & Carlsson, 2018). These
tools are constantly evolving, and are becoming more user-friendly, which allows BI
applications to operate with a larger group of end-users and to switch from management-
focused decision support to easy-to-use decision support for strategic, tactical and operational
users at all levels of an organization (Poonnawat & Lehmann, 2014). A self-service tool's
most important feature is its user-friendly drag-and-drop or click-based interface. These tools
shouldn’t require any programming language or technical expertise to use, which makes them
ideal for business users.
In a Study of Swedish Business and Data Science, conducted by Svahn and Ax (2020), it was
discovered that in 2020 Tableau and Qlik were two of the leading trio in tools for reporting
and analysis. It was stated that it’s obvious that the suppliers trigger each other as the
development progresses rapidly and partly follows each other. They continued by explaining
that Augmented Analytics has been the dominant trend in recent years, i.e. that the tools
become even more user-friendly. More and more organizations utilize self-service tools,
which leads to users getting more support in finding insights (Svahn & Ax, 2020).
In general, SSBI refers to BI activities that enterprises can conduct by themselves without the
assistance of experts. The tools should provide the end-user support to reduce the complexity
without losing functionality (Behringer, Hirmer & Mitschang, 2017). In other words, the
concept behind these instruments is that they give the consumer more freedom and
responsibility to achieve total independence and self-sufficiency, resulting in BI being
decentralized within an organization (Poonnawat & Lehmann, 2014). Bani-Hani, Tona &
Carlsson (2018), who investigates the similarities between SSBI and SST, state that in SSBI,
the IT/BI department develops and maintains a service platform (SSBI) which users utilize to
service themselves. On a more comprehensive level, the user uses data analytics to draw
conclusions and make business decisions. Although, they bring up the importance of data
quality and governance, as incorrect or uneducated self-service measures in data selection and
analysis would almost certainly result in incorrect business decisions.
To be fully self-reliant in the data mining process casual users have to be able to do more than
visualizing data and creating reports. They must be capable of performing data processing,

12
such as handling incomplete data, data analysis, data correction, and data integration. And
also, to collect data, which is set in the context of the user's ability to access data sources and
combine data sources, they need to have an understanding of data sources and manipulate
data sources (Bani-Hani, Tona & Carlsson, 2018; Lennerholt, van Laere & Söderström, 2020).
Accordingly, SSBI tools are supposed to support business users in the data mining process,
Figure 1 is an illustrative visualization compiled of the business data mining process and how
the users and SSBI is implemented in the process, put together from the reviewed information
of current literature. The arrows are showing the causality and the double-arrows remarks the
iterative processes within these chains. The lines between the different chains show how they
are linked together, with the people and SSBI tools.

Figure 1: SSBI in the business data mining process (Abbasi, Sarker & Chiang, 2016;
Barnard & Simon, 1947; Lennerholt, van Laere & Söderström, 2020; Sharda, Delen &
Turban, 2014)

2.4 Users
As aforementioned, there are different users and roles when it comes to SSBI. Most research
distinguishes between three different ones: Casual users, Power users and IT experts (Alpar &
Schulz, 2016; Behringer, Hirmer & Mitschang, 2017; Lennerholt, van Laere & Söderström,
2020).

2.4.1 Casual users


Casual users, also called business users, have basic and domain-based skills (Bani-Hani, Tona
& Carlsson, 2018) and are the Information consumers (Imhoff & White, 2011). They are
usually organizational personnel, managers and users who lack technical BI expertise and
therefore rely on pre-defined reports or dashboards produced and supplied by power users to
support their decision-making (Lennerholt, van Laere & Söderström, 2020). Behringer,
Hirmer & Mitschang (2017) signifies that the casual consumer is generally satisfied with

13
predefined reports or the possibility of changing visualization or analyzed attributes.
Accordingly, Alpar and Schulz (2016) continue to explain that for performing analytical
activities, casual users are correlated with scarce information. Usually, they gain access to
already created information or only have to determine parameters before processing them.

2.4.2 Power users


Power users, such as data analysts and data scientists, are the Information producers (Imhoff
& White, 2011), who cannot only access information but also build new sources of
information. Hence, in order to develop and operate BI effectively, power users have the
requisite technical skills, such as analytical skills on how to build ad-hoc reports, dashboards
and what-if scenarios (Bani-Hani, Tona & Carlsson, 2018). Likewise, they can pick and use
data, create reports, and evaluate content on their own or present it to casual users so that they
can make choices (Lennerholt, van Laere & Söderström, 2020). This user is limited in current
approaches, having basic knowledge about data mining techniques or data characteristics but
no programming skills. They can perform a lot of BI activities through a visual user interface
(Behringer, Hirmer & Mitschang, 2017).

2.4.3 IT experts
In order to minimize the threshold for novice users, the technical experts are still relevant as
they have deep knowledge of technical problems and can therefore build new data sources or
operations (Behringer, Hirmer & Mitschang, 2017). IT experts, such as engineers and
developers, are responsible for providing the infrastructure required to operate with data and
thus serve as enablers for the use of such systems. Engineers' responsibilities range from
saving huge amounts of data in a database to data management and governance. Developers
are focused on implementing data-intensive systems in the form of software or tools to
facilitate the process of data enrichment and analysis. That involves designing, developing or
changing systems that enable data analysts or data scientists to carry out their tasks (Schuff et
al., 2018). Imhoff and White (2011) explains the role of the BI implementation team, which is
to build the infrastructure that allows data to flow freely from all of these sources. They can
then monitor data access and use, ensure the environment's optimum efficiency, enforce
acceptable security and privacy protocols, and assist the business community with the
creation or publishing of BI reports, analytics, and other reports as required.

2.5 User-related challenges of SSBI


With the objectives in mind, scholars have found some issues within the research related to
users and their use of SSBI tools. The results of user-related challenges in the case study
conducted by Lennerholt, van Laere & Söderström (2020), show several problems related to
the use of SSBI tools. Below is a short description of the identified challenges which have a
direct connection to using SSBI tools:

2.5.1 Access and data usage


To allow SSBI to run effectively, it is necessary to access data when desired and required.
According to the findings, meeting this requirement is not easy. They found several access
and data use challenges, such as: Difficulties in knowing what data sources are available;

14
difficulties in locating data; difficulties in using data; difficulties using different data sources
simultaneously; support is required to add data.

2.5.2 Low user skills


Another challenge associated with becoming a self-reliant user demonstrates how SSBI
necessarily requires skills that many users lack. Within this category, they mention three
identified challenges: Limited competence level; Difficulties when interpreting report content;
and limited general IT skills.

2.5.3 Difficulties when using tools


This category is related to being a self-reliant user, and it illustrates how users experience
SSBI tools as too difficult to use. Accordingly, casual users have difficulties using SSBI tools;
they create isolated solutions; and they feel like they don’t have the right tools.

2.5.4 Difficulties creating and changing content in reports


This category illustrates the difficulties of building and modifying content in SSBI reports. It
states that casual users have difficulties creating SSBI reports; they have difficulties changing
content; and that it requires a lot of time and manual work.

2.5.5 Difficulties to assure quality of reports


This category is also related to creating SSBI reports, and consists of four challenges. Those
challenges are as follows: Casual users have difficulties assuring quality of reports; casual
users create redundant reports; there is no governance of SSBI reports; and the use of
unsupported/other tools.

2.5.6 No formal education


Moreover, they suggest that in the concept of SSBI education, there are two categories, and
one of them is no formal education. This refers to the planning of education and provided
training to users when implementing SSBI. The challenges conducted were: that no formal
education is given; Users forget how to use SSBI; and not using the SSBI after education.

2.5.7 Low interest in the SSBI tools


Lastly, the second identified category of SSBI education has proven that organizations have
users who have low interest in SSBI. This is an issue which has a bad impact on the usage of
SSBI. The challenges which were found in this category are the following: Users do not see
the benefits of SSBI; and Users have different technical backgrounds.

15
3 Method
The methodology and approach that underpin the analysis are described in detail in the
following sections. The chapter starts with a discussion of the method's suitability for
performing this research, continuing with describing the sampling methodology and
respondents, as well as the data collection, analysis, the study's research quality in terms of
validity and reliability, accompanied by a discussion of the research ethics principles used.

3.1 Research design


The purpose of this study is to examine the challenges occurring when casual users use SSBI
tools, and to answer the RQ’s ”In what ways are casual users self-reliant when it comes to
performing BI activities using SSBI?” and ”What are the occurring challenges for casual
users when using SSBI tools in the data mining process?”. Based on the research aims, there
is a need to increase understanding of the problem area. This means the experiences and
knowledge of relevant individuals is needed. And in quantitative research, experimental
models are often unable to draw meaning and useful conclusions from experimental studies,
and that the research methods influence the outcomes. Quantitative methods have become
artifacts that hinder the natural sentiments and behaviors (Marshall & Rossman, 2011), hence
a quantitative approach does not fulfill the purpose of this study. A more suitable approach is
the qualitative which involves collecting and analyzing non-numerical data. It fits research
that seeks to understand concepts, opinions, or experiences, and can be used to gather in-
depth insights of a problem or generate new ideas for research (Creswell & Creswell, 2017).
Accordingly, this research will follow the practices of the interpretivism paradigm, which
aims to understand phenomena through the meanings people assign to them. The objectives
focus on the full complexity of human sense-making as a situation emerges (ibid.). Hence,
this study is conducted with a qualitative approach, and the starting point of the research was
the examination of respondents' personal experiences and opinions. The qualitative method is
suitable in areas without much previous research, like this one, and it also provides an
opportunity for a more open design of the research (Denscombe, 2019).
Moreover, case study is a commonly used research method within the interpretivist worldview
(Creswell & Creswell, 2017). Case studies are a model of investigation used in diverse fields,
particularly assessment, in which the researcher performs an in-depth analysis of a case, often
an activity, behavior, process, or an individual or group of people. Accordingly, the objective
of this paper is to collect detailed data about challenges that occur when casual users use the
SSBI tools and to understand how these tools support them in the data mining process, hence,
the case study is considered a suitable research method. Case studies start from a holistic
perspective and try to get as comprehensive information as possible. Case studies often come
to use to examine processes and changes. Moreover, data-driven or inductive approaches are
particularly useful when exploring new or immature terrain, and in this specific predicament
there is a lot to be explored for the qualitative researcher (Clarke & Braun, 2016).

3.2 Sampling methodology


Mack et al. (2005) mention one of the most common sampling techniques is purposive
sampling, which groups participants based on pre-determined criteria related to a specific
research question. Purposive sampling was applied in this study and the sample size, which
was not set prior to data collection, were determined by the study's goals, as well as the

16
resources and time available. Theoretical saturation is often used to assess purposeful sample
sizes. Purposive sampling is most efficient when data review and interpretation are carried out
simultaneously with data collection (Mack et al., 2005).
Also, in case studies one can choose cases that are as different from each other as possible in a
number of important respects. When selecting respondents for this study, two selection
criteria have been set up to be able to answer the research problem. By limiting the research
scope and enabling reasonable and accurate conclusions you create selection (Denscombe,
2019). The first selection criterion of this research meant that the respondent would have
experience of SSBI tools, specifically Tableau or Qlik. The motive for this selection criterion
was that the respondents would have an insight of how SSBI supports casual users in the data
mining process and what challenges arise when casual users use SSBI. The second selection
criterion was that these individuals had to be either casual users themselves, or power users/IT
experts, who have worked with or around casual users. The reason for this selection criterion
was the fact that the respondents who are different users of the tools enable the possibility of
greater knowledge gain of different views of challenges and how these tools are used. The
amount of experience and time these people have worked with these tools vary, some are
beginners and others are professionals. Accordingly, putting the different perspectives from
the people involved provides an in-depth understanding of what kind of problems can occur
while casual users are utilizing SSBI tools. Hence, multiple cases with different backgrounds
and experience supports the results validity while avoiding possible prejudices (Creswell &
Creswell, 2017). The research is based on casual users and how they use these tools. To gather
data from casual users and other users' perspectives was an appropriate empirical source to
achieve the purpose of the study and gain an understanding of the practice.
There have been seven conducted interviews with different users from five different
enterprises in Sweden, where the initial contact was made via email or through LinkedIn.
These organizations were found by initial Google search and by networking. When contacting
the respondents, they were informed about the research and a short introduction about the
study was to be conducted, with the following question if they were willing to participate with
an interview. After they accepted to participate an informed consent were sent to them, as well
as the interview protocol. All interviews were kept in Swedish and proceeded 35-50 minutes.
In the compilation of the results the interviewees are represented by a number to keep them
anonymous. The enterprises, where the respondents work, are large in size, and are within the
retail and finance industry. They use one of the SSBI tools Tableau or Qlik, which have been
implemented in their organizations. As mentioned, in section 2.2, these tools are two out of
the three most popular for reporting and analysis (Svahn & Ax, 2020). Tableau and Qlik are
interactive visualization tools providing users the opportunity to link graphical user interfaces
with drag-and-drop (ibid.). Table 1 presents a short overview of the interviewees, their
position, experience with SSBI, and the industry of the organizations they work for.

3.3 Data collection


Marshall and Rossman (2011) explain that qualitative research's strengths should be
demonstrated for exploratory or descriptive research that emphasizes the significance of
context, environment, and participants' frames of reference. A concise yet strong argument
that is strongly grounded in the study's conceptual context and justifies the particular data
collection methods should be included in a well-reasoned and persuasive justification for

17
qualitative methods. The reasoning should explain how the method selection is related to the
research questions (Marshall & Rossman, 2011).
Table 1: Interviewees

Respondent Industry Type of User Position SSBI tool Experience

Respondent 1 Retail Casual user Store manager Tableau 1 year

Respondent 2 Finance Casual user Senior associate Qlik 2.5 years

Respondent 3 Finance Casual user Anti-Money Laundering Qlik 10 months


Analyst
Respondent 4 Finance Power user Senior associate Qlik 3.5 years

Respondent 5 Finance IT-expert BI consultant Tableau and Qlik 3 years

Respondent 6 Retail IT-expert Head of BI Tableau 16 years

Respondent 7 Retail IT-expert Director of big data and Tableau 10 years


advanced analytics
technology

When conducting a qualitative study there are a lot of data collection methods to choose from.
For example one could use observations, interviews, focus groups, document, survey or
photo/video (Creswell & Creswell, 2017). Denscombe (2019) suggests that the best method
for investigating people's opinions, understandings and experiences is through interviews. The
researcher can get more detailed and deeper answers (Mack et al., 2005), than if the data
collection were to be carried out through e.g. surveys. Interviews in qualitative research are
performed by the researcher through face-to-face, telephone, or participation in a focus group
with six to eight interviewees in each group. Such interviews contain unstructured and usually
open-ended questions that are few in number and are intended to evoke the participants' views
and opinions (Denscombe, 2019). For this particular study, the key data collection technique
chosen is semi-structured interviews, since the emphasis is to define in-depth experiences of
challenges when casual users are using SSBI and to understand how SSBI tools assist casual
users in participating in the conversion of data into knowledge. The interviews were carried
out with individuals through video chat or telephone call. Considering the interviewees have
diverse backgrounds and experiences, opposed to predefined questions or unstructured
interviews, semi-structured interviews are more suitable (ibid.).
The design of the interview protocol was made in such a way so that it would provide as
many and detailed answers to the research questions as possible. It was reviewed by a
supervisor who approved the questions before the interviews were conducted. The interview
protocol was sent, one or two days before the interview, to the respondents to give the
participants the opportunity to go through the questions. This is something Denscombe (2019)
believes makes the respondents feel more relaxed and thus give truthful answers and not the
answers that they think the interviewer wants.

3.4 Data analysis


The interviews will result in non-numerical empirical material which have to be analyzed.
Hence, after the collection of data, there will be an analysis process, where the purpose is, in

18
general, to make sense of the empirical material. It includes segmenting the information and
taking it apart and putting it back together (Creswell & Creswell, 2017). In qualitative
research, data analysis will proceed conjointly with other sections of the qualitative study,
namely data collection and the writing of findings. Since text and image data are so dense and
rich, in a qualitative analysis, all of the information can not be used. Therefore, researchers
need to sift the data when reading it, a method of concentrating on some of the data and set
aside other parts of it. And then it is time to start coding the data, which is the method of
organizing data by bracketing chunks or parts and putting them in categories (ibid.).
Some qualitative studies may have a deductive orientation, whereas the research can be theory
-guided. Theory is used upfront as a guide, and sensitizing concepts can help researchers see
what they otherwise might not have noticed, as it gives a general sense of reference and
guidance in collecting data and analyzing it (Bowen, 2020). On the other hand, Thematic
analysis (TA) can be used to capture both explicit and latent meaning in both inductive and
deductive analyses, and can be applied in smaller case studies with 1–2 participants to large
interview studies with 60 or more participants. Accordingly, TA is a way to analyze both large
and small data sets in an iterative manner. TA is a method for systematically identifying,
organizing and offering insight into themes across a dataset. By focusing on meaning across a
data set, TA allows researchers to see and understand collective or shared meanings and
experiences (Clarke & Braun, 2016). Moreover, in the analysis process of this research, the
empirical material has been reviewed, compiled and criticized. The research questions were
structured and guided from current research, which formed a generally structured way of
gathering the empirical material. Accordingly, the analysis that took part of the collected
material aiming to find themes within the data, was partly deductive as the question structure
was used in the analysis process as a reference, but also seeking to add dimensions to the
current knowledge, the analysis took inspiration from TA.

3.5 Validity, reliability and generalizability


Denscombe (2019) emphasizes that it is of great importance that the chosen method of the
research has been applied correctly and that it is reliable for the result of the study to be
considered valid. Validity and reliability are critical components of any study. Ensuring that
the results are accepted as reliable and trustworthy by other scientists is especially important
in qualitative research, where subjectivity can easily cloud data interpretation and research
results are often challenged or regarded with skepticism by the scientific community (Brink,
1993).
Creswell and Creswell (2017) recommend defining and addressing one or more methods
available to check the accuracy of the findings. In this particular study a couple methods were
used to assure the validity of the results, which is essential for producing trustworthy and
believable findings in qualitative research. The use of two or more data points, methods,
investigators, theoretical perspectives, and research approaches in the study of a single
phenomenon and then validating the congruence among them is referred to as triangulation
(Brink, 1993). Triangulation was performed by generating a thorough literature review and
collecting empirical material through interviews, and affirm the conformity among the data
sources. Another method used to convey the findings was by writing rich and detailed
descriptions. The descriptions may transport readers to the scene and add a sense of shared
experience to the discussion. I.e. the findings become more realistic and richer when

19
qualitative researchers give comprehensive explanations of the environment or provide
several perspectives on a subject. This technique will improve the validity of the results
(ibid.). Also, by presenting negative or contradictory information that contradicts the themes,
adds on to the validity of the study. Since real life is made up of a variety of viewpoints that
don't always coincide, discussing contradictory information and presenting data that
contradicts the theme's overall perspective adds to an account's credibility (ibid.).
In qualitative analysis, reliability refers to the consistency of responses to multiple coders of
data sets (Denscombe, 2019). The reliability was improved by taking extensive field notes,
recording them, and transcribing the digital files. Furthermore, to assure the reliability of
qualitative research one should examine transcripts to ensure that they do not contain any
obvious transcription errors. This can be done by matching data to the codes on a regular
basis and writing memos about the codes and their meanings (Creswell & Creswell, 2017). In
this predicament the reliability was assured by conducting a thorough, iterative analysis
process.
In qualitative research, there is a different approach when it comes to generalizability as the
studies are based on a smaller number of respondents (Denscombe, 2019). Also, qualitative
generalization is a concept which refers to qualitative research in a restricted sense, since the
purpose of this type of study is not to generalize results to people, locations, or places other
than those under investigation. The importance of qualitative analysis, in reality, lies in the
precise definition and themes formed in the context of a specific situation (Creswell &
Creswell, 2017).

3.6 Ethical considerations


In order to maintain good morals, it is important to follow the basic principles on which good
research practice is based. It is required that all work done and the result produced must be
reliable. This means ensuring the quality of research, which is reflected in the design, method,
analysis and utilization of resources. Honesty is also important when it comes to developing,
implementing, reviewing and reporting and informing about research in an open, fair,
complete and objective way (Wallén, 1996).
In addition, respect must be shown for colleagues, other research participants, society and the
environment. The participants were informed about the purpose of the research, see Appendix
C informed consent, as well as their involvement in the study and how their involvement
contributes to the research. Their identity is confidential and they can at any time withdraw
their participation. The researcher takes responsibility for research from idea to publication
and for possible consequences.
The researcher is the tool of qualitative analysis. Her participation in the lives of those who
have agreed to take part in the research is essential to the methodology (Marshall & Rossman,
2011). Hence, it is also important to try to be as objective as possible as a researcher, even in
studies where values play a major role. It is important not to imprint data with one's own
perception (Wallén, 1996). Accordingly, the researcher has her own experiences,
characteristics and opinions, which characterize how she looks at things. Therefore, these
should be set aside by trying to be as objective as possible and thus not shape the study
according to one's own perception, in the role as a researcher. During interviews, the
questions asked were neutral and not leading in any way.

20
4 Result
The findings of the analysis, gathered through seven qualitative interviews, are presented in
the following sections. The aim of this section is to express the respondents' perspectives on
how SSBI tools are currently being used in organizations by casual users in the data mining
process, but also to find the challenges in these use areas.

4.1 Motivation and ambition


It appears that it is of utmost importance for companies to use new digital technology, as the
processes are becoming more complex. The competition is turning more and more to agile
methods, which leads to digitalization becoming a basic prerequisite for being able to keep up
and act quickly on things that are changing.
When asking about the motivation and ambition of using SSBI tools, the goals were clear.
The implementation of SSBI tools facilitates the analysis process and allows casual users to
accomplish a better, more accurate job. Hence, one of the main motivations is to be able to
use the full dataset and easily check all numbers of large organizations and not rely on
samples, to offer a better accuracy of reports, but also to optimize profit. Moreover, SSBI also
shortens the time it takes to get a useful report, and makes the analysis process more efficient,
as well as reducing dependence of IT-experts.
Furthermore, value creation is the most affected aspect in this context, mainly referring to the
quality improvements. SSBI greatly streamlines certain parts of casual users work, at the
same time as it increases the quality of the analysis immensely. SSBI tools allow to quickly
find what is of business value, and are therefore value-creating. Moreover, the structural
aspect comes as a ”bonus”, since all data is gathered in one place and allows for security at a
personal level.

4.2 Ease of use


4.2.1 Easy to use SSBI tools
When using these tools the casual users claimed that they encounter challenges and they need
support to move forward. They explained that the visualization feature and interface of these
tools are easy to both use and understand. Although, Respondent 2 stated that to do more than
visualizing data and creating reports, the tools require some knowledge to use, and that there
will be bottlenecks, since they only have five people in the whole company who know the
tool at that level.
”Jag tycker det är svårt att använda. Det kräver en del kunskap, speciellt om man ska bygga
modeller och om man vill ändra någonting. Jag måste alltid fråga om hjälp, t.ex. ’Jag kunde
inte ladda in datan, kan du se vad det är för fel?’” (I find the tool difficult to use. It requires
some knowledge, especially if you want to build models and if you want to change something.
I always have to ask for help, e.g. ’I couldn’t load the data, can you see what’s wrong?’) -
Respondent 2
Another thing he mentioned was that the overall feeling in the enterprise when they had to
work with Qlik was ”Oh no, not Qlik”, which says a lot as well. Respondent 1 continued on
the same track as he mentioned that he thinks it’s difficult to use SSBI because he didn’t get a

21
proper education. The tool has too many functions, making it hard to know what exists in the
tool, and the information overload also makes it tougher getting a grip.
”Jag har tänkt flera gånger på saker som skulle vara jättebra om jag kunde göra, sen har jag
tänkt på det en stund, bara för att senare inse att det var superlätt att göra” (I have thought
several times of things that would be great if I could do, then been thinking about it for a
while, just to be proved later it was super easy to do) - Respondent 1
The power user and IT-experts stated that the interface of visualization and creating reports
are easy to grasp and don’t demand much skill, and also that the various functions for
visualization were simple to use and highly valued. They all claimed that SSBI tools are
relatively easy for casual users to use.
”Det bästa med verktyget är att det är mycket visuellt, du kan använda cirkeldiagram etc.
vilket ger en helhetsbild på ett annat men bättre sätt. Det möjliggör enklare sätt att hitta
enstaka transaktioner” (The best thing with the tool is that it is very visual, you can use pie
charts etc. which gives you an overall picture in a different but better way. It allows for easier
ways to find single transactions) - Respondent 4
Accordingly, the power user and IT-experts all agreed that casual users have grasped the
visualization feature quite well, and the overall agreement is that visualization is the prime
feature of these tools. It is efficient and it does it nicely. Yet, Respondent 4 continued to
explain that it is an issue how these tools are marketed as easy to use, that they’re supposed to
be built to be user-friendly, yet five to six people who work only with these tools are needed
in a company, which says a lot.
”Oftast är det datan som är svår när det kommer till byggandet och det kommer man aldrig
undan. Inte med något SSBI verktyg. Datan är ju komplex och man behöver rensa den,
felsöka osv. Men de flesta casual users visualisera om man ger dem data och tabeller de
behöver” (It is usually the data that is difficult when it comes to construction and you can
never get away with it. Not with any SSBI tool. The data is complex and you need to clean it,
troubleshoot, etc. But most can visualize if you give casual users the data and tables they
need) - Respondent 4
According to Respondent 5, depending on the organization's data maturity, common problems
may be that the visualization tool lacks the flexibility found in Excel, or it may be that casual
users feel that someone more technically savvy needs to confirm how the data is supposed to
be interpreted. He continued by mentioning that SSBI tools are relatively simple, and that it is
very important for those who have the technical skills to show how easy it is, which was
brought up by more respondents. Respondent 7 agreed and also mentioned that there is a
certain threshold to get started, and that it is important to have both ongoing training and
available support for users.
Although, on the contrary, Respondent 5 went on to describe that one mistake that developers
do is making the development more laborious than it needs, i.e. that they overwork a
technically complex solution, which actually is too hard for casual users. He also brought up
UX design, which he states is underestimated when it comes to SSBI development.
”UX är så underskattat, jag vill få in mer UX-fokus i SSBI-utveckling, så det är definitivt
något jag försöker konkretisera i mitt arbete” (UX is so underestimated, I want to bring more

22
UX focus into SSBI development, so it is definitely something I try to concretize in my work) -
Respondent 5

4.2.2 Creating dashboards


New dashboards are created when new data is entered in the tool but also when you want to
display the same data in a different way. Casual users without SQL skill and data knowledge,
like Respondent 1 and 3, can not make their own dashboards.
”Nej de finns skapade redan åt oss. Det är inget jag råder över eller kan skapa själv” (No, the
dashboards are already created for us. It's not something I can control or create on my own) -
Respondent 1

Yet, Respondent 2, who seemed to have a little more knowledge of the backend architecture,
mentioned that he can create his own dashboards but says that it is really hard to do.
”Om du vill skapa en lite mer komplex dashboard så kräver det mer kodning, och backend är
hemskt. Men enklare dashboards är lättare att skapa och är något jag kan göra på egen
hand” (If you want to make a slightly more complex dashboard, it requires more coding, and
the backend is awful. But simpler dashboards are easier to create and are something I can do
on my own) - Respondent 2

The experiences from the power user and IT-experts were similar as they stated that they have
not seen casual users create their own dashboards to a large extent. It requires additional
technical knowledge and data knowledge for a good result.
”Även om de kan skapa sina egna dashboards blir de inte bättre än vi kan skapa datakällor.
Det är lätt att gå vilse i att aggregera en redan aggregerad siffra. Och då är man fast, för du
måste ha bra koll på vad din grunddata var” (Even if they can create their own dashboards,
they do not get better than we can create data sources. It's easy to get lost in aggregating an
already aggregated number. And then you're stuck, because you have to have a good idea of
what your basic data was) - Respondent 6

4.2.3 Data analysis and creating own reports


All casual users were unanimous, stating that performing analysis is something they/casual
users can do with SSBI. Respondents 1, 2 and 3 stated that the analysis process is very easy
and these tools are pedagogic, which let them easily explore the data.
”Jag kan absolut analysera data och skapa egna rapporter. Innan Tableau var vi begränsade,
och det ärliga svaret är att om jag ville veta något om mina kunder, var jag tvungen att fråga
IT-personalen på analysavdelningen. Jag skickade ett mejl om vad jag ville veta och sedan
skickade de ett mejl tillbaka med en rapport” (I can definitely analyze the data and create my
own reports. Before Tableau we were limited, and the honest answer is that if I wanted to
know something about my customers, I had to ask the IT-staff at the analytics department. I
sent an email about what I wanted to know and then they sent me an email back with a report)
- Respondent 1

The power user and IT-experts were all on the same page as they stated that casual users can
use these tools for analyzing data and creating reports. Respondent 4, 5 and 7 simply said that

23
this is something they have seen them do and that is working well. Accordingly, there was a
unanimous perception that casual users also can create reports on their own.
”Alla casual users hanterar analysdelen, visualiseringen är enkel och tydlig. Visualiseringen är
den främsta egenskapen i SSBI” (All casual users handle the analysis part, the visualization
is simple and clear. Visualization is the main feature of SSBI) - Respondent 5

4.2.4 Efficient analysis process


The respondents all agreed that the analysis makes their work more effective. Since casual
users can do the analysis themselves with SSBI, they don’t have to wait on IT-experts or
power users to do it for them.
”Ja, det är mycket effektivt, det hjälper oss i vårt arbete väldigt mycket. Den faktiska
dataanalysen tar oftast inte alls lång tid, beroende på hur komplicerat caset är” (Yes it is very
effective, it helps us in our work very much. The actual data analysis usually doesn’t take long
at all, depending on how complex the case is) - Respondent 3
Another thing that was enlightening was that the end product is very good, but the way there
is difficult. Accordingly, Respondent 2 mentioned that the analysis will be very effective
when they have a finished model, but the model takes some time to create. He explained that
they spent a lot of time, almost a year, developing a model for a company. But after the model
is done it saves them a lot of time.
”Vi skulle förmodligen inte ha kunnat göra den typen av jobb på så kort tid utan den här
modellen. Så nu är vi beroende av Qlik på det sättet, för utan det tar analysen mer än tre
gånger så lång tid” (We probably wouldn’t have been able to do that kind of job in such a
short amount of time without this model. So now we are dependent on Qlik in that way,
because without it the analysis takes more than three times as long) - Respondent 2
Respondent 1 brought up that the minor analysis goes faster, since he immediately can go into
the tool and check the data himself instead of having to wait three days to receive an email
from the data analysts. Yet, now when he has to perform the work himself he thinks it’s not as
efficient, because specialists are better than him at producing and analyzing data, and the
larger analyzes take longer because he doesn’t always know what to look for.
”Om jag t.ex. vill kontrollera hur mitt kundunderlag har förändrats, var de bor och var
försäljningen ligger, vet jag inte var jag ska börja. Det tar ett tag att ta reda på hur jag ställer
in alla parametrar för att få det jag vill veta. Och så går något fel och jag måste börja om. Jag
tror att de som har kunskap i verktyget gör det snabbare” (E.g., if I want to check how my
customer base has changed, where they live and where the sales are, I don’t know where to
start. It takes me a while to figure out how to set all the parameters to get what I want to
know. And then something goes wrong and I have to start over. I think that if you have
knowledge of the tool it goes faster) - Respondent 1
Henceforth, the power user and IT-experts think that the analyzing process is more efficient
for casual users when using these tools.
”Du kan snabbt hitta saker som tar mycket lång tid att hitta i vanliga finansiella siffror eller
bokföringsprogram. Du får en mycket snabb översikt över KPI:er etc.” (You can quickly find

24
things that take a very long time to find in regular financial figures or accounting software.
You get a very quick overview of KPIs etc.) - Respondent 4
Accordingly, Respondent 7 explained that the analysis process definitely is more efficient
with SSBI tools, and that it solves one part of the large area that analytics is. It provides a
flexibility that is close to the user and the business. But, he continued to mention that one
must take into account that it can never replace good data management and structured work
around data storage and processing.

4.2.5 Labor-intensive and time-consuming


Respondent 2 and 3 had only worked with Qlik which is their current tool and had nothing to
compare with. The same applied to Respondent 1, but he did say that Tableau still was quite
extensive, compared to other tools.
”Vi använder inte andra verktyg men jämfört med de som används i uppföljning av
butiksekonomi är det rätt omfattande, deras verktyg känns effektivare, men de är utformade
för att endast kolla på försäljning, marginaler osv.. Det är färre klick och inte lika mycket
information runt omkring” (We do not use other tools but compared to those used in store
finance follow-up, it is quite extensive, their tools feels more efficient, however, they are
designed to only look at sales, margins, etc. There are fewer clicks and not as much
information around) - Respondent 1

When asking IT-experts and the power user, they agreed that SSBI tools are time consuming
and labour intensive. The amount of manual labour required are more than other tools,
especially if casual users want to do more than visualizing and reporting.
”Den färdiga dashboarden tar inte längre tid att använda än en standardrapport. Men om
casual users använder SSBI-plattformen för nya analyser hela tiden tar det lång tid, det är
analytikernas uppgift” (The published dashboard does not take longer than a standard report
to use. But if casual users use the SSBI platform for new analyzes all the time, it does take
time, that’s the analytics’ responsibility) - Respondent 7
Respondent 6 mentioned that it was a much bigger uphill than they thought with the self-
service parts. And that right now the biggest threat to getting casual users to use the tool is
that they have so much to do with their daily tasks, and that it takes a while before they get
back the investment they make in a tool. They should gain time using SSBI instead of other
tools, e.g. Excel.

4.2.6 Creating own solutions


The participating casual users stated that they do get stuck using SSBI, and when they do they
use other tools or ways to fix the problem. Apparently various problems occur quite often, and
the respondents explained different solutions they take to, when running into difficult
situations. Respondent 3 mentions that they very seldom take help from power users, and
instead they try to figure out solutions themselves when they encounter problems.
”Om vi inte hittar en transaktion som vi letar efter måste vi titta i de andra systemen, eller så
frågar vi kunden. Det händer att vi inte hittar saker vi behöver i Qlik, men det är alltid möjligt
att lösa problemen på andra sätt” (If we do not find a transaction that we are looking for, we

25
have to look in the other systems or we ask the customer. It happens that we do not find things
we need in Qlik, but it is always possible to solve the problems in other ways) - Respondent 3
On the contrary, Respondent 1 states that he doesn’t use any other tools when he gets stuck,
and doesn’t create his own solutions. Instead he asks for help from power users or IT-experts.
”Om jag fastnar skickar jag ett mail och frågar personalen på analysavdelningen om de kan
producera datan/informationen åt mig” (If I get stuck, I email in and ask if the people at the
analytics department can produce data/information for me) - Respondent 1

When talking to the power user and IT-experts they all mention that it happens all the time,
casual users create their own solutions. As stated by Respondent 4, technological development
is progressing and individuals have to keep up. But some people may find it a little difficult to
do things themselves in SSBI. Moreover, it has been shown that Excel is without a doubt the
main escape route casual users take when they get stuck, and they think it will continue to
happen.
”Ja, vid minsta problem går de tillbaka till Excel. Det är deras utväg” (Yes, at the slightest
problem, they go back to Excel. It's their way out) - Respondent 6

4.2.7 Adapted to work tasks


The casual users all mentioned that they thought their tools were customized for their work
processes, they could perform the analysis they required in their work. But they also
explained that there were a lot of functions that they didn’t use, and that they felt like they
only used a small part of the tool.
”Den är byggd så att du kan använda den i fler områden och inte bara i mitt arbete. Det finns
saker som jag inte har nytta av men det spelar ingen roll” (It is built so that you can use it in a
few more areas and not just for my work. There are things that I have no use for, but it doesn’t
matter) - Respondent 1

The IT-experts and power user all said that the tool isn’t adapted for the casual user’s work
tasks. Respondent 7 brought up that the tool is not adapted from the beginning, and that it
requires that casual users learn how to work with them to be able to integrate it into their work
processes. Respondent 4, 5 and 6 unanimously stated that the tools are quite general, but that
dashboards and models are created to fit their work processes. In that way it’s customized,
since they are made after their needs.
”Vi tar hänsyn till deras arbetsprocess. Vi producerar data som är helt unika för deras analyser
och arbetsprocesser” (We take into account their work process. We produce data that is
completely unique for their analyzes and work processes) - Respondent 6

4.2.8 Requiring new skills


All casual users agreed that SSBI requires knowledge of the tools, which was obvious.
Accordingly, Respondents 1 and 3 have only had to learn the tool and its functionalities. They
mentioned that they have done so through basic training, which didn’t give them much, and
stated that they learned more on their own, by clicking around and exploring the tool.

26
”Vi fick tid att lära oss verktyget. Det finns träningstillfällen så snart någon behöver, men
utbildningen var inte bra alls. Jag säger till nya kollegor/nybörjare att klicka runt och utforska
på egen hand istället, för det är det bästa sättet att lära sig” (We got time to learn the tool.
There are training sessions as soon as someone needs it, but the education was not good at
all. Instead, I tell new colleagues/beginners to click around and explore on their own, because
that’s the best way to learn) - Respondent 3

However, they are not able to do more than ad hoc analysis and reporting, unlike Respondent
2. He mentioned that he can process, load and add data on his own, but it required basic SQL
skills which he had to learn. The coding skills were also needed to create dashboards.
”Jag förväntade mig inte att jag skulle behöva lära mig att skriva grundläggande SQL-kod
innan jag började arbeta som revisor. Jag hade inte den färdigheten när jag började, men nu är
det en liten del av mitt jobb. Andra saker som jag var tvungen att lära mig är hur man
använder modeller, hur den byggs, samt hur man filtrerar på ett visst sätt för att få specifika
detaljer och hur man växlar mellan olika instrumentpaneler. Allt var helt nytt” (I did not
expect to have to learn to write some basic SQL code before I started working as an
accountant. I did not have that skill when I started, but now it's a small part of my job. Other
things I had to learn are how to use models, how it is built, as well as how to filter in a certain
way to get specific details and how to switch between different dashboards. Everything was
brand new) - Respondent 2

When asking the power user and IT-experts they explained that these tools require some new
skill to be able to use them to their full potential. Respondent 4 explained that these tools are a
little more technically complicated than other solutions, but it’s for a reason.
”Om man vill analysera data behövs bara visualiseringsverktyg, där alla kan göra vad som
helst. Om man däremot vill göra fler saker behöver man en grundläggande förståelse för vad
data är. Om man bara vill kontrollera rapporter kan man göra det i Excel eller med hjälp av
diagram” (If you want to analyze data, you only need the visualization tool, where anyone can
do anything. If you want to do more things, you need a basic understanding of what data is. If
you just want to check reports, you can do it in Excel, or using charts) - Respondent 4

Accordingly, he stated that to be able to use the tool, users must have some understanding of
data, the rest of them agreed that it’s the main new skill required. To summarize, it is required
that casual users learn to use the tool and above all learn more about how data and the
underlying data warehouse works. Yet, to retrieve and locate certain data requires SQL skills
that casual users do not have.

4.3 Data availability


4.3.1 Knowledge of available data
In this study, most of the casual users agreed that they know the data they need are in the tool,
and what data sources it’s coming from.
”Jag skulle säga att jag vet vilken data som är tillgänglig, åtminstone den data jag behöver
använda i mitt dagliga arbete och jag vet var jag hittar den. Men det händer ändå att jag måste

27
fråga hur jag kan hitta olika data” (I would say that I know what data is available, at least the
data I need to use in my day-to-day business and I know where to find it. But it still happens
that I have to ask how I can find different data) - Respondent 1

Unanimously they also mentioned that they know there is other data available but not what
kind, and they wouldn’t know what to use it for. They also agreed on that with more training/
experience working with the tool they could probably find and use more data.
As a power user, Respondent 4 said it’s common to receive questions from casual users such
as: Is this data available? This was also something IT-experts have encountered, and it’s clear
that casual users have a lot of problems knowing what data is available, and often reach out
for support. Accordingly, the casual users usually only have knowledge of a small part of the
data sources available, but sometimes not even that.
”Det finns definitivt de som inte förstår verktyget eller vilken data som finns tillgänglig även
om det är tydligt och skapat efter deras behov” (There are definitely those who do not
understand the tool or what data is available even though it is clear and created according to
their needs) - Respondent 4

4.3.2 Locating available data


According to the casual users, they have quite different experiences about locating available
data. Respondent 2 mentions that the data sources they use come from their clients, which he
adds data to the tool himself, but he has to write some SQL code to do it. Respondent 3
mentioned that she can get to the data she needs and knows where it comes from. Although,
she has never explored data outside of the dashboard they prepare for them. On the contrary,
Respondent 1 explained that he felt a little restricted by not knowing how to locate and fetch
available data. He also brought up that laws and regulations limit him in his work.
”Det finns data som jag inte känner till. Eller snarare, jag vet att den finns men jag vet inte hur
man får tag i den. Det finns också data som jag inte kan ta fram på grund av GDPR. T.ex. om
jag vill skicka erbjudanden till kunder som har barn i åldern 0-2 år kan jag inte ta reda på var
de bor” (There is data that I do not know about. Or rather, I know that the data exists but I
don’t know how to get it. There is also data that I can not produce, due to GDPR. E.g. if I
want to send offers to customers who have children aged 0-2 years, I can not find out where
they live) - Respondent 1

When speaking to IT-experts and the power user, they all mention that this is a major issue
and is something they work with a lot, trying to make casual users able to gather their own
data. All of them agree that they partly can locate the available data. It was also brought up
that it depends on which data source they want to connect to, and how well the data is
documented.
”Vi har kallat vissa datakällor för self-service datakällor så att alla mätvärden och parametrar
står på svenska och det finns förklaringar för de olika mätvärdena, och de vanligaste
kalkylerna har vi redan förberett. Det behövs bara drag-and-drop för att använda dem” (We
have named certain data sources self-service data sources so that all measurement values and
parameters are in Swedish and there are explanations for the various measurement values,

28
and have already prepared the most common calculations. All you need to do is drag-and-
drop to use them) - Respondent 6

Yet, even with the thorough preparation of data and data sources, some casual users still aren’t
able to locate the data they need. They have difficulty with traceability of data, and the reason
for this is suggested to be partly due to the lack of metadata. Accordingly, as a power user,
Respondent 4 said it’s common to receive questions such as "Where can I find this particular
data?” from casual users, which show that they do not know how to find the data on their
own. Correspondingly they can’t be self-reliant in the data mining process, if they can’t
process or collect data, but it’s not uncommon that casual users think that SSBI will lead to
them becoming completely independent and able to do everything self-sufficient in the data
mining process when implementing SSBI.
”Det är inte ovanligt att casual users tror att SSBI kommer att leda till att de blir helt
oberoende och kan göra allt på egen hand, dvs. gå hela vägen från ett datalager/källsystem till
analys eller rapport på egen hand. Men det är något jag inte har sett än” (It’s not uncommon
that casual users think that SSBI will lead to them becoming completely independent and able
to do everything on their own, i.e. going all the way from a data warehouse/source system to
analysis or report on their own. But that is something I haven’t seen yet) - Respondent 5

4.3.3 Data definition


The casual users mentioned that the data in the SSBI tool is clearly defined, for the most part,
and the descriptions are easy to understand. It is visualized in columns, which makes it easy to
see what data is available, which they think is pedagogical. Yet, they also agree that some
parts are not very clear. Some data lacks further explanation which is crucial for the analysis.
”Det finns vissa ställen där det inte riktigt är tydligt om det är försäljning inklusive eller
exklusive moms” (There are some places where it is not really clear whether it is sales
including or excluding VAT) - Respondent 1

According to the power user and IT-experts, the definition of data is a major challenge when
users access data on their own. Differences in definitions can appear when using the same
word, hence companies have started to make definitions visible and train users to increase
their understanding of interpreting data depending on where they get it from.
”I detta område kan jag se att casual users har tagit ett steg framåt, för diskussionen är nu i stil
med ’vems data är korrekt’, istället för ’det finns inga data’” (In this area, I can see that
casual users have taken a step forward, because the discussion is now in the style of "whose
data is correct”, instead of "there is no data”) - Respondent 5

In other words, it’s important to be extra clear when it comes to defining the data for casual
users to understand it. But sometimes it doesn’t matter how specific and careful they are when
naming columns, or how thoroughly they mention details of the data in model specifications,
some casual users still don’t know how to use it. Respondents 6 and 7 said they encounter
mines almost daily regarding definitions, because there’s so many definitions and business
rules hidden in the concepts, making it difficult for casual users to know how to use them
correctly. Accordingly, they all agreed that it’s very important to ensure clear definitions for

29
the data, so that two independent individuals can draw the same conclusions when looking at
the same dataset, but it’s easier said than done.
”Förmågan att tolka data beror på hur tillgängliga definitionerna är, och om man har skapat
tillräckligt bra förutsättningar för att kunna tolka och förstå data. Det är vanligt att personer
som är bekanta med data har långa diskussioner om definitioner, så det är inte alltid helt
rimligt att förvänta sig att casual users enkelt ska kunna tolka och förstå data” (The ability to
interpret data depends on how accessible the definitions are, and if you have created good
enough conditions to be able to interpret and understand data. It is common for people who
are familiar with the data to have long discussions about definitions, so it is not always quite
reasonable to expect users to be able to interpret and understand the data easily enough) -
Respondent 5

4.3.4 Adding missing data


Casual users have problems adding data by themselves in the tool. And the overall perception
to this question was that they can’t add missing data, because they don’t know how to do it or/
and don’t have permission to get to the data source.
”Verktyget är styrt, jag kan bara använda det som finns tillgängligt. Jag kan filtrera, analysera
och klippa i data men inte lägg inte till något som inte redan finns” (The tool is controlled, I
can only use what is available. I can filter, analyze and cut data but not add something that
does not already exist) - Respondent 1

When casual users need more data they either have to ask power users/IT-expert for help.
Respondent 3 can not add data herself, but says she never had to ask or know anyone who has
asked for more data.
”Vi lägger inte till våra egna data i systemet, vi använder bara verktyget för att granska och
analysera den data som redan finns” (We do not add our own data to the system, we only use
the tool to review and analyze the data that is already there) - Respondent 3

On the other hand, Respondent 2, who possesses basic SQL knowledge, can add data to
simpler models, but need support when there are bigger errors or more complex data models.
According to the power users and IT-experts, casual users who are used to consuming data
often want to dig deeper, or combine data with other data that is not available, but it's not
something they can do on their own. Adding data is something you have to learn, since it’s
more advanced than visualization, hence a need for power users arises. Casual users can’t add
missing data without SQL knowledge.
”Det är inte så vanligt att en casual user kan hitta data direkt i ett datalager, det kräver ofta
SQL kunskaper, och tabellerna har vanligtvis tekniska namn och är lite svåra att tolka för
någon som inte är van vid det språket. Data i BI-verktyg finns vanligtvis i datamodeller, men
det är förmodligen mer en kombination av säkerhet/dataskydd som hindrar dem från att
komma åt ett datalager” (It is not very common that a casual user can find data directly in a
data warehouse, it often requires knowledge of SQL, the tables usually have technical names
and are a little difficult to interpret for someone who is not used to that language. Data in BI

30
tools is usually found in data models, but it is probably more of a combination of security/
data protection that prevents anyone from accessing a data warehouse) - Respondent 5

Accordingly, Respondent 6 also mentioned that one reason for casual users not being able to
add missing data is because they don’t have permission.
”Utanför huvudkontoret ger vi bara access till data per Apotek, dvs. deras egna butiksdata,
vilket är det mest relevanta” (Outside the head office, we only provide access to data per
Pharmacy, i.e. their own store data, which is the most relevant) - Respondent 6

On the contrary, Respondent 7 mentions that some casual users can add data by themselves,
but if they do they are responsible for the data being useful.

4.3.5 Managing and filtering data


Filtering is a way of cutting the data, and users almost always manage it themselves. It is
important to be able to visualize the data in different manners, i.e. aggregating data in slightly
different ways. All casual users expressed that it is possible for them to filter the data to adjust
the content of the tool depending on what they want to look at. Yet, there is a limit to what
they can filter on. A unified response mentioned that to filter on something more complex
they need support.
”Om jag till exempel vill titta på kontantbetalningar kontra inkomst för en viss månad då
måste jag be om hjälp” (For example, if I want to look at cash payments vs. income for a
specific month then I need to ask for help) - Respondent 2

Although, as mentioned by the casual users, there still is some manipulation which is too
difficult for them to handle. E.g. the more complex filtering requires creating new dashboards,
which is not something casual users commonly can do.
Manipulating the data in simpler ways is one of the main features of SSBI, all IT-experts
agree on this. SSBI developers have managed to produce simple interfaces making it very
easy to filter and group the data in different ways. The view of the filtering can be sent to
someone else, so they can see exactly the same report. Hence, filtering and communications
about the findings are areas where self-service shines/stands out.
”Inuti rapporter kan användare välja att bli av med outliers och filtrera på viss data, de kan
spara filtret och det blir då standardinställningen. Filtrering och visualisering är det som sätter
dessa verktyg i framkant. Man kan också dela dina rapporter, lägga till kommentarer i
rapporterna osv. Du kan filtrera på flera sätt, spara den som en vy och sedan dela den med
någon annan, så kommer de att se exakt vad du tittade på. Därför är det väldigt enkelt att
kommunicera om resultaten i data” (Inside reports users can choose to get rid of outliers and
filter on certain data, they can save the filter and it will become the default setting. The
filtering and visualization is what puts these tools at the forefront. Also you can share your
reports, add comments in the reports etc.. You can filter in several ways, save it as a view, and
then share it with someone else, and they will see exactly what you were looking at. Hence,
communicating about the findings made in the data is very simple) - Respondent 6

31
4.4 Data quality
4.4.1 Report quality
The casual users mention that they do trust the data contained in the tool. And because of that
they believe the reports they make are of good quality. Yet, the greater part of casual users
suggest quality check reports before using them.
”Jag skulle säga att jag alltid säkerställer kvaliteten på mina rapporter med power users innan
jag skickar iväg något, även om rapporten korrekt i 95% av fallen.” (I would say that I always
ensure the quality of my reports with power users before sending anything away. The report is
correct in 95% of the cases) - Respondent 2

Accordingly, Respondent 3 mentions that sometimes errors occur when she’s performing
analysis in the tool, and it forces her to double check the results with the other tools or reports
that they use. This makes her question the quality but not because of the data, but because of
the underlying model and functionality that’s not working correctly.
”Ibland vet jag inte vad som orsakar felen. Uppgifterna i Qlik är inte alltid tillförlitliga och det
händer att siffrorna är helt fel” (Sometimes I don’t know what causes the errors. The data in
Qlik are not always reliable, and it happens that the numbers are completely off) -
Respondent 3

IT-experts have seen that the reports aren’t always made with good quality. There seem to be
different underlying factors causing this to happen. In the enterprise where Respondent 6
works, he states that only half the reports released are of high quality, which quickly create
big gaps between high and low quality. Respondent 7 mentions that the freer dashboards,
where the responsibility lies entirely with the business, have a greater spread of quality.
Moreover, Respondent 5 mentions that the quality aspect is very individual, and that the value
of a created report could be different depending on who you ask. E.g. there are dashboards
that may not be so good from a developer perspective, but the business can see a very high
value in that particular layout.
”Jag har arbetat med att utse en ’företagsägare’ för varje SSBI-lösning som utvecklas, och den
personen bestämmer vilket material som anses vara bra för affärsändamålet. Vår ambition
med SSBI har alltid varit att utvecklare och analytiker ska ha så lite engagemang som möjligt
i resultaten” (I have worked on appointing a "business owner" for each SSBI solution that is
developed, and that person decides which material is considered good for the purpose of the
business. Our ambition with SSBI has always been that developers and analysts should have
as little commitment to the results as possible) - Respondent 5

On the other hand, Respondent 4 mentions that the analysis casual users do is done properly
and comes out good. If the in-data is right the model will work and generate good quality
reports.
”När det kommer till analys har det m.h.a. dessa verktyg blivit möjligt att minska
tidsåtgången och kostnader men också höja kvaliteten betydligt jämfört med vad som har
gjorts tidigare. Så de gånger de fungerar bra är kvaliteten mycket hög men det finns också

32
brister” (When it comes to analyzing, with the help of these tools it has become possible to cut
a lot of time and costs and also raise the quality significantly compared to what has been
done before. So the times they work well, the quality is very high but there are also
shortcomings) - Respondent 4

4.4.2 Governance of reports


The casual users all had different experiences when it came to governance of reports.
Respondent 1 trusts that the data added in the tool is correct, and that the reports he creates
are correct. No-one checks the reports he makes, but he still uses them as a basis for decision-
making.
”Jag litar på att den datan som finns i verktyget är rätt. Ingen kontrollerar rapporter utan vi
använder det som vi skapar” (I trust that the data contained in the tool is correct. No one
checks reports but we use the ones we create) - Respondent 1
Accordingly, Respondent 2 stated that in their company they have basic directives, steps to
follow to ensure data is right and used correctly. Although, there’s no strict governance to
quality check reports. Yet, he mentioned that even if everything is correct he still checks the
report with a power user before sending it off to a client, to ensure the quality. Respondent 3
doesn’t make reports in Qlik, but she did mention that everyone works differently, there’s no
directives on how to use the tool.
The power user and IT-experts all agree that reports are not verified sufficiently. Respondent 4
mentioned that the competence level has been too low to be able to question models and
analyzes, even though it is very important that the reports are correct.
”Det har inte varit tillräckligt hög nivå av styrning och kvalitetssäkring” (There has not been a
sufficiently high level of governance and quality assurance) - Respondent 4
Respondent 6 explained that they do basic tests on the data before they release it, but
sometimes when products are added in the wrong category it can affect several KPI’s, which
can hit hard on reports. Therefore it’s essential to establish governance of data, metadata and
reports to ensure high quality, but that’s not usually the case, merely half the time. In a perfect
world only reports of good quality would be accepted. Thus today, there’s no time or
resources for validation. Respondent 7 mentioned that the reports that require high quality are
examined and monitored to ensure quality, but reports made freely by casual users are not
verified to the same extent.
”Rapporter som behöver vara av hög kvalitet undersöks och övervakas för att säkerställa
kvalitet. Men friare dashboards, där ansvaret helt ligger hos verksamheten, har en större
spridning av kvalitet och blir inte validerade i samma utsträckning” (Reports that need high
quality are examined and monitored to ensure quality. But freer dashboards, where the
responsibility lies entirely with the business, have a greater spread of quality and does not get
validated to the same extent) - Respondent 7
Respondent 5 mentioned that the data in the tools are correctly prepared and therefore the
reports can be trusted, without having to check each and every report. The governance falls
under checking that the pre-made dashboards are correctly created. Also, to minimize IT-
experts involvement in ad hoc reporting and analysis, he stated that they remove the

33
responsibility from developers to govern reports, without consideration to enterprises data
maturity.
”Det enklare alternativet är att öka kundens behörigheter och låta dem skära i det irrelevanta.
Affärsområdet är kundens ansvar och de har bättre kunskap om vad som är relevant” (The
easier option is to increase the client's permissions and let them cut in the irrelevant. The
business area is the clients responsibility, and they have better knowledge in what is relevant)
- Respondent 5

4.4.3 Redundant reports


All casual users mentioned that they create the same reports, but make smaller changes in the
data that they want to look at. Hence, they stated that they don’t create redundant reports, they
make similar reports but tweak the data.
”Jag skapar samma rapporter ganska ofta, med mindre förändringar, för om jag vill skapa en
rapport igen beror det på att jag tror att något har förändrats eller att jag vill lägga till
parametrar. Så det händer att jag skapar samma rapport flera gånger men de är aldrig
identiska” (I create the same reports quite often, with minor changes, because if I want to
create a report again, it's because I think something has changed or that I want to add
parameters. So it happens that I create the same report several times but they’re never
identical) - Respondent 1
When asking the power user and IT-experts there’s a common perception that casual users do
create the same reports on the same dataset.
”I ett projekt var vi två utvecklare som utvecklade en SSBI-lösning som hade 110
återkommande användare. Bara cirka 5 av dessa blev helt oberoende power users och vi
hamnade i en situation med 70 rapporter om samma dataset” (In one project we were two
developers who developed an SSBI solution that had 110 recurring users. Only about 5 of
these became completely independent power users and we ended up in a situation with 70
reports on the same dataset) - Respondent 5
Respondent 6 mentioned that casual users use the most popular reports and create their own
variations, but not dangerously frequent. Yet, as users gain more experience, they will make
their own variations of existing reports to a greater extent, but they will also have the
knowledge of finding and looking at already made reports instead of redoing the same work
all over.

4.5 Identified user-related challenges


While the results show that SSBI enables casual users to create ad-hoc reports and allows
them to perform simpler analyses of the data through data visualization and uncomplicated
filtering, there are a lot of challenges and hinders which restrains them from using SSBI in a
self-sufficient matter in the data mining process. Bellow, in Table 2, is a summarization of the
user-related challenges, which have been discussed in this section.

34
Table 2: User-related challenges

Ease of use • Not adapted to work-processes


• Difficult to use and demand a lot of work
• Can’t create own dashboards or complex filtering
• Can’t locate data or add missing data
• Can’t perform data processing or data collection
Data availability • Data access permissions
• Demanding data understanding and technical skill
• Un-sufficient data definitions
Data quality • Poor report quality
• No governance

35
5 Discussion
The study's findings are examined in the following parts based on previous studies, such as
that presented in the literature review. The similarities and differences between theory and
practice are discussed in this section.

5.1 Method
For the purpose that the study and question intend to answer, the qualitative method was best
suited, as the conditions that the survey aims to explore required to elicit experiences and
views through conversations with respondents. The qualitative method is best suited to the
studies that examine these types of cases in order to gain a deeper understanding in the field
(Denscombe, 2019).
The choice of method allows for the results of the collected empirical material to answer the
research question. The semi-structured interviews give the researcher the chance to steer the
questions in a direction that provides relevant answers to the research, at the same time as it
opens up opportunities for interesting further discussion during the interview (Denscombe,
2019). Yet, the use of multiple data collection methods and sources enhances the credibility of
results and allows for the use of various interpretations and definitions in data analysis.
Multiple data collection methods are commonly used in case study studies, and data is
gathered from a variety of sources (Creswell & Creswell, 2017). Accordingly, implementing
more data collection techniques could improve the collection of empirical material and lead to
greater results than by only conducting interviews. E.g. observations can generate data that
confirms the respondents' answers and prove that the interview material was truthful. Also,
the observations could gain additional data relevant to the study. Moreover, qualitative data
collection methods are more suitable for this study than e.g. surveys that are better suited
when there is a need for standardized data or uncomplicated information (Denscombe, 2019).
Thus, by examining source evidence and using it to construct a logical justification for
themes, this process can be argued as adding to the validity of the study if themes are
identified based on the convergence of multiple sources of data or perspectives from
participants (Creswell & Creswell, 2017). Furthermore, the literature review serves a variety
of research purposes. It validates the study's eventual results in a narrowly descriptive study
and supports the relevance of the research’s emphasis (Marshall, & Rossman, 2011).
The respondents hand-picked are different users and have different experience of SSBI. The
different perspectives from different user groups did provide a wider and deeper
understanding of what kind of problems can occur while casual users are utilizing SSBI
tools. The research is based on casual users and how they use SSBI tools. To gather data from
casual users and other users' perspectives was an appropriate empirical source to achieve the
purpose of the study and gain an in-depth understanding of the process. The number of
respondents in this research is only seven, and there wasn’t enough participants for saturation.
In qualitative analysis, saturation is a key concept. It is used to assess whether a study's data is
sufficient to establish a comprehensive and accurate understanding of the phenomenon under
study (Brink, 1993). More respondents would most likely lead to more data, which could be
generated into more knowledge of the area. Also, there was only one power user participating
in this study, which was not the intention from the start. To get an equal amount of answers
from each respondent group would have given even more trustworthy data. Moreover, the

36
areas of industry where the respondents work were limited to retail and finance. Accordingly,
this research is mainly pointed at the enterprises in those areas, limiting the relevance of this
research in other industries.
Generalizability is based on the statistical probability that a certain aspect of data will be
found elsewhere, and that it is a probability that depends on large samples as representative of
a wider population. In qualitative research, there is a different approach when it comes to
generalizability as the studies often are based on a smaller number of respondents
(Denscombe, 2019). The participation of all respondents provided relevant empirical material
that helped to examine casual users’ usage of these tools to find challenges and use areas,
which the study focused on. Accordingly, the participants' response helped the survey to
provide answers to what the study intended to investigate.
The analysis process, which was a mix of deductive while taking inspiration from TA, helped
me as a researcher to create an understanding of the data collected. The theory foundation was
a support in creating the questions for interviews, to make sure there were no important parts
that got left missing, whereas the TA inspiration aided in the identifying, organizing and
offering insight into themes across a dataset. Moreover, in qualitative research, it is more
difficult not to have an impact on the survey as a researcher than it is in quantitative studies.
Because they are a product of an interpretation process, it has consequences for the possibility
of objectivity (Denscombe, 2019). As a researcher, I have tried to be objective in my work by
analyzing the material produced from interviews and observations with an open mind and
putting my own experiences, characteristics and opinions aside. However, it is difficult to
completely avoid coloring data when you as a researcher in qualitative studies are involved in
the study (Denscombe, 2019).

5.2 Result
Vendors of SSBI implies these tools allow casual users to generate their own reports and
dashboards, without the dreaded bottleneck, in a very simple and timesaving manner, and
improve time to value (Imhoff & White, 2011). Hence, users should be able to quickly access
and translate raw data into information (Bani-Hani, Tona & Carlsson, 2018). And it’s not
uncommon that casual users think that SSBI will lead to them becoming completely
independent and able to do everything self-sufficient in the data mining process, i.e. going all
the way from a data warehouse/source system to analysis or report on their own. But the
results show that this is not the case. SSBI provides a flexibility that is close to the user and
the business. But, one must take into account that it can never replace good data management
and structured work around data storage and processing.
It’s essential to get a better and deeper understanding of the motivation and ambition,
underlying factors, of implementing SSBI in enterprises where the respondents work, to
understand the usage of these tools. Accordingly, following is a detailed examination of the
results, including explanations of the significance of the conclusions. To put the results in
context, the discussion is focusing on the use areas and main challenges of SSBI found in the
empirical material and relating them to the objectives and established challenges, which are
compiled in the theory section.

37
5.2.1 SSBI tools are not sufficiently user-friendly
One of the objectives is Easy to use SSBI tools, and not only must BI results be simple to
collect and refine, but the tools that produce them must also be easy-to-use. When
organizations transition from traditional BI reports to SSBI, the tools need to become more
user-friendly, as users do not possess the same competence as the IT experts or even power
users (Imhoff & White, 2011). SSBI relies heavily on ease of use. One of the SSBI premises
is to reduce operational complexity so that a business consumer with limited analytical or
technological expertise can still use a range of resources and participate in data analytics.
Users should be able to quickly access and translate raw data into information. (Bani-Hani,
Tona & Carlsson, 2018).
Accordingly, the results show SSBI solutions of today have gotten far when it comes to the
analysis itself, and created pedagogical Data Visualization tools, and also achieved easy
reporting solutions for the business user. With Data visualization software, the perception is
that casual users can perform ad hoc analysis on their data and create reports, with higher
quality performance. The prime of SSBI tools is the visualization, as well as the way casual
users can filter the data independently in different ways to gain valuable insight. Also, simpler
interfaces, such as drag-and-drop, allows for some casual users to explore data sources and
manipulate the data in a more complex way independently.
These tools should not require any programming language or technical skills to use, which
makes them ideal for business users. Yet, research shows that SSBI tools are difficult to
understand and use for nontechnical casual users (Lennerholt, van Laere & Söderström,
2020). The previous findings are in-line with the results from this research. SSBI tools solve
one part of the large area that analytics is, but not all of it. Hence, there still are areas of SSBI
which creates a barrier between casual users and performing data mining tasks. The tools are
not adapted to casual users’ work from the beginning, and casual users are required to learn
how to work with them to be able to integrate it into their work processes. Hence, as it looks
today, the users need to adjust their work processes to the tools, not the other way around. The
overall feeling is that casual users think these tools are difficult to use and demand a lot of
work, which pushes them away from SSBI, and instead use tools creating their own solutions.
There’s also several operations they can’t perform on their own, making them dependent on
support. E.g. to perform more tasks in the data mining process they need additional skills,
such as more data knowledge and SQL skills. Without these additional skills they can’t create
their own dashboards, perform complex filtering, locate data sources, add missing data, etc.,
which is not ideal and how SSBI is being marketed. These tasks are today being performed by
or together with IT-experts and power users.

5.2.2 ”DIY” data preparation and collection


To cut costs, improve time to value, and help increasingly extreme data processing, SSBI
needs to imply examining alternative ways to distribute data. When users must collect data on
their own, the SSBI tools must provide adequate efficiency and scalability for analytical
workloads and broad data sets. By utilizing SSBI, users can be directly involved in the
production of data which means that business units can distribute their own applications, to
meet their specific requirements (Imhoff & White, 2011).

38
To be fully self-reliant in the data mining process casual users have to be able to do more than
visualizing data and creating reports. They must be capable of data processing, such as
handling incomplete data, data analysis, data correction, and data integration. And also, to
collect data, which is set in the context of the user's ability to access data sources and combine
data sources, they need to have an understanding of data sources and manipulate data sources
(Bani-Hani, Tona & Carlsson, 2018; Lennerholt, van Laere & Söderström, 2020). This
contradicts the results of this particular study. The Data Warehouse solutions of today are well
developed and sophisticated, providing for easy management and fast deployment for data
mining. Technical users can utilize these efficient and scalable systems in a simple manner.
Yet, casual users, who do not possess technical skill, such as SQL knowledge, don’t have the
ability to use data sources to explore or gather data for analysis in a self-sufficient manner.
Accordingly, when people discuss what distinguishes SSBI platforms from their conventional
counterparts, immersive data visualizations for business users are typically at the top of the
list. Data visualization, on the other hand, is just part of the story. Data integration and
preparation, is the real technological challenge in allowing Do-It-Yourself (DIY) analytics for
business users.
And the results show that casual users who are used to consuming data often want to dig
deeper, or combine data with other data that is not available, but it's not something most of
them can do on their own in these SSBI tools. As the rapid sales growth of these vendors in
recent years attests to the growing popularity of self-service analytics (van der Lans, 2015),
there will be more and more casual users with more experience who need to be able to dig out
their own data for interpretation and analysis. And therefore the next step is to incorporate
self-service data preparation to enable business users to mix and match data sources
interactively, instead of relying on power users or IT-experts to bridge data silos. Hence, with
the current SSBI solutions, casual users have problems when it comes to accessing, locating
or adding data on their own. They’re not able to be self-reliant in the whole data mining
process, and require more user-friendly tools to be able to go all the way from a data
warehouse/source system to analysis or report on their own, without the help or support from
power users/IT-experts.

5.2.3 Whose data is correct?


Another objective of SSBI is to make it easy to collect and improve BI results. From a
business perspective, it's of utmost importance that the information is presented in an SSBI
environment in such a way that it is easy to understand. In addition to understanding the data,
SSBI must provide a BI environment that facilitates the discovery, access, and exchange of
reports and analyses (Imhoff & White, 2011). Users must be able to customize their
dashboards for data presentation and visualization so that the information is displayed in a
consistent manner for each specific situation (ibid.). Moreover, Imhoff and White (2011)
suggest without business context, SSBI is unlikely to succeed. Specific business concepts that
are easily accessible, as well as data lineage that is recorded and registered, are required for
easy to consume and enhance BI.
The results show that the data and results are not always easy to understand for casual users,
and that it is an area in need of improvement. It appears that casual users are gaining more
experience, wanting to dig deeper in the data, because the data usage and understanding
appears to move forward as the discussion is now in the style of "whose data is correct”,

39
instead of "there is no data”. The ability for casual users to interpret data depends more on
how accessible the definitions are, and if the conditions are good enough to be able to
interpret and understand data. It is common for individuals who are familiar with the data to
have long discussions about definitions, therefore it’s not quite reasonable to expect casual
users to be able to interpret and understand the data easily enough. Accordingly, one of the
largest challenges IT-experts and power users are facing when it comes to SSBI is to make
casual users understand the data. Data added in pre-made dashboards etc., seems to be clearly
defined and not as big of a problem as when casual users start to access data on their own,
then the definition of data is a major challenge. Differences in definitions can appear when
using the same word, so companies have started to make the definitions visible and train users
so they know how to interpret the data depending on where they get it from. It is a lot of
work, yet very important to define the data in such a way that casual users can understand it.
The goal is to ensure that there are clear definitions for the data, so that two independent
individuals can draw the same conclusions when looking at the same dataset. Thus, it seems
that SSBI solutions are incorporating business context to a large extent, with minor flaws
referring to business rules.
On the other hand, in compliance with the objectives, as shown in the results, collecting BI
results is simple using SSBI. Even if data understanding also can be a limitation when sharing
reports between casual users, the communication of data findings made in the data is very
simple and something that has made casual users less dependent on power users when it
comes to ad hoc reporting and analysis. The communication of reports, together with the
visualization and filtering options embedded in the tools have proven to also be one of the
major parts that puts SSBI in the forefront, casual users can look at the data in different ways
themselves and share their outcome. Although, depending on the organization's data maturity,
common problems may be that the visualization tool lacks the flexibility found in Excel, or it
may be that casual users feel that someone more technically savvy needs to confirm how the
data is supposed to be interpreted.

5.2.4 Data access permissions limiting data exploration


Easy accessed data sources is another objective of SSBI, and Imhoff and White (2011) have
uncovered that if data sources are not easy to access, SSBI won’t take place. Unlike BI, in
SSBI users get access to external data, outside of the Data Warehouse, to be able to build
analyzes and create reports. All data must be completely available to all users, as the capacity
of information workers to comprehend the full picture is becoming increasingly important
(Imhoff & White, 2011).
Governing access, through permission limitations, is a way to ensure security/data protection,
as you can control who can access what in a data warehouse. Yet, the security has a backside
when business users are limited in what data they can access and use, leaving them unable to
explore all data sources, opposing the objectives of easy access data sources. It’s a tricky
situation figuring out how to have open data sources and still keeping the data safe.
Like aforementioned, casual users who are used to consuming data often want to dig deeper,
or combine data with other data that is not available, and the permission limitations to data
sources are often an obstacle hindering them from doing that on their own. From the study of
Lennerholt, van Laere and Söderström (2020) they saw that users of SSBI who are aware that
data is available have difficulty locating it and have difficulties knowing who to contact in

40
order to get access. This is in-line with the results of this research, which proved that casual
users are having difficulties locating data and don’t have access to data sources. Accordingly,
out of the results there seems to be a bigger concern of keeping the data safe than making it
available to casual users. Laws and regulations, such as GDPR, is also an obstacle when it
comes to data access, limiting the casual users of having all data available to them. The
limitations don't allow them to be self-reliant in the data mining process, and in turn it does
create data silos.

5.2.5 No governance leads to poor report quality


As aforementioned, a key element of SSBI is to encourage casual users to become more self-
reliant and less dependent on power users, but also the IT departments. But to effectively
enforce and use SSBI, good governance is needed, similar to the data governance that
organizations have been implementing for years (Lennerholt, van Laere & Söderström, 2020).
In line with the challenge of SSBI report governance (Lennerholt, van Laere & Söderström,
2020), the finding displays that enterprises have a hard time finding a balance when it comes
to governing casual users’ use of data and created reports. As more and more developers hand
over the responsibility of governing reports to enterprises, to relieve themselves of the
involvement in ad hoc reporting and analysis, it comes with a cost of report quality.
Depending on the data maturity in the enterprise, the quality of reports will have more or less
spread of quality. The bigger spread of quality refers to the freer dashboards created by casual
users, when casual users have access to data sources, and where the responsibility of report
governance lies entirely within the business. Casual users don’t have adequate knowledge to
validate reports as well as power users and IT-experts, to make sure the data in reports is
being used correctly. Insufficient validation can lead to severe consequences, a small mistake
can ruin reports and make them unusable or even have enterprises make decisions on faulty
information, which support the findings in Bani-Hani, Tona & Carlsson (2018) study who
discuss the importance of quality and governance. The low level of quality in reports
generated by casual users is a concern to enterprises, and it becomes a problem when SSBI
makes casual users self-reliant at the cost of good quality of reports.
Accordingly, this predicament creates a vicious circle. When the issues regarding reports not
being verified sufficiently creates room for quality deficiencies, the solution is for validation
of casual users self-generated reports. Which in turn creates a need for support, and that is
exactly what IT-experts are trying to move away from.

5.2.6 Lack of good education and required knowledge


SSBI education has been listed in the literature as a challenge. SSBI is often misunderstood as
being simple, when in reality it’s not. Enterprises implementing SSBI should recognize that
comprehensive training is required when casual users with minimal technical and analytical
skills must learn to perform complex data analysis tasks (Lennerholt, van Laere &
Söderström, 2020). Accordingly, the results show that SSBI tools require thorough and
continuous training for casual users to be able to use them in a self-reliant manner. Yet, in
compliance with the challenges of no formal education (Lennerholt, van Laere & Söderström,
2020) found that casual users lack the thorough education and training required to be self-
sufficient in the current SSBI solutions. The lack of education may hinder SSBI
implementation, causing them not to have enough knowledge to perform the required tasks in

41
the tools. Enterprises have not understood the importance of education, that they need to make
sure casual users have the best foundation to comprehend and optimize the use of SSBI tools.
Accordingly, to leave the responsibility of governance to the business users, and for them to
be able to locate and add data from data source/data warehouse, they need more education in
data definition and greater data knowledge. It would help them understand and perform each
step in the data mining process better, and also how to utilize these tools, use and analyze
data, and create reports in an optimal way. Moreover, the way SSBI tools are marketed,
vendors mean they shouldn’t require any programming language or technical expertise to use,
which makes them ideal for business users. Yet, contradicting these statements, it has been
indicated that for casual users to be able to use SSBI tools fully, they have to learn SQL
coding, a skill which shouldn’t be necessary for nontechnical users to possess.
As it looks right now, different users contain different levels of skill. Casual users are business
users who consume information. Power users are the data analysts and data scientists who
produce information and can perform complex data analysis in SSBI tools, thus, they do not
have the programming skills like IT-experts do, such as BI developers. However, the research
shows that it’s a necessity for casual users to gain deeper data knowledge and additional skill
to be independent using SSBI. Hence, with the current SSBI tools, it isn’t enough to be an
information consumer to be self-reliant in the data mining process, and if casual users have to
learn new skills and gain more knowledge, such as deep data understanding and SQL
programming, are they still casual users or do they become power users?

5.2.7 Dependency requirements switched focus


To decrease the dependency of power users and IT experts, a lot of organizations are looking
to technology as a solution: Self-service BI (SSBI). These tools allow organizations to still be
able to compete even if their staff does not consist of big data specialists. SSBI tools are
designed to help organizations reach their big data goals, and the emphasis on encouraging
casual users to become more self-reliant and less dependent on power users but also the IT
departments, is a key element of SSBI (Lennerholt, van Laere & Söderström, 2020). Hence, if
enterprises may not manage to mitigate the request and response scenario, which is one of the
key issues SSBI seeks to address as opposed to conventional BI, the intent of implementing
SSBI is challenged (ibid.).
Yet, recent research have shown that the dependency of support haven’t stopped with SSBI,
instead new request-response dependencies between casual users and power users/IT-experts
have appeared, moving from the traditional dependencies of BI, to more explicit ones
(Lennerholt, van Laere & Söderström, 2020). The switched focus of support is consistent with
the finding in this study. And with the dependency exchange actually has reached the goal of
IT-staff getting rid of the responsibility of ad hoc reporting. In spite, the new request-response
relationships are tasks which they don’t necessarily want to spend their time on, e.g. creating
dashboards, locate and get access to data, adding data or other more complex operations
casual users should be able to perform themselves. Additionally, there’s still a need for
checking ad hoc reports made by casual users (see section 5.2.5), to make sure they maintain
a high standard. Just as mentioned in the literature, this shows that the support from power
users and IT-experts doesn’t cease to exist, it just switched focus. Likewise, the dreaded
bottleneck is still a huge challenge, as the supply of power users even now doesn’t meet the
demand.

42
5.2.8 Summary
The results show how the SSBI tools are used within enterprises today and what challenges
they bring to casual users trying to be self-reliant in the data mining process. Casual users
have problems performing tasks outside of ad hoc reporting and analysis, and they can’t
explore data on their own. The majority can only use SSBI tools if the data is prepared,
extracted, stored, processed, put at their disposal, secured, etc. by someone else. And the data
are often only as good as the team that does the preparation. Accordingly, the data preparation
stage is not something they can handle on their own.
When it comes to the challenges, the results from this study are in line with the results from
recent research. The challenges are many, but from the results the main issues focus on ease
of use tools, complex operations, data access, locate and add data, and lastly report
governance and quality. Yet, one challenge that was found in the data, which wasn’t
enlightened to the same extent by recent studies, was data definitions. Hence, the definitions
of data is a major issue for casual users being self-sufficient whereas it creates a lot of
problems, e.g. wrongfully made reports based upon faulty interpreted data, but it also create
new request-response dependencies between casual users and power users/IT-experts. The
goal is to ensure that there are clear definitions for the data, whereas two independent
individuals can draw the same conclusions when looking at the same dataset. A lot of work is
required around data preparations, especially considering definitions, however, casual users
also need to gain a deeper understanding of data and the definitions. Greater data
understanding would also improve the report quality and diminish the need for governance.
With the current SSBI solutions, casual users also need to possess SQL skills to be self-
reliant. Most casual users are only information consumers and lack the technical skill, hence
they’re dependent on power users. As of today, the supply of power users doesn’t meet the
demand, which in turn creates a bottleneck in the request–response relationship between them
and casual users. Therefore, a lot is pointing at casual users having to become more self-
sufficient as SSBI tools are used more broadly to a greater extent. For this to happen, the
inadequate education has to be addressed, as it overturns the use of SSBI, as well there have
to be improvements when it comes to the SSBI tools. SSBI need to become more user-
friendly to be simple enough for casual users to utilize in a self-reliant manner in the data
mining process. Like aforementioned, other major issues is the data access, processing and
collection. Accordingly, for casual users to become fully independent in the data mining
process the sophisticated solutions for data processing and collection have to become more
user-friendly, a great challenge for SSBI developers.

43
6 Conclusion
In the last chapter the author present findings from the discussion, which are related to the
study's aim. In addition, the study's research question is addressed. The chapter concludes
with study recommendations.

6.1 Conclusions
The aim of this study was to answer the research questions ”In what ways are casual users
self-reliant when it comes to performing BI activities using SSBI?” and ”What are the
occurring challenges for casual users when using SSBI tools in the data mining process?”.
Accordingly, this research contributes an elaborate picture to what extent SSBI tools are used
by casual users in the data mining process, and what challenges emerge when using them.
This study have shown that casual users are self reliant when it comes to ad-hoc reporting and
uncomplicated analysis of the data. The prime of SSBI tools is the visualization, the simple
filtering of data and communication of the results. With Data visualization software, as
mentioned the perception is that casual users can perform ad hoc analysis on their data and
create reports, independent of power users. But to be fully self-reliant in the data mining
process casual users have to be able to do more than visualizing data and creating reports.
They must be capable of data processing and data collection (Bani-Hani, Tona & Carlsson,
2018; Lennerholt, van Laere & Söderström, 2020). The findings display that SSBI tools aren’t
developed for casual users to be self-reliant in the data mining process. Thus, the solutions
aren’t user-friendly enough for casual users to perform more than data visualization and
reporting, therefore, complex data collection and integration require additional technological
skill and comprehensive data knowledge. Accordingly, SSBI tools solve one part of the large
area that analytics is, but not all of it.
The study have also shown that tools are still way too difficult for casual users to handle in a
self-sufficient matter, yet, it has proven to be a very big difference between how involved
casual users are in the analytical process, and how they participate in the transformation of
data into information. The difference, again, lies in the level of knowledge they possess.
Although, the majority of casual users work with pre-defined models and dashboards, thus,
the conclusion is that SSBI tools are still mainly used as visualization tools, where the pre-
made data is the foundation of their work. Nevertheless, even though the ad hoc analysis and
reporting have become SSBI tools’ prime area, it still comes with difficulties. The major parts
where SSBI tools are limiting casual users to be self-reliant are; the tools are not not
sufficiently user-friendly, poor definition of data and lack of data knowledge, limited data
access, indigent validation of reports and inadequate education.
Furthermore, experience, data knowledge and SQL coding skills appear to be the main factors
influencing the level of casual users’ independence using the SSBI tools. SSBI is limiting
casual users who don’t possess the required skills of coding or deeper data knowledge needed
to go from data source to finished report on their own. To stop being dependent on help from
power users/IT-experts, casual users need to accumulate additional knowledge and skill, to
handle everything from data processing and collecting to complex reporting. However, this
raises a question: will casual users still be nontechnical information consumers with all the
additional qualities or will they become power users?

44
Accordingly, there is still a major need for support in a lot of areas of the data mining process,
to perform tasks and to ensure the use and outcome is performed and generated in the right
way. As information consumers, casual users will have a dependency of power users since
these tools are too complex for them to utilize. Yet, the level of dependence of power users
and IT-experts depends on the level of knowledge casual users possess, knowledge of the tool,
data knowledge and SQL skill.

6.2 Contribution
As Big data and its technologies is an area which recently have blossomed, the research isn’t
mature on this specific topic. The purpose of this study is to examine the challenges of casual
users being self-reliant when using SSBI tools, and the way they are using them in the data
mining process, and the contributions are more knowledge in an area that is in need of more
research. The results display a clear description of how SSBI is used by casual users in the
data mining process, and what challenges they are up against. It is a complement to the
already existing literature, contributing to the existing body of knowledge.
Vendors of SSBI claim casual users can become self-reliant if they implement SSBI, but this
research provides a more elaborate picture of SSBI tools in reality. In that matter, this study
provides a clearer picture for enterprises, mainly in the retail and finance industries, of what
they can expect when implementing SSBI. Enterprises need to understand that these tools
won’t make casual users fully independent in the data mining process. The SSBI solutions of
today do not allow for casual users without additional skill to do more than ad hoc analysis
and reporting. Hence, the results can be utilized by organizations to facilitate and improve the
usage of SSBI tools and what to be aware of implementing such technology, e.g. that it's
necessary to understand the need for support and training and plan accordingly.
The results of this study can also be a contribution to developers of SSBI tools in the shape of
improvement areas and requirements in the technology. There are a lot of areas where these
tools are not user-friendly and where non-technical users struggle to use and understand them.
A lack of UX in the development of these tools was expressed by IT-experts, something that
can be interesting to look at in future research.

6.3 Future Research


Accordingly, there’s a lot of challenges when it comes to casual users using SSBI, and it
would be interesting to dive deeper into each issue to see what the underlying factors are.
Also, more research needs to be done on how to tackle the arising challenges. For casual users
to become self-reliant, the issues need to be understood and also handled.
It's obvious that the focus of developing SSBI has been on the functionality, but lately the
interface has improved a lot when it comes to analysis and reporting. Yet, SSBI has proven to
still be too complex for nontechnical users, and needs improvement. Even though SSBI
should provide the end-user support to reduce the complexity without losing functionality
(Behringer, Hirmer & Mitschang, 2017). The next step for casual users to become more self-
reliant is to make data sources more accessible, and as far as we can see from this study, even
the more developed interfaces still need improvement. Hence, diving deeper into the UX of
SSBI to unravel the real weaknesses could improve the design of these tools, allowing casual
users to become more independent in the data mining process.

45
References
Abbasi, A., Sarker, S. and Chiang, R.H., 2016. Big data research in information systems: Toward an
inclusive research agenda. Journal of the Association for Information Systems, 17(2), pp.3.

Alpar, P. and Schulz, M., 2016. Self-service business intelligence. Business & Information Systems
Engineering, 58(2), pp.151-155.

Bani-Hani, I., Tona, O. and Carlsson, S., 2018. From an information consumer to an information
author: a new approach to business intelligence. Journal of Organizational Computing and Electronic
Commerce, 28(2), pp.157-171.

Behringer, M., Hirmer, P. and Mitschang, B., 2017. Towards interactive data processing and analytics-
putting the human in the center of the loop. In International Conference on Enterprise Information
Systems, (2). pp. 87-96. SCITEPRESS.

Barnard, C. and Simon, H. A., 1947. Administrative Behavior: A Study of Decision-making Processes
in Administrative Organization. Macmillan, New York.

Bowen, G.A., 2020. Sensitizing concepts. SAGE Publications Limited.

Braun, V. & Clarke, V., 2012. Thematic analysis. In H. Cooper, P. M. Camic, D. L. Long, A. T. Panter,
D. Rindskopf, & K. J. Sher (Eds), APA handbook of research methods in psychology, (2): Research
designs: Quantitative, qualitative, neuropsychological, and biological, pp. 57-71. Washington, DC:
American Psychological Association.

Brink, H.I., 1993. Validity and reliability in qualitative research. Curationis, 16(2), pp.35-38.

Clarke, V. & Braun, V., 2016. Thematic analysis. The Journal of Positive Psychology, 12(3), pp. 297–
298.

Cremona, L., Ravarini, A. and Vom Brocke, J., 2019. Visualizing big data: The impact on sense-
making and decision-making. In Organizing for the Digital World, pp. 23-31. Springer, Cham.

Creswell, J.W. and Creswell, J.D., 2017. Research design: Qualitative, quantitative, and mixed
methods approaches. Sage publications.

Denscombe, M., 2019. Forskningshandboken - för småskaliga forskningsprojekt inom


samhällsvetenskaperna (4). Lund: Studentlitteratur.

Elgendy, N. and Elragal, A., 2016. Big data analytics in support of the decision making
process. Procedia Computer Science, 100, pp.1071-1084.

mhoff, C. and White, C., 2011. Self-service business intelligence. Empowering Users to Generate
Insights, TDWI Best practices report, TWDI, Renton, WA.

Janssen, M., van der Voort, H. and Wahyudi, A., 2017. Factors influencing big data decision-making
quality. Journal of Business Research, 70, pp.338-345.

Lennerholt, C. and van Laere, J., 2019. Data access and data quality challenges of self-service
business intelligence. Proceedings of the 27th European Conference on Information Systems (ECIS).

Lennerholt, C., van Laere, J. and Söderström, E., 2020. User-Related Challenges of Self-Service
Business Intelligence. Information Systems Management, pp.1-15.

Llave, M.R., 2018. Data lakes in business intelligence: reporting from the trenches. Procedia computer
science, 138, pp.516-524.

Mack, N., Woodsong, C., MacQueen, K., Guest, G. and Namey, E., 2005. Qualitative Research
Methods: A Data Collector’s Field Guide. Family Health International (FHI), USA.

46
Marshall, C., & Rossman, G. B., 2011. Designing qualitative research, (5). Thousand Oaks, CA: Sage.

Michalczyk, S., Nadj, M., Azarfar, D., Maedche, A. and Gröger, C., 2020. A State-of-the-Art
Overview and Future Research Avenues of Self-Service Business Intelligence and Analytics. In
Proceedings of the 28th European Conference on Information Systems (ECIS).

Poonnawat, W. and Lehmann, P., 2014. Using Self-service Business Intelligence for Learning
Decision Making with Business Simulation Games. In CSEDU, (2). pp. 235-240.

Schuff, D., Corral, K., Louis, R.D.S. and Schymik, G., 2018. Enabling self-service BI: A methodology
and a case study for a model management warehouse. Information systems frontiers, 20(2),
pp.275-288.

Sharda, R., Delen, D. and Turban, E., 2014. Business intelligence and analytics. System for Decesion
Support.

Sidani, D. and Sayegh, M., 2019. Big Data at the Service of Universities: Towards a Change in the
Organizational Structure and the Decision-Making Processes. In ICT for a Better Life and a Better
World, pp. 299-312. Springer, Cham.

Svahn, T. , & Ax, E., 2020. Den Svenska Business Intelligence & Data Science-Studien. In Swedish;
The Swedish Business Intelligence & Data Science Study.

van der Lans, R.F., 2015. Strengthening Self-Service Analytics with Data Preparation and Data
Virtualization.

Wallén, G., 1996. Vetenskapsteori och forskningsmetodik, (2). Lund: Studentlitteratur.

47
Appendices
Appendix A: Interview questions for casual users
About you
What is your position?
How long have you been working with SSBI tools?
What is the tool used for where you work today?
What year was SSBI implemented?
Is it of importance for the company to adopt to new digital technologies? Why?
Is the company actively working to find and implement new technologies to further digitalize
the company? Could you give any examples?
Motivation for SSBI usage
What is the motivation and ambition for implementing SSBI?
Which of these three aspects: financial, changes in value creation and structural changes, has
been most affected from the implementations of the SSBI in your company?
Ease of use
Is it hard to use the SSBI-tool? Can you create your own dashboards? Can you create reports/
manage analysis of the data?
Does SSBI make the analysis process more effective? Do you save time through using SSBI?
How do you experience the time required and how laborious is this tool compared to other
tools?
Do you/have you created other solutions instead of using SSBI?
Is the tool adapted to your work processes?
Have the implementation of the SSBI created any need to learn new skills for you?
Data availability
Do you know what data is available?
Is the available data appropriate, is it relevant to your work?
Can you use the data that are available? Is the data comprehensible, clearly defined?
Can you locate available data?
Can you add missing data?
Can you change the data content in the tool? Can you filter the data?
Data quality

48
What does the governance of SSBI reports look like?
Do you think that the reports made are of high quality?
Do you know what reports are created? And can you use already created reports?
Other questions
What are power users competencies? What is their area of responsibility/tasks?
How do you think you can get casual users more independent of support from power users
and IT?

Appendix B: Interview questions for power users/IT-experts


About you
What is your position?
How long have you been working with SSBI tools?
What is the tool used for where you work today?
What year was SSBI implemented?
Is it of importance for the company to adopt to new digital technologies? Why?
Is the company actively working to find and implement new technologies to further digitalize
the company? Could you give any examples?
Motivation for SSBI usage
What is the motivation and ambition for implementing SSBI?
Which of these three aspects: financial, changes in value creation and structural changes, has
been most affected from the implementations of the SSBI in your company?
Ease of use
Is it hard for casual users to use the SSBI-tool? Can they create your own dashboards? Can
they create reports/manage analysis of the data?
Does SSBI make the analysis process more effective? Do casual users save time through
using SSBI?
How do you experience the time required and how laborious is this tool compared to other
tools?
Do casual users create other solutions instead of using SSBI?
Is the tool adapted to casual users work processes?
Have the implementation of the SSBI created any need to learn new skills for casual users?
Data availability

49
Do casual users know what data is available?
Is the available data appropriate, is it relevant to casual users work?
Can casual users use the data that are available? Is the data comprehensible, clearly defined?
Can casual users locate available data?
Can casual users add missing data?
Can casual users change the data content in the tool? Can they filter the data?
Data quality
What does the governance of SSBI reports look like?
Do you think that the reports made are of high quality?
Do casual users know what reports are created? And can they use already created reports?
Other questions
What are power users competencies? What is their area of responsibility/tasks?
How do you think you can get casual users more independent of support from power users
and IT?

Appendix C: Informed consent


TITLE OF STUDY
Is Self-Service BI a hoax?
A study of casual users’ independence using SSBI

INFO
Sandra Hansson
Linneuniversitetet
0738763965
sh223gt@student.miun.se

You've been invited to participate in a research project. Before you decide to take part in this
report, it's important that you understand why it's being conducted and what you'll be doing.
I am writing my thesis on Self-Service Business Intelligence (SSBI), and the main motivation
for implementing SSBI is the increased flexibility that the system offers users, which makes
them more independent and thus improves the organization's operational efficiency.
Therefore, the purpose of this study is to investigate whether end users can use SSBI in an
efficient and independent way. What I investigate in this work is user experiences when using
SSBI and how well the information system helps them in their work processes.
Your participation in the interview will help me gather empirical material so that I can
investigate the phenomenon on a deeper level. I want to know how casual users experience
the use of SSBI and how they can use it.

50
The interview will be semi-structured, which means that there are prepared questions but that
the conversation itself can also take us into unprepared parts of discussion. You who
participate in this interview will be completely anonymous. You may decline to answer any or
all questions and you may terminate your involvement at any time.
In order to avoid documenting during the interview, I have planned to record the interview
(voice recording) in order to be able to transcribe it at a later stage. If this is something you do
not want me to do, please let me know. The voice recordings will be destroyed after
transcription.
Please contact me if you have any questions about this study or if you experience any
negative effects as a result of participating in this study. Accordingly, the participation in this
study is completely voluntary. You are free to withdraw at any time after taken part of the
research, and all the data you have returned will be destroyed.
Reading this information, and moving on with your participation states that you understand
the provided information and are willingly taking part of this study.
Best regards,
Sandra Hansson

51

You might also like