Eindhoven University of Technology
BACHELOR
What do we already know and what do we not know about decision making based on data in entrepreneurship?
Entrepreneurs driven by data: A systematic literature review of data-driven entrepreneurship
Merkx, Kai
Award date:
2022
Awarding institution:
Tilburg University Link to publication
Disclaimer
This document contains a student thesis (bachelor's or master's), as authored by a student at Eindhoven University of Technology. Student theses are made available in the TU/e repository upon obtaining the required degree. The grade received is not published on the document as presented in the repository. The required complexity or quality of research of student theses may vary by program, and the required minimum study period may vary in duration.
General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.
What do we already know and what do we not know about decision making
based on data in entrepreneurship?
Entrepreneurs driven by data: A systematic literature review of data-driven entrepreneurship
January, 2022
Author: Kai Merkx
Student number: 1397583
Supervisor: Dr. W.J. Liebregts
1 Abstract
As the importance of data analysis techniques is growing and academical re- search on the process of decision making in entrepreneurship is little, there is need of synthesis. This study aims to fill this gap by creating a descriptive scop- ing review on the subject: the use of data and data analysis techniques in the process of entrepreneurial decision making. This systematic literature review shows a clear division of three domains concerning data analysis techniques in the decision making process: business analytics, big data analytics and data- driven decision making. Results show a positive, direct or indirect, influence of all three methods on the decision making quality or firm performance when applied correctly. Skilled workers and a data-driven organizational environment are essential factors in order to optimize and exploit the benefits of data anal- ysis techniques in the process of decision making. Without proper calibration of organizational resources and an appropriate level of fit between employees, tasks and data analysis techniques, the benefits of using these data analysis techniques will diminish drastically. Prior investment in related departments like IT and IPC facilitate the rapid implementation and integration of different data analysis techniques across the entire company. Gaps that have been iden- tified concern research about the type of decisions that are being made based on data, research focusing on the employees who make the decisions and research on the influence of firm size and industry type.
Contents
1 Abstract 2
2 Introduction 4
3 Background Research 5
4 Approach 9
4.1 Search strategy . . . 10
4.2 Study selection . . . 11
4.3 Data extraction and analysis . . . 14
5 Results 15 5.1 Comparison between concepts of analysis techniques . . . 15
5.2 Business analytics . . . 17
5.3 Big data and big data analytics . . . 19
5.4 Data-driven decision making and support systems . . . 22
5.5 Overall synthesis . . . 24
6 Discussion and Conclusion 28 6.1 Gaps in the literature & Ideas for further research . . . 28
6.2 Limitations . . . 32
6.3 Conclusion . . . 33
References 35
2 Introduction
Over the years, more and more enterprises and businesses have started gather- ing and using data to retrieve new insights into their business processes. This kind of business has led to a new kind of entrepreneurship, called data-driven entrepreneurship. The general definition of data-driven entrepreneurship can be described as making use of data-driven techniques and technologies in shaping entrepreneurial activities (Erzurumlu et al., 2018). This paper has a specific focus on a subsection of this kind of entrepreneurship: data and data analysis techniques involved in entrepreneurial decision making. The goal of this paper is to collect all relevant scientific articles about this specific topic of data-driven entrepreneurship. This will lead to a complete synthesis that contains all knowl- edge and includes each important aspect about decision making based on data in the entrepreneurial scene. First of all, the influence of data and correspond- ing analysis techniques on the decision making performance will be discussed.
The determinants of effective decision making based on data and data analysis techniques will be discussed as well. Also, other aspects will be included, like mediating factors between those analysis techniques and decision making qual- ity, the effect of environmental factors, what kind of decisions are being based on data-driven insights and the influence of external factors, like firm size and industry sector.
In order to achieve this, a systematic literature review (SLR) about this topic has been performed and the results will be shown in this paper. The kind of review that has been written is a descriptive SLR as it addresses the current state of research that has been done about this topic. This SLR will be a scoping review. The aim of a scoping review is to extract as much relevant data from every literary piece. With all this information, its goal is to create a snapshot of the field and to create a comprehensive synthesis of the topic (Xiao & Watson, 2017). On top of this, scoping reviews enable to detect research gaps which can lead to future research (Peters et al., 2015). These results match with the objectives of my research. This had led to the following research question:
”What do we already know and what do we not know about decision making based on data in entrepreneurship?”
The remainder of this paper will continue as follows. Section 3 contains back-
ground research that has been done on the topic: decision making based on data and data analysis techniques in entrepreneurship, which results in the motivation for this literature review, supported by literary works. Section 4 will contain the methods and approach of this research. Especially, the review protocol that has been followed throughout the execution of the SLR will be explained. Section 5 contains the results and a synthesis of all findings of the performed literary search. Section 6 will contain the conclusions, limitations, gaps in the literature and ideas for further research.
3 Background Research
First of all, it is important to understand the definition of data-driven en- trepreneurship and what it entails. To understand the term data-driven en- trepreneurship, the term entrepreneurship should be considered first. How- ever, several studies have shown there is no general definition of the term en- trepreneurship (Kobia & Sikalieh, 2010). Because of this, it is impossible to for- mulate one general agreed upon definition for the term data-driven entrepreneur- ship. Despite this lack of a general detailed definition, in its broadest sense data-driven entrepreneurship can be described as making use of data-driven techniques and technologies in shaping entrepreneurial activities (Erzurumlu et al., 2018). This paper will not take a deeper dive into the definition of data-driven entrepreneurship, but will have a specific focus on one aspect of data-driven entrepreneurship: the role of data and data analysis technologies in the process of entrepreneurial decision making.
The ability of adequate decision making is a core skill that every entrepreneur should possess (Gustafsson, 2006). Despite the fact decision making is a crucial factor in entrepreneurship, several studies have shown that research about this topic is lacking. Although decision making has been researched in other fields like psychology and marketing, there has been no further investigation into decision making in entrepreneurship (Shepherd et al., 2014). Another article expressed its hope in multilevel research on entrepreneurial decision making, as this field is pretty untouched (Aguinis et al., 2010). An article published in 2019 also confirmed that the use of data and big data analysis in entrepreneurship has not received the amount of research it should have (Obschonka & Audretsch, 2019). This absence of research is unexpected as decision making can be con-
sidered a core element of entrepreneurship.
Over the last decades, a very useful asset has emerged that can support en- trepreneurs with this decision making, which is data. Various techniques have been developed that enable analysis of this data. One of those techniques is called data-driven decision making (DDDM). The general definition of DDDM can be described as using facts, metrics and data to guide strategic business de- cisions that align with initiatives, goals and objectives (Grant, 2021). Research has shown that forty percent of the managerial decisions are based on gut feeling and intuition (Davenport et al., 2010). By implementing DDDM, people can avoid making these split-second decisions or decisions that are solely based on gut feeling, experience or intuition. This kind of evidence-based decision making has numerous advantages over decisions that are not evidence or information- based. Using data and data analysis techniques, improves decision making over the long run (Brynjolfsson et al., 2011), reduces uncertainty, helps managing risks (Davenport et al., 2010) and provides a strategic edge over competitors who do not make use of data and data analysis techniques in the process of decision making (Fast et al., 2021). As DDDM brings such great advantages, the use of DDDM in entrepreneurship has increased enormously. Many small and large businesses try to exploit data tools in order to get an advantage over competitors (Provost & Fawcett, 2013).
Although DDDM improves decision making, it has not yet fulfilled its potential, according to a study that surveyed 84 software practitioners (Svensson et al., 2019). A vast majority of the participants concluded that DDDM will be of great importance and highly valued in the future. While only few participants appoint a wide-spread use of DDDM in their current decision processes, most participants agreed upon an increasing influence in the future considering higher level and more general decision making (Svensson et al., 2019). As the amount of data increases every day, the amount of information you can get out of this data increases with it. This will result in an even larger source of knowledge.
Between 2005 and 2010 the adoption of DDDM has increased by thirty percent (Brynjolfsson & Mcelheran, 2016a). This article states, although DDDM is al- ready economically important, there is still plenty of room for further diffusion and use of DDDM. Because DDDM has not yet reached its potential and its importance is very likely to increase in the future, this study will be very help- ful by addressing the current state of this topic. Identifying gaps or areas in
which implementation or further diffusion of DDDM is possible, will improve entrepreneurial decision making performance even more.
The use of data and data analysis to support decision making can also be placed under the concept of business analytics (BA). BA can be described as ”the extensive use of data, statistical, and quantitative analysis, explanatory and predictive models, and fact-based management to drive decisions and actions”
(Davenport & Harris, 2007). The development of information technology (IT) has made it possible for entrepreneurs to generate and collect large amounts of data. These large amounts of data, which can be referred to as big data, result into profitable business opportunities for organizations to retrieve new insights into their business processes (Watson, 2014). The application of BA enhances the ability to not only process these large amounts of data, but also derive useful insights out of that data which will in the end lead to better decisions and an increasing organizational performance (Lavalle et al., 2011). Another advantage of BA is that it enables faster decision making. This is needed in order to face increasing competition due to the fact of globalization and technological devel- opment (Kiron et al., 2014).
Because of these advantages, the implementation of BA across organizations has increased significantly (Lavalle et al., 2011). All of the advances combined in IT, DDDM and BA lift up the level of decision making and turn it into a data-driven activity, while lessening the level of intuition needed in the process of making these decisions. On top of this, research has shown that the advan- tages of using data analysis are very likely to expand in the future, including faster decision making and an improvement of decision quality (Acito & Khatri, 2014). Although the influence of data analysis on decision making is proven by various experiments, the consideration of these topics as still rising fields of expertise is small (George et al., 2014). Because of this, little is known about the current development of those analysis techniques, organizational decision making and the relationships between those elements. This paper aims to fill this gap by creating a complete overview of existent literature about this topic:
organizational decision making based on data and data analysis techniques.
Because of the confirmation by several studies on the need of research on the pro- cess of decision making in entrepreneurship (Aguinis et al., 2010) (Obschonka &
Audretsch, 2019) in combination with the benefits and increasing importance of
using data in the process of decision making (George et al., 2014) (Brynjolfsson
& Mcelheran, 2016a), this SLR has a contribution to the already existing re- search. Although, there has been done some research on decision making in entrepreneurship and the increasing importance of decision making based on data has been proven, a clear overview of this topic is missing. Because of this, it can be concluded that the research that has been done on this topic is fragmented and is in need of synthesis. The objective of this research is to create a synthesis of all current knowledge about data analysis in combination with business related decision making. As such an overview of this topic will be created, it will function as a status update on which further research can built on. Also, this research will have a contribution to enterprises in substantiating information on implementing, adopting or using analysis techniques in order to improve organizational decision making.
This literary review will include all aspects that are involved in the process of entrepreneurial decision making based on data and data analysis techniques.
One component of this research will focus on the effect of using data and data analysis techniques on the effectiveness and quality of organizational decision making. Another aspect that will be investigated, concerns the determinants for optimal use of data-driven decision making in an organization. Particular set- tings or variables might influence the outcomes of using data analysis techniques in the process of decision making. Other aspects that might be discovered can relate to the type of decisions that are being made based on data as different techniques might be connected to specific kinds of decisions. The influence of people or boards who actually make the decisions based on data, will be investi- gated as well. Data-driven decision making can support various departments of a business, so the people who make the decisions based on data can differ as well.
Other external factors will be considered as well, like firm size, industry type and the influence of external parties. Besides creating an all-inclusive overview of this topic, its goal is to identify possible gaps in the literature. By developing a synthesis about the topic, it enables to detect certain topics that might have been underdeveloped or even untouched at all. Those undiscovered topics can result into ideas for future work or even entirely new research domains. In order to formulate an answer to all of those sub-domains, one comprehensive research question has been formulated:
”What do we already know and what do we not know about decision making
based on data in entrepreneurship?”
In the next section of this paper, the methodology of conducting the SLR will be explained. More specifically, the review protocol that has been developed will be presented.
4 Approach
A systematic literature review has been conducted on published empirical re- search. The type of review that has been written is a descriptive scoping SLR.
It addresses the current state of research that has been done about this topic and it tries to extract as much relevant data from every literary piece (Xiao
& Watson, 2017). By doing this, the SLR will represent a complete synthesis about the topic. On top of this, scoping reviews enable to detect research gaps which can lead to future research (Peters et al., 2015).
The three stages of conducting a SLR (planning, conducting and reporting the review (Kitchenham & Charters, 2007)) can again be divided into 8 smaller steps: (1) formulating the research problem, (2) developing and validating the review protocol, (3) searching the literature, (4) screening for inclusion, (5) as- sessing quality, (6) extracting data/information, (7) analyzing and synthesizing data, (8) reporting the findings (Xiao & Watson, 2017). Figure 1 shows how these steps are divided among the three stages that are involved in writing a systematic review. Step 1 of this process has been done in section 2 and 3 of this paper, ”Introduction” and ”Background research” respectively. Step 2,
”Develop and validate the review protocol” will be shown and explained in this section. This will include the chosen search engines and search terms used, in- clusion and exclusion criteria, quality assessment criteria and data extraction methods. Step 7 and 8 of this process will be presented in section 5 of this paper: Results.
Process of conducting a SLR
Figure 1: Three stages of conducting a SLR divided into 8 sub-steps (Xiao &
Watson, 2017)
4.1 Search strategy
For this SLR, only digitally published literary works have been considered.
This choice has been made as the topic of data-driven decision making in en- trepreneurship is quite a digitally aged topic. Because of this, the assumption has been made that little to none promising articles would be available any- where else than in a digitized form. As this study is a digital topic, only articles published later than the year 2000 will be considered. This has been decided, because the goal of this paper is to get a present status update about this topic. Therefore, there has been decided to ignore all studies executed before the year 2000. Only English and American articles have been considered as the time constraints would not enable to properly translate papers which have been written in any other language. Other inclusion criteria has been aimed more at the content of the articles. Organizational decision making should be a clearly studied pillar in the executed research. At the same time, this should be related to data or data-driven technologies supporting this decision making.
When an article did not consist of decision making in combination with data or data driven technologies, the study has been excluded from this review. To include as many relevant articles and to limit location bias as much as pos- sible, several literary databases have been included, which are the following:
Google scholar, ScienceDirect, SpringerLink, Research Gate and IEEE explore.
These literary databases have been chosen as they are open-access and include
the most important full-text journals about this technical subject. As this is a scoping review, and its goal is to get an all-inclusive overview of the topic, grey literature will be considered as well. This will be done to make sure no relevant information will be left out.
Search terms have been divided into two different domains: entrepreneurial- related terms and terms related to data-driven decision making. For data-driven decision making the following terms have been used: ”Data-driven decision mak- ing” OR ”Information-driven decision making” OR ”decisions based on data”
OR ”Data analysis involved in decision making”, for entrepreneurship these terms have been used: ”Entrepreneurship” OR ”Entrepreneur” OR ”Enter- prise” OR ”Business” OR ”Firm”. Search terms within the same domain will be connected using the operator OR, search terms between different domains will be connected using the ”AND” operator. The decision has been made to keep the domain related to ”data-driven decision making” terms as broad as possible. This has been done in order to include every possible data analysis technique that has been used for entrepreneurial decision making. As the goal of this SLR is to create a comprehensive synthesis, the search terms should not lead to one specific data analysis technique, but should remain as broad as pos- sible in this studied field. For the entrepreneurial related terms, there has been decided to use various synonyms. This has been done as different studies might vary in the terms related to entrepreneurship.
The literature search comes to an end when repeated searches result in no new relevant articles. When no new information of references show up, the search for new literature can stop (Levy & Ellis, 2006). Table 1 shows the number of search results per search engine after initial title screening.
4.2 Study selection
Initially, papers have only been scanned by their title. After this initial scan, the selection process has proceeded in a two-way selection process. First, the ab- stracts of the journals have been considered. When the abstract did not provide enough information about whether or not to include the journal, the conclusion has been consulted. This has been done as there was no possibility to discuss
Table 1: Number of articles per search engine after title screening Search engine Number of Journals
Google Scholar 70
IEEE 7
ScienceDirect 15
Research Gate 4
SpringerLink 10
Total 106
the inclusion of an article with a fellow assessor. When in doubt, the article did go through to the second stage of the selection process, which is the full text analysis. During the full text analysis, the quality of the journal has been assessed. When a journal does not seem relevant at all, it can still be left out.
With a scoping review, the quality of the review does not matter as much as with other types of review. As a scoping review is about representing the full breadth of the topic, studies of all levels of quality should be considered (Whittemore & Knafl, 2005). However, not all articles suffice to the level of quality that is needed to be included in this SLR. So, there has been performed a certain quality assessment for each article. According to the quality of the study, the article will be included or excluded of this SLR.
In addition to reviewing the paper based on its content, external factors will be considered as well. One aspect that will be looked into is the number of times a paper has been cited by other articles. When an article gets cited a lot by other related papers, the quality of the paper is likely to be of high quality. When an article gets cited little by other papers, the quality of the paper is more likely to be of lower quality. Also, the kind of journal in which an article has been published, tells a lot about the quality level of a paper. When an academic paper gets published in a highly rated journal, it is likely that the article itself is of high quality as well and vice versa. After the quality assessment has been performed, an additional number of papers can be discovered by performing both forward and backward citation search.
Figure 2 shows a flow chart that summarizes the process of the literature search during the execution of this SLR. The figure shows the selection process of this
SLR. After the initial search (title screening), 106 articles have been registered as relevant for this review. However, within those 108 articles there have been identified 8 duplicates. These duplicates have been removed immediately, which resulted in a total number of 98 articles. After abstract screening, while in some cases the conclusion has been consulted as well, 63 articles have been excluded from this review. Several articles did not satisfy the inclusion criteria, while other studies focused on different aspects that were irrelevant for this SLR.
This has resulted in 35 remaining articles for the full text analysis. During this phase, another 14 articles have been removed from this study. Several articles slightly deviated from the desired research topic. Other articles did not satisfy the desired level of quality. Forward citation search has resulted into 1 addi- tional article, which has set the total number of articles reviewed in this SLR to 22.
Funnel procedure of a systematic literature review
Figure 2: The process of conducting a systematic literature review
4.3 Data extraction and analysis
The information of the articles has been extracted based on the objective of this study, which is to create a complete overview about this topic. In order to get an overview of all relevant papers, several sections of the papers have been used for data extraction. These sections are the following: introduction (objectives), results and conclusions. These parts have been used, as these parts contain most of the focus points and findings of each study. The methodology section plays an important role when it comes down to the quality assessment of an article. However, it does not contain the main findings or conclusions, which are needed in order to create an overview about this topic. That is the reason why this section does not take part in the data extraction and data analysis part.
As the objective of this study is mainly focused on the outcomes of the studies, all findings have been extracted from every paper. Once all the findings had been extracted, these findings have been categorized together with comparative findings of other papers. When all findings had been categorized, these cate- gories have been synthesized into one all-inclusive and comprehensive overview of this topic.
5 Results
When conducting this SLR about the use of data and data analysis techniques in organizational decision making, three separate areas of research emerged.
These three topics could be distinguished by considering the data analysis tech- nique that was studied. The articles have been categorized based on these data analysis techniques. The three areas of research that evolved were: (1) business analytics (BA) / business intelligence & analytics (BI&A), (2) big data / big data analytics (BDA) and (3) data-driven decision making (DDDM) / data- driven decision support systems.
All three domains consist of two different streams of research. One stream of research focuses on the influence of that specific analysis technique on the decision making quality or organizational performance. The other stream of research focuses on the determinants and optimal organizational conditions or settings in order to benefit optimally from data analysis techniques in the de- cision making process. These studies consider factors like the most important environmental variables in order to benefit from data analysis techniques, which parts or people of an organizations should use data in their process of decision making, for what type of decisions data can be used best and in what way an en- terprise should adopt its processes or employees. Next to this, mediation factors between those techniques and decision making performance will be discussed as well. In the end, all three domains will be combined in order to draw the right overall conclusions. First, an overview of several definitions encountered during this SLR about different data analysis techniques, will be presented. This pro- vides insights on the meaning of certain terms in comparison with other terms.
5.1 Comparison between concepts of analysis techniques
During the execution of this SLR, several terms considering the use of data and data analysis techniques in the process of decision making have been encoun- tered. To understand all of those definitions, a two dimensional figure has been created containing those definitions. This figure helps to understand each defi- nition along two scales. The y-axis shows the level of difficulty of situations in which the technique will be applied from simple to complex. The x-axis shows
whether the concept is being used as a descriptive or a prescriptive type of anal- ysis. Although this graph represents information about the concept itself, more importantly, it shows in what position it is located compared to other terms.
This comparison enables to detect the differences between all those terms and so get an overview of all concepts in this environment. This graph has been shown in Figure 3.
Comparison between concepts of data analysis techniques
Figure 3: Definitions of various data analysis techniques along two scales
What can be concluded first is that information technology (IT) and informa- tion processing capability (IPC) function as the basis of further data analysis (Brynjolfsson et al., 2011) (Liu et al., 2012) (Chatterjee et al., 2021). These top- ics facilitate other data analysis techniques in more complex situations. Because of this, these terms have been placed in the left bottom corner as smaller circles than the other data analysis techniques. The terms business analytics (BA), business intelligence & analytics (BI&A) and data analytics have been concep-
tualized as one domain during this research, which will be referred to as business analytics. All three definitions function as a broad expression including multi- ple techniques of handling data (Davenport & Harris, 2007), (H. Chen et al., 2012), (Ghasemaghaei et al., 2017). However, as those definitions are not used for extremely complex or prescriptive situations, these terms have been placed relatively central in the environment. Big data analytics (BDA) has been placed higher along the y-axis. This has been done as handling enormous amounts of data coming from heterogeneous sources increases the level of difficulty in pro- viding the appropriate outcome (H. Chen et al., 2012), (Awan et al., 2021). So, BDA will be used to solve more complex situations. On the other hand, data- driven decision support systems and data-driven decision making (DDDM) have been placed below BDA on the y-axis. However, these terms have been placed more towards the right on the x-axis. This has been done as these concepts function more as a prescriptive method. These methods already provide several options that should be considered or one specific type of action that should be taken, functioning as a prescriptive method (Princes & Kosasih, 2021). The def- initions: Forecasting/Predicting, Deep learning and Artificial intelligence have not been encountered during this research. These three concepts have been added to visualize the definitions in a larger environment, so the concepts that have been faced during this review could be compared against multiple domains.
As the definitions of the different terms have been explained in more detail, the results of the systematic literature review can be shown. First, the most important findings of the domain business analytics (BA) will be presented.
5.2 Business analytics
Business analytics can be described as ”the extensive use of data, statistical, and quantitative analysis, explanatory and predictive models, and fact-based management to drive decisions and actions” (Davenport & Harris, 2007). In a study performed by Chen and colleagues, BA has been referred to as the use of analysis methodologies and applications to analyze data in order to better understand its business and from there make better timely business decisions (H. Chen et al., 2012). As can be concluded from these definitions, BA can be considered as the overarching term when it comes down to implementing
relatively simple data analysis techniques for better decision making.
Several studies have shown the influence of BA on entrepreneurial decision mak- ing. Davenport and colleagues showed that integrating BA in an organization improves its data processing capabilities. The insights that can be derived from this improved data processing, enable to make effective decisions and so result in an improvement of firm performance (Davenport et al., 2001). Another study, followed up on the aspect of information processing capability (IPC). executed by Cao et al, confirmed BA positively enhances IPC through the mediation of a data-driven environment (DDE). Then, IPC has a positive effect on data-driven decision making (DDDM) which positively influences decision making effective- ness (Cao et al., 2015). This study concluded that if a firms DDE was properly developed, BA can be enabled to improve firms decision making.
This has been confirmed by a study performed in 2018, which proved that BA provided new data insights for decision making (Acharya et al., 2018). These new insights would result in better decision making (Lavalle et al., 2011). An- other study examined this mediating factor of data-driven insights between BA and decision making even further. Awan and colleagues concluded that BA is positively associated with data-driven insights, which then leads to improvement of decision making quality (Awan et al., 2021). However, this study also found that BA solely relies on data-driven insights, to connect with decision making.
The relationship between BA and decision making itself is insignificant. Other studies did detect a direct link between data analytics and decision making. One study who researched the role of data analytics on a firms agility, found that data analytics enable to detect changes in the market faster, and thus leads to improved response speed and improved decision making efficacy (Ghasemaghaei et al., 2017). This study has been in accordance with the study performed by (Cao et al., 2015), considering a firms DDE. Ghasemaghaei and colleagues con- firmed the importance of the level of fit between data analytics capabilities and other key related firm elements in order to result in more responsive and ef- fective decision making. This study even found that when the level of fit is low between data analytical tools and other organizational sources, like data, employees or business tasks, it might hurt the business in terms of agility. The mere use of data analytics does not provide any value to organizational decision making or firm performance, unless there is proper calibration of organizational resources (DDE) (Ghasemaghaei et al., 2017). This statement has been sup-
ported by Hedgebeth, who stated there should be an organizational emphasis on data analytics in order to successfully implement it in a company (Hedgebeth, 2007). As the importance of BA has been proven across various studies, this does not provide insights into what aspect of BA cause these benefits. However, a study executed in 2018 has provided some insights into the individual aspects of BA. Ghasemaghaei and colleagues have concluded that data quality, analyti- cal skills, domain knowledge, and tools sophistication significantly impact both decision making quality and decision making efficiency (Ghasemaghaei et al., 2018). This study also found that bigness of data increases the decision making quality. However, it does not influence decision making efficiency.
Other studies have tested the influence of firm size on the use of analytical tools in order to improve decision making. The results showed that there is no difference between large and medium firms, when it comes down to ways of integrating data analytics for optimal decision making (Ghasemaghaei et al., 2018) (Davenport & Harris, 2007). There has been done no research on small firms. The influence of the sector or industry type is also proven to be of no significance in using data analysis to improve decision making (Ghasemaghaei et al., 2018) (Davenport & Harris, 2007). However, not all sectors and types of industry were included in these studies. Although these aspects have been investigated, these topics have not been of primary focus. So, further explo- ration of these topics is necessary to confirm these findings, show the results for smaller companies and discover the effect of all industry types.
5.3 Big data and big data analytics
The concept of big data has emerged over the last years, as more and more data of heterogeneous sources could be collected and stored. Big data can be differ- entiated by four values: volume, velocity, variety and veracity (Ko´scielniaka &
Putoa, 2015). These enormous amounts of data could be combined and analysed which resulted in big data analytics (BDA). BDA can best be described as a holistic approach by managing, processing and analyzing big data characterized by the following four values: volume, velocity, variety and veracity (Shamim et al., 2020).
Several studies have researched the effect of BDA on decision making quality or firm performance. One study investigated the effect of big data on manufac- turing performance. This study concluded that BDA is very likely to improve the organizational ability to make decisions (Dubey et al., 2019). This study also concluded that BDA has emerged into one of the most important factors when it comes down to delivering meaningful insights which influence decision making. This finding has been confirmed by a study performed by Awan amd colleagues. Awan et al. positively associated BDA with data-driven insights, which enhances decision making quality (Awan et al., 2021). This study also researched both the indirect link through data-driven insights and the direct link between BDA and decision making performance. This study concluded there is no indirect link between BDA and decision making performance. How- ever, there is a significant direct and positive link between BDA and decision making performance (Awan et al., 2021). A study performed by Kopanakis and colleagues confirmed this by concluding big data has a positive effect on firm performance. Kopanakis et al. found that big data provides decision makers with the capability to innovate and increase organizational performance gaining a competitive advantage against rivals (Kopanakis et al., 2016). This finding has been supported by a study executed by Fast and colleagues. This study investigated this competitive advantage of big data against business rivals. The results showed that the advantage may not be necessarily transitory. Dominant firms can establish long-term market power due to the collection and use of big data (Fast et al., 2021). Another study detected a positive link between BDA and decision making quality through the mediating role of contractual and re- lational governance (Shamim et al., 2020). This study showed the importance of strong contractual and relational governance, which results in data of high quality. This results into improved BDA, which enhances organizational deci- sion making performance. So, from all outcomes of these different studies, it can be concluded that big data and BDA have a positive effect on organizational decision making.
Other studies focused on the current status of knowledge within firms about big data in the decision making process. A study performed by (Davenport, 2014) found that most managers are familiar in handling relatively easy data analysis techniques. However, managerial knowledge about big data and BDA is lacking (Davenport, 2014). The biggest problem of handling big data is combin- ing data from heterogeneous sources and formats (Ko´scielniaka & Putoa, 2015).
This study also proposed several ideas on how to familiarize organizations with this BDA. The most important aspect to improve the quality of decisions based on big data is efficient and effective use of available data sets (Ko´scielniaka &
Putoa, 2015). In order to achieve this, this study proposed making use of score- cards among employees in combination with explicit business rules. This will make it easier to detect individual shortcomings, which then can be fixed by coaching and training to improve the enterprise performance (Ko´scielniaka &
Putoa, 2015). Dubey and colleagues confirmed this finding of Ko´scielniaka et al., by stating that an organizational data-driven culture is essential to exploit BDA capabilities to its full potential (Dubey et al., 2019).
Also, there has been done research on what kind of entrepreneurial decisions get involved in BDA. While traditional data analytics are meant to advise senior management on internal business decisions, BDA provides help regarding other departments in an organization, mainly focusing on customer facing products and services (Davenport, 2014). Big data and BDA can facilitate in decisions regarding customer satisfaction, managing supply chain risk, competitive intelli- gence, pricing of goods and services and discovering experimenting (Davenport, 2014). Although big data startups benefit most from BDA (Davenport, 2014), it enables firms of any size to improve organizational performance (Ko´scielniaka
& Putoa, 2015) (Davenport, 2014).
A study performed by Merendinoa and colleagues dived deeper into the effect of big data on managerial decision making (Merendinoa et al., 2018). Three levels of board level decision making have been researched: level of individual directors, level of the board itself and at the level of the entire organization and stakeholders. Results showed that individual directors show a gap of cog- nitive capabilities to cope with such enormous amount of data. In the end, this leads to an information overload for the board, with the result that this excessive amount of data does not lead to an improved competitive advantage (Merendinoa et al., 2018). The decision makers are unable to extract relevant in- formation out of these significant amount of data. In order to prevent this from happening, individual cognitive capabilities need to develop, so these amounts of data can be processed at an earlier stage, not leading to an information over- load for the board. At the level of the board itself, big data disrupts its board cohesion. This directly affects the decision making process, due to the pres- sure to adapt the way strategic decisions are made. This pressure is caused by
shortened timescales because of constantly new and refreshing data. Likewise individual directors, boards and decision makers need to develop cognitive ca- pabilities. Strong internal co-ordination, internal integration and a data-driven culture among decision makers are essential in order to benefit optimally from big data. Rapid decision making and accepting the tensions resulting from this, enable firms to effectively make use of big data (Merendinoa et al., 2018). Con- sidering the entire organization and its external stakeholders, this study found that third parties can play an important role in handling big data and develop- ing a companies BDA. Competitors, consultancy companies, digital experts and policy makers have been identified as key stakeholders influencing the use of big data. While the importance of knowledge-based capabilities for decision making rises, it is extremely important to surround a company with stakeholders who enable or support the use of big data (Merendinoa et al., 2018).
5.4 Data-driven decision making and support systems
Likewise BA and BDA, various experiments have been performed investigat- ing the influence of DDDM on the quality of the decision making process and firm performance. A study performed by brynjolfsson and colleagues in 2011, have extensively tested the influence of DDDM on three different levels of firm performance: productivity, business profitability and market value. The pro- ductivity test has revealed that integrating data-driven decision making into a businesses main processes, increases their productivity output by five to six percent compared to their original investment and information technology usage (Brynjolfsson et al., 2011). The business profitability test revealed a positive correlation between DDDM and return on assets. Also, DDDM and asset uti- lization have been found positively correlated with each other (Brynjolfsson et al., 2011). This has been confirmed by a study in 2016, who confirmed that DDDM is positively correlated with entrepreneurial productivity (Brynjolfsson
& Mcelheran, 2016a). The market value test revealed evidence to conclude that companies with a larger adoption of DDDM represent a higher market value (Brynjolfsson et al., 2011). In 2016, Brynjolfsson and colleagues have conducted another research about DDDM in U.S. manufacturing. This study concluded that more diffusion of DDDM always results into better performance (Brynjolfsson & McElheran, 2016b). Even businesses who adopt DDDM at a
later time than a competitor, will have a three percent increase in their average company value higher than non-adopters. Furthermore, this study detected a decrease over time in DDDM-related performance differentials for both early and later adopters of DDDM (Brynjolfsson & McElheran, 2016b).
A study performed by Princes and colleagues disagreed with the sole use of data analysis techniques in decision making which is claimed in the present lit- erature. Although DDDM is considered a must in the present entrepreneurial world, the research showed that intuition derived from experience is required in any condition (Princes & Kosasih, 2021). DDDM is essential in minimizing uncertainties and mistakes, but is incomplete for making final decisions au- tonomously. This study proposed to use DDDM as a supportive tool providing several options, after which the final decision will be made with help of intuition and experience of the decision maker (Princes & Kosasih, 2021). So, this study promotes the use of analysis techniques as data-driven decision support systems, in which intuition, experience and data-driven insights complement each other.
In contrast to other studies, (Brynjolfsson & McElheran, 2016b) investigated the influence of data-driven decisions made by front-line workers . Although some industries seem to benefit more from DDDM when certain decisions are delegated to front-line workers, most sectors do not. Instead of joint decision making between front-line workers and managers, centralized coordination is of great importance in order to effectively increase a firms productivity using DDDM (Brynjolfsson & McElheran, 2016b). Princes and colleagues confirmed this finding by stating DDDM approach is best suited for the middle man- agerial level and long term planning. As DDDM makes it possible to retrieve information out of collected data generated by customers, clients or processes, it can lead to interesting insights, which will help entrepreneurs make evidence based decisions in product design, delivery and in the development of inno- vation (Troisia et al., 2020). This study, executed by Troisia and colleagues, emphasized the importance of employees’ abilities who work with DDDM. This finding has been confirmed by other studies, which observe an important role for employees (Brynjolfsson & McElheran, 2016b). Employees who have re- ceived a college education are far more likely in dealing with high levels of DDDM (Brynjolfsson & Mcelheran, 2016a). Also, communication and data transfer among different departments of an organizations are found to be im- portant factors for improving decision making effectiveness (Troisia et al., 2020).
DDDM is integrated faster and more prevalent among multi-unit firms who have higher levels of employment than it is among smaller single-establishment firms (Brynjolfsson & McElheran, 2016b) (Brynjolfsson & Mcelheran, 2016a). An- other major determinant of successfully integrating DDDM in an organization is prior investment in IT (Brynjolfsson & Mcelheran, 2016a). Higher levels of IT capital in combination with important attributes of firm culture reinforce DDDM (Brynjolfsson & McElheran, 2016b).
5.5 Overall synthesis
By conducting this SLR, three domains of data analysis techniques have emerged:
business analytics, big data analytics and data-driven decision making. Within these domains, there has been another clear division of two research domains.
Within each domain, one stream of research has been focusing on the direct or indirect effect of this data analysis technique on the performance or quality of the organizational decision making process. The other stream of research has been focusing on the determinants of effective decision making based on data analysis and the optimal settings of integrating or using these data analysis techniques in the entrepreneurial decision making process. A table has been created containing every key finding of each article. These can be found in ta- ble 2. This table has been categorized by the analysis technique studied in the article.
Although there has been a clear division of three separate domains, there are several general findings that apply to the general domain of decision making based on data. The influence of implementing and using data and data analysis techniques in the process of decision making has been identified as significant (Awan et al., 2021). Organizational integration of data analysis has a positive effect on decision making quality (Dubey et al., 2019) (Cao et al., 2015) and firm performance (Brynjolfsson et al., 2011) (Davenport et al., 2001). Prior invest- ments in IT and other related departments like IPC, have been found crucial in exploiting and optimizing the benefits of using data analysis (Brynjolfsson et al., 2011) (Chatterjee et al., 2021) (Cao et al., 2015). Furthermore, the abili- ties of the employees who deal with data in the process of decision making are key elements (Davenport & Harris, 2007) (H. Chen et al., 2012) (Brynjolfsson
& Mcelheran, 2016a). However, studies showed that decision makers do not possess those desired abilities (Merendinoa et al., 2018) (Davenport, 2014). An- other important factor to optimize the use of data analysis in the process of decision making is the organizational environment (Dubey et al., 2019) (Troisia et al., 2020). An organizational emphasis on the use of data analysis is required in order to optimally benefit from it (Ghasemaghaei et al., 2017). All of these observations, together with more domain specific findings, have been summa- rized in table 2.
During the execution of this SLR, certain aspects have been identified as un- derdeveloped research. Those specific aspects have not been scientifically re- searched enough or have not received any academical attention at all. These lacking parts in the literature have been identified as gaps and will be presented in the next part of this report together with some ideas for further research. The limitations of this research will be discussed as well and the final conclusions will be summarized at the end.
Table 2: An overview of each article with its key findings
Author(s) Year Journal Technique Research Focus Key Findings
Acharya, Singh, Pereira
& Singh
2018 International journal of information management BA
The effect of data on decision making in the world of fashion
- Implementation of BA results in new insights for decision making. These insights will improve decision makin performance
Awan, Shamim, Khan, Ul Zia, Shariq
& Khan
2021 Technological Forecasting
& Social Change
BA
& BDA
The role of data-driven insights on the relationship between BDA & BA and decision making
- BA is positively associated with data-driven insights, which then lead to improvement of decision making quality.
- BDA has a direct effect on decision making quality, while BA fully relies on data-driven insights as a mediating link - BDA has a more direct and indirect relationship (through the mediating factor data-driven insights) with decision making than BA.
Cao, Duan
& Li
2015 IEEE transactions on engineering management BA
Linking BA to decision making effectiveness
- BA has a positive effect on IPC, which has a positive effect on DDDM, which improves decision making effectiveness.
- For BA to be productive, an organization requires a DDE simultaneously to support and enable BA activities - DDE improves IPC, which improves DME.
Chen, Chiang
& Storey
2012 Management Information
Systems BA
Characteristics and influence of BA on organizations
- Intelligence from BA leads to insights on consumer opinion, customer needs and recognizing new business opportunities - For BA to provide useful insights and decision making support, employees must be capable of understanding the business issues and framing the appropriate analytical solution.
Davenport
& Harris 2007 Harvard Business School
Review Press BA
Competitve advantages of BA and essential resources to exploit it
- Companies who have been identified as aggressive analytics competitors are clear leaders in their fields
- Employees hired for their expertise with numbers or trained to recognize their importance are armed with the best quantitative tools. As a result, they will make the best decisions.
Davenport, Harris, De Long
& Jacobsen
2001 Sage Journals BA
Primary succes factors needed to transform data into knowledge and then into business results
- Integrating BA in an organization improves its data processing capabilities. These insights, enable to make effective decisions and so result in an improvement of firm performance - For important decisions, skilled workers are essential when applying data analysis.
Ghasemaghaei, Ebrahimi,
& Hassanein
2018 Journal of Strategic Information Systems BA
The influence of data analytics competency on firm decision making.
- Data analytics competency is significantly associated with decision making performance.
- Data quality, analytical skills, domain knowledge and tools sophistication significantly impact decision quality and decision efficiency.
- Bigness of data significantly increases decision making quality, but it does not significantly influence decision making efficiency.
Ghasemaghaei, Hassanein
& Turel
2017 Decision Support Systems BA
The role of fit between data analytics and other organizational resources on a firms agility
- The often-take emphasis on the mere use of analytics is somewhat hype and proper calibration of organizational resources related to data analytical tools is critical for improving organizational outcomes through the use of data analytics
Hedgebeth 2007 VINE BA
The development of BA and its effect on organizations
- There should be an organizational emphasis on data analytics in order to successfully implement it in a company
Lavalle, Lesser, Shockley, Hopkins
& Kruschwitz
2011 MIT Sloan
Manage. Rev BA
The way organizations try to exploit data and make use of data analytics
- New insights generated through BA lead to better deicison making.
- BA must be implemented enterprise-wide in order to result into value rapidly.
Davenport 2014 Strategy & Leadership BDA
The support of BDA on internal business decisions
- Managerial knowledge about big data and BDA is lacking - Big data and BDA can facilitate in decisions regarding customer satisfaction, managing supply chain risk, competitive intelligence, pricing of goods and services and discovering & experimenting.
Dubey, Gunasekaran, Childe, Blome &
Papadopoulos
2019 British journal
of management BDA
The importance of big data resources for skills, big data culture for improving firm performance
- Data-driven culture is essential to exploit BDA capabilities to its full potential
- BDA is one of the most important factors for generating meaningful insights
- BDA is very likely to improve the organizational ability to make decisions
Fast, Schnurr
& Wohlfahrt 2021 Association for
Information Systems BDA
Business advantages of big data in digital markets
- Benefits of big data and BDA are not necessarily transitory, but can lead to an established long-term power based on the collection and use of big data.
- As the importance of big data increases, new policy measures should be taken to control or mitigate power of dominant parties.
Kopanakis, Vassakis
& Mastorakis
2016 Management of
Innovative Business BDA
The influence of big data on innovation and performance of businesses
- Big data provides decision makers with the capability to innovate and increase organizational performance gaining a competitive advantage against rivals.
Ko´scielniaka
& Putoa 2015 Management and
Information Technology BDA
Method of integrating the use of big data in the process of decision making in enterprises
- The quality of decisions taken in the contemporary operations of enterprises is affected by efficient and effective use of available data sets
- Four stages of supporting decision making based on big data are determining the authorized source of data, using individual scorecards among employees, explicit management of business rules and the use of coaching to improve the effects of enterprise activity
Merendinoa, Dibba, Meadowsa, Quinna, Wilson, Simkina
& Canhotoc
2018 Journal of Business
Research BDA
The effects of big data on the process of decision making for managers, boards and entire organizations
- Individual directors show a gap of cognitive capabilities to cope with such enormous amount of data.
- At the level of the board, big data disrupts its board cohesion.
- External parties and stakeholders can play an important role in handling big data and developing a companies BDA.
Shamim, Zeng, Khan
& Ul Zia
2020 Technological Forecasting
& Social Change BDA
The role of contractual and relational governance on BDA and decision making performance
- Contractual and relational governance are positively associated with decision making performance. The indirect relationship through BDA is significant as well.
- Strong relational and contractual governance leads to better data, which improves BDA, which enhances decision making
performance
Brynjolfsson,
Hitt & Kim 2011 SSRN Electronic
Journal DDDM
The effect of DDDM on business performance
- Integrating data-driven decision making into a businesses processes, increases their productivity output by 5-6 percent.
- Return on assets and asset utilization have been found positively correlated with DDDM.
- Companies with a larger adoption of DDDM represent a higher market value.
Brynjolfsson
& Mcelheran 2016a American Economic
Association DDDM The adoption of
DDDM
- DDDM is positively correlated with entrepreneurial productivity.
- College educated workers are far more likely in dealing with high levels of DDDM.
- DDDM is integrated faster and more prevalent among multi-unit firms who have higher levels of employment.
Brynjolfsson
& Mcelheran 2016b SSRN Electronic
Journal DDDM
Performance effects of DDDM in U.S.
manufacturing
- More diffusion of DDDM in a company will always result into a better organizational performance.
- Centralized coordination is of great importance in order to effectively increase a firms productivity using DDDM.
Princes
& Kosasih 2021 Journal of Southwest
Jiaotong University DDDM
The importance of intuition in the process of decision making
- Intuition derived from experience is required in any condition and DDDM is incomplete for making final decisions autonomously.
- DDDM should be used as a supportive tool in the decision making process.
Troisia, Maionea, Grimaldia,
& Loiab
2020 Industrial marketing
management DDDM
The influence of DDDM on decision making
- DDDM is used for decisions in product design, delivery and in the development of innovation.
- Communication and data transfer among different departments of an organizations are found to be important factors for improving decision making effectiveness.
.
6 Discussion and Conclusion
This study contributes to the existing literature by providing a status update about data and data analysis techniques in the process of entrepreneurial de- cision making. This literature review synthesized all current knowledge about this topic. At the same time, gaps in the literature have been identified. Those gaps have led to new research ideas, which can result into additional literary knowledge. Also, this research will have a contribution to enterprises in sub- stantiating information on implementing, adopting or using analysis techniques in order to improve organizational decision making.
In this final section, the gaps in the literature will be addressed first. From these gaps, ideas for further research have been developed. This part will con- tain propositions for further research, resulting from current findings in this SLR. This will be followed by a few limitations that have been faced during the execution of this literature review. Lastly, the main findings and conclusions will be presented once again to finalize this SLR.
6.1 Gaps in the literature & Ideas for further research
By creating a synthesis of this topic, various gaps in the literature have been detected. These can be areas of research that have not received the deserved amount of attention, that have not been studied at all, or fields of study that have emerged out of previously conducted research.
The first gap that has been noticed is that almost every study only focuses on board or managerial level decision making. At first, this does not seem strange, as managers are supposed to make the most influential decisions in an organizations and should have the optimal understanding and knowledge of the entire organization. However, nowadays there are more people in an organization responsible for decision making and it is not only a managerial task anymore. Often, the use of data and corresponding analysis techniques has dispersed across several or even all departments of a company (Hufnagel
& Vogel-Heuser, 2015). Consequently, deciding based on data and data-driven insights has spread over the entire company as well. Especially, by using data and data analysis techniques, other people will be as important when it comes down to decision making or assisting the managers with their decision making (Landry, 2020). A study performed by Lee and colleagues already noticed the organizational urgency of hiring employees specialized in data, like data officers, information officers and analytics officers (Lee et al., 2014). Those people are hired to improve the use of big data in the process of decision making. So, this immediately results into ideas for future research. As research on data-driven decision making can be conducted on other employees in an organization, like IT workers, data scientists/engineers or front-line workers, this might change the decision making quality or firm performance. Researching the influence these employees have on the decision making process, might lead to insightful results.
A study executed by Brynjolfsson and colleagues, already investigated the ef- fect of joint decision making between front-line workers and managers. Results showed that this decentralized type of decision making does not benefit the company (Brynjolfsson & McElheran, 2016b). As this study only investigated the U.S. manufacturing sector, and this subject was not of primary focus, more research is needed to confirm this finding. However, this finding has led to the following proposition:
Proposition 1: Joint decision making between managers and other employees does not benefit the organization.
At the same time, this brings a methodological gap in the conducted research.
Almost every study researching the topic decision making based on data, re- trieves its data by sending surveys to businesses or organizations. However, all of these surveys are sent to board members or managers as they are ought to have the best organizational-level business-related and technology-related view.
However, as discussed earlier, this is not always the case anymore considering data-driven insight. As a result, experimental findings based on surveys can become biased towards the managers. An option to prevent this, is to retrieve information by sending not only surveys to managers, but to all or most of the companies employees. Most of the time, lower ranked employees have another view or opinion than the manager about organizational processes (Straz, 2016).
This may result into different outcomes.
Although almost every study researching this topic uses surveys, this is not the best way to retrieve the data. As most of the time, those surveys are sent to business owners or managers, the results can become biased towards the company. On top of this, each study uses its own preferred scale, which may influence the results as well. An idea for future research, might be to retrieve the data in another way. One way might be to generate the data more empir- ically. This can be done by measuring the amounts of data a business uses, or the number of decisions are being based on data in an organization. This way, the results might become less biased towards the people at the company.
Another gap that has been detected is about the type of decisions being made based on data and data analysis techniques. There has been very little re- search on this specific topic. Some studies mention strategic business decisions, organizational decisions or more responsive decisions. However, this does not tell us anything about the type of decisions and what this exactly means for the company. An idea for future research can thus be to investigate for what kind of decisions companies use their data analysis. These can be opportunity assessment decisions, entrepreneurial entry or exit decisions, exploiting oppor- tunity decisions or decisions regarding employees. At the same time, the type of business processes can be considered. For what business processes data-driven decision making will be used, can be quite divergent as well. Will data mainly be used for decisions involving sales, profit maximization, cost minimization, customer satisfaction, process efficiency or any other business process. These two elements combined can form an interesting topic to investigate for future research. Previous research has linked data analysis techniques to certain types of decisions. According to a study performed by Troisia and colleagues, DDDM is mainly being used for decisions concerning product design, delivery and de- velopment of innovation (Troisia et al., 2020). Another study linked BA to decisions regarding customer opinions, customer needs and new business op- portunities (?, ?). BDA has been linked to customer satisfaction, managing supply chain risks, competitive intelligence, pricing of goods and services and discovering & experimenting (Davenport, 2014). However, confirmation of these findings is absent. If, indeed, specific data analysis techniques could be linked to certain departments of a company or even connected to specific decision choices, data analysis techniques could be applied more effectively in different situations.
Companies could choose a specific analysis technique that suits the situation, resulting in an optimal outcome. Because of this, it would be an interesting
