Data-driven decision-making for efficient & sustainable production

(1)

INOM

EXAMENSARBETE INDUSTRIELL EKONOMI, AVANCERAD NIVÅ, 30 HP

STOCKHOLM SVERIGE 2021,

Data-driven decision-making for efficient & sustainable production

ARVID BROMS

SIMON LILJENBERG OLSSON

KTH

(2)

(3)

Data-driven decision-making for efficient &

sustainable production

by

Arvid Broms

Simon Liljenberg Olsson

Master of Science Thesis TRITA-ITM-EX 2021:228 KTH Industrial Engineering and Management

Industrial Management SE-100 44 STOCKHOLM

(4)

Datadrivet beslutsfattande för effektiv och hållbar produktion

Arvid Broms

Simon Liljenberg Olsson

Examensarbete TRITA-ITM-EX 2021:228 KTH Industriell teknik och management

(5)

Master of Science Thesis TRITA-ITM-EX 2021:228

Data-driven decision-making for efficient &

sustainable production

Arvid Broms Simon Liljenberg Olsson

Approved

2021-06-08

Examiner

Matti Kaulio

Supervisor

Pernilla Ulfvengren

Commissioner

KTH-Industrial

Transformation Platform

Contact person

Zuhara Chavez Lopez,

Maheshwaran Gopalakrishnan

Abstract

As a result of digitalization, previously analog systems in the manufacturing industry have become digitalized, including the decision-making processes. Companies are, therefore, becoming more dependent on data for strategic decisions. However, because of the rapid development of digitalization, companies are left blindfolded in the path towards smarter manufacturing which often leads to unsuccessful technological implementations.

Therefore, the thesis will explore this problem by asking: What are the required initiatives for successfully implementing digital data-driven decision-making to improve efficiency and sustainability by Swedish manufacturing companies?

To answer the research questions, an exploratory multiple case study approach was conducted, where interviews with informants from the industry as well as researchers within the context of smarter manufacturing were made. The findings were then used to derive propositions which worked as the foundation of a conceptual model which

functionality would be to illuminate the results in the form of a strategy map.

Findings suggest that it is not always necessary for companies to implement technologies linked to large investments to enable digital data-driven decision-making. However, for those that do, there needs to be a clear organizational plan and agenda before executing the projects since they otherwise often lead to insufficient results. That means, the

technological aspects are often not the culprit in failed digital data-driven decision-making projects. Additional findings suggest that there are synergies connected to digital data- driven decision-making such as data-sharing possibilities that have the potential of becoming a major aspect within the context of sustainability and efficiency.

(6)

Examensarbete TRITA-ITM-EX 2021:228

Data-drivet beslutsfattande för effektiv och hållbar produktion

Arvid Broms Simon Liljenberg Olsson

Godkänt

2021-06-08

Examinator

Matti Kaulio

Handledare

Pernilla Ulfvengren

Uppdragsgivare

KTH-Industrial

Transformation Platform

Kontaktperson

Zuhara Chavez Lopez,

Maheshwaran Gopalakrishnan

Sammanfattning

Som ett resultat av ökad digitalisering har analoga system i tillverkningsindustrin blivit digitaliserade, vilket inkluderar beslutsfattandet. Företag har därför börjat förlita sig allt mer på data för sina strategiska beslut. Men på grund av den snabba utveckling av digitalisering har tillverkningsföretagen lämnats utan klara riktlinjer för hur de bör gå tillväga för att implementera digitalt datadrivet beslutsfattande på ett effektivt men hållbart sätt. Avhandlingen kommer därför att undersöka detta problem genom att fråga:

Vilka är de initiativ som krävs för att framgångsrikt implementera digital datadrivet beslutsfattande med målet att förbättra effektiviteten och hållbarheten hos svenska tillverkningsföretag?

För att svara på forskningsfrågorna användes en undersökande metod med flera

fallstudier, där intervjuer gjordes med informanter från industrin såväl som forskare inom ramen för smartare tillverkning. Resultaten användes sedan för att härleda förslag som därefter användes till konstruktionen av en konceptuell model vars huvuduppgift var att illustrera resultaten i form av en strategikarta.

Slutsatserna pekar på att det inte alltid är nödvändigt för företag att implementera teknik kopplad till stora investeringar för att möjliggöra digitalt datadrivet beslutsfattande. Men för de som valt att implementera sådana system behövs en tydlig organisationsplan innan projekten genomförs eftersom de annars ofta leder till ofördelaktiga resultat. Detta tyder på att de tekniska aspekterna oftast inte är vad som orsakar misslyckade datadrivna beslutsprojekt. Dessutom tyder resultaten på att det finns synergier kopplade till digitalt datadrivet beslutsfattande, till exempel möjligheter att dela data som har potential att bli en viktig aspekt inom hållbarhet och effektivitet.

(7)

List of Figures

1 A scheme defining the fundamentals of a managerial process according to Hartmann et al. (2014). . . 17 2 A Strategy Map according to Hartmann et al. (2014) . . . 18 3 A defining the investigation areas of the project . . . 23 4 The fundamentals of a managerial process according to Hart-

mann et al. (2014) with the conceptual model domain high- lighted in blue. . . 55 5 The conceptual model . . . 57

List of Tables

1 Disposition of the thesis, with short summary of each chapter. 5 2 Conducted interviews with a short summary of topic and aim. 25

(10)

Nomenclature

• AI - Artificial Intelligence

• BSC - Balance Scorecard

• CPS - Cyber Physical Systems

• DDDM - Data-Driven Decision-Making

• ERP - Enterprise Resource Planning

• EMS - Environmental Managements Systems

• HSE - Health, Safety and Environment

• IoT - Internet of Things

• KPI - Key Performance Indicators

• KRI - Key Results Indicators

• MMEI - Maturity Model for Enterprise Interoperability

• OEE - Overall Equipment Effectiveness

• PI - Performance Indicators

• ROI - Return on Investment

• SCM - Supply Chain Management

• TPM - Total Productive Measurement

• VR - Virtual Reality

(11)

1 Introduction

In this chapter, the current state of the manufacturing industry in Sweden is introduced and subsequent challenges that requires research to be resolved.

Then the purpose of thesis is presented in short and the following research questions are formulated. The chapter thereafter ends by presenting the de- limitation of this thesis.

1.1 Background

The decisions one makes are often based on the information gathered to reach a rational conclusion, in other words, data-driven decision-making (DDDM).

Throughout history, that data has been collected from observations, intuition, or from other people, and as the concept of organization has evolved, structures have been developed in order to simplify the flow of data from the different parts and branches of the organization to the decision-making authority of the organization. However, as the organizations have grown larger, and the amount of information increased, it has become harder for the decision-makers to gather relevant information, subsequently increasing the risk of inadequate decisions which, depending on the context, might affect the outcome of the company in a negative way.

However, as the phenomenon of digitalization has arisen, its influence has caused new solutions to earlier analog systems, DDDM being one of them.

Organizations have realized that through digital data, more information can be gathered at a faster pace, and with less mistakes since the human error is left out to a greater extent. The manufacturing industry is no exception, resulting in new technologies and concepts being introduced as the digital tools become more mature and accepted. Today, manufacturers are using technology to increasingly automate activities in their processes (Engwall et al. 2017). A buzzword commonly used to describe the application of these technologies is Industry 4.0, however, because of its immaturity, several researchers tend to define it differently (Müller et al. 2018, Bärring et al. 2018, Lindström et al. 2019). Moreover, the concept of Industry 4.0 is the notion of using the very powerful technologies that exist today in a way that will transform the current industry paradigm into the next. Calling it smart manufacturing can ease the need for a stricter definition hereafter. The essence is, however, that smart manufacturing is the idea of using facts and metrics to guide decisions rather than intuition, and thus data-driven decision-making

(12)

can be seen as a more practical term for what smart manufacturing entails.

Moreover, associated with the technologies of the new industry paradigm are the goals of using it to conquer current challenges in the world. The sustainable aspect has become more prevalent in the manufacturing industry, leading to more emphasizes being put to that area. Threats such as increased competition and volatile demands are still seen as strong drivers for organizations to acquire new technology that enables digitization of data to make manufacturers more competitive (M¨uller et al. 2018). However, in the crossfire for global dominance between competitors lies the ecological and societal aspects of the manufacturing industry. United Nations, the European Commission and the World Economic Forum all agree that global warming is a fact, and that the world needs to change how it han- dles resources for humanity to survive long term. The ecological and societal aspects of sustainability are, therefore, two of the main dimensions of this thesis, since smarter manufacturing could result in higher resource efficien- cies in many ways. For example, by using data analytics and cyber-physical systems (CPS) for higher energy efficiency and longer life cycles with the use of shared data, improved employee motivation with education and increased material efficiency by using data analytics.

Even though the manufacturing industry is experiencing benefits through the implementation of these new technologies, there are still major difficulties connected to them since there currently are no clear guidelines on how companies are supposed to act or how to successfully implement these tools when deciding to invest in digital tools for their data-driven decision- making. Research by (Zdravković et al. 2018) has shown that the implementation of a digitization project often fails to capture the intended goal of such projects, resulting in unsuccessful ventures. Furthermore, scholars have identified unwanted trajectories in the realm of efficiency in Swedish manufacturing, namely, Overall Equipment Efficiency (OEE). Instead of an increasing OEE, which could be expected due to improved digital tools, the Swedish manufacturing industry has experienced a declining OEE over the last decades(Ylipää et al. 2017, Ljungberg 1998).

For the potential of smarter manufacturing to be realized, a successful adoption of digital data-driven decision-making is needed. Managers in organizations can, therefore, benefit from a model visualizing the benefits of certain initiatives and strategies to reach this potential. The model aims to show in what ways data-driven decision-making can improve efficiency and sustainability in manufacturing organizations and what initiatives organiza-

(13)

tions can take to make data-driven decision-making a successful part of their organization.

Anchored in an exploratory multiple case study, the research aims to explore digital data-driven decision-making (DDDM). By investigating the challenges and possibilities in current practises and future ambitions, initiatives for successful implementation of digital DDDM will be developed.

1.2 Purpose

The purpose is to develop a model that explains the phenomena of digital data-driven decision-making (DDDM) within the Swedish manufacturing industry, which can guide managers in their quest towards more efficient and sustainable production through the usage of digital DDDM.

1.2.1 Research question

To address the purpose of the thesis, the following main research question (MQ) was derived, and its consequent sub-questions (SQ), where the SQ:s are designed to answer the MQ in an analytical approach by breaking down the MQ into two smaller SQ:s.

• MQ1 - What are the required initiatives for successfully implementing digital data-driven decision-making to improve efficiency and sustainability by Swedish manufacturing companies?

• SQ1 - What are the driving incentives for implementing digital data- driven decision-making in the Swedish manufacturing industry?

• SQ2 - What are the challenges in implementing digital data-driven decision making in the manufacturing process?

1.3 Delimitations

The thesis is limited to only look at the Swedish manufacturing industry, thus, only gathering data from manufacturers that have production in Swe- den. The reason is that by limiting the scope to just one country, the compar- ison between organizations can be done more fairly because of, for example, laws and regulations that might give organisations an uneven set of rules to

(14)

play by. However, because of the exploratory nature of the study, no limitations have been set regarding the type of manufacturing that the company has, nor the size of the organization.

Furthermore, the concept of data-driven decision-making is woven into the decision one makes daily which means that it comes in different forms, for example, digital or analog. However, this thesis will delimit itself to the digital aspects of the subject, meaning, automated processes, digital solutions etc. The reasoning behind that choice is mainly due to the current trajectory towards ”smart” manufacturing, as defined in the introduction. Therefore, the thesis will refer to the digital aspects of data-driven decision-making throughout.

(15)

1.4 Disposition

The disposition is displayed in table 1 where the outlining of the thesis, together with a summary of the chapter is depicted.

Chapter Chapter description

1. Introduction Introduces the background of the project which is followed up by the problem formu- lation with the idea of funneling the reader to the purpose of the study. Thereafter, the research questions are presented and terminates with the delimitations of the project.

2. Literature Review Gathering relevant secondary data from prior literature related to the areas of data-driven decision-making and sustainability.

3. Method Depicts the chosen methodology used through-

out the thesis. It then further explains how the data was collected, analysed and, thereafter, used to derive a result.

4. Findings Presents the empirical data gathered from interviews.

5. Discussion Discusses the findings, its implications on the research area, and a conceptual model which was the purpose of the thesis, as well as the limitations of the study.

6. Conclusion Concludes the thesis by presenting the main findings, followed by a presentation of sug- gested future research questions which ultimately terminates the thesis.

Table 1: Disposition of the thesis, with short summary of each chapter.

(16)

2 Literature review

In this chapter, literature on the subject of data-driven decision-making (DDDM) within the context of manufacturing is presented. Both an overview of the technical aspects of DDDM and the sustainable aspects of it is presented.

Lastly, fundamental concepts of managerial processes is analyzed to build a foundation on which the final model of the thesis is built upon.

2.1 Digital aspects of data-driven decision-making

To understand the phenomenon of data-driven decision-making and the digital tools that exist within the field of manufacturing, a review of current methods and tools is conducted.

2.1.1 Data sources

As technology has progressed over the past century, data has become an increasingly implemented factor within the manufacturing industry. As of 2010, the manufacturing industry was storing approximately 2 Exabytes of new data which was the largest amount in any industry (Nedelcu 2013).

However, the strength and possibilities that enters along with ”big data”, meaning data that comes in large volumes, is not solely tied to the size of the information, albeit valuable. The greatness lies within the information and the knowledge that it can provide(Tao et al. 2018). Tao et al. (2018) continues by presenting five types of data sources that are used in the field of manufacturing:

1. Management data from manufacturing information systems. Included in this category are systems that are handling data regarding every aspect of supply chain from product planning to customer service. Ex- amples of systems are Enterprise Resource Planning (ERP) and Supply Chain Management (SCM).

2. Internet data, which is information that has been gathered from online platforms. The information collected can be used as guidance as to further develop customer experience, for example.

3. Equipment data, where information is collected in real time through technologies such as IoT to have access to data such as maintenance history and conditions of the machinery.

(17)

4. Product data, which is also collected through similar technologies as the equipment data. Examples could be temperature, and humidity to ensure the quality of the products.

5. Public data, where information is provided through open sources such as data bases or from the government. In the context of production, public data could be regulations which are needed to ensure compliance with manufacturing regulations and laws.

2.1.2 Data collection

Data collection methods vary, however, in order to decide what data needs to be collected, one must identify certain parameters. Znamen´ak et al. (2016) presents three key ratios to consider:

1. KRI - Key Result Indicators. Long term data, such as financial measures such as earnings or non-financial ratios such as customer satisfac- tion. Cost per part unit is proposed by Znamen´ak et al. (2016).

2. PI - Performance Indicators. Data that indicates what the production should do and therefore works as a middle ground between KPI and KRI. Proposed by Znamen´ak et al. (2016) are Flow time, Scrapped number of products and number of finished products.

3. KPI - Key Performance Indicators. Data that entails core information about the production’s performance. Short term data that are non- financial that has significant impact. Znamen´ak et al. (2016) propose several parameters such as, Reject ratio, Overall Equipment Efficiency (OEE), Downtime, and Lead time.

However, there are challenges connected with these different types of collection methods. Firstly, there are different formats in which the data can be stored. Thus, causing incompatibility when trying to integrate the data.

Zhong et al. (2016) presents the example of two companies that are trying to merge their transactions, the integration is prone to be unsuccessful due to the lack of standardization. Furthermore, due to the sheer size of the data volume that Big Data introduces, signal collisions are often caused which result in incomplete data collection. This suggests that Big Data still is in its early phase of maturing, which means that further development is needed before Big Data can be used to its full extent and potential.

(18)

2.1.3 Data storage

Challenges regarding data storage are currently one of the greatest hurdles with regards to Big-data decision making (Zhong et al. 2016). Since the data flows are large, there are difficulties for the hardware to handle the traffic.

Zhong et al. (2016) presents two possible solutions: Cloud-based services and smart storage mechanism. The possible benefits that are presented by these technologies are not only the ability to handle large amounts of data. It also provides intelligent methods such as self-optimization and self-management in order for the storage system to work even more efficiently. Wei & Li- hui (2017) argues that data which is collected from manufacturing processes needs to be addressed and stored differently with regard to its size, and therefore suggest three categories: cloud data, local network data and local data.

The local data is information stored in distributed computers and includes historical and real-time data of the machinery, processes, and quality mea- surements as examples. Local network data analysis performance regarding its potential and is therefore analyzing historical data. Cloud storage is also covering historical data, however, focusing more on capacity and time.

2.1.4 Data processing

Data processing refers to the process of analyzing data flows to extract knowledgeable information that can be exploited to enable higher efficiency, in this context, an example could be a greater Overall Equipment Efficiency (OEE).

Algorithms used to analyze big data are several. Wei & Lihui (2017) presents the following six: Cluster Analysis, Factor Analysis, Analysis of correlation and dependence, Regression analysis, A/B testing and Data mining. How- ever, one major aspect is the exclusion process of redundant data (Tao et al.

2018). Thus, through data cleaning and reduction, the data can be exploited through the techniques listed by Wei & Lihui (2017).

2.1.5 Data visualization

One of the key areas of big data implementation is to excavate data that can be used to improve processes in a way where the workers at the site can easily understand and interpret the data. However, even though it has become easier to implement tools that collect data, visualization techniques are crucial to actually be able to understand the information which the data can grant. Zhong et al. (2016) presents various visualization strategies used

(19)

by different manufacturing companies, however, stresses the fact that the tool needed to fully extract the correct and full potential of the data, one needs to tailor the tool to the context of the type of manufacturing. Open-source data visualization is preferred by the end user and should be easily linked to other already existing tools and programs at the site.

2.1.6 Data transmission

Since Big data subsequently results in high volumes of data traffic, it is essential that to manage the large amount of data, then there needs to be a transmission infrastructure that can cope. The current network of 4G internet, is not capable of handling those large numbers. However, modern technologies such as 5G can change that complication. Furthermore, small errors connected with the transmission can result in major errors since it can lead to butterfly effects, which means that small problems lead to major complications at a later stage (Zhong et al. 2016). These problems call for more advanced technologies and systems that are reliable. Tao et al. (2018) suggests that modern technologies such as IoT and as previously mentioned, 5G will result in faster and more reliable transmissions.

2.1.7 Data application

With new technologies implemented that enable real-time data to influence production, companies can be more agile in its production. Meaning, the production can through real time data-driven decision-making cut down on energy consumption and plan maintenance more efficiently, subsequently resulting in higher OEE. Furthermore, the sustainable aspect, which is becoming a highly prioritized subject for manufacturers, can be positively affected by this trajectory as well (Tao et al. 2018).

2.1.8 Data-sharing

An area which is linked to the context of data collection, is the possibility to share data. Sharing data is not only limited to sharing information within one company. Instead, it can be used to trade information with other companies as a tool to, for example, benchmark ones performance compared to the industry standard. Dreller (2018), Davis et al. (2015) suggests that there are several possibilities within the realm of data sharing, which has been explored to some extent in the world of science, it has not become as adopted

(20)

within the business world. On the contrary, many companies do not see the benefits that could come with it, and instead choose to further distance themselves from the topic. Challenges that also were identified to be applied to the subject were lack of standards, rules, and agreements. Furthermore, Zdravkovi´c et al. (2018) stresses the importance of interoperability to achieve a sustainable development of IoT ecosystems. To combat this challenge, Zdravkovi´c et al. (2018) suggests using maturity criteria, namely, a maturity model for enterprise interoperability (MMEI).

2.1.9 Performance data

In the context of measuring the performance of a factory, there are currently several methods one can use. Overall Equipment Efficiency (OEE), which has been developed from the Total Productive Measurement (TPM) concept presented by Nakajima (1988), is one example of a performance measurement, which is commonly used by manufacturers. It is built upon three factors:

availability, performance, and quality of the output and is used to get a grasp of how well the production is functioning with regards to how much it is supposed to do. However, this model has aspects, which need to be addressed to be fully functional in practice. Firstly, as presented by Muchiri

& Pintelon (2008), OEE focuses on internal operation-related errors, which affect production. By definition, it does not include business related such as organizational problems nor external errors into account. The way in which the data to the OEE is collected is also an important factor. Manual data collection is not reliable, since there are human errors which can affect the results. Instead, it is preferable if automatic collections are implemented to reduce human-related errors.

Moreover, through a recent study by Ylip¨a¨a et al. (2017), conducted on manufacturers in the Swedish industry, the OEE was found to be approximately 51.5%. The result illustrates an area of improvement within the Swedish manufacturing industry. Furthermore, prior research shows that the OEE has decreased from 55% (Ljungberg 1998) within the industry when the contrary would be expected since technology has been developed in that field since 1998. Ljungberg (1998) continues by presenting the differences in efficiency when comparing new processes to old, where new processes had an OEE of 45% and old processes 65%. The results indicate that there are problems connected with new processes which result in a lowered OEE.

Furthermore, Parida et al. (2014) argue that the largest problem in the

(21)

present situation of the manufacture industry is a low OEE, since it at current levels remain lower than what is targeted, which is around 15-25% higher.

2.2 Environmental, social, and economical aspects of digital data-driven decision-making

To get an overview of the the three main aspects of sustainability in the context of manufacturing, a review of literature is conducted.

2.2.1 Environmental aspects

Progress towards smarter manufacturing is also progress toward a more sustainable industry from the environmental perspective. In the context of digital data-driven decision-making Stock & Seliger (2016) suggest to view the whole product life cycle as a closed loop within different manufacturers where shared data, materials, and energy is a new competitive advantage, and they call it industrial symbiosis (cross-company) which co-aligns with Dreller (2018), who suggests that the realm of data sharing has great potential. More efficient usage and allocation of resources such as electricity, fresh water and materials can be created with a connected value creation chain (Stock & Seliger 2016, Kusiak 2019). A closed loop approach to the manufacturing industry is also proposed as an important aspect of reaching sustainable manufacturing. Ma et al. (2020) argues that a circular econ- omy plays an integral part of energy intensive industries that aims to be contributing to a more sustainable development. Energy intensive manufacturing includes pulp and paper, ceramic and steel production Ma et al. (2020) while discrete manufacturing such as milling, and drilling are modest users of energy Kusiak (2019). Considering that 38 percent (2018) of Sweden’s total energy consumption is used by the industry and that 51 percent (2018) of that goes to pulp and paper manufacturing (Swedish Energy Agency 2019) goes to show that a closed loop strategy probably will be an important one for the Swedish manufacturing industry.

Another promising solution that Stock & Seliger (2016) mention is to retrofit typical capital-intensive machinery with sensors and actuators to achieve a cyber-physical system in a cost-effective and environmentally friendly way, this method is tested successfully in their paper as a use case. With smarter manufacturing, a higher flow of data can be achieved and thus also information regarding the run-time of specific tools will increase, so planning

(22)

and predicting maintenance will thus increase the lifespan of tools, which in turn improves both the economic and the ecologic sustainability Sj¨odin et al.

(2018).

The opportunity with smart manufacturing at the bigger scale is what Stock & Seliger (2016) calls the essential advantage of smart manufacturing, namely the usage of smart grids. Smart grids where factories with self- sufficient renewable energy supply can be used jointly with an external smart grid. Moreover, together with factories’ own energy management systems, the feedback and supply to and from the grid can lead to synergies between companies that can both act as a supplier of energy and a user depending on the situation, an example is a pulp manufacturer that can use waste products for energy that can be sold to other manufacturers. The two viable sources of renewable energy of today is energy from solar (both thermal and photo-voltaic) and wind (Kusiak 2019). Kusiak (2019) also mentions that with smart planning and scheduling, reducing the idling time and more efficient processing can save energy. He also proposes the idea of using “la- bels of environmental friendliness” like nutrition stickers on foods to increase awareness.

2.2.2 Societal & cultural aspects

There is no question that company culture and their employees play an essential part when adopting new ways of working, especially in the manufacturing industry. Shahbazi et al. (2016), G¨urd¨ur et al. (2018), Alieva & Haartman (2020), Linnenluecke & Griffiths (2010) all mention human resources and culture as an integral part in handling some sort of change in the industry. Digitization, sustainability focus, and data analytics are all relatively contemporary subjects within the manufacturing industry. Linnenluecke &

Griffiths (2010) examines a link that scholars have for long time seen as the truth, that is, if an organization wants to adopt sustainability principles, they need to adopt or create a sustainability-oriented culture within the organization itself. They investigated if organizations can display a sustainable culture in a unified way or there are subgroups within the organization that displays different cultures. Although it can exist subcultures with differenti- ated views among each group, the results show that adoption of sustainability principles at the surface level, for instance in corporate annual reports or employee training, can induce change in employees values, beliefs but also core assumptions. However, Linnenluecke & Griffiths (2010) also provides with a

(23)

number of barriers and other limitations for these kinds of changes, especially in the context of what the dominant culture is, for instance, one culture that is focused on stability and control uses a lot of precise communication and data-driven decision-making can be very rigid in terms of cultural change.

The existence of subcultures in the organizations is in itself seen as a barrier for change. But to fully respond to the changes regarding social and environmental challenges, organizations will have to transform and change their culture significantly (Linnenluecke & Griffiths 2010).

In a paper by Shahbazi et al. (2016) that investigates material efficiency in the Swedish manufacturing industry, found that productive materials (i.e.

metals) had a much bigger economic incentive to be handled correctly as scrap could easily be sold in relation to the less valuable plastics, cardboard, and similar residual materials. The study found the cultural and HR empirical barriers for improving material efficiency to be lack of goals, white-collar oversight, oversight from employees caused by indolence and weariness, lack of life-cycle mindset and ultimately lack of training. Gürdür et al. (2018) argues that companies will have a hard time achieving any sort of success with any change without addressing the corporate culture in which the change is to be used in. In the same paper by Gürdür et al. (2018) it is argued that the industry has a high resource readiness in terms of human resources and tools, and high readiness in terms of the cultural aspects of workers acceptance of data-driven decision-making and high information system readiness. The paper suggests that the industry should leverage the strength of resources, information, and culture to explore the impact data analytics can have on the business side of operations and thus foster organizational readiness for structural changes.

There is also the case that untrained and unqualified people cannot use the gathered data as intended. Thus, as mentioned by Alieva & Haartman (2020), a new form of waste can be defined as digital waste, also known by the Japanese word for waste, muda. Furthermore, Alieva & Haartman (2020) defines digital muda as data that is collected but not processed or processed but not analyzed or not collected at all when possible. But for cases where data is analyzed although unsuccessfully can be seen as a value-adding ac- tivity as long as it is a learning process. As mentioned in the article by R¨ußmann et al. (2015), manufacturers must prioritize upgrading the compe- tence of the workforce to handle the technological shift. To solve the issue of unqualified staff, which can be one of the main causes for not processing the gathered data, Stock & Seliger (2016), Sj¨odin et al. (2018) suggest retraining

(24)

workers with the use of virtual reality. Sj¨odin et al. (2018) investigated what types of challenges there were in a manufacturing organization with multiple factories and came to the conclusion that nurturing digital savvy people was a key enabler when implementing smart manufacturing when it came to the cultural level.

2.2.3 Economical aspects

As previously mentioned, Mart´ınez & Silveira (2012) recognized that it is possible to produce more with less. However, these results only show one side of the coin. The inefficiencies housed in Sweden have moved towards other parts of the supply chain, and the energy needs have gone up, resulting in an increasing nuclear energy mix in Sweden (Mart´ınez & Silveira 2012).

However, one of the main drivers to achieve higher efficiency has been the financial incentives to lower energy usage and lower carbon emissions by gov- ernmental measures, such as taxes and tariffs. Mart´ınez & Silveira (2012) argues that financial incentives has led to a higher number of investments in technology and improvements for efficient energy usage. Basic capital investment theory says that if there are multiple purposed investments to consider, prioritizing the one with the highest return is the logical thing to do. However, as Sa et al. (2017) argues, there are other factors that play an important role when making a strategic investment, such as sustainability influenced culture, managerial interests, and how the investment links to the core business and core strategy of the organization. In the same paper, a survey of around 100 companies in Australia came to the conclusion that 35 percent of energy efficiency investments were not adopted because of their weak connection to the main business. Sa et al. (2017) also lists typical decision-making barriers for investment in energy managements systems which include; access to capital (typically a more relevant issue for smaller organizations), time and expertise (lack of time resources and trained people, awareness and uncertainty (lack of information regarding energy usage and lack of benchmarking of best practices) and the complexity of the industry (more relevant issue for larger companies).

Furthermore, IT projects are often over budget, delayed, and more often than not result in less value than predicted(Zdravkovi´c et al. 2018). What Davis et al. (2015) suggests is that there is a lack of incentive for companies to invest in technologies that do not directly affect costs, and continues by concluding that it seems to be flawed practises among companies when eval-

(25)

uating the possibility of new technology investments. Therefore, Davis et al.

(2015) suggest that companies need to enable new opportunities through modernized performance metrics.

When implementing IT-technology with the purpose of using the information that it provides to either improve financial or efficiency aspects, the outcome does not always lead to improvements (Galy & Sauceda 2014). Sys- tems such as Enterprise Resource Planning (ERP) demands complex infras- tructures and high degrees of freedom to be fully successful from a financial perspective. Galy & Sauceda (2014) continues by explaining how technological infrastructure, as well as the absorptive capacity within a company is believed to be two major variables for an ERP o be effective and thus also financially viable. But by investing in digitization technology the ability to deploy strategies in servitization, supply-chain integration, and short lead times that can lead to sustainability indirectly are enabled by these technologies (Chiarini et al. 2020, Waibel et al. 2017).

2.2.4 Measuring sustainability

What gets measured gets done as the popular saying goes in management literature. As mentioned by Rosen & Kishawy (2012) in a paper investigating key contributors of sustainable manufacturing identified sustainability indicators and controls as a key driver for sustainable manufacturing. Therefore, by using sustainability focused Key Performance Indicators (KPI:s) driven by data, there will be sustainable development within the manufacturing industry, and thus the result from Mart´ınez & Silveira (2012) confirms that it is possible to have both economic growth and sustainability growth, while at the same time reducing the pressure on resources and the carbon output from these industries in Sweden. In a paper by Shahbazi et al. (2017), they concluded that current KPI:s used in the manufacturing industry mainly focuses on financial targets and thus prioritising costs- and quality-related figures rather than environmental figures such as waste, end-of-life, waste segregation etc. The same study also highlights in the discussion that even though material efficiency is clearly linked to overall carbon emissions, energy use, and thus also global warming. Companies do not make these holistic connections. The reason is that the factory floor managers who have limited accountability extending only to inside the factory, making it hard for the rest of the organization to see the sustainability benefit of having more efficient material use.

(26)

The ability to share data as previously mentioned by Stock & Seliger (2016), Dreller (2018) can also enable companies to share standardized sustainability KPI:s. A paper by Prause (2015) argues that sharing data can be the future of what he calls fractal companies, where a network of companies that are self-organizing and self-optimizing are linked to a common access point of information. To achieve that, sharing data in the form of standardized KPI:s will be essential.

One popular standard that some Swedish manufacturers have is the ISO 14001, a standard aimed at implementing an Environmental Management System (EMS), which has been increasingly popular in the last 15-20 years.

Zobel (2015) found that there is no statistical significant effect on waste management in relation to manufacturing firms that do not use this system and a possible explanation is that firms which have these systems in place probably are more focused on other KPI:s that are more popular such as carbon emissions and energy use.

As mentioned by Parida et al. (2014) it is essential to not only measure the performance but to manage it as well. The paper suggests the use of a management system such as a Balanced Scorecard (BSC) that can be used for everyday operations such as maintenance but also be used when making investment decisions. Hartmann et al. (2014) mentions that the evolution of the BSC has led to another strategic tool called the Strategy Map, further discussed in the next section. In the context of performance management, Parida et al. (2014) also mentions the concept of eMaintaiance that takes real-time digital data and uses it for management and maintenance with the ability to use it remotely, making it possible for managers to work from home.

2.3 Concept of building a model directed towards man- agers

To understand how the final model of this thesis should be designed to be as practical for managers as possible, one must understand the fundamental concepts of how managerial processes work in the context of control. Hart- mann et al. (2014) presents a scheme which depicts the core fundamentals throughout the managerial control process, see figure 1. What it illustrates is the role of managers and how they treat information when having a set goal. Furthermore, the figure illustrates the importance of data in the decision processes. However, the data can be obtained in different ways and in

(27)

different forms. However, as mentioned in the delimitations, this thesis will focus on the digital aspects of data-driven decision-making.

• Manager: managerial input that brings capabilities, knowledge, inten- tions, etc., into an organization.

• Decisions: all actions and decisions managers make as a part of their duties.

• Results: performance measures that managers seek to meet.

Figure 1: A scheme defining the fundamentals of a managerial process according to Hartmann et al. (2014).

Moreover, a Strategy Map¹, see figure 2, is a way of aligning initiatives and measures with goals and by designing the model in a similar way to a strategy map, the chances of the model being practical for managers is increased. The strategy has been derived from the commonly used Balanced Scorecard (BSC) (Kaplan & Norton 1992) by linking an organization’s strategy to practical implementations (Kaplan & Norton 2008). Furthermore, strategy maps enhances managers’ ability to gather strategically relevant information, subsequently filtering out the irrelevant parts. That results in

1A Strategy Map is a tool for managers to link measures in a strategic way towards company goals. The system typically uses four perspectives, for example the employee-, process-, customer- and, owner-perspective. (Management Control Systems by Robert N.

Anthony et al.)

(28)

managers reducing the risk of information overload. Additionally, strategy maps increases the managers’ ability to decide the appropriateness of strate- gical decisions(Cheng & Humphreys 2012). However, a typical issue with a strategy map is that the lack of top management commitment and the lack of employee involvement in the strategy map design, but by involving employees as well as top management in the design process the chances of an successful implementation of the strategy map is increased (Hartmann et al.

2014).

Figure 2: A Strategy Map according to Hartmann et al. (2014)

(29)

Chapter summary

Since the thesis aims to construct a conceptual model used by organizations in their quest to develop their path towards smarter manufacturing, the literature review commenced by investigating the managerial processes and tools that can be used to develop the conceptual model. Thereafter, an exploratory investigation on prior literature within the field was made which identified areas of interest.

As mentioned in previous sections, the field of data in the manufacturing context has increased significantly during the past years. As a result, companies have started to implement different type of technological solutions in order to survive the trajectory towards smarter manufacturing. However, scholars have identified several issues connected to different areas of the solutions. Furthermore, not only does technological obstacles occur due to immaturity of the digital tools themselves, prior literature also identified hindrance in the organizational aspect.

Additionally, since the project aims to also analyze the sustainable aspect of data-driven decision-making, an analysis of prior research within the field of sustainability in smart manufacturing was made. Ultimately, identifying opportunities and the state-of-the-art in the area.

(30)

3 Method

The following chapter depicts the method used throughout this thesis. Firstly, the research design is explained followed by a depiction of the method used in the data collection. Secondly, the data analysis is explained, followed by the ethical considerations used throughout the project. The chapter terminates by discussing the limitations of this thesis.

3.1 Research Design

In order to derive a satisfying result that fully answers the research questions and the purpose of the thesis, an exploratory perspective was chosen in accordance with Saunders et al. (2015) since the research question tries to explore and clarify the phenomenon of digital data-driven decision-making. Another reason why this thesis was designed with an exploratory approach was because it was commissioned by KTH´s Industrial Transformation Platform to explore the subject to get a more precise understanding of the research area for further research within the platform. Moreover, an inductive multiple case study approach was chosen. As the gathered data were used to explore a phenomenon and identify themes to build a theory the inductive approach was deemed suitable in accordance with what Saunders et al. (2015) argues.

A multiple case study approach was deemed to be suitable by following the criteria presented by Baxter & Jack (2008), Yin (1994), namely, the methodology is suitable when exploring differences and when the goal is to derive a theory which can be predicted on other cases from the sample investigated. (n.d.) further argues that a multiple case study enables a stronger base for theory building and a more appropriate level of abstraction is achieved. Additionally, constructs and relationships are better delineated.

Findings from the interviews were then discussed, where differences and similarities between the cases were illuminated. Thereafter, those findings were juxtaposed in relation to the data gathered from the literature review to reach a synthesis.

By having strong links to theory through the literature review, it enabled the construction of a clear research framework (Gibbert et al. 2008). Further- more, by building a rigorous methodology over the data collection phase, as well as a framework, see Figure 3, to design questions which were used in the interviews, it depicted the entire process from the research question to the final result, aligned to the suggestions of Gibbert et al. (2008), Yin (1994).

(31)

Also, to conduct a transparent study, the twelve-point checklist presented by Aguinis & Solarino (2019) was adopted throughout the project to a large extent. The exact efforts made are described below.

1. Kind of qualitative method - the type of method used has been depicted throughout.

2. Research setting - the context of the study has been illuminated.

3. Position of researchers along the insider-outsider continuum - none of the involved have close relationships which could alter the information retrieved from the informants.

4. Sampling procedure - how the data has been collected and thereafter analyzed is described.

5. Relative importance of the participants/cases - all informants have been classified as equally important throughout. However, the differences from researchers compared to company informants have been lifted.

6. Documenting interactions with participants - how the interviews were conducted has been explained in the method chapter.

7. Saturation point - Not fully satisfied since the time limit of the thesis prohibited more interviews to be conducted, which could have increased the result of the study.

8. Unexpected opportunities, challenges, and other events - due to the narrow time limit, no major events occurred which heavily affected the thesis. However, due to the pandemic of Covid-19, many companies declined to participate as a result of lack in resources.

9. Management of power imbalance - The Royal Institute of Technology has funded as well as created the project which most likely affected companies’ views when approached about participating.

10. Data coding and first-order code - the coding was based upon the linking areas from the framework presented, see figure 3.

11. Data analysis and second- and higher-order code - higher-order coding was not implemented throughout the coding process.

(32)

12. Data disclosure - data has been made as transparent as possible. How- ever, because of anonymity reasons, no recordings nor transcripts have been shared outside the research group.

3.1.1 Data gathering

As previously mentioned, an inductive multiple case study approach was deemed to be suitable.

The data gathering consisted of different types of sources to ensure data tri-angulation and generalizability (Gibbert et al. 2008), methods used to achieve more perspectives on the findings as a way to ensure validity. Fur- thermore, both primary and secondary sources, in the form of literature review and, primarily, interviews were used as main forms of data sources.

3.1.2 Framework of the investigated phenomena

As data gathering from interviews was the primary source of information in the context of this thesis, a framework was developed to structure the data gathering, see Figure 3. The framework presents four domains that were of interest in this thesis. The domains were current operations, data, sustainability, and improved operations. Starting with the link between current operations and data. These links aimed to explore how digital data-driven decisions were implemented today and what challenges there were in organizations connected to these systems. Next couple of links were between data and improved operations, these links explored the organizational ambitions related to data-driven decisions, and thus what the future might look like within this domain. Challenges to these ambitions were also of interest.

Then the relations between current operations and sustainability were derived. These links, like the ones between current operations and data, aimed to explore what sustainability implementations were in place today and what challenges were connected to that subject. Those links were followed by the coupling between sustainability and improved operations, which investigated the organizational ambitions of improving sustainability measures, what developments are currently under way, and the challenges connected to improved sustainability measures. Lastly, the link between data and sustainability was analyzed, where the aim was to find the direct synergies between digital data-driven decision-making and sustainability. The unknown answers to the links between these domains were thus the core of the thesis

(33)

and answering them was the purpose of the interviews. Furthermore, the framework and its links helped in the process of structuring the literature search by illuminating research areas.

Figure 3: A defining the investigation areas of the project

3.1.3 Selection of companies

The selection of companies was not limited to size of the organization. Thus, enabling a result applicable to as many organisations as possible within the manufacturing industry by enabling perspectives from different contexts through tri-angulation (Gibbert et al. 2008). Furthermore, through the context of a multiple case study, one can provide a stronger base for theory building which ultimately is the purpose of this study in the form of a conceptual model( n.d.). Public lists of companies within the manufacturing industry provided by the Statistical Central Bureau of Sweden were used as tools to identify companies that function within the industry. Thereafter, companies were systematically contacted through contact forms with an invi- tation to an interview. Those who accepted, were sent further instructions on how the interviews were conducted as well as an invite to an online meeting.

What ultimately governed the selection of companies was merely subjects that could illuminate and extend the relationship among constructs, and not necessarily fully representative of the whole population ( n.d.). Furthermore, the role of the informants were decided by the companies themselves since

(34)

they wanted to provide an informant which was knowledgeable within the research area.

3.1.4 Primary sources

The interviews were conducted with representatives of Swedish manufacturing companies which ranged from small to large in enterprise size. Moreover, researchers within the field were interviewed to get an academic perspective.

The interviews regarded partially current state practices to get an understanding of how the state-of-the-art methods are currently implemented.

Furthermore, questions regarding future projections were used to identify what methods and practices are currently being investigated and prioritized by practitioners and researchers. As this paper was exploratory in its nature, and inductive in its approach, (Saunders et al. 2015) suggests that semistructured interviews with a focus on open-ended questions should be used. The reason is that a semistructured interview opens the possibility of asking probing questions to dive deeper into areas of higher interest for the paper and to get some insights into areas that were not previously considered but might have significance to the study. Another reason that this method of data gathering works well for an exploratory study as this one is the nonverbal communication such as pauses for gathering thoughts and the act of thinking out loud, which gives even richer data in the form of nuances.

However, this also means that caution should be taken to the interaction with the interviewee to not influence the answers in any way. Saunders et al.

(2015) also suggests in-depth or semi-structured interviews to be conducted rather than structured ones such as surveys when there are many questions of varying complexity and varying logical order which this paper inherently had as the exploration started with interviews.

As far as practicality goes, all interviewees were promised anonymity rather than confidentiality, which would preclude any data of interest. The subjects were also able to skip questions and terminate the interview when- ever at will. These two rules had two effects, the first being to ensure that they felt secure and confident to talk freely. The second was for ethical reasons.

A major problem when conducting interviews is bias. Therefore, an approach that limits the bias was wanted. (n.d.) suggest that to avoid that, one must interview individuals with different perspectives of the research area.

Those could be informants from different hierarchical levels or different sec-

(35)

tors of the investigated industry. Subsequently, by interviewing informants with different roles as well as different sized companies, combined with researchers to enable the academic perspective, one could better ensure that bias was limited.

The interviews conducted are illustrated in tabular form below in table 2.

Interviewee,

Date Role Characteristics Aim Duration Company

A

3/03-21 Business Developer Pulp producer Exploring current state

and future of a large corp. 45 min Company A B

10/03-21 Chief Operating Officer Vehicle manufacturer Exploring current state

and future of a large corp. 45 min Company B C

11/03-21 Chief Executive Officer Metal component manufacturer Exploring current state

and future of SME 40 min Company C

D

15/03-21 Business Developer Aluminium parts assembler Exploring current state

and future of SME 35 min Company D

E

16/03-21 President over operations Engine manufacturer Exploring current state

and future of a large corp. 45 min Company E F

8/04-21 IT Business Developer Retail consumer products Exploring current state

and future of a large corp. 45 min Company F G

30/03-21 Researcher Industry and environmental sustainability Focus on problems, and

future in sustainability and technology 30 min University A H

6/04-21 Researcher Digital technologies in manufacturing Focus on problems, and

future in sustainability and technology 30 min University A

Table 2: Conducted interviews with a short summary of topic and aim.

3.1.5 Literature review

A literature review was used to reach two goals. The first one was to gather current knowledge to be able to ask appropriate questions in the interview process. However, there is a balance of how much to know before asking questions. As Gioia et al. (2012) argues, there is a value of not knowing everything and thus staying ignorant to the literature. Because, up to the stage of conducting interviews, too much knowledge of the current research might create tunnel vision and thus turn an inductive approach into a type of abductive approach with leading questions. There is also a need for balance between exploration and confirming some previous bias. Then there is the case for the opposing effect, that asking questions readily available in the literature will just waste time. The second goal of reviewing the literature was to juxtapose the literature findings with the empirical findings. From the interviews, themes would emerge and to find what themes that are more novel than others, a literature review helped. Focusing on novel concepts rather than on incumbent ones are of higher interest because of the exploratory and inductive theory approach that was used (Gioia et al. 2012). It can be argued that reading everything published around a topic is an impossible feat,

(36)

however, this was not the goal of the literature review Saunders et al. (2015).

The goal was to find the most significant and relevant literature on the topic, thus finding a unique angle to which proceed with the subsequent empirical data gathering without putting blinders on the exploration of new concepts and thus hinder theory building (Gioia et al. 2012). With this strategy in mind the search for literature started with phrases such as sustainable manufacturing and smart manufacturing Sweden. Moreover, by primarily focusing on resent papers of the last ten years, an understanding of commonly used citing in the analyzed literature was found, but more importantly, getting the sense of buzzwords and jargon used. With a better understanding of the jargon used in the literature, one could extend the search even further with more precision on topics of interest such as data-driven decision-making and Industry 4.0 Sweden.

3.2 Data analysis

The process of analysing the collected data was through several different phases. Firstly, all interviews were recorded with both video and audio rather than just using notes to ensure that all data were gathered. Thereafter, the interviews were transcribed and analyzed manually. The analysis process started by identifying themes and then the transcripts were color coded according to those themes. Subsequently, identifying key factors which juxtaposed with the findings from prior literature were used to derive a synthesis.

The approach used to analyze the data was inductive, which from Saunders et al. (2015) definitions indicates that one does not commence the study with a previous known theoretical framework in mind as with a deductive approach, instead one intends to develop a theory from the data that has been collected. Since theory building in the form of a conceptual model was wanted, the inductive approach was deemed suitable.

As the coding commenced, triangulation was achieved by coding the transcripts twice with different writers, in accordance to Gibbert et al. (2008), Yin (1994), as an attempt to find unanimous areas.

3.3 Ethical considerations

Ethical aspects have been considered throughout this master thesis. Firstly, because of the sensitive information discussed in the conducted interviews, steps have been taken to ensure the informant confidentiality. In accordance

(37)

to the recommendations of Vetenskapsr˚adet (the Swedish Research Council) (Vetenskapsr˚adet 2002); (1) the information requirement, (2) the consent requirement, (3) confidentiality requirement, and lastly, (4) the good use requirement.

1. The purpose of the research was made transparent for the interviewee before the interviews were conducted. Thus, informing them of what is expected as well as what they will contribute to by participating in the study.

2. The informants chose themselves whether to participate or not. Fur- thermore, they were in the beginning of the interview presented with the option to not answer questions as well as to leave the interview at any point if that was what they wanted. Lastly, since what was said during the interview was of great importance for the findings of the study, each interviewee was asked for consent to record the meeting in order for the interview to be transcribed. Those recordings were deleted by the end of the study.

3. Confidentiality has been assured throughout this paper by anonymity of the company which the informant represent. They are, therefore, presented as company X. Furthermore, the informants themselves are referred to as interviewee X. However, a short description of their role is disclosed in table 2. Confidential information gathered throughout the project has only been shared amongst the research team and has, therefore, not been disclosed in this report.

4. All information gathered has a sole purpose of fulfilling the ambition of this project, in accordance to the good use requirement.

3.4 Limitations of method

Saunders et al. (2015) introduces the concept of biases in the context of conducting interviews such as interviewer bias, response bias, and partici- pation bias. The first being a bias cast by the interviewers, where the use of non-verbal behavior and framing questions could and, most likely, would alter the quality of the data. With the knowledge of nonverbal communication beforehand, the interviews were conducted with this in mind, however, to control subconscious behaviour can be argued to be an impossible feat.

(38)

The second bias is where the interviewee’s responses are biased for different reasons such as not disclosing the whole truth for secrecy of sensitive information or other reasons. By using multiple sources, the effect a singular informant can have on the overall product of this thesis is limited. The third and last bias is about the sample of responses, the sample should represent a good proxy for the generalized population that this paper intends to cater towards. However, as limited resources such as time, the gathered sample did not represent a perfect proxy for the whole manufacturing industry in Swe- den. Furthermore, due to the ongoing pandemic of Covid-19, all interviews conducted have been via online platforms, mainly Microsoft Teams. Since an important factor of a semi-structured interview is to enable the interviewee to feel relaxed, thus encouraging more exploratory answers, that could have affected the answers that were gathered compared to if the interviews had been conducted face-to-face.

(39)

Chapter summary

In this chapter, a depiction of the used methodology throughout this thesis has been presented. The decision of using an exploratory research design together with an inductive approach has been argued to be preferred because of the scope of the project as well as the relatively limited prior research within the research area. Furthermore, the chapter explains the data collection process.

The main data sources were from semi-structured interviews with informants from various companies within the industry, researchers within the field of study, as well as prior literature. The data that was wanted had been derived from the framework connecting the different areas that had been investigated.

Throughout this thesis, effort has been made to achieve a rigorous methodology from criteria defined by prominent scholars. Those specific actions have been illuminated throughout the chapter.

(40)

4 Findings

In this chapter, a thematic analysis of the findings is presented and structured in accordance with the framework used, see figure 3, for designing the interview guide used to collect these empirical findings. The five themes are:

Current use of data in operations, Current sustainability implementations, Future use of data in operation, Future sustainability implementations and Data-driven sustainability.

4.1 Current use of data in operations

Current usage of data as a tool to make decisions within the Swedish manufacturing industry differs from the size but also the context in which the manufacturing company functions in. Researcher G mentions that from experience, the context of the field in which the company exists is the driving factor for how much data is used throughout the factory floor. The com- parison between the automotive industry and the marine sector is brought up where researcher G argues that there is a significant difference in the implementation of data usage between those sectors.

I have done some projects in the marine sector and those systems were primitive and I was surprised. I thought that making boats is very advanced but it was very primitive (Researcher G, 2021)

Researcher G further argues that the technology exists and often is in place for companies, but that there are organizational problems and difficulties in how the data is actually used, which hinders the extent to which the data is being exploited. Interviewee B, which represents a large corporation, agrees with that view by explaining that the problems that their company faces in the context of data in their manufacturing is not the technological part, but instead in how it is being used. As outlined by interviewee B:

Technology is easy, humans are hard (Interviewee B, Chief Operating Officer, 2021)

and continues by arguing that there is a threshold that needs to be surpassed in order for companies to fully rely on what suggestions the data is giving.

To reach that level of maturity, there needs to be a joint goal within the organization and an aspiration to actually implement data-driven decision- making. Interviewee B continues by saying that company B and almost every

Data-driven decision-making for efficient & sustainable production

Data-driven decision-making for efficient & sustainable production

ARVID BROMS

SIMON LILJENBERG OLSSON

Data-driven decision-making for efficient &

sustainable production

by

Arvid Broms

Simon Liljenberg Olsson

Datadrivet beslutsfattande för effektiv och hållbar produktion

Arvid Broms

Simon Liljenberg Olsson

Contents

List of Figures

List of Tables

Nomenclature

1 Introduction

1.1 Background

1.2 Purpose

1.3 Delimitations

1.4 Disposition

Chapter Chapter description

2 Literature review

2.1 Digital aspects of data-driven decision-making

2.2 Environmental, social, and economical aspects of digital data-driven decision-making

2.3 Concept of building a model directed towards man- agers

Chapter summary

3 Method

3.1 Research Design

3.2 Data analysis

3.3 Ethical considerations

3.4 Limitations of method

Chapter summary

4 Findings

4.1 Current use of data in operations