Anne-Mari Järvenpää Developing data analytics capabilities of circular economy SMEs  ACTA WASAENSIA 496 Copyright © Vaasan yliopisto and the copyright holders. ISBN 978-952-395-043-6 (print) 978-952-395-044-3 (online) ISSN 0355-2667 (Acta Wasaensia 496, print) 2323-9123 (Acta Wasaensia 496, online) URN https://urn.fi/URN:ISBN:978-952-395-044-3 Hansaprint Oy, Turenki, 2022. ACADEMIC DISSERTATION To be presented, with the permission of the Board of the School of Technology and Innovations of the University of Vaasa, for public examination on the 2nd of December, 2022, at noon. Article based dissertation, School of Technology and Innovations, Industrial Management Author Anne-Mari Järvenpää https://orcid.org/0000-0001-6612-0150 Supervisor Dr. Iivari Kunttu, Adjunct Professor University of Vaasa. School of Technology and Innovations Häme University of Applied Sciences. Dr. Petri Helo, Professor University of Vaasa. School of Technology and Innovations. Custos Dr. Iivari Kunttu, Adjunct Professor University of Vaasa. School of Technology and Innovations Häme University of Applied Sciences. Reviewers Dr. Nina Helander, Professor Tampere University. Management and Business, Information and Knowledge Management. Dr. Anna Visvizi, Associate Professor SGH Warsaw School of Economics. International Political Economy Department, Institute of International Studies. Opponent Dr. Nina Helander, Professor Tampere University. Management and Business, Information and Knowledge Management. Dr. Kaisa Henttonen, Associate Professor University of Eastern Finland. Faculty of Social Sciences and Business Studies. https://orcid.org/0000-0001-5451-9746 V Tiivistelmä Kestävyyskriisi vaatii toimia, joilla kulutus ja tuotanto sovitetaan luonnon kantokykyyn. Kiertotalouden odotetaan ratkaisevan kestävyyshaasteen, jossa kierrätys- ja jätehuoltoteollisuudella on suuri rooli materiaalien kierrossa ja neitseellisen materiaalitarpeen vähentämisessä. Digitalisaatio ja datan kasvava määrä tarjoavat avaimet tehokkaiden materiaalikiertojen kehittämiseen. Siksi on olennaista kysyä, miten kiertotaloudessa toimivat yritykset, erityisesti pk- yritykset, pystyvät hyödyntämään dataa. Tämä väitöskirja yhdistää kiertotalouden, dataan pohjautuvan päätöksenteon ja data-analytiikan kyvykkyydet hakien vastausta tutkimuskysymykseen: Millaisia kyvykkyyksiä, tarpeita ja haasteita kiertotalouden pk-yrityksillä on datan hyödyntämiseen ja miten kyvykkyyksiä voitaisiin parantaa yhteistyössä korkea- koulujen kanssa? Tämä kysymys on jaettu neljään alakysymykseen, joita käsitellään neljässä tutkimusartikkelissa. Tutkimus toteutettiin laadullisena tapaustutkimuksena keräämällä empiiristä dataa seitsemästä pk-yrityksestä, jotka toimivat kiertotalouden alalla, tarkemmin kierrätyksen ja jätehuollon parissa. Tutkimus toteutettiin vuosina 2018–2022 Suomessa. Tämän tutkimuksen tulokset lisäävät ymmärrystä siitä, miten kiertotalouden pk- yritykset hyödyntävät dataa päätöksenteossa, mitkä ovat niiden käytännön haasteet datan hyödyntämisessä ja miten pk-yritykset voisivat parantaa kykyään hyödyntää dataa. Tulokset paljastavat, että kiertotalouden pk-yritysten datan hyödyntämistarpeet liittyvät toiminnan suunnitteluun ja tulevaisuuteen varautu- miseen kilpailun lisääntyessä, muuttuviin trendeihin, kasvaviin ympäristö- vaatimuksiin, investointeihin sekä hiilidioksidipäästöjen vähentämiseen. Haasteena on yhdistää relevanttia dataa useista sisäisistä ja ulkoisista lähteistä sekä käyttää analytiikkaa. Nämä haasteet liittyvät tiedonhallintaan, resurssien ja kyvykkyyksien puutteeseen sekä sääntelyyn. Lisäksi kiertotalouden pk-yrityksiä näyttää ohjaavan enemmän sääntely kuin data. Asiasanat: tietoon pohjautuva päätöksenteko, data-analytiikan kyvykkyydet, kiertotalous, pk-yritys, korkeakoulu-yritysyhteistyö. VI Abstract The sustainability crisis demands action to match consumption and production to the limits of the nature. The circular economy is expected to solve the sustainability challenge, in which the recycling and waste management industry are envisioned to play a major role maintaining materials in a cycle to reduce the need for virgin materials. There are high expectations that digitalization and increasing amounts of data will provide the keys to develop efficient circular material cycles. Thus, it is relevant to ask how companies operating in the circular economy, especially SMEs, are able to utilize this data. This dissertation integrates the circular economy, data-driven decision-making and data analytics capabilities, and seeks answers to the research question: What kinds of capabilities, needs and challenges do SMEs in the circular economy have for data utilization, and how these capabilities could be improved by means of collaboration with universities? This question is divided into four sub-questions, which are addressed in four research articles. The research was conducted as a qualitative case study by collecting empirical data from seven case companies that operate in the field of the circular economy—namely in recycling and waste management. This research was conducted during 2018–2022 in Finland. The results of this study increase the understanding of how circular economy SMEs utilize data in decision-making, what their practical challenges are regarding data utilization, and how SMEs in the circular economy could improve their data utilization capabilities. The results reveal that the data utilization needs for SMEs in the circular economy relate to planning operations and preparing for the future in terms of increasing competition, changing trends, high environmental standards, investing in facilities and capabilities and reducing carbon emissions. The challenge is to combine relevant data from several internal and external sources and use analytics. These challenges relate to data management, lack of resources and capabilities, as well as to regulation. Moreover, circular economy SMEs are more driven by regulation than by data. Keywords: data-driven decision making, data analytics capabilities, circular economy, SME, university-industry collaboration. VII ACKNOWLEDGEMENT This was a great opportunity that provided me with the way to grow from a lecturer to a researcher and enabled me to combine research and teaching. It is time to thank those who contributed my journey. First and foremost, I am grateful to Dr. Vesa Salminen, the former Director of the HAMK Smart Research Unit, for giving me this amazing opportunity. Thank you, Professor Jussi Kantola, for your support and guidance. I am grateful to my employer for being able to combine the work of a lecturer with a work of a project manager in the HAMK Smart Research Unit, and I want to thank the Director, Dr. Jukka Pulkkinen for his support and for the opportunity to prepare new projects. This dissertation was conducted as a part of the TULEVA (2018-2020) and VÄLKKY (2021-2022) projects and I wish to thank the Regional Council of Häme for funding these projects. It has been a great privilege to collaborate with circular economy SMEs, whose representatives had a positive attitude towards my research allowing me access to gain research data and insight from the companies I studied. I have great respect for Dr. Iivari Kunttu and Dr. Jari Jussila, who provided me with the best “research school” I can imagine. Thank you for your guidance, collaboration, sharing your wisdom and supporting me to grow my academic network. I would like to thank the reviewers, Professor Nina Helander and Associate Professor Anna Visvizi, for their constructive feedback. Colleagues and friends, thank you for your encouragement, peer support and therapy. To my dear relatives, thank you for the childcare. To my husband, thank you for respecting my ambitions and supporting me to achieve my goals. Now relax, there won’t be another huge project like this, or at least I suppose so at the moment. To my wonderful children, thank you for letting mommy do her "school stuff”. September 2022 Anne-Mari Järvenpää VIII “If we knew what it was we were doing, it would not be called research, would it?” ― Albert Einstein IX Contents ACKNOWLEDGEMENT ............................................................................ VII 1 INTRODUCTION ................................................................................. 1 1.1 Background ............................................................................. 1 1.2 Research gap ........................................................................... 2 1.3 Research questions .................................................................. 4 1.4 Research context ..................................................................... 6 1.5 Key concepts ........................................................................... 6 1.6 Structure of the dissertation .................................................... 7 2 THEORETICAL BACKGROUND ............................................................. 8 2.1 Data-driven decision-making in SMEs ....................................... 8 2.1.1 Data .......................................................................... 9 2.1.2 Data-driven decision-making ................................... 10 2.1.3 Business intelligence ............................................... 12 2.1.4 Knowledge management ......................................... 13 2.2 Data analytics in circular economy SMEs ................................ 15 2.2.1 Circular economy .................................................... 15 2.2.2 Digitalization and the smart circular economy ......... 17 2.2.3 Data analytics in circular economy SMEs .................. 19 2.3 Developing the data analytics capabilities of circular economy SMEs ...................................................................................... 21 2.3.1 Opportunities data provides .................................... 22 2.3.2 Capabilities needed in data utilization ..................... 23 2.3.3 Developing capabilities ........................................... 25 2.4 Summary ............................................................................... 27 3 METHODOLOGY ............................................................................... 29 3.1 Research philosophy and paradigm ....................................... 30 3.2 Research approach ................................................................ 31 3.3 Methodological choice and nature of research ....................... 31 3.4 Research strategy .................................................................. 32 3.5 Time-horizon ......................................................................... 33 3.6 Data collection and analysis ................................................... 34 3.7 Quality criteria ....................................................................... 35 3.8 Summary ............................................................................... 36 4 RESEARCH ARTICLES ........................................................................ 38 4.1 Article 1 Using foresight to shape future expectations in circular economy SMEs .......................................................... 38 4.2 Article 2 Data-driven decision-making in circular economy SMEs in Finland...................................................................... 41 4.3 Article 3 Barriers and practical challenges for data-driven decision-making in circular economy SMEs ............................ 42 4.4 Article 4 Developing data analytics capabilities for circular economy SMEs by Design Factory student projects ................ 44 X 5 DISCUSSION .................................................................................... 47 5.1 Answers to the research questions ........................................ 48 5.2 Theoretical contribution ........................................................ 50 5.3 Managerial recommendations ................................................ 51 5.4 Practical implications ............................................................. 52 5.5 Limitations and future research ............................................. 54 REFERENCES ......................................................................................... 55 PUBLICATIONS ...................................................................................... 66 Figures Figure 1 Data analytics capabilities for data-driven decision- making in the circular economy. ...................................... 4 Figure 2 Generating knowledge from data ................................... 10 Figure 3 Steps to adopt business intelligence in SMEs. ................. 13 Figure 4 Descriptive, predictive and prescriptive analytics. ........... 21 Figure 5 Questions to guide data visualization ............................. 24 Figure 6 Resources needed to develop data analytics capabilities. 26 Figure 7 Research design according to a research onion model .... 29 Tables Table 1 Summary of articles. ........................................................ 5 Table 2 Research design. ........................................................... 36 Table 3 The number of case companies. .................................... 37 Table 4 Foresight activities of the case companies. .................... 39 Table 5 Future expectations of the case companies. ................... 40 Table 6 Data utilized in decision-making in the case companies. 42 Table 7 Data utilization challenges in the case companies. ......... 43 Table 8 Developing data analytics capability in university-industry collaboration. ................................................................ 45 XI Publications Article 1. Järvenpää, A.-M., Kunttu, I., & Mäntyneva, M. (2020). Using foresight to shape future expectations in circular economy SMEs. Technology Innovation Management Review, 10(7), 41–50. https://doi.org/10.22215/timreview/1374 Article 2. Järvenpää, A.-M., Kunttu, I., Jussila, J., & Mäntyneva, M. (2021). Data-Driven Decision-Making in Circular Economy SMEs in Finland. Springer Proceedings in Complexity, 371–382. https://doi.org/10.1007/978-3- 030-84311-3_34 Article 3. Järvenpää, A.M, Jussila, J., Kunttu, I. (2023) “Barriers and practical challenges for data-driven decision-making in circular economy SMEs”, in: Visvizi, A., Troisi, O., Grimaldi, M. (eds) (2023) Big Data and Decision- Making: Applications and Uses in the Public and Private Sector, Bingley, UK: Emerald Publishing, ISBN: 978–1803825526 Article 4. Järvenpää, A.-M., Jussila, J., & Kunttu, I. (2022). Developing data analytics capabilities for circular economy SMEs by Design Factory student projects. The XXXIII ISPIM Innovation Conference “Innovating in a Digital World,” June. https://doi.org/10.22215/timreview/1374 1 INTRODUCTION Circular economy companies face increasing targets to reduce, reuse and recycle. This is affected by strict legislation and growing environmental requirements, e.g. concerning the reduction of carbon emissions. Increasing demands caused by the European Commission, national governments as well as customers are also bringing changes to the business environment. The circular economy is expected to solve the sustainability crisis with help of digitalization and increasing amounts of data. This in turn requires education and continuous learning to build the needed capabilities in companies. To achieve the targets of the circular economy in society, it is important to understand the challenges of companies to be able to support them in the development as well as to provide education on the skills needed in industry. This dissertation aims to increase the understanding of how small and medium- sized enterprises (SMEs) operating in the circular economy business utilize data in decision-making, what the challenges are in utilizing the data and how the SMEs could be supported to develop their capabilities. 1.1 Background Today, we are facing a sustainability crisis as production exceeds the limits of nature (Prieto-Sandoval et al., 2019). The circular economy aims to provide a solution by fitting material consumption within nature’s boundaries according to the 3R principle, aiming to reduce material consumption, increase reuse of materials and recycling (Kirchherr et al., 2017; Prieto-Sandoval et al., 2018). This transition requires promoting high-quality material cycles, digitally exploitable information, research, experimentation, and legislation that enables innovations. The European Commission’s new action plan for the circular economy (European Commission, 2020) expects that the global consumption of materials will increase 100% by 2060, while annual waste generation will increase 70% by 2050. The European Union aims to double the use of circular materials by 2030 and closed- loop material cycles could increase the profitability of manufacturing companies and protect them from price fluctuations. To achieve this, the European Union should increase recycling by 100 megatons per year (Tisserant et al., 2017). The transition towards a circular economy and circular material flows requires data to enable efficient recycling, component harvesting, material value assessment (Charnley et al., 2019) and the creation of closed-loop production- 2 Acta Wasaensia consumption systems (Auh et al., 2022). The best time to transform a traditional linear material flow into a circular process is the decision-making phase in product and service development, as well as raw material production and procurement (Kauppila et al., 2022). Breakthroughs in the circular economy can be found in ecosystems, for example, in industrial symbioses between companies, where side stream or waste are exploited to save resources, and to reduce costs and environmental impacts (Järvenpää et al., 2021). There is need for a data-driven circular economy and digital solutions to manage and utilize resources efficiently. This in turn requires skills, education, continuous learning, and innovation. Data can help the efficient use of resources in the whole value chain by optimizing material flows, and minimizing the used materials and environmental effects. Technology is available to develop digital solutions in the circular economy (Niska & Serkkola, 2018), however, the challenge lies in linear supply chains where information is not shared (Salmenperä et al., 2021), and where demand and supply do not meet (Charnley et al., 2019; Cramer, 2018). Furthermore, the increased utilization of data increases energy and material consumption as well, which is the opposite of the aim of the circular economy (Bressanelli et al., 2022). As the circular economy is expected to provide a solution to the sustainability challenge at least partly through the use of digitalization and data, it is relevant to ask, how industry, especially SMEs, can utilize data to achieve the goal. 1.2 Research gap In the transition towards a circular economy and fulfilling the aims of strategies set by the EU (European Commission, 2020; Tisserant et al., 2017) and governments, there are opportunities and threats for SMEs. Empirical foresight research in Europe is focused on large companies (Jannek & Burmeister, 2007) while SMEs have received less attention (Stonehouse & Pemberton, 2002). Many SMEs work in an environment that does not require foresight activities (Jun et al., 2013), but SMEs operating in rapidly changing environments do need foresight capabilities (Uotila et al., 2012) to ensure success and survival (Rohrbeck, 2011). The circular economy is rapidly changing the business environment, providing both threats and opportunities that require foresight activities. Thus, there is a need to develop a greater understanding on how circular economy SMEs collect data for foresight activities and what they expect from the future. Prediction and forecasting require data from the business environment. Utilizing data requires competence and resources, but despite the benefits of data based Acta Wasaensia 3 decision-making, many SMEs face challenges understanding how to use analytics and they may face difficulties such as the lack of understanding of data, as well as lack of in-house experts and tools to analyze data (Iqbal et al., 2018; Kim et al., 2003; Ormazabal et al., 2018; Parra et al., 2019; Ransbotham et al., 2016; Rizos et al., 2016). Previous research has concentrated mainly on large-scale companies and somewhat on SMEs, but circular economy SMEs remain unexplored. This research aims to generate practical understanding at the grassroots level. The transition towards the circular economy is driven by digitalization, thus utilizing digital tools and data are considered as one of the key capabilities facilitating the circular economy (Parida & Wincent, 2019). However, circular economy SMEs face challenges utilizing digitalization and the increasing amount of data (Iqbal et al., 2018), despite the fact that data enables companies to develop material flows, logistics and customer behavior (Lacam, 2020). Thus, there is a need to develop more understanding on the development needs concerning the data analytics capabilities of SMEs as well as how these capabilities could be improved. To fill the aforementioned gaps in the existing literature, this dissertation integrates the literature on the circular economy, data-driven decision-making and data analytics capabilities (Figure 1) to extend the research by studying how circular economy SMEs collect data for decision-making, how they utilize data, what practical challenges they face, and what opportunities exist to develop data analytics capabilities. 4 Acta Wasaensia Figure 1 Data analytics capabilities for data-driven decision-making in the circular economy. 1.3 Research questions The primary objective of this dissertation is to address the following research question: What kinds of capabilities, needs and challenges do circular economy SMEs have for data utilization, and how could these capabilities be improved by means of collaboration with universities? This research question is divided into four sub-questions: RQ1: How do industrial actors and service providers operating in the circular economy foresee future changes in their operational environment? RQ2: How can SMEs operating in a circular economy utilize data to support their decision-making? RQ3: What are the practical challenges to data-driven decision-making in circular economy SMEs? RQ4: How are the data analysis capabilities in SMEs developing in joint action between university and companies? By seeking answers to the abovementioned research questions, this dissertation aims to increase the understanding on how circular economy SMEs collect and utilize data, what challenges they face, and how they could develop their Circular economy Data analytics capabilities Data-driven decision- making Capabilities of data- driven decision- making in the circular economy Acta Wasaensia 5 capabilities to utilize data in collaboration with educational institutions. For academia, this study aims to provide an understanding of how to develop joint activities with SMEs to support collaborative learning. For public administration and regional development organizations, this work provides an understanding of what it takes from circular economy companies to utilize data to boost circular economy targets and environmental benefits. Each research question is covered in one article. The first research question (RQ1) in article 1 aims to identify the data collection practices and activities to predict the future as well as expectations for the future. The second research (RQ2) question in article 2 aims to identify the types of data and analytical methods that circular economy SMEs typically utilize. The third research question (RQ3) aims to seek barriers and practical challenges to utilizing data in circular economy SMEs. The fourth research question (RQ4) explores how circular economy SMEs’ capabilities may increase during the collaboration with universities. A summary of the articles is presented in Table 1. Table 1 Summary of articles. Article 1 Article 2 Article 3 Article 4 Focus area Data collection practices for organizational foresight activities Data sources and methods used in decision-making Barriers and challenges to utilizing data in decision-making Development of data analytics capabilities Theory Organizational foresight Data-driven decision-making Data-driven decision- making Dynamic capabilities, relationship learning Research Strategy Comparative multiple case study Comparative multiple case study Comparative multiple case study Comparative multiple case study Research Context Foresight activities and future expectations in circular economy SMEs Data-driven decision-making and management practices in circular economy SMEs Barriers and practical challenges for data-driven decision-making in circular economy SMEs Opportunities to develop data analytics capabilities of circular economy SMEs Data Collection Methods Interviews Interviews Interviews, group interview Interviews, group interview 6 Acta Wasaensia Article 1 Article 2 Article 3 Article 4 Main Findings 1. Activities to predict the future 2. Expectations for the future 1. The most used types of data 2. The most used analytical methods 1. The barriers to utilizing data 2. Practical challenges to utilizing data SMEs capabilities increased during collaboration with university students 1.4 Research context This research involves the Finnish circular economy industry, namely SMEs providing recycling services, recovery of sorted materials, waste management and biogas production. Companies in this field often lack digitalization and data utilization, but they have increasing pressure to do gain these capabilities, as the environmental, operational, and business requirements are increasing all the time. As digitalization and data are expected to boost the circular economy and bring environmental benefits, it is important to study the situation from the point of view of SMEs operating in the circular economy. As SMEs represent a major part of companies in Europe, their development is important (Coleman et al., 2016). 1.5 Key concepts This subchapter explains the key concepts related to the research questions. • The circular economy refers to minimizing the demand for resources and recovering value from waste. • Data refers to the objective facts that can be used to produce information. • Data-driven decision-making uses data to guide the decision-making process. • Data analytics capabilities refers to the skills, abilities and knowledge to create value from data. • Small and medium-size (SME) enterprise refer to the size of a company in terms of the number of staff and turnover. Acta Wasaensia 7 1.6 Structure of the dissertation This dissertation is structured in two parts. The first part consists of five chapters that provide both a theoretical and practical background to the dissertation. The first chapter of part one introduces the background and objectives of this dissertation. The starting point is the sustainability crisis and strategies that expect that the circular economy will be able to help solve the sustainability challenge. It is expected that digitalization and ever-increasing amounts of data will drive the development of the circular economy, and it is important to research and explore the reality in the context of circular economy SMEs. The second chapter introduces the theoretical framework of the study, relating to data, data-driven decision-making, business intelligence, knowledge management, the circular economy, data analytics and capability development. This constructs a theoretical base for empirical exploration and reflection. The third chapter introduces the research design and methodology. This study is conducted as a comparative qualitative case study. Data was collected by interviewing SMEs operating in circular economy businesses in Finland. Each interview round sought a cross-sectional snapshot for one research question. The fourth chapter provides a summary of the articles, each article covering one research question. The article summaries provide an overview of the related literature, explain the research methods used, and present the main results. The fifth chapter answers the research questions and discusses the contribution and limitations of the study, and provides some suggestions for future research. Part two contains the four dissertation articles. In all the articles, the author of this dissertation is the primary author, who had the main responsibility for the data collection, analysis, composing, and writing the articles, and also for managing the review processes. 8 Acta Wasaensia 2 THEORETICAL BACKGROUND This chapter introduces the theoretical background for this study, relating to data, data-driven decision-making, business intelligence, knowledge management, the circular economy, data analytics and capability development. These concepts form a theoretical base for the study to empirically explore circular economy SMEs and to answer the research question. The circular economy is a developing business area, and the operational environment provides both, threats, and opportunities. For this reason, data- driven decision-making is important for companies in this business field. Predicting future changes enables businesses to anticipate threats and opportunities (Korreck, 2018; Rohrbeck & Gemünden, 2011; Uotila et al., 2012) and to get prepared for the future or even shape it (Cuhls, 2003). As SMEs have fewer resources and shorter planning horizons than large companies, their foresight goals relate to planning operations and managing innovations (Jannek & Burmeister, 2007). Activities often occur when SMEs are forced into the product development (Bidaurratzaga & Dell, 2012; Jannek & Burmeister, 2007). However, SMEs are typically focused on certain markets, and they might overlook new opportunities (Coleman et al., 2016). 2.1 Data-driven decision-making in SMEs “Where a firm can go is a function of its current position and the paths ahead. Its current position is often shaped by the path it has travelled.” (Teece et al., 1997, p. 522) Data-driven decision-making enables cost reduction, increased operational efficiency, customer loyalty and communication (Pulkkinen et al., 2019; Troisi et al., 2020). It requires managing data to make decisions to prescribe actions, predict development, and drive change (Troisi et al., 2020), as well as the capability to harness data into value, innovations, or competitive advantages (Troisi et al., 2021; Watson, 2016). In the case of SMEs, there is lack of understanding of data, data analytics infrastructure, and the necessary expertise to select appropriate solutions (Iqbal et al., 2018; Parra et al., 2019; Ransbotham et al., 2016). Companies can utilize data from several internal or external sources, and data can be generated by machines, humans or business (Olshannikova et al., 2017; Saggi & Jain, 2018). As data can be utilized in many ways, the company must understand Acta Wasaensia 9 the feasibility of available analytic methods (descriptive, predictive, or prescriptive) that can provide answers to business questions. For circular economy companies, data analytics provides opportunities to optimize the material flow and manage the supply chain, that can enable a leap to the operational efficiency and improved sustainability (Kristoffersen et al., 2019). However, even there are simple tools available, SMEs may not be capable to utilize them (Parra et al., 2019). 2.1.1 Data A huge volume of real-time data is generated by digital platforms every day. For this reason, many organizations cannot use all of the generated data effectively (Lavalle et al., 2011). The data does not contain any meaning itself, it just consists of objective facts that can be used to produce information by contextualization (Nykänen et al., 2016) for economic benefit. Data is defined as a carrier to store and transfer information and knowledge, but data becomes information or knowledge only through interpretation (Kock et al., 1997). Information quality plays an important role in the data environment in terms of achieving business value, customer satisfaction and company performance (Fosso Wamba et al., 2019). The quality of data reflects the completeness, currency, format and accuracy of information. However, the quality of data is affected by processes where the data is handled and analyzed (Ferraris et al., 2019). Data comes from different sources either inside or outside the organization, and it exists in structured, semi-structured or unstructured forms, thus, the organization must identify important data and utilize it (Nykänen et al., 2016). Structured data can be stored in databases and therefore can be computed. For this reason, data analysis tools mainly focus on numerical business data (Baars & Kemper, 2008). Semi-structured data is becoming important for companies, but there are difficulties applying analytics to it (Nykänen et al., 2016). In recent years the amount of data has increased due to the Internet, mobile devices and integrated databases generating big data (Olszak & Zurada, 2020). Big data is often described in terms of high-volume, high-velocity and high-variety (Hartmann et al., 2016), veracity, value (Ferraris et al., 2019), high complexity (Coleman et al., 2016), valence, and variability (Saggi & Jain, 2018). Data can be generated by machines, humans or business information systems (Saggi & Jain, 2018). Despite the definitions, big data needs to be processed to gain information and insight. Data is only a set of facts (Nykänen et al., 2016) or a set of signs. Information refers to empirical knowledge, whereas knowledge refers to the meaning in a context 10 Acta Wasaensia (Zins, 2007). Information describes the past and the present, while knowledge enables predictions of the future (Kock et al., 1997). A prerequisite to fully benefit from data is to make it understandable for all employees, so they can be committed to utilizing it in decision-making (Ferraris et al., 2019). The steps (Figure 2) to generate knowledge from data include: the selection of the data set, pre-processing, transformation, data mining, and interpretation (Fayyad et al., 1996). The pre-processing step includes cleaning the data, while the transformation step involves reducing the number of variables, whereas data mining identifies patterns that the final step of interpretation visualizes (Fayyad et al., 1996; Hartmann et al., 2016). Figure 2 Generating knowledge from data (Fayyad et al., 1996; Hartmann et al., 2016). Large companies are making advances in utilizing data, while SMEs appear to be rather slow adopters (Coleman et al., 2016). Data related challenges for SMEs include heterogeneity, scale, timeliness, privacy, and human collaboration (Wang & Wang, 2020), as well as “extremely low understanding of data analytics by SME representatives”, shortage of qualified data analysts, difficulty to choose the suitable solution and to understand the European General Data Protection regulation (Coleman et al., 2016). As SMEs form a great part of the economy in Europe, their development requires attention (Coleman et al., 2016). Evaluation of a company’s maturity for the strategic use of data is the starting point for development. Maturity model asses data adoption by business strategy, data management, analytical skills, technological infrastructure, engagement in data- driven management, leadership and culture, and data governance (Coleman et al., 2016). 2.1.2 Data-driven decision-making Data provides a strategic asset for companies by improving data-driven decision- making (Qaffas et al., 2022). Top management requires scenarios and simulations Data selection Pre- prosessing Transfor- mation Data minining Interpre- tation Acta Wasaensia 11 to be able to quickly determine optimal solution based on complex business data (Lavalle et al., 2011). Data-driven culture refers to the ways that companies deal with data in data creation, collection, consolidation, analysis, and sharing, as well as in decision- making and managerial support (Medeiros & Maçada, 2022). A data-driven culture is required to realize the potential of data, where organization members make “decisions based on the insights extracted from data” (M. Gupta & George, 2016, p. 1053). Decision-making is supported by knowledge sharing, and includes actions to disseminate knowledge enabling access to relevant information as well as data analytics tools that provide a source for sharing knowledge (Ghasemaghaei, 2019). Dissemination is important, as reporting has a strong effect on performance (Bianchini & Michalkova, 2019). Data-driven decision-making can increase performance through data-driven development, processes, marketing and organization management, as well as data products and data-intensive products (Bianchini & Michalkova, 2019). LaValle et al. (2011) reported that organizations that use business information and analytics differentiate themselves in their markets, and they utilize analytics for small and large decisions, as well as for future strategies and daily operations. SMEs need different approaches in decision-making than large companies as there are several differences between them (Salles, 2006): e.g. decision-making processes might be poorly formalized in SMEs and their decision-makers are required to make decisions at different levels. Small companies often have a flat organization, and they are informal and non-bureaucratic, while their management style encourages innovation and owner-managers are often in the central position. For these reasons, the decision-making might be limited to one person (Durst & Edvardsson, 2012; Valkokari & Helander, 2007). Companies must move from experience-based decision-making toward data- driven decision-making. This can be disruptive and cause discomfort in the organization (Auh et al., 2022) because the analytical information derived from big data will reduce the value of the experiential knowledge of central decision- makers as big data tends to improve decisions and performance with analytics capabilities and a knowledge management orientation (Ferraris et al., 2019). Even though the technology is available, challenges may arise in managerial issues, as achieving real gains requires the willingness of employees to utilize facts based on analytics (Auh et al., 2022). Analytics are currently used more to validate actions afterwards, not used in proactive decision-making and a major barrier is the lack of understanding on how to use analytics in decision-making (Auh et al., 2022). 12 Acta Wasaensia Strategic decisions are non-repetitive decisions, where the decision makers have insufficient information (Salles, 2006). The analytical readiness to utilize data in strategic decision-making can be evaluated by cultural readiness to integrate people towards the same goal, leadership commitment, strategic alignment, structures and systems, and talent capacity (Auh et al., 2022). For example, existing technology can hamper an SME’s ability to meet their customers’ needs. When managers are making decision to adopt new information technology solutions, they need to understand how it would improve productivity and whether it would be compatible with existing processes. The challenge is to predict the advancement and the value that new solution provides (Eze et al., 2018). 2.1.3 Business intelligence Decision-making process are facilitated by business intelligence, providing quality, timely and accurate data, considering past events, present and future (Gauzelin & Bentz, 2017). There are two approaches to defining business intelligence, that emphasize either the technology that makes data available, or that emphasize processes to transform data into information (Nykänen et al., 2016). From the technology point of view, business intelligence creates information from data and refers to processes and software used to collect, analyze, and disseminate data aiming at better decision-making (Davenport, 2005). Intelligence refers to finding unseen contexts from data (Herschel & Jones, 2005). Business intelligence encompasses an organization’s large-scale decision support system and is the most important information technology application for an organization to support decision-making (Arnott et al., 2017). The whole business can be built by collecting and analyzing data (Davenport, 2005). The benefits provided by business intelligence lead to the opportunity to reduce costs, and increase revenues and profits (Gauzelin & Bentz, 2017). Gaining a competitive advantage requires understanding the data generated in the business, e.g. SMEs can improve their understanding of business processes by analyzing historical data to search for unknown patterns (Guarda et al., 2013). Competing in the dynamic business environment can be hard. Competitive intelligence is a part of business intelligence providing understanding of the competition in the market (Pirttimäki, 2007). Competitive intelligence collects information and develops insights into the external environment including competitors, customers, suppliers, and technology (Calof, 2020, p. 565; Gauzelin & Bentz, 2017). Market analyses identify changing and emerging trends. In Finland, business intelligence is common among large-scale companies that invest in sophisticated business intelligence tools to improve their decision- Acta Wasaensia 13 making and enhance their competitiveness (Hannula & Pirttimaki, 2003). The most common reasons for large Finnish companies to use business intelligence are to increase their business knowledge, improve operational efficiency and improve their decision-making (Nykänen et al., 2016). SMEs have been slower to adopt business intelligent systems than large companies, and they tend to believe that business intelligence systems are effective only for large-scale companies that are able to invest in technology and have skilled staff to work with it (Gauzelin & Bentz, 2017). Nykänen et al. (2016) noted, that it is not easy for companies to get data out of its repositories for use as an analysis tool, as business intelligence systems do not provide adequate integration with data sources and other applications. In addition to that, business intelligence systems were not perceived to be user friendly. The adoption of business intelligence in SMEs might require adaptation of processes, as information must be available at the right time for the right people (Guarda et al., 2013). Utilizing large amounts of data is a challenge to SMEs and there are no promises that business intelligence itself will bring success (Gauzelin & Bentz, 2017). Challenges that SME might face with business intelligence relate to the high cost of the systems and availability of skilled employees (e.g. with mathematical and IT skills). SMEs can adopt business intelligence in four steps (Figure 3), including defining the needed data, choosing suitable technology and tools, defining the critical success factors, metrics and alerts, and lastly, interpreting and sharing the results (Guarda et al., 2013). Figure 3 Steps to adopt business intelligence in SMEs (adapted from Guarda et al., 2013, p. 189). 2.1.4 Knowledge management As business intelligence concentrates on explicit knowledge, knowledge management covers both, tacit and explicit knowledge (Herschel & Jones, 2005). Knowledge management enables the creation of “big knowledge”, where human What data is needed to form the basis for business intelligence? What business intelligence tools will be used to collect, process and analyze data? What needs to be evaluated or measured? How to disseminate? 14 Acta Wasaensia knowledge defines how the generated information will be utilized at the operational, tactical, and strategical level (Wang & Wang, 2020). Knowledge management provides the opportunity to increase the understanding from the organization’s own experience and it is “a systematic process of finding, selecting, organizing, distilling and presenting information in a way that improves employees’ comprehension in a specific area of interest“ (Herschel & Jones, 2005). Quintas et al. (Quintas et al., 1997, p. 387) defined knowledge management as “the process of continually managing knowledge of all kinds to meet existing and emerging needs, to identify and exploit existing and acquired knowledge assets and to develop new opportunities”. Demarest (1997, p. 379) suggest that knowledge management “is the systematic underpinning, observation, instrumentation, and optimization of the firm’s knowledge economies”. Jennex et al. (2009, p. 183) defined knowledge management as a multidimensional concept of “capturing the right knowledge, getting the right knowledge to the right user, and using this knowledge to improve organizational and/or individual performance”. SMEs manage knowledge differently than large organizations—it is not about scaling down practices, as SMEs have their own uniqueness. SMEs have a lack of explicit knowledge repositories and they utilize external sources of knowledge. Their domain specific knowledge is deep and wide, and they are capable of avoiding knowledge losses in the case of an employee leaving the organization (Desouza & Awazu, 2006). The key elements in knowledge management are knowledge construction, knowledge embodiment, knowledge dissemination and knowledge use (Mcadam & Reid, 2001). The critical success factors for adopting knowledge management in SMEs in prioritized order are management leadership and support; culture; strategy and purpose; resources; processes and activities; training and education; human resource management; information technology; motivational aids; organizational infrastructure; and measurement (Wong & Aspinwall, 2005). Knowledge management challenges are different in large and small companies and researchers tend to apply approaches to SMEs that were created for large companies, which can reduce SMEs’ capacity to act (Durst & Edvardsson, 2012). Compared to large companies, many SMEs do not have a policy for strategic knowledge management, and they conduct knowledge management at the operational level (Durst & Edvardsson, 2012). SMEs are found to be weaker in knowledge construction due to the lack of systematic and formal social interaction as they tend to have an unstructured and short-term approach to organizational learning, and managers might try to block knowledge sharing to prevent Acta Wasaensia 15 knowledge leakage from the company (Durst & Edvardsson, 2012). Knowledge sharing takes time and requires trust. SMEs may have a limited terminology for knowledge, and they may define knowledge as “useful information or a list of scientific facts” (Mcadam & Reid, 2001). A knowledge management model of data for SMEs emphasizes knowledge management over sophisticated information technology and the volume of data. This knowledge management model includes the strategic use of data and long- term planning, a knowledge-guided definition of data requirements, information technology solutions suitable for SMEs, and new outcomes as knowledge, new product or decision-making rules (Wang & Wang, 2020). Companies must integrate the processes of data and knowledge management and to understand the managerial reason for doing so. 2.2 Data analytics in circular economy SMEs Despite the benefits of data-driven decision-making, many companies face challenges understanding how to use data analytics (Iqbal et al., 2018; Lavalle et al., 2011). The managerial and cultural barriers to adopting analytics are larger than the technological obstacles (Lavalle et al., 2011). To take advantage of data, LaValle et al. (2011) reported that better performing organizations utilize analytics at a five-fold rate compared to others. The literature reports numerous challenges for SMEs in terms of utilizing data (Iqbal et al., 2018; Kim et al., 2003; Parra et al., 2019; Ransbotham et al., 2016), such as the lack of capabilities and resources, including lack of financial or technological resources (Ormazabal et al., 2018; Rizos et al., 2016). 2.2.1 Circular economy The expanding global needs, including population, urbanization and climate change mean that social prosperity and the resilience of nature require new management strategies toward the implementation of circular economy (Prieto- Sandoval et al., 2019). The circular economy concept is created by different disciplines such as ecology, economy, engineering, design and business (Prieto- Sandoval et al., 2018). This might be one reason the circular economy has several definitions and concepts which cause challenges for research (Kirchherr et al., 2017) and dissemination (Kalmykova et al., 2018). Until 2012, the circular economy research was focused mainly on China, as China had adopted this concept in its national strategy. Subsequently, Europe started to develop the concept, 16 Acta Wasaensia especially the European Commission, and governments or non-governmental organizations in European countries (Kalmykova et al., 2018). The circular economy is often explained by the 3R principle of reduce, reuse and recycle, that can be applied in the cycle of production, consumption and return of resources (Kirchherr et al., 2017; Prieto-Sandoval et al., 2018). 3R refers to the recirculation of resources, minimizing the demand for resources, and recovering value from waste (Prieto-Sandoval et al., 2018). In addition, the literature explains the 4R framework including dimensions of reduce, reuse, recycle and recover (e.g. Directive 2008/98/EC) and 9R including refuse, rethink, reduce, reuse, repair, refurbish, remanufacture, repurpose, recycle and recover (Kirchherr et al., 2017). The circular economic model obtains supplies from waste and other side streams, makes better use of resources, and is able to reduce the negative impact of industries (Prieto-Sandoval et al., 2019). For doing this, companies must have sufficient information about waste streams. The circular economy aims to be restorative, meaning not only to prevent environmental impacts, but to also repair damage (Murray et al., 2017). However, there are criticisms that the circular economy is weakly linked to sustainable development (Kirchherr et al., 2017), it has overly simplistic aims (Murray et al., 2017), and that there is a lack of economic theory to guide the transition towards a sustainable circular economy (Velenturf & Purnell, 2021). For example, logistics causes environmental impacts from CO2 emissions, which raises the need for optimized collection system, and these require data. The circular economy aims to close the loop of resource flows between production and consumption and to reduce the need of virgin materials and the generation of waste. It is generally viewed as a cycle to extract, transform, distribute, use, and recover materials (Prieto-Sandoval et al., 2018). Customers are increasingly aware of environmental issues, and they require sustainability. This causes the need for companies to verify and report their sustainability and reduced environmental impacts. This can only be done with sufficient data and analyses. Recycling is a fundamental part of the circular economy (Murray et al., 2017), namely recycling is the most common component in circular economy definitions, the second is reuse (Kirchherr et al., 2017). To achieve the targets of the Action Plan for Circular Economy by 2030, the European Union should increase recycling by 100 megatons per year and reduce landfilling by approximately 35 megatons per year (Tisserant et al., 2017). Circular economy strategies contain risks for industry that relates e.g. to the uncertainty of fluctuating demand and supply (Charnley et al., 2019). Cost- efficient waste management requires monitoring data and novel analytic approaches. Containers can be monitored by sensors and transportation control Acta Wasaensia 17 systems, and combining this data with other data sources helps to create data- driven insight into waste flows (Niska & Serkkola, 2018). This insight is important in the circular economy, where uncertainties relating to the volume, quality, time and location of returned end-of-use products may reduce profitability (Bressanelli et al., 2022). The four key conditions in recycling material flows are a sufficient collection system, sufficient volumes, market demand, and the quality of the recycled material, while important drivers involve collaboration in the circular material chain and information exchange (Cramer, 2018). The waste management industry could promote the maintenance of material value in the cycle by providing services for manufacturers and demonstrating the economic value of recycled materials and by sharing waste data (Salmenperä et al., 2021). Regardless of the development and strategies of the circular economy, regulation, taxation and policy systems have been criticized not fully supporting the aims of the circular economy (Bressanelli et al., 2022). Salmenperä et al. (2021) reported that barriers to the circular economy were viewed differently by different actor groups and there is a lack of systemic thinking and barriers exist in the material supply chain. Kirchherr et al. (2018) found that technological barriers were not ranked as a top barrier for the circular economy, but the most mentioned barriers were cultural e.g. lack of consumer interest, a hesitant company culture, operating in a linear system, and limited willingness to collaborate in the value chain. These are serious challenges for the exploitation of external data, which prevents the development of an efficient circular economy system. 2.2.2 Digitalization and the smart circular economy Digitalization offers new opportunities for SMEs to innovate, grow and go to the global markets (Bianchini & Michalkova, 2019). This transition involves large amount of data, which challenges SMEs to access and analyze relevant data, as they might face internal and external barriers. Data-driven circular economy relates to the Industry 4.0 technologies, the term refers to the digitalization that 4th Industrial Revolution enables. It is about smart factories and real-time information for decision-making, where data flows are used in recycling, harvesting components, value assessment of materials and the value of supply chain as well as to inform end-of-life behavior (Charnley et al., 2019). In the smart circular economy waste data forms a resource to be exploited, the focus is in the ecosystem instead of single organization and integrated digital technologies are assessed by their environmental impact (Bressanelli et al., 2022). 18 Acta Wasaensia Manufacturing companies are forced to develop circular economy practices due to the increasing scarcity or resources, digitalization could help to create closed-loop production-consumption systems (Awan et al., 2021). By data analytics, productivity, and resource efficiency as well as waste to resource process can be improved. Identified categories for data utilization in the circular economy include behavior of customers, lifetime of products and services, performance of systems and value chain network and flow of materials (Kauppila et al., 2022). In Finland, there are several publicly available data sources for the circular economy such as registers, statistics and databases containing data about raw materials, products, side streams and production capacity for food, carbon, batteries, textiles, plastics, mining, and industrial production. These data sources can be utilized to design solutions for the circular economy. However, one challenge is that there are no centralized data standards or governance for data collection as the data has several owners from the private and the public sector and there is no systematic data collection or automatic update (Kauppila et al., 2022). These issues lead to challenges in information quality (Fosso Wamba et al., 2019), and thus affect business value and company performance. The smart circular economy is defined as “an industrial system that uses digital technologies during the product life-cycle phases to implement circular strategies and practices, aiming at value creation through increased environmental, social, and economic performance” (Bressanelli et al., 2022, p. 9). The smart circular economy is supported by digital technologies such as the Internet of Things (IoT), data, and analytics. Digitalization enhances sustainability by replacing physical flows with information flows to avoid over-transportation or over-production, both causing costs and environmental impact. Operational efficiency and sustainability can be increased if relevant information is available for the right actor at the right time. One solution to demonstrate material and information flows are digital twins, that is a virtual and real-time illustration of physical objects (Rocca et al., 2020) that can provide relevant information to the right actor at the right time, presenting relevant information to manage and control product life- cycles (Preut et al., 2021) or material flows. In the circular economy, digital twins can provide information for recyclers about materials and instructions for disassembly. For logistics partners it can provide information needed for transportation such as condition and location. Acta Wasaensia 19 2.2.3 Data analytics in circular economy SMEs In the circular economy, the drivers to implement analytics include the visibility of material flows, operational efficiency, and collaboration between supply chain partners (Kazancoglu et al., 2021). Data analytics refers to the business intelligence, while analytics refers typically to data mining and statistical analysis (Chen et al., 2012). Raw data from various sources do not generate value automatically, but data analytics can make sense from data, for example by identifying hidden patterns and relations. Before conducting analytics, raw data must be cleaned, standardized, consolidated, and organized (Bianchini & Michalkova, 2019). SME often underutilize data as they suffer a lack of information technology resources for data collection and analyses (Wang & Wang, 2020). SMEs face challenges because of their limited resources to utilize data for competition by analyzing costs and profits, customer’s purchases, marketing campaigns and long- term risks (Wang & Wang, 2020). Even if the technology is available, challenges might arise due to managerial issues (Auh et al., 2022). For instance, achieving real gains may require the willingness of employees to utilize facts that are produced by analytics (as discussed earlier). The key internal barriers preventing SMEs from adopting data analytics are the lack of managerial awareness and skills to utilize analytics for improving the business, lack of specialists to conduct analyses at an advanced level, inability to assess and prevent digital risks, and limited data sources or amounts of data (Bianchini & Michalkova, 2019). The key external barriers are financial constraints, accessing complementary external data, complex regulation of personal data, and lack of solutions suitable for SMEs. The transition to utilizing analytics requires leadership from executives with quantitative desire; it is known that this is a hard job for business unit leaders as they lack perspective and cross-functional scope to change the company culture (Davenport, 2005). A leader needs to understand quantitative methods and their limitations to apply analytics to business. LaValle et al. (2011) recommended for organizations that are developing analytics, to focus on the highest value opportunity and to start with questions instead of gathering all the available data, and by doing so, wasting resources on cleaning and converting data. Starting with questions means first defining the needed insight to meet business targets, and only after this, is the needed data identified. Companies may act reactively or pro-actively. Reactive companies conduct analyses due to negative situations, while pro-active companies conduct analyses 20 Acta Wasaensia regarding positive situations (Guarda et al., 2013). Companies can use analytics for describing, predicting, and improving operations. For example, in the waste management sector, analytics allows players to predict forthcoming waste volumes from certain customers and recognize exceptions in waste generation, while advanced analytics relates to machine learning and complex behavior prediction (Niska & Serkkola, 2018). Analytics can be used to simulate and optimize material flows in the supply chain, identify the most profitable customers, detect quality issues, and to recognize financial performance drivers (Davenport, 2005). Companies should integrate their internal and external data to entirely utilize the potential of data analytics (Qaffas et al., 2022). For example, waste monitoring data may include the waste quantity, time stamp and location. By combining this data with external data, such as socio-economic data, it may be possible to create predictive models for waste generation (Niska & Serkkola, 2018). With descriptive analytics, each waste producer can be grouped with similar producers and required operations can be planned for them (Niska & Serkkola, 2018). Descriptive statistics shows aspects such as the size of an average order, while predictive modelling enables users to recognize the most profitable potential customers by pooling internal and external data sources (Davenport, 2005). Description aims to achieve human-interpretable patterns, while prediction uses variables to predict future values. These aims can be achieved by data-mining methods such as classification (data to predefined classes), regression (to predict e.g. the volume), clustering (identification of categories), summarization (compact description e.g. by tabulation), dependency modelling (describing dependencies between variables), and change and deviation detection (discovering changes) (Fayyad et al., 1996; Saggi & Jain, 2018). Descriptive analytics helps to understand what is happening at the current moment using dashboards and scorecards (Ghasemaghaei, 2019; Ghasemaghaei & Calic, 2019). Descriptive analytics summarizes data to provide easy access and understanding using graphics and statistical metrics (Coleman et al., 2016). Figure 4 summarizes the explanations for descriptive, predictive, and prescriptive analytics. Predictive analytics helps to determine what is likely to happen in the future and offers forecasts (Ghasemaghaei, 2019; Ghasemaghaei & Calic, 2019). Predictive analytics allows forecasts that are based on historical data (Coleman et al., 2016). Prescriptive analytics helps to identify needed actions for the optimal result, including optimization and simulation (Ghasemaghaei, 2019; Ghasemaghaei & Calic, 2019). Prescriptive analytics uses results of descriptive and predictive analytics and converts them into decision-making (Coleman et al., 2016). Acta Wasaensia 21 Figure 4 Descriptive, predictive and prescriptive analytics. As an example of the circular economy, a waste management company can utilize external data such as maps, orthophotos, road data, property data, civil data, traffic data and weather data to generate insight for managers, transportation planners and drivers (Strand & Syberfeldt, 2020). The analytic value of external data provides predictive opportunities such as avoiding overloaded trucks and estimating the duration of collection routes. Prescriptive opportunities include optimized routes for waste container collections, in addition to optimal fleet usage and configuration. 2.3 Developing the data analytics capabilities of circular economy SMEs Dynamic capabilities refer to a company’s “ability to integrate, build, and reconfigure internal and external competences to address rapidly changing environment” reflecting the ability to gain a competitive advantage—companies cannot buy capabilities, they must build them (Teece et al., 1997). Dynamic means “the capacity to renew competences” in the changing business environment, while capabilities refer to strategic management choices to adapt, integrate and reconfigure skills, resources and competencies (Teece et al., 1997). The strength of dynamic capabilities affects the speed at which the customer’s needs can be met (Teece, 2018). Dynamic capabilities are one of the most studied topics in the intersection of innovation and the circular economy (Sehnem et al., 2022). Digital tools and data are key capabilities in the circular economy, as it is driven by digitalization (Parida Descriptive •Summarise big data to provide easy access and understanding for human •Helps to understand, what is happening now, e.g. the size of an average order •Includes graphics and statistical metrics, dashboards and scorecards Predictive •Use variables to predict future values e.g. the most profitable potential customer •Helps to figure out, what is likely to happen in the future and allows forecasts based on historical data •Includes machine learning and data mining from databases Prescriptive •Uses results of descriptive and predictive analytics, converts them into decision-making •Helps to identify needed actions for the optimal result, •Includes optimization and simulation 22 Acta Wasaensia & Wincent, 2019). Nevertheless, utilizing data and digitalization poses challenges to circular economy SMEs, who would benefit from data enabled development in material flows and customer behavior (Lacam, 2020). Educational organizations can act as development partners for SMEs to enhance the dynamic capabilities of SME. This can take place in co-operation with students, who can facilitate shared knowledge creation, learning and innovation (Kunttu & Neuvo, 2019; Perkmann et al., 2013). 2.3.1 Opportunities data provides In the growing digital environment, a data analytics capability offers opportunities for performance improvement, quality of decision-making and opportunities for innovation. It is difficult to sustain a competitive advantage without analytical capabilities, as they are needed to guide decisions and operations. Analytical capabilities are required to gain insight into processes, customers, supply and demand—to understand changes and take action (Medeiros & Maçada, 2022). The understanding provided by data analytics can transform the means of competition (Ferraris et al., 2019). Performance and data analytics capabilities are related (Ferraris et al., 2019; M. Gupta & George, 2016) as SMEs with higher technological and managerial data analytics capabilities have been found to increase their performance, while knowledge management increases the effect of data analytics capabilities (Ferraris et al., 2019). Data analytics capabilities affect decision-making. Data analytics tools increase knowledge sharing in company, but knowledge sharing does not increase the quality of decision-making without adequate data analytics competency (Ghasemaghaei, 2019). The quality of decision-making is also affected by the quality of the data and sophistication of the analytics tools. Data analytics capabilities must be considered not only from the technological perspective, but as a strategic skill to develop an open innovation process to increase company’s performance, to promote the development of innovations, and to improve customer satisfaction (Arias-Pérez et al., 2022). Data analytics capabilities require skills, abilities and knowledge to enable creating value from data (Visvizi et al., 2021). Circular economy needs collaborators and requires systems thinking - otherwise the supply chain with several actors suffers lack of coordination and conflicting interests as different actors aims to optimize their own activities and they are not Acta Wasaensia 23 viewing the big picture of the whole chain which leads to inefficiency and decreased profitability. All stakeholders should be viewed as a system with a shared goal, that in turn requires trust between actors. Once there is trust, information sharing takes place and data analytics can support coordination with common stakeholder view of shared sustainability goals for circular economy system. (S. Gupta et al., 2019). Circular economy ecosystem with many actors increase the complexity, and the complex interaction with different objectives can be analyzed by game theory (Palafox-Alcantar et al., 2020). Data analytics capabilities do not directly bring sustainable performance, but can lead to circular economy practices and supply chain flexibility (Edwin Cheng et al., 2021). Analytics capabilities need to be connected to data-driven circular economy strategies to achieve sustainable supply chain flexibility that in turn leads to sustainable supply chain performance. 2.3.2 Capabilities needed in data utilization A data analytics capability is an important organizational capability to achieve a sustainable competitive advantage and to enable company performance. It is challenging for companies to recruit employees with high analytic capabilities (Davenport et al., 2001). Data analytics capabilities are positively related to circular economy performance (Awan et al., 2021). The quality of decision-making is driven by data analytics capabilities that enable companies to transfer insight to recyclable and reusable products. LaValle et al. (2011) explain the three level of analytics capability: aspirational, experienced, and transformed. By aspirational, they refer to companies that pay attention to automating processes to gain cost savings—these companies are short of resources that are needed for analytics. Experienced organizations develop analytics to optimize the organization, while transformed organizations aim to improve customer profitability. Investing in data analytics is not enough on its own, the capability must be developed as well, as companies need capabilities that competitors find hard to match (M. Gupta & George, 2016), and the data analytics talent capability is “a significant enabler for firm performance” (Qaffas et al., 2022). A competitive advantage can be gained by developing organizational capabilities for the targeted use of analytics including the expertise to organize data analytics resources, separating data analytics capabilities from data-enabled capabilities, the rationality of data analytics in relation to the quantity and quality of data, utilizing 24 Acta Wasaensia analytics insight in practice, and building managerial trust towards analytics insight (Mikalef et al., 2018). Data analytics competency describes the capability to utilize data analytics, which is a requirement for using analytics tools (Ghasemaghaei, 2019). A data analytics capability refers to the skills required to collect, store, process, analyze and visualize data to create information, including dimensions of data integration and interpretation (Medeiros & Maçada, 2022; Sabharwal & Miah, 2021, p. 9). A data analytics capability includes technical knowledge (as operational systems, programming, statistics), technology management knowledge (such as data resource management), business knowledge (such as business functions and environment), and relational knowledge (such as collaboration) (Qaffas et al., 2022). Visualization is a way to communicate insight extracted from data in a graphical form (Saggi & Jain, 2018). When creating visualizations, the first step is to identify the main questions that need to be answered and the relevant analysis unit (Figure 5). Then, different data types need to be integrated in analysis and visualization. Visualization should be linked during ongoing data collection and it should provide interactions for dynamic visualization (Tay et al., 2018, pp. 664–665). As discussed earlier, it is important to provide information for employees so it is understandable and applicable for their needs. Figure 5 Questions to guide data visualization (adapted from Tay et al., 2018, p. 664). Davenport et al. (2001) identified key competencies and key roles for developing analytics capabilities. The key roles working together are a database administrator, business analyst and data modeler, decision maker and outcome manager. Key competencies relate to the technology skills, statistical modelling and analytic skills, knowledge of the data, knowledge of the business, and communication and partnering skills. De Mauro et al. (2018) identified data related work roles such as business analysts, data scientists, developers, and data engineers. Business analysts transform insight into business impact, data scientists transform data using analytical methods into insight, data developers design data solutions, while What is the question the user aims to answer using this data? How different data types will be integrated in analysis? How visualization is linked to ongoing data collection? What dimensions of data allows visual interactivity? Acta Wasaensia 25 data engineers build infrastructure to store and process big data. Developers and engineers are technology experts who focus on systems and applications, while business analysts and data scientists are business experts connecting data analyses to value creation. These roles and required skills need to be understood in companies. SMEs have a lack of understanding of data and analytics, as they are dominated by field specialists, the general management functions might be poorly covered, and the domain specific culture and conservatism are not interested in management trends (Coleman et al., 2016). For this reason, SMEs could benefit from case study examples and appropriate consulting and analytics services. Ghasemaghaei (2019) found that companies rarely mentioned success in data analytics investments—the explanation was given that companies did not know what was required to apply data analytics tools. Another explanation for the lack of success in data investments is that companies are not ready or do not utilize the gained insight in decision-making (M. Gupta & George, 2016, p. 1049). SMEs find challenges in data projects as long-term data storage, integration of internal and external data, and the fact that SMEs are “more worried about the unstructured nature of the data rather than the volume of data” (M. Gupta & George, 2016). Data analytics readiness in industry can be measured in terms of resources, information systems, culture, and organization (Gürdür et al., 2019). Resource readiness is measured by data analytics tools and human resources working on data analytics. Information systems readiness is measured by data-related policies linked to the definition, collection and utilization of data as well as by easy access to data. Cultural readiness measures the importance of data analytics for employees and the organization. Organizational readiness examines the roadmap towards data analytics and the business impact. Even though the readiness in terms of resources, information systems and culture may be high, the organizational structures enabling the adaption of data analytics can be low (Gürdür et al., 2019). For this reason, companies are recommended to focus actions on the business impact of data analytics by educating managers and employees to utilize analytics on a daily basis. 2.3.3 Developing capabilities There is increasing demand for experts in the field of descriptive, predictive and prescriptive analytics to serve the needs of decision-making and to communicate knowledge to business experts (Qaffas et al., 2022). As SMEs might not be able to 26 Acta Wasaensia recruit data analytics experts, they should build the skills internally (Coleman et al., 2016). As digitalization accelerates the transition towards the circular economy, companies need to develop business analytics resources (Kristoffersen et al., 2021). The development of data analytics capabilities is complex and requires a combination of tangible, intangible and human resources (M. Gupta & George, 2016; Kristoffersen et al., 2021). Tangible resources relate to data, technology, time and finance (M. Gupta & George, 2016; Kristoffersen et al., 2021). Human resources include managerial and technical skills (M. Gupta & George, 2016), systems thinking and data science skills (Kristoffersen et al., 2021). Intangible resources refer to a data-driven culture, organizational learning (M. Gupta & George, 2016), a circular-oriented innovation culture, and openness and co- creation (Kristoffersen et al., 2021). As companies can buy tangible resources, the importance of intangible and human resources are highlighted (M. Gupta & George, 2016). Figure 6 Resources needed to develop data analytics capabilities (M. Gupta & George, 2016; Kristoffersen et al., 2021). Education for analytics and business intelligence needs to be interdisciplinary. Information technology and analytics skills are not enough, analytics specialists need a general understanding of management, logistics, operations management, accounting, finance, and marketing (Chen et al., 2012) to be able to communicate and interact with others. SMEs can use open-source tools and Massive Open Online Courses (MOOC) to develop analytics capabilities or to increase business knowledge among IT personnel, but the challenge is the lack of time allocated to learning and the need for intuitive software with short learning curve (Coleman et al., 2016). Successful business intelligence and analytics education implements the “learning by doing” principle in hands-on projects carried out by student, internships, and industry-guided practicum, as data analytics requires experimentation, meaning Tangible Resources • Data • Technology • Time • Finance Intangible Resources • Data-driven culture • Circular-oriented innovation culture • Organizational learning • Openess and co-creation Human Resources • Managerial skills • Technical skills • Data science skills • Systems thinking Acta Wasaensia 27 trial and error (Chen et al., 2012, p. 1183; Teece et al., 1997). Thus, education needs a strong relationship with industry to promote experiential learning in practice. “Learning is often a process of trial, feedback, and evaluation. If too many parameters are changed simultaneously, the ability of firms to conduct meaningful natural quasi experiments is attenuated. If many aspects of a firm’s learning environment change simultaneously, the ability to ascertain cause-effect relationship is confounded because cognitive structures will not be formed and rates of learning diminish as a result.” (Teece et al., 1997, p. 523) Organizational learning happens by processing information to develop insight and association between actions in the past and the future (Selnes & Sallis, 2003). Organizations try to make sense of information, but they reject some information as they do not have the capability to make sense of it. Organizational learning can happen in a relationship between organizations, where the relationship can be a source and a target for learning that are dependent of the organizations’ willingness to collaborate. Selnes & Sallis (2003) reported findings that relationship learning increases the relationship performance and commitment in collaboration. Interestingly, even though relational trust is a prerequisite for collaboration, a high level of trust can decrease the outcomes of learning for several reasons, such as avoiding negative information and not risking the relationship, not seeking critical information to question the current situation, as well as falling into group thinking. Mutual trust can be created by close and personal-level interactions between key stakeholders that in addition increases commitment towards the collaboration (Kunttu & Neuvo, 2019). Data analytics capabilities can be developed in relationships between industry and universities by applying co- creation pedagogy to apply knowledge in practice (Lahdenperä et al., 2022). 2.4 Summary Data-driven decision-making provides opportunities to reduce costs and increase efficiency but it requires managing and harnessing data into value. In the case of SMEs there is still a lack of understanding, technological infrastructure and expertise in this area. As SMEs form a great part of the economy in Europe, their development in the area of data-driven decision-making needs attention. As the circular economy aims to achieve the efficient use of resources and minimize environmental emissions, data provides valuable opportunities to optimize the material flows. Hence, the key internal barriers in SMEs are a lack of managerial awareness and the lack of skills to utilize analytics. 28 Acta Wasaensia Today there are increasing amounts of data available, but the data itself does not contain any meaning. Data typically originates from different sources inside and outside an organization in many forms, and the most important data needs to be identified and utilized. However, without appropriate data processing there is no information and insight. To fully benefit from data it is necessary to make it understandable for companies, but SMEs face challenges such as the shortage of qualified data analysts, difficulty in choosing suitable solutions, and in understanding data protection regulations. However, investing in data analytics tools is not enough, SMEs must build their analytics capabilities as well. It must be highlighted that data analytics tools increase knowledge sharing, but they do not increase the quality of decision-making without data analytics competency. Dynamic capabilities can be built and reconfigured rapidly to answer to the changing requirements in the business environment, which is an active topic in the intersection of innovation and the circular economy. Data analytics capabilities include the skills required to collect, store, process, analyze and visualize data. As SMEs have a lack of understanding of data and analytics, they could benefit from case study examples and analytics services. As SMEs might not be able to recruit data analytics experts, they should build the skills internally. This requires a combination of tangible, intangible, and human resources. Education on analytics must be interdisciplinary, as there is a need to understand business, management, logistics, finance, and marketing aspects. Successful analytics education could implement “learning by doing” principles to carry out hands-on projects and experimentation. Acta Wasaensia 29 3 METHODOLOGY This chapter discusses the research design, shown below according to a research onion (Saunders et al., 2012) model, which explains the research philosophy and approach, methodological choices, research strategy, time-horizon, data collection and analysis, as well as research quality. The overview of approaches and choices are visualized in Figure 7. Figure 7 Research design according to a research onion model (adapted from Saunders et al., 2012). This research seeks answers to questions of “what” and “how” that can be studied by qualitative research. It is pragmatic research seeking to contribute to practical phenomena from the grassroots level, by finding out how things work from companies. Philosophy: pragmatism Approach: abduction Methodological choise: monomethod qualitative Strategy: case study Time-horizon: cross-sectional Data collection: interview 30 Acta Wasaensia 3.1 Research philosophy and paradigm Research philosophy relates to the development and the nature of knowledge, and it considers what knowledge is acceptable and how it should be developed. Ontology refers to the nature of reality with two aspects: objectivism and subjectivism. According to objectivism, “social entities exist in reality external to and independent of social actors” (Patton, 2015; Saunders et al., 2012, p. 131). Subjectivism refers to the concept “that social phenomena are created from the perceptions and consequent actions of social actors” (Patton, 2015; Saunders et al., 2012, p. 132). In the subjective meaning, individuals in an organization represent their own experience (Bell et al., 2019). The ontology of this dissertation is subjectivism, as the study focuses on SMEs by interviewing company representatives, obtaining their views, opinions and experiences. Epistemology refers to what constitutes acceptable knowledge. Positivism requires observable reality to find regularities and causalities to produce generalizations. Realism refers to the concept “that objects have an existence independent of human mind” (Saunders et al., 2012, p. 136). Interpretivism contrasts with positivism (Bell et al., 2019) and refers to social phenomena and subjective meaning. It is focused on details and related reality. Interpretivisms uses small samples and aims to carry out an in-depth investigation using qualitative data. Pragmatism refers to practical applied research and to multiple ways of interpreting reality and conducting research, as from one single view it is impossible to see the entire picture (Patton, 2015; Saunders et al., 2012, p. 130). Pragmatists may adopt both subjective and objective views, while values are used in interpreting the results, and they collect data by methods that enables well founded and relevant data. The epistemology of this dissertation is pragmatism, as it represents practical applied research that aims to explore SMEs in depth using qualitative data. This study does not aim to observe causalities or produce wide generalizations. A research paradigm “is a way of examining social phenomena from which particular understandings of these phenomena can be gained and explanations attempted” (Saunders et al., 2012, pp. 140–141). The four paradigms can be arranged by dimensions of subjectivist - objectivist and radical change - regulation. The functionalist paradigm is a combination of objectivism and regulatory dimensions. This paradigm is interested in rational explanations and developing recommendations. The interpretive paradigm is a combination of subjectivism and regulation. This paradigm is more interested in understanding and explaining things, than making changes. The interpretive paradigm focuses on building understanding from the experience of individuals (Bell et al., 2019). The radical Acta Wasaensia 31 humanist paradigm is a combination of subjectivism and radical change. The radical structuralist paradigm is a combination of objectivism and radical change. Radical paradigms are interested in achieving a fundamental change according to the analysis of phenomena. This dissertation follows the interpretive paradigm, as the study aims to understand and explain the situation in companies. The paradigm could be a functionalist paradigm, if the aim would be stronger in terms of rational explanation and the creation of recommendations. The paradigm is clearly not radical, as it does not aim to make fundamental changes. 3.2 Research approach There are three research approaches: deductive, inductive and abductive (Bell et al., 2019; Patton, 2015; Saunders et al., 2012). The deductive research approach is used for testing a theory from a theory-based hypothesis, the inductive approach is used to generate theory as a result of data collection and data analysis, while an abductive approach combines both theory generation and theory testing to generate and to test new theories. Deductive research begins with a theory developed from the literature and the aim is to test the theory. When research aims to explore a phenomenon and generate theory, the approach is inductive, and the research starts with data collection. In an abductive approach, data is collected to explore the phenomenon, identifying categories to generate a theory that will be tested through supplementary data collection. This dissertation uses an abductive approach as it combines theory and empirical findings to collect data, identify categories and generate theory. The research group involved in this study has previous experience and has researched the situation in SMEs. 3.3 Methodological choice and nature of research Quantitative research refers to the numerical data collection and analysis methods, while qualitative research refers to the data and methods used with non-numerical data (Bell et al., 2019; Patton, 2015; Saunders et al., 2012). However, quantitative, and qualitative methods can be combined in mixed-methods research. Multi- method research uses more than one, either quantitative or qualitative, data collection and analysis technique. Quantitative research is typically associated with the research philosophy of positivism and a deductive research approach with experimental and survey research strategies (Saunders et al., 2012, pp. 162–163). Qualitative research is generally associated with the research philosophy of interpretivism and an abductive approach with a variety of research strategies, 32 Acta Wasaensia such as action research or case study research (Saunders et al., 2012, p. 163). Quantitative research examines the connection between variables with numerical measures and analyses by using statistical techniques (Bell et al., 2019; Saunders et al., 2012, p. 162). Qualitative research examines meanings and relationships between them, exploiting a variety of analytical procedures to generate a conceptual framework, where success in research relies on building trust to gain data (Saunders et al., 2012, p. 163). This dissertation represents qualitative research that collects and analyzes non-numerical data, with research questions starting with “what” or “how”. A qualitative method suits interpretivism and abduction and can be used with case study research strategy. The nature of research can be exploratory, descriptive, or explanatory (Saunders et al., 2012, pp. 171–172; Yin, 2018, p. 8). Exploratory research discovers what is happening by asking open questions and it aims to gain insight and understanding into a problem. Descriptive research aims to gain an accurate profile of what is being studied. Explanatory research aims to study causal relationships between variables. This dissertation represents exploratory research, as it discovers the situation of SMEs operating in the circular economy and it collects research data by interviewing respondents with open questions. 3.4 Research strategy A research strategy is a plan of how the research questions aim to be answered, and it forms a methodological link between the research philosophy, data collection, and data analysis methods. Typical research strategies in qualitative research are experiment, survey, archival research, case study, ethnography, action research, grounded theory, and narrative inquiry (Saunders et al., 2012, p. 173). According to Yin (2018, p. 15), a case study is an empirical method that “investigates a contemporary phenomenon in depth and within its real-world context, especially when the boundaries between phenomenon and context may not be clearly evident”. Research design is a logic that connects empirical data to research questions and conclusions. Case study design includes five components: research questions, propositions, cases, logic links from the data to the propositions, and the criteria to interpret the findings (in statistical analyses) or to identify rival explanations for the findings (Yin, 2018, pp. 27, 33). However, exploratory research does not necessarily have any proposition, or the propositions may be a research outcome providing direction and determining what should be examined next (Yin, 2018, p. 28). The research questions show what Acta Wasaensia 33 information needs to be collected. The cases being studied could be a person, an organization, or organizational learning—if the study includes many cases, it is a multiple-case study (Yin, 2018, p. 29). Eisenhardt (1989) described a process of building theory from case study research. Case study research aims to understand the dynamics present in a single setting. The definition of research questions is tentative and research questions can shift during the research. Random selection of cases is not required. Eisenhardt says that theory-building research should start with no theory under consideration, but specification of potentially important variables based on the literature is needed. When selecting cases, random selection is not necessary. Cases can be chosen to fill theoretical categories or to provide examples of polar types. There are no set numbers for the relevant number of cases, but Eisenhardt suggests that a number of cases between 4 to 10 works well. Case analyses should be done within-case and it is important to get familiar with each case before trying to generalize patterns across cases. According to Dubois & Gadde (2002), case studies enable the development of theory by exploiting in-depth insight of phenomena and its context. Abduction- based case studies necessitate an integrated approach, and there is difficulty to manage several interrelated elements. For this reason, the researcher moving between research activities from theory to empirical observation can enlarge understanding. The authors further stated that “theory cannot be understood without empirical observation and vice versa” (Dubois & Gadde, 2002, p. 555). The research strategy in this dissertation is a case study, as this research aims to investigate the in-depth situation in SMEs. As Eisenhardt suggests, cases in this study are not randomly selected. The cases are companies, and the study focuses on their practices, challenges, and development needs. As this research includes several companies, it is a multiple-case study. 3.5 Time-horizon The time horizon in research can be either cross-sectional or longitudinal (Bell et al., 2019). The cross-sectional horizon provides a snapshot of events in a chosen time, while a longitudinal horizon provides “a representation of events over a given period” (Bell et al., 2019; Saunders et al., 2012, p. 190). The time-horizon in this dissertation is cross-sectional, as it studies the practices, challenges, and development needs of case companies at the moment. 34 Acta Wasaensia 3.6 Data collection and analysis An interview is a method to collect research data. There are different ways of conducting interviews: structured, semi-structured and unstructured or in-depth interviews (Patton, 2015; Saunders et al., 2012, p. 374). Structured interviews are constructed by a predetermined set of questions. Structured interviews use questionnaires, and each question must be asked the same way in each interview. This type of interview collects quantifiable data. Semi-structured interviews contain a list of themes to be covered in an interview, where questions and their order may be different in each interview. Unstructured interviews are informal and in-depth interviews without a list of questions, and they are used for in-depth exploration. These types of interviews are often referred to in qualitative research. The strength of interviews in case studies is the opportunity to target the focus on the study topic and the insight for explanations. Weaknesses stemming from interviews are possible bias if the questions are articulated poorly or if the interviewee tells the interviewer what they expect to hear (Yin, 2018, p. 114). Interviews in a case study are guided conversations with a list of themes to be covered. A case study interview typically takes about an hour (Yin, 2018, p. 119). This dissertation uses semi-structured interviews and group interviews, including a list of themes and questions to be covered during the interview. The aim is to have a conversation with the interviewee, as they would provide the big picture to the topic discussed. It is impossible in exploratory research to give exact and strict questions as there is not yet enough information about the current situation in the cases—this is to be explored in the research being conducted. The research data was collected for each research question. The interviews were conducted during 2019-2021. In addition, secondary data was used from company websites and datasets that the companies provided. There are four general analysis strategies: to follow the propositions, to work with data from the ground up, to develop case descriptions and to examine competing explanations (Yin, 2018, pp. 168–172). When working with data from the ground up, there are no propositions, and the aim is to find useful concepts from the data. Cases can be described according to a descriptive framework that comes from the literature. According to Yin (2018, pp. 175–195), there are five analytic techniques for case study research: pattern matching, explanation building, time-series analysis, logic models, and cross-case synthesis. Pattern matching is one of the most desirable techniques when analyzing cases as it compares empirically-based patterns with a predicted one. A pattern search serves generalizability in case study research, and it can be done by constructing an array to search for similarities or Acta Wasaensia 35 differences. Explanation building describes aspects such as how and why something happened. Cross-case synthesis is for multiple-case study research when the aim is to compare or synthesize cases (Patton, 2015; Yin, 2018, p. 196). The analysis strategy in this dissertation is developing case descriptions and conducting pattern matching, which are then summarized (RQ1, RQ4) or compared (RQ2, RQ3) in the research question specific articles within a selected framework from the literature. The analysis me