On relating big data analytics to supply chain planning: towards a research agenda Open Access

Data has never captured as much attention as it currently does. Having emerged rapidly in the last two decades, big data and big data analytics (BDA) have widely affected the way companies conduct their businesses. The knowledge and insights excavated from the mass amount of data generate competitive advantages for the organisations that master this technological innovation (Gunasekaran et al., 2017; McAfee and Brynjolfsson, 2012; Waller and Fawcett, 2013). There is little doubt that the emergence of big data has an unmistakable impact on the amount and speed of data processed in supply chains.

Supply chain planning (SCP) is a data-driven process that focuses on the activities of developing plans to operate supply chains, translating requirements to feasible programmes and optimising outcomes under given constraints (Supply Chain Council, 2012). Big data and BDA demonstrate significant relevance and applicability to the SCP activities (Brinch, 2018), helping to balance requirements and resources and to determine planned capabilities for setting demand forecasts, inventory levels, material location and allocation and production schedules (Stadtler and Kilger, 2005). For instance, Walmart has stretched its demand forecasting to an hourly basis by analysing the customer-generated big data (ProjectPro, 2017). Amazon is piloting the “anticipatory shipping” that moves products close to their customer prior to the order placement (Mitchell, 2015). Recently studies also discussed the use of a BDA-based solution by start-ups to support decision making when facing grand sustainability challenges such as waste prevention in food supply chains (Ciccullo et al., 2021).

Contributions on BDA application in supply chains have been steadily growing over the last decade. A handful of special issues, sections and literature reviews have hit the top-ranking supply chain-related journals elucidating the implications of BDA to the supply chain domain (Table A1). Yet, the extant literature still leaves behind several gaps. Firstly, despite the significance of BDA in SCP, current research often focuses broadly on supply chain management while falling short in providing an overview of how BDA changes the planning activities (Brinch, 2018; Jonsson and Holmström, 2016). Secondly, literature concerning the implementation of SCP-related innovation is scarce (Jonsson and Holmström, 2016). While, according to information system management literature, the adoption of technological innovation involves technological, organisational and environmental dimensions (Baker, 2012; Tornatzky and Fleischer, 1990), the technical advancement of BDA is often isolated from its managerial concerns in extant operations and supply chain literature (Hofmann and Rutschmann, 2018).

Therefore, this paper addresses how BDA will reshape the future supply chains with a specific focus on planning. By means of a systematic literature review, we provide a comprehensive overview of the influence of big data and BDA on SCP looking into the big data sources, objective of BDA adoption and changes to the SCP processes. We then bridge this discussion to the implementation of BDA, revealing how organisations and supply chains can reap the benefit from BDA through the technology-organisation-environment (TOE) framework (Baker, 2012; Tornatzky and Fleischer, 1990). In particular, the following research questions (RQ) will be answered:

This study joins the conversation on the contribution of BDA to supply chains and the management of BDA adoption in the SCP context concerning endogenous and exogenous factors. In this paper, the BDA adoption is considered as the initial evaluation of the technological innovation (Fichman, 2000; Zhu et al., 2006). While this paper serves managers and policymakers as a reference for understanding the potential and the management levers of BDA adoption in SCP, a research agenda is proposed for further scholarly development in the field.

The remainder of the paper is arranged as follows: section 2 explicates the conceptual background. Section 3 outlines the research design and review methodology; Sections 4 and 5 present respectively the descriptive result and the findings respectively, which are followed by the discussion and a future research agenda in sections 6 and 7.

Due to the high number of records and attributes in business nowadays, data being collected and processed are often large in size (i.e. volume), with high speed and frequency of generation and exchange (i.e. velocity), which give rise to the potential in elaborating real-time insights (Hofmann, 2017; Russom, 2011; Tiwari et al., 2018). In contrast to traditional systems, big data are collected from a wide range of diversified sources with various perspectives and data formats (i.e. variety) (Richey et al., 2016; Russom, 2011).

Applying advanced analytics to big data, BDA aims to extract meaningful patterns and insights to inform decision-making (Arunachalam et al., 2018; Wang et al., 2016). In the context of BDA, the data sources are no more limited to structured data (e.g. numbers and strings related to transactions) but also encompass semi-structured and unstructured ones (e.g. texts, audio and video in social media, geospatial information and Internet log) (Dubey et al., 2018). BDA can be classified into descriptive, predictive and prescriptive analytics depending on its scope (Souza, 2014; Wang et al., 2016). Descriptive analytics address “what is happening”, identifying problems and opportunities; predictive analytics answer “what will be happening”, forecasting future trends based on the analysis of historical data; while prescriptive analytics explain “what should be happening” to optimise business performance by assessing alternative scenarios (Souza, 2014; Wang et al., 2016). The underpinning BDA models extend across classification, regression, clustering, association, visualisation, semantic analysis, graphical analysis, optimisation and simulation (Nguyen et al., 2018).

The competitive advantage of supply chain management is achieved substantially though SCP (Jonsson and Holmström, 2016). SCP involves multiple functional areas where the processes and activities can be broadly distinguished by: (1) the focal supply chain process – that is sales, procurement, production and distribution, and (2) the planning horizon (Mauergauz, 2016; Stadtler and Kilger, 2005). Taking the supply chain planning matrix as a reference (Stadtler and Kilger, 2005), the processes in SCP span across strategic network planning, demand planning, demand fulfilment & ATP, master planning, production planning and scheduling, purchasing and material requirement planning and distribution and transport planning. Each of the abovementioned SCP processes comprise a series of sub-activities.

BDA has demonstrated high relevance and applicability to the SCP activities (Brinch, 2018). Advanced analytics, such as forecasting and optimisation techniques, provide fundamental support to demand planning, production planning, inventory plans and logistics planning by improving planning accuracy and flexibility (Russom, 2011; Seyedan and Mafakheri, 2020; Wang et al., 2016).

While there is little doubt that BDA has noteworthy business value, not all organisations embracing BDA solutions have observed expected performance improvement (Maroufkhani et al., 2020). The difference mainly lies in how such innovation is incorporated into the organisation, that is the technological innovation adoption process (Cooper and Zmud, 1990; Hazen et al., 2012; Kapoor et al., 2014). One classical framework differentiates the adoption process into three stages: (1) initiation – evaluating the potential benefit, (2) adoption – deciding to use the technological innovation and (3) routinisation – widely integrating the innovation into the organisation's value chain. An alternative view differentiates the stage of pre-adoption and post-adoption, where the latter includes acceptance – steady implementation, routinisation – adjustment in the organisational governance system, and assimilation – full diffusion into organisational functions and processes (Hazen et al., 2012).

Rooted in the innovation diffusion literature, technological innovation adoption studies often rely on the diffusion of innovation (DOI) theory (Rogers, 1983), originally concerns the patterns and stages of innovation diffusion among a network of individuals, and the attributes of innovations. On the other hand, the TOE framework is broadly used in complementary to the DOI theory in explaining drivers, barriers and the context for a wide range of technological innovation adoption in inter-organisational networks (Maroufkhani et al., 2020; Tornatzky and Fleischer, 1990; Zhu et al., 2006).

This paper adopts the standard process of systematic literature reviews (Seuring and Müller, 2008; Tranfield et al., 2003) while considering the idiosyncrasies of the supply chain domain (Durach et al., 2017). Three macro stages were conducted: (1) planning the review, where RQs are developed and research protocol is discussed and shared among the authors; (2) conducting the review, where material collection and selection are carried out and literature analysis was performed by full-article coding; (3) reporting, where findings are reported in descriptive analysis and thematic analysis (see Table 1).

A list of search keywords was identified based on the research scope (Durach et al., 2017; Tranfield et al., 2003) consisting of three groups: (1) SCP-related, (2) “big data” and (3) “management” and “planning”. SCP-related keywords originated from the SCP matrix processes (Stadtler and Kilger, 2005) and were then validated with the search query employed in extant review papers on related topics (e.g. Nguyen et al., 2018; Tiwari et al., 2018; Wang et al., 2016). “Management” and “planning” were attached in search of managerial implications in the primary studies. Similarly, the choice of not including other terminologies related to BDA (e.g. artificial intelligence, machine learning) was aimed at focusing on managerial implications rather than diving into the technical aspects in distinguishing the various technologies (Ardito et al., 2019).

The literature collection was performed in two scientific publication databases – Scopus and Web of Science – which show a high literature coverage in science, management and technology disciplines (Lamba and Singh, 2017) from diverse publishers (e.g. Emerald, Science Direct, Wiley). The search was restricted to articles, reviews and editorials in peer-reviewed journals in English to ensure material quality consistency (Arunachalam et al., 2018). Meanwhile, by checking a sample of conference proceedings, we observed that the insightful conference papers typically manage to become journal publications in a more extensive form with limited time lag. We regularly update the paper database during the review (Durach et al., 2017), and the final consolidation in October 2020 resulted in 1,719 articles.

The selection process adapts the three-step approach by Brinch (2018).

1. Step 1, compliance with the research topic. Articles are screened by title, dissemination outlet and keywords for checking the research scope, removing papers from off-topic disciplines (e.g. urban planning, tourism management, energy transportation). Editorials were removed after careful assessment if they serve as an introduction to research papers with limited original contributions (Lamba and Singh, 2017).

2. Step 2, compliance with the research objective. Based on the abstract and skimmed-through reading, articles were examined if both BDA and supply chain were discussed. Papers were removed if BDA was mentioned solely for future research, or if the focus on supply chain management was very limited.

3. Step 3, compliance with the research questions. In this step, we also limited the scope to contributions from top-ranking journals, referred to as the Q1 and Q2 journals from the JCR ranking 2020. A reading cut was performed to check the contribution of the papers to the RQs, and articles were removed if they presented very limited managerial implications besides technical considerations (see details in Table 1). This process resulted in 72 articles for full-text analysis.

Multiple authors were involved in the selection process to cope with selection bias, and a shared database was used to track the entire selection history of the articles. Any mismatch in the decisions was thoroughly discussed and reviewed among the authors (Durach et al., 2017).

The selected articles show an evident growth over the years with a peak in 2018 (Figure A1), which coincides with the publication of several journal special issues as anticipated in the introduction (Table A1). A few journals dominate the contribution with six papers respectively (i.e. Production and Operations Management, International Journal of Production Research and International Journal of Production Economics), while a long tail is made up of twenty-two journals presenting only one or two publications (Table A2).

Conceptual and theoretical research outweigh the empirical ones among the primary studies (Table 2), which matches the common pattern of new and unexplored research fields (Seuring and Müller, 2008). Forward-looking research is predominant as the adoption of this technological innovation is still sparse. Among the conceptual pieces, conceptual model or framework, as the most popular research method, is often applied for presenting theoretical discussions, envisioning potential use-cases and developing conceptual models of BDA architecture (Babiceanu and Seker, 2016). Analytical models are mostly used to quantify the potential benefit of BDA in supply chains (Hou et al., 2017). In empirical research, survey research takes the lead that is often applied for testifying to the impact of BDA technology and BDA capabilities (Mandal, 2018) on organisational performances (Wamba et al., 2019), while grounded methods are commonly used to explore and elucidate the future of BDA in supply chains (Brinch et al., 2018; Roßmann et al., 2018).

Table 2 presents the use of theories in the primary studies. The result shows that most articles do not take explicit theoretical perspectives, supporting the discussion in previous studies (Durach et al., 2017). Among the theoretical pieces, the resource-based view (RBV) and dynamic capabilities share the lead. These theories are commonly applied to understand the relationship between BDA capabilities and supply chain performance. For instance, Fosso Wamba and Akter (2019) investigated the impact of supply chain analytics capability on company performance with RBV, under the moderating effect of supply chain agility. Mandal (2018) explored the link between BDA personnel capability (i.e. technical knowledge, technology management knowledge, business knowledge, relational knowledge) and supply chain agility performance with dynamic capabilities. Richey et al. (2016) argued that big data constitute a company's dynamic capability by improving organisational responsiveness through reconfiguration and adaption of company resources. The other theory-based studies include Sodero et al. (2019) that use the sociotechnical system to investigate BDA, while viewing BDA itself that accommodate technological capabilities; and Roßmann et al. (2018) that uses the information processing theory, and states that BDA reduce uncertainty and increase decision-making speed while requiring organisational changes for their successful application.

We adapted the unit of analysis in supply chain literature review (Durach et al., 2017) as the focal process of SCP analysed in the primary studies. Five clusters emerged in the content coding.

Process level refers to the cases when the investigation of BDA is restricted to a single SCP process, or when the discussion on BDA is drawn respectively on single isolated SCP processes even if multiple processes are commented. For instance, Zhong et al. (2015) expounded a big data approach to estimate shopfloor manufacturing delivery time, while Boone et al. (2018) proposed an analytical model for reducing forecast errors in demand forecasting.

Process level + contains papers that explicitly elucidate implications and interrelationships between diverse processes, while the primary focus is still on a single SCP process. For instance, while exploring offline order prediction with clickstream data and social media comments, Huang and Van Mieghem (2014) and Choi (2018) related the consequence of improved demand planning to superior inventory performance in supply chains.

Organisational level highlights the essence of a cross-functional perspective of SCP. The link between various processes is better clarified, and SCP is considered as a holistic issue within the company. For instance, Liu et al. (2019) raised a cyber-physical system-based big data model to simultaneously support decisions in smart manufacturing, intelligent logistics and in-shop service. Dubey et al. (2019) investigated big data capabilities required taking an organisational standpoint.

Supply chain dyadic level considers a specific dyad in the supply chain when investigating the SCP problem. Hofmann (2017) assessed the impact of three big data dimensions on the bullwhip effect, considering a supply chain of one retailer and one manufacturer.

Holistic supply chain level emphasises the complexity of multiple supply chain relationships. For instance, Kache and Seuring (2017) investigated the challenges and opportunities of BDA in supply chain management treating the supply chain as an integrated system consisting of multiple partners. Giannakis and Louis (2016), viewing SCP as a cross-organisational activity, denoted the implications of BDA in collaborative planning.

Intersecting with the research methodology, conceptual and analytical modelling has the tendency to take a process-specific or relationship-focused view, while empirical studies, such as survey and grounded methods, are more developed under a holistic perception of supply chains (Table 2). Although literature on relating BDA to supply chains often views SCP processes in functional silos, there is an emerging trend to emphasise the interrelations between multiple SCP processes.

When it comes to big data terminologies, supply chain literature has the tendency to take them for granted without providing explicit definitions. We found the primary studies refer to big data in the following sense:

1. the dimensions a dataset must exhibit that differentiate it from the traditional ones (in 20 papers). These dimensions are typically referred to as the “V”s of big data, ranging from volume, variety and velocity (e.g. Boone et al., 2019; Hou et al., 2017; Nguyen et al., 2018) to veracity (Richey et al., 2016; Sodero et al., 2019) and value (Fosso Wamba and Akter, 2019; Lai et al., 2018; Nguyen et al., 2018; Yu et al., 2018). These features of big data exceed the management and processing capability of traditional data systems, thus presenting substantial challenges.

2. the abilities and attempt to manage and process large and complex datasets (Mandal, 2018; Wang et al., 2016) leading to value creation and development of competitive advantage. The technological capabilities and the ability to manage data pools and validation tools are some of the examples (Sodero et al., 2019).

3. a new paradigm of computing that involves the entire span of activities to extract knowledge from the fast, diverse and massive amount of data, comprising data collection, processing and analysis (Lee et al., 2018).

The lack of consensus in the use of terminologies, together with the constant evolution of the concepts, has made it difficult for supply chain researchers to agree on a common boundary (Barbosa et al., 2017). Hence, drawing on the extant literature, we attempt to propose the following definitions in seeking to develop a coherent use of terminologies in future studies:

1. Big data for supply chains are extremely large (i.e. volume) information assets that are continuously generated from diversified sources (i.e. velocity) and not restricted by the format (i.e. variety). While exceeding the capturing, storage, handling and analytical capability of traditional systems, they hold fact-based actionable knowledge and insight (i.e. value, veracity) for supply chain decision-making.

2. BDA for supply chains are the application of advanced analytic models and techniques on big data with the aim of extracting valuable knowledge and insight, by identifying trends, detecting patterns, assessing scenarios and gleaning invaluable information to facilitate data-driven supply chain decision-making.

Big data and BDA can be applied to various SCP processes and activities with different scope (Table 3).

BDA employs user-generated big data (e.g. product review, user data, search data) to understand the pattern in purchasing decision from final consumers (Boone et al., 2019; Hou et al., 2017; Lau et al., 2018; See-To and Ngai, 2018). Analysis of data from social media (Choi, 2018), point of sales (Boone et al., 2019), query (Bertsimas et al., 2016; Papanagnou and Matthews-Amune, 2018) and clickstream (Huang and Van Mieghem, 2014) helps to improve mid-term sales and demand forecasting accuracy and flexibility, and thus, informs inventory decisions and enhances agility in highly uncertain contexts (Ren et al., 2019). For online product sales, the use of sentiment and neural network analysis on customers' reviews (Hou et al., 2017; Lau et al., 2018), user characteristics (Hou et al., 2017), product-level customer reviews (See-To and Ngai, 2018) and search keywords (Boone et al., 2019) help to improve forecasting accuracy (Hou et al., 2017; Lau et al., 2018), reduce out-of-sample forecast error and support available to promise (See-To and Ngai, 2018).

Supplier selection in long-term purchasing planning benefits from the wide range of non-traditional sources that BDA can handle (e.g. news data, supplier option and alternatives) (Maghsoodi et al., 2018). Moreover, sustainability attributes can be fully integrated into the decision-making process of materials programme planning and supplier selection (Gholizadeh et al., 2020) when BDA is introduced.

BDA in production planning is rather mature compared to the other processes, justified by a higher rate of models and algorithms proposed and tested. Owing to the high velocity of big data, research has mainly focused on improving short-term production planning sensors and RFID-equipped smart objects collect mass real-time data from the shop floor and manufacturing processes that can be used to identify potential bottlenecks, predict cycle time, perform dynamic production scheduling and manage shop floor material flows (Zhong et al., 2015).

BDA supports distribution planning primarily in short-term planning. Through the use of local and network data (Ilie-Zudor et al., 2015), information on weather and traffic condition (van der Spoel et al., 2017), transport planning gains higher efficiency and transparency where predictive analytics can even forecast arrival time for individual trucks.

Big data and BDA influence SCP based on three distinctive roles – supportive facilitator, source of empowerment and game-changer – which differ from each other by the following dimensions: (1) what type of big data can be introduced to support SCP (i.e. source of big data), (2) why integrate BDA for SCP (i.e. objective of adoption) and (3) how BDA can be integrated in existing SCP processes (i.e. changes to SCP).

BDA as supportive facilitator in SCP primarily aims to assist and facilitate improvement in extant SCP processes which consequently improves SCP performance (Gunasekaran and Ngai, 2004), such as demand forecasting accuracy (Andersson and Jonsson, 2018; Hou et al., 2017), production and transportation planning efficiency (Wu et al., 2018; Zhong et al., 2017) and effectiveness of ordering decisions in spare parts inventory management system (Zheng and Wu, 2017). These BDA initiatives overcome the limitations of traditional systems, often targeting the short- or mid-term, and are capable of collecting, storing and processing richer and more granular data stemming from diversified sources. They could be attained in parallel with extant planning processes to affirm planning decisions or as add-on modules to extend existing planning capacities.

BDA as source of empowerment in SCP enable new processes and capabilities in SCP processes compared to traditional planning systems, empowering decisions such as short-term ATP for fast delivery programmes, internal process improvements, sourcing strategy analysis and supplier evaluation and negotiation (Gholizadeh et al., 2020). Most sources of these big data that are new to the SCP (e.g. weather forecast, supplier performance and customer purchasing behaviour) are used to enable process-oriented improvements, such as reducing dependency on forecasting performance (Boone et al., 2019), acquiring higher flexibility and improving the speed and frequency of the decision-making process (Hofmann, 2017).

Finally, BDA as game-changer target radical changes to SCP by integrating new objectives with significant strategic implications in parallel to the traditional planning goals, such as risk detection (Nguyen et al., 2018), disruptions management in global supply chains (Boone et al., 2018), uncertainty-oriented capability development (Wang et al., 2016) and flexibility and agility reinforcement (Fosso Wamba and Akter, 2019; Giannakis and Louis, 2016). A handful of literature has unfolded the link of BDA to sustainable supply chain management (Ren et al., 2019; Singh and El-Kassar, 2019), not limited to the enhancement of energy efficiency (Feng and Shanthikumar, 2018). The integration of the additional goals to extant SCP leads to more complex, multi-objective planning problems, and a range of new data sources must be introduced (e.g. social network data, customer reviews, product lifecycle information). The initiative of game-changer in SCP should be a long-term endeavour that requires full integration of BDA into the existing planning process and practices, extending active coordination and collaboration to the entire supply chain.

The determining factors are identified and classified into the TOE framework (Baker, 2012; Tornatzky and Fleischer, 1990) together with the examples from the primary studies (Table 4). For survey-based papers, we also highlighted if the factor was supported in the original paper. The application of the TOE framework shows that, besides technological ones, also the organisational and environmental factors are also relevant drivers of BDA adoption in SCP.

Technological factors include relative advantage, compatibility with the current system, complexity, trialability, observability, stability and availability of the technology, as well as data quality. When the implication from extant business cases is unclear or no particular benefit is interpreted, and the complexity of the BDA technology is high, organisations will experience reluctance in adopting BDA for SCP (Kache and Seuring, 2017; Richey et al., 2016). However, existing literature shows conflict results in assessing the impact of relative advantage and technology complexity.

Organisational factors cover organisational structure, readiness, strategy and competence at both organisation and personnel level. Organisational structure and supply chain structures affect BDA adoption decision in terms of organisational complexity (Lamba and Singh, 2018; Sodero et al., 2019). Organisational readiness refers to the preparedness of an organisation to accept the technological innovation from the cultural, managerial, financial and human resource perspectives (e.g. top management commitment, presence of slack human resources) (Dubey et al., 2019; Lamba and Singh, 2018). BDA adoption should also seek appropriate organisational competence, which stands for BDA knowledge, big data management capabilities and current IT system in place on a company level (Kache and Seuring, 2017; Lai et al., 2018; Queiroz and Telles, 2018), and BDA technical knowledge, BDA technology management knowledge, business knowledge and relational knowledge (Arunachalam et al., 2018; Dubey et al., 2019; Mandal, 2018) at a personnel level.

Environmental factors consider the general operational context external to the organisation, where security and privacy concerns (Kache and Seuring, 2017; Queiroz and Telles, 2018) and the presence of BDA service providers (Schoenherr and Speier-Pero, 2015) are outlined. Meanwhile, BDA adoption of competitors and the regulatory environment potentially trigger or prohibit the diffusion of BDA in industry depending on the incentive (Lai et al., 2018; Schoenherr and Speier-Pero, 2015).

There is little doubt that the technological advancement of BDA substantially drives its application. Data quality, data availability and technological peculiarities have a significant influence on the relative advantage that BDA can bring in comparison with the existing SCP processes. However, in addition to the organisational and environmental aspects, the determining factors for BDA adoption decision differ depending on the role that big data and BDA play in SCP (Figure 1).

As a supportive facilitator, BDA solutions are either internally developed or acquired to stimulate evolvement of existing SCP processes for better performance (Schlegel et al., 2020). Technological capabilities are vital for the effective use of the emerging technology (Sodero et al., 2019), including the extraction and cleansing of the desired data from the organisation's system (Sodero et al., 2019), and for coping with the high volume of data (Lai et al., 2018). Data preparation, automated dashboard and data visualisation are typically standard descriptive BDA solutions (Schlegel et al., 2020) that can be acquired directly from the market. With a relatively low maturation level in the journey towards BDA-driven SCP, they require the least integration across supply chain functional areas (Jonsson and Holmström, 2016) where organisational and environmental factors are not the central concern of the adoption decision.

BDA solutions as source of empowerment require the integration of new data sources and new processes – for example supply chain partners integration (Richey et al., 2016) and cross-functional planning (Schlegel et al., 2020), individualization and tracking of the material flow (Jonsson and Holmström, 2016) – as combinatorial interventions to the extant processes and systems. In order to allow the enabling mechanisms to extend the capability and scope of SCP (Jonsson and Holmström, 2016), and unfold its value in more dynamic and complex situations (Schlegel et al., 2020), it is necessary to develop a favourable data-driven culture with a precise understanding of the business problem and appropriate top management support (Dutta and Bose, 2015). The presence of clear norms and values towards supply chain information sharing (Roßmann et al., 2018), availability of technology partners and the regulatory environment are of significant importance to ground domain knowledge in the ad hoc BDA solutions in order to integrate the new processes and data sources to the specific operational context (Feng and Shanthikumar, 2018; Schoenherr and Speier-Pero, 2015; Sodero et al., 2019).

BDA as a game-changer assist strategic turnaround based on the reconsideration of organisations and supply chain strategy and processes, resulting in the adaptation of the SCP objectives. Instant coordination of changes along supply chain activities (Jonsson and Holmström, 2016) and efficient decision-making on sustainability issues (Wang et al., 2016) are examples of technological innovations that bring the scope of SCP beyond its current level. These initiatives require the rethinking of the strategic alignment of BDA solution and SCP scope where all TOE dimensions act as strategic levers. While companies may decide to outsource the unfamiliar technological activities to professionals, for example development of BDA solutions concerning the tools for data collection and analysis, the management issues as well as the integration process cannot be outsourced (Lai et al., 2018). Decision recommendations and cross-functional impact assessment by BDA should be based on SCP processes and activate redesign, to appropriately reflect the new objectives (Schlegel et al., 2020). The lack of organisational support and human inaction will leave the decision making for value creation “continued reflecting [on] existing, pre-established goals” even if the underpinning technology changes (Sodero et al., 2019). Issues on data security and data ownership will also become obvious when it comes to necessary data sharing in supply chains (Richey et al., 2016).

While our study highlights that BDA can contribute to SCP in various forms, three major research gaps are revealed. Firstly, extant literature has mainly covered BDA application in a few SCP processes, while how BDA can support other SCP processes is underexplored. Secondly, current research primarily leverages on big data to achieve short-term oriented improvements, owing to its breakthrough in granularity and timeliness. Despite its contribution to the tactical SCP activities, current knowledge falls short in understanding how BDA influences strategic planning decisions focusing on the mid- to long-term, and whether the long-term benefit is simply the accumulation of short-term improvements. Lastly, on relating to BDA, extant literature commonly assumes SCP in functional silos involving single organisations and planning processes. Research needs to clarify how BDA will transform SCP as a holistic process that extends across the boundary of multiple organisations.

Future research should firstly consider the development of conceptual and empirical pieces in investigating how BDA can be applied to support mid- to long-term planning processes (e.g. capacity planning, plant location, supply network planning), potentially integrated with related technologies (e.g. artificial intelligence, IoT and cyber-physical systems).

1. How can BDA assist planning decisions of plant location?

2. How can BDA outperform existing methods in capacity planning?

With regards to the supply chain level, future research may delve into the implication of BDA in collaborative SCP and the coordinating planning actives at the boundary of organisations. The focus can range from the clarification of applications to the assessment and quantification of expected benefit.

1. How do BDA create shared value for supply chain partners?

2. How to manage the coordination of BDA-enabled anticipatory shipping?

Lastly, studies are needed to further elucidate the three roles of big data and BDA developed in this research, thus presenting examples and discussions on BDA in assisting, empowering and radically changing the traditional SCP processes. Future studies should specify the objective of BDA adoption, the consideration of big data sources and how integration to the existing processes can be orchestrated.

1. How can BDA support the integration of sustainability considerations in SCP?

The research field to date is dominated by contributions discussing the technical aspects of BDA in supply chains (Tiwari et al., 2018). Most studies focus on the pre-adoption stage, neglecting how organisations and supply chains manage the phases from adoption onwards (Hazen et al., 2012; Zhu et al., 2006). Practitioners struggle with the implementation of BDA in a SCP context to generate desired outcomes (Jonsson and Holmström, 2016; Schlegel et al., 2020). Despite the fact that a wide range of determining factors for BDA adoption decisions have been researched, it is not clear how they affect the adoption mechanism and how they may interrelate with the distinctive roles of BDA in SCP. Thus, future research is left with ample space to crystalise the adoption and implementation process (Schlegel et al., 2020). The following research directions could address these issues:

Firstly, with a particular focus on a single organisation, empirical research may explore the mechanism of how the factors influence BDA adoption decisions for each role of BDA. We have highlighted the relevance of the TOE factors on a categorical level, while, in connection with the three roles, future work may also explore the hierarchical or casual relationships among the individual factors.

1. How do the determining factors affect BDA adoption in the context of supportive facilitator, source of empowerment and game-changer?

2. How do BDA adoption determining factors contribute to the development of tangible, human-related and intangible BDA capabilities?

Moreover, it is necessary to establish a BDA technology adoption roadmap for SCP, elucidating the behavioural aspects on how organisations react and interact during the post-adoption of the technological innovation. As empirical research is scarce on BDA implementation, especially in specific contexts (Jonsson and Holmström, 2016; Schlegel et al., 2020), future studies may highlight the paths of transformation for organisations and supply chains into a BDA-driven SCP paradigm, scrutinising how such transformation may relate to the development of tangible, human-related and intangible BDA capabilities (Schlegel et al., 2020).

1. How do strategies, processes, systems and people in supply chains adapt with a view to BDA adoption for SCP?

2. How does the behavioural aspect work in the BDA adoption process for different roles of BDA in SCP?

3. How do BDA capabilities interact with the BDA adoption process?

Finally, taking a holistic view of supply chains, it is possible to develop research in addressing the multi-stakeholder perspective on BDA adoption in SCP, addressing, for example: What are the potential synergies and conflicts among the supply chain actors in the management of BDA adoption?

This study presents a systematic literature review to establish a holistic overview on relating BDA to SCP from prior publications. It reveals the contributions of BDA on SCP and the factors determining the adoption decision of this technology innovation. We synthesised definitions for big data and BDA in the supply chain domain in search of a consensus on big data-related terminologies in future supply chain literature. As extant research on relating BDA to SCP mostly focuses on short-term oriented planning problems in isolation, our study questioned the relevance of BDA for long-term SCP activities. Moreover, we argued that, to make effective use of BDA in SCP, it not only requires strong technical infrastructures but also appropriate BDA adoption management actions concerning the organisational and environmental changes. Successful BDA adoption management for SCP should follow a long-term, inter-organisational perspective, orchestrating the determining factors of BDA adoption once the scope of BDA has been clarified.

The theoretical contribution of this paper is mainly threefold. Firstly, it shows that big data and BDA can be used to improve the SCP performance, i.e. with the role of supportive facilitator or source of empowerment, but can also be leveraged to determine a radical innovation in the supply chain, i.e. when it plays the role of game-changer, by enabling SCP to be guided by new objective functions, for example environmental sustainability, or be the compass of the supply chain in case of disruptions. Secondly, it endeavoured to show that the BDA adoption decision is mainly driven by technological factors only when the aim is to improve SCP performance. Organisational and environmental dimensions instead play a determinant role in ensuring that BDA is adopted to reach radical innovation, thus showing the relevance of managerial concerns over technological ones when it comes to extracting value from the adoption of technology innovations. Lastly, this paper has identified several research directions to be addressed by future research.

This work has some limitations. While the material collection and selection process aimed to be inclusive and representative, it may still fail to capture some relevant contributions. Firstly, the explicit search of “big data” in the primary studies could result in omitting relevant papers that coined the technology differently. For instance, artificial intelligence, machine learning, IoT, digital twin and cyber-physical systems all overlap somewhat with what we referred to as big data. The inclusion of these additional keywords could result in an expansion of the journal base. Moreover, we deliberately concentrated on journal papers targeting quality research, while this choice could have led to discarding some recent discussions from conferences. However, despite the potential shortcomings, we still believe that our study offers an accurate and distinctive angle to the existing work while stimulating future conversations on relating BDA to SCP.