Digital innovation: a bibliometric review and research agenda

With the accelerated digital transformation and intensified global societal challenges, digital innovation (DI) has gained increasing interest among scholars (Yoo et al. 2010; Nambisan et al. 2017; Beltagui et al. 2020). Its transformative impact on business, economy, society, and daily life also shows the ubiquitous nature of digital and the quintessential role of innovation for societal welfare and progress. Organizations increasingly invest in DI products, services, and processes. As per (Møller et al. 2022), global company leaders state that their digital investments in 2021, in the middle of Covid-19 pandemic, improved customer experiences (55%) and supported the launch of digital products and services (58%). The rapid advancement of disruptive technologies like artificial intelligence (AI), blockchain, big data analytics, and cloud computing has created new opportunities and challenges in DI for businesses and society. This is also visible in the recent European Commission’s Digital Compass, introducing actions in the EUʼs digital strategy for 2030, as specified by (EBN—European Business and Innovation Centre Network 2022). However, despite plurality of definitions, typologies, nomenclatures in the literature on DI, the ambiguity and complexity of concepts prevent conceptual clarity, and thus weaken the potential for managerial action and policy intervention.

The most common research topics in DI scholarship include the perspective of outputs, which can take the form of digital business model innovation (Frank et al. 2019; Trischler and Li-Ying 2022), product innovation (Yoo et al. 2010), service innovation (Lehrer et al. 2018; Urbinati et al. 2019; Yoo et al. 2010), process innovation (Fichman et al. 2014; Malhotra and Majchrzak 2022) and platform innovation (de Reuver et al. 2018). Another perspective which has also gained increasing interest is DI as an outcome (Nambisan et al. 2017; Henfridsson et al. 2018), providing a better understanding of the effects of DI on organizations, industries, cities, and society (Malhotra and Majchrzak 2022). Furthermore, DI research has extensively focused on the scale of implementing DI, ranging from incremental to radical and local vs. global, as well as benefits and risks of DI (Teece 2018; Parker et al. 2017).

Despite multiple scholarly efforts to describe what constitutes DI and distinguish it from traditional innovation (Nambisan et al. 2017; Yoo et al. 2010), there are still conceptual ambiguities and lack of understanding on how DI research evolves over time. Even recent research conducted across disciplinary boundaries on DI only captures conceptual framing (see Hund et al. 2021) or solely shows the prominent research themes (Cheng et al. 2023), limited to the specific research field or context, such as e.g., DI in entrepreneurial firms (Felicetti et al. 2024). Recognizing the gap that previous research has not captured comprehensively the intellectual structure of DI research over time, missing clarity of connections between the research papers, interrelationships between the concepts, as well as weak interpretations of research agendas and managerial problems in DI, our study addresses the following research questions (RQ):

To answer these research questions, we used bibliometric and textual analysis techniques to unpack the knowledge structure of DI research, following best practices in management and business research (e.g., Ferreira et al. 2019; Piñeiro-Chousa et al. 2020; Donthu et al. 2021). The bibliometric methodology involves using quantitative techniques, such as citation analysis, to examine bibliometric data like publications and citations (Donthu et al. 2021). We applied BiblioShiny (R-package) and VOSviewer tools to conduct performance analysis and science mapping of the DI research field. Afterwards we used NVivo to conduct content analysis of 40 papers with the highest bibliographic relatedness, enabling us to identify four major themes in DI research: (1) antecedents, (2) execution, (3) outcomes, and (4) levels of DI. Additionally, we qualitatively analysed the managerial problems in DI, to understand better what researchers should focus on next empirically. Hence, based on synthesis of bibliometric findings, our study offers four theoretical contributions.

First, we contribute to the literature on dynamics of DI research (Cheng et al. 2023; Hund et al. 2021) by outlining the intellectual structure of DI research that has evolved around themes such as: (1) value creation from DI, (2) capabilities from DI, (3) sustainability- and impact-enabling DI, (4) affordances of DI, (5) organizing forms and processes in DI, and (6) strategic orientation of DI.

Second, our study contributes with the conceptual network structure of DI research, showing the conceptual inter-connection of key phenomena in DI research that goes beyond studies focusing primarily on technologies for DI (Nambisan et al. 2019). Instead, we visually showcase the key motor themes, niche teams, as well as emerging and declining themes in DI over time and discuss their conceptual inter-connection.

Third, this research provides a better understanding of the social network structure in DI research, by mapping the key authors and their inter-connections. With this contribution we outline the need for more integrative DI research that bridges geographical, disciplinary and research fields’ boundaries.

Our fourth contribution is based on four emerging themes identified through content analysis, which enable us to map the next future research agenda of DI research, by suggesting multiple research questions, as well as to outline managerial problems, that scholars should address. While previous research derives a future research agenda on DI conceptually, e.g., based on concepts such as innovation structure, innovation technology and innovation strategy (Cheng et al. 2023), or paradoxes and knowledge recombination in DI (Hund et al. 2021), we take evolutionary and analytical perspectives. Evolutionary perspective means that we study DI over time (by analysing intellectual, conceptual, and social evolution of the phenomenon). The methodological novelty of our study is reflected in our analytical perspective to derive future research directions and questions based on sensemaking of bibliometric synthesis, following seven factors suggested by Mukherjee et al. (2022) for conducting rigorous bibliometric research. This approach allowed us to roadmap future research agenda and structure it accordingly to: (1) antecedents, (2) execution, (3) outcomes, and (4) levels of DI. Relying on the new logics of theorizing about digitalization of innovation, introduced earlier by Nambisan et al. (2017), our bibliometric review research adds a more detailed perspective of understanding the dimensions of DI, such as antecedents, execution, outcomes, and levels. We suggest that it would be worth to examine how these dimensions are interconnected and how for instance emerging technologies drive DI research agenda. Furthermore, with our research we developed a set of insights about the managerial problems, such as organizing, managing, and leveraging the impact of DI. This is the first effort to systematize these DI problems and future research should provide a more nuanced empirical understanding about success and failures in addressing these problems.

The purpose of this section is to present shortly what DI is, what we know and don’t know about it. We define DI, state its characteristics, types, the benefits, and risk associated with it, as well as success conditions that matter for DI processes, outputs, and outcomes. In line with the latest scientific guidelines for conducting bibliometric research, bibliometric analysis as in our case, is considered already a literature review tool, thus we build on (Öztürk et al. 2024, who suggested the following structure): (1) Defining the aim of the research; (2) Collecting data on the relevant literature; (3) Analysis and visualization; (4) Interpreting the findings and results. Following these guidelines, our theoretical background section is focused on DI, for the purpose of presenting the key contributions on this concept.

Our definition of DI is based on the recent grounded theory review research conducted by Hund et al. (2021), describing “the creation or adoption, and exploitation of an inherently unbounded, value-adding novelty (e.g., product, service, process, or business model) through the incorporation of digital technology” (p. 6). This definition integrates three perspectives offered earlier by Nambisan et al. (2017), such as the output perspective in form of new products, services, platforms, created value and experiences; the integration perspective of digital technologies and infrastructure perspective (e.g., digital platforms, AI, blockchain, 3D printing, quantum computing) enabling DI processes; and finally contextual perspective, outlining the possibilities of innovation to be applicable across different contexts and contributing to real world change. For more enhanced understanding, we will refer in this paper to DI processes, outputs, and outcomes, which build upon those perspectives. This view aligns with latest research of Yoo et al. (2024), who suggest that it is important to use the conceptualization of DI as it “implicates using various digital resources to create value, differentiating it simultaneously from physical and IT.” (p.4). Compared to IT, that has been mostly focused on technological capabilities, the term digital has more emphasis on value creation in innovation processes. Researchers also argue that with the widespread of emerging technologies, such as AI tools, there is a big shift in conceptualizing technology beyond the traditional IT function (see e.g., Malhotra and Majchrzak 2024), that is why we focus on DI and outline its characteristics next.

Following the process of conducting bibliometric reviews directed to scholars in management studies (Zupic and Čater 2015; Aria and Cuccurullo 2017; Donthu et al. 2021) and more specifically business research (e.g., Ferreira et al. 2019; Piñeiro-Chousa et al. 2020), this section describes the methodological choices applied during the collection and analysis of data (Fig. 1).

Bild vergrößern

This section reports the findings captured from the bibliometric and content analysis. We start first with summarising the descriptive results of key citation and publication metrics, second, we uncover results from science mapping including co-citation, bibliographic coupling, co-word, and co-author analysis. Last, identified research themes and future research directions are presented based on interpretation from the content analysis of the 40 papers with highest bibliographic relatedness among each other.

Through bibliometric review and analysis of the sample of 315 articles on DI, we make four contributions.

First, in response to the RQ1, we show how the intellectual structure of DI research has evolved over time. We argue that central concepts in DI research form six thematic clusters, such as: (1) value creation from DI, (2) capabilities from DI, (3) sustainability- and impact-enabling DI, (4) affordances of DI, (5) organizing forms and processes in DI, and 6) strategic orientation of DI. As these dynamics highlight inter-connections between the clusters, with this contribution, we extend conceptual and empirical research in DI, by highlighting the need for integration, combining e.g., research on value creation with research on sustainability and impact enabling DI. Some recent studies emphasize already such dynamics, by bringing the role of digital sustainability and entrepreneurship in DI (George et al. 2021), or conceptualizing relationships between affordances and capabilities of DI (such as connecting affordances to capability of openness and generativity, see Nambisan et al. 2019). Also, research on DI has been evolving with the speed of ongoing digital transformation. The central themes of interest for scholars remain management, technology, performance and strategy. However, increasingly, over the last three years, scholars focus has shifted from pure focus on technology towards understanding of DI performance related to “accuracy, flexibility, scalability, and networkability” e.g., in product design (Lee and Berente 2012, p. 1433). With the integration of central concepts/theories from the management research such as dynamic capabilities or actor-network theory or resource-based theory to DI literature, conceptual discussions are becoming more complex (Lokuge et al. 2019; Khin and Ho 2018). For example, Thomson et al. (2022) show how industrial equipment manufacturers can align to the development of technology, business models and ecosystem relationships to advance performance of autonomous solutions. This complexity has implications for the future of scholarship on DI, requiring consideration of interplay between the concepts and disciplines when studying for instance the context or impact of enabling technologies like AI, big data analytics and blockchain on DI processes, outcomes and outputs.

Second, aligned with RQ2, we identify the conceptual network structure of DI research and discuss the thematic evolution of current and emerging DI concepts over time. As a contribution to DI literature, we outline that the trends of studying DI have shifted towards managerial and strategic effectiveness of DI on performance and more recently scholars explore outcomes and dynamic strategies of DI (Vega and Chiasson 2019). Such dynamics highlight the importance of new approaches to studying DI, driven by digital transformation, emerging technologies and critical need to streamline research focus on motor themes, niche teams and emerging themes rather than declining or basic themes. This research provides examples of concepts which can be studied within each theme and thus contributes to DI research, by enhancing the conceptual understanding of the phenomenon amongst unfolding digital transformation (Nambisan et al. 2019). For example, technology strategy entrepreneurship has been a motor theme in recent years, meanwhile, DI management, governance and firm performance have been the niche teams (Endres et al. 2022) and emerging themes that have an increasing scholarly interest are centred on emerging technologies such as e.g., AI, machine learning, robotics, blockchain in DI (Malhotra and Majchrzak 2024; Yoo et al. 2024). Meanwhile, declining themes, as identified by this study are associated with knowledge management, and one of the reasons as argued by recent research is that digital platforms and emerging technologies provide new opportunities for knowledge, value and impact co-creation in DI and thus there is the need for new empirical studies, that explore e.g., the role of DI processes supported by AI in enabling crowdsourcing (Malhotra and Majchrzak 2022).

We also find interestingly that most of the analysed studies focus on the combination of DI antecedents and outcomes, outcomes and levels, or execution and outcomes. To understand better how DI leads to the desired or unintended outcomes, we suggest for the future research agenda to investigate those aspects in more detail, considering the level of analysis, such as individuals, teams, networks, ecosystems, society, or environment (de Reuver et al. 2018; Nambisan et al. 2019).

Third, in synthesizing knowledge to address RQ3, we discuss the evolution of social network structure of the DI research, by outlining the key authors, who significantly advance knowledge in DI field through collaboration with other scholars and have high citation-related metrics, such as e.g., Ola Henfridsson, Kalle Lyytinen, Ann Majchrzak, Jan vom Brocke, Satish Nambisan. Dominating studies of the DI literature stream across six knowledge clusters (identified within intellectual structure) included contributions from scholars, who are primarily in the information systems research field, such as Nambisan et al. (2017, 2019), de Reuver et al. (2018), Fichman et al. (2014), Hinings et al. (2018). These authors mainly outline the key role of digital technologies in innovation and the need for future research to map connections at different levels as well as integrate ideas from different disciplines/areas.

Finally, with the fourth contribution, in response to RQ4 we propose a new research agenda (next section) and describe what type of managerial DI problems the future scholarship should address. Specifically, we add new knowledge on the evolutionary dynamics and future development of DI research, considering scholarly perspectives and managerial problems that need to be addressed. We notice also that recent review studies discuss only limited number of future research areas. For example, Hund et al. (2021) identified only two areas for future research, such as paradoxes in DI and knowledge recombination for DI. Cheng et al. (2023) identified several avenues such as e.g., conceptual boundaries of DI and its closely related phenomena, configurational aspects of technologies or means of provision of DI. In contrast, our study enriches DI research thematically by providing a holistic overview of future research areas, methodological perspectives, geographical gaps, theoretical avenues and examples of research questions. Furthermore, the identified managerial problems of organizing DI, managing DI and evaluating the impact of DI can be useful both for future research and innovation managers.

In conclusion, this research paper employed a hybrid approach, combining bibliometric and textual analyses, to unravel the knowledge structure of DI research. Leveraging tools such as BiblioShiny, VOSviewer, and NVivo, we analysed 315 papers sourced from the Web of Science database. We applied performance analysis, which involved citation and publication metrics, alongside science mapping techniques such as co-citation, bibliographic coupling, co-word, and co-author analyses. Further to this, we used content analysis of 40 papers with the highest bibliographic relatedness to uncover major themes and identify future research directions. Despite the acknowledged research limitations, our findings shed light on the evolving intellectual, conceptual, and social network structures within the topical DI research.

Methodological limitations of this study must be acknowledged. Firstly, the use of BiblioShiny introduces a specific limitation as it only indicates corresponding authors during the visualization of networks, such as in co-citation analysis or similar processes. Although this is also a constraint for VOSviewer, it is important to note that VOSviewer relies on the GCS to generate network maps, introducing an additional limitation. Beyond these identified constraints, it is crucial to recognize broader methodological challenges inherent in bibliometric studies. One overarching limitation lies in the reliance on citation and publication metrics as indicators of research impact, which may not fully capture the depth and influence of scholarly contributions. Additionally, the dynamic nature of DI may present challenges in capturing the rapid shifts and emerging trends within the field, urging researchers to adopt more real-time and adaptive bibliometric methodologies. Not to mention, while traditional bibliometric analyses often overlook the qualitative nuances of research, this limitation is mitigated through the incorporation of content analysis. By scrutinizing publications with the highest bibliographic relatedness among each other, we delve deeper into the contextual relevance of citations and the evolving nature of DI research themes over time. That said, although bibliographic coupling technique allows for newer and niche studies to gain prominence, we still acknowledge that recent and prominent research streams may have been excluded from content analysis due to the applied citation threshold that is set to make the selected sample size tangible. As a result, the findings may disproportionately reflect established knowledge while underrepresenting emerging trends, an important limitation that scholars need to be cognizant about. Lastly, although this study relies on WoS due to its recognized data quality and frequent preference in bibliometric research (Aria and Cuccurullo 2017; Alaassar et al. 2022), we acknowledge that Scopus also provides extensive coverage of academic publications. While our applied search strings yielded a similar number of articles in both databases, the exclusion of Scopus remains a limitation. Future research may consider incorporating both WoS and Scopus to explore potential variations in data coverage and citation trends, despite the manual consolidation challenges associated with combining multiple databases (Donthu et al. 2021).