LEADING IN THE DIGITAL AGE: CONCEPTUALISING DIGITAL LEADERSHIP AND ITS INFLUENCE ON SERVICE INNOVATION PERFORMANCE

Introduction

With a service-oriented pivot in the global economy and services comprising over 70% of the global gross domestic product (Tajeddini et al., 2020), firms have recognised service innovation as a key source of competitive advantage (Hofmeister et al., 2022). Its strategic role in handling disruptive organisational change, such as digital transformation and COVID-19 triggered, addresses an altered understanding of service innovations as an imposed rather than a discretionary management activity. This means that firms now consider service innovation as a strategic necessity to respond to dynamically changing market conditions by improving existing services or creating novel service solutions and experiences (Heinonen and Strandvik, 2021). Research reflects several manifestations of service innovation (e.g., new customer interaction or revenue model), which require organisational capabilities for their delivery (e.g., technology) and management (e.g., co-producing and orchestrating). These innovations may lead to transformed business models for both service and non-service organisations (den Hertog et al., 2010). Whereas service firms innovate their portfolios regularly to stay competitive (Heinonen and Strandvik, 2021), manufacturing organisations increasingly infuse their product offerings with services to exploit new ways of creating value (Khanra et al., 2021; Paschou et al., 2020). These developments illustrate that both kinds of organisations undergo significant changes, not only towards business model setups (Rapaccini et al., 2020) but also in terms of adjusted leadership capabilities.

Strategically, managing organisational capabilities for service innovation becomes ubiquitous for modern leaders to achieve a high level of service innovation performance (SIP) (Plattfault et al., 2015). Leaders also take a vital role in creating nurturing work environments and, thus, influencing employees’ innovative work behaviour (IWB), explaining increased degrees of SIP (Li et al., 2019). These responsibilities underpin the multidimensional character of leadership in digital transformation contexts (Brunner et al., 2023b; Tigre et al., 2023), which the concept of digital leadership (DL) addresses. Eberl and Drews (2021) understand DL as a multidimensional construct capturing a leader’s capabilities relevant to successful employee management and a digital organisation.

Despite an increasingly academic appraisal (Tigre et al., 2023), scholars consider DL vaguely conceptualised (Banks et al., 2022; Benitez et al., 2022), posing challenges in capturing relationships with other constructs such as e-leadership (Avolio et al., 2014). Therefore, research calls for a deeper understanding that conducting multidimensional analyses of the DL construct could generate valuable insights (Borah et al., 2022). Due to its focus on enhancing innovation performance by leveraging a firm’s digital assets, research links DL to innovation at the organisational and individual levels (e.g., Benitez et al., 2022; Erhan et al., 2022). Thus, some studies explore the influence of DL on IWB (e.g., Erhan et al., 2022) as well as IWB’s relation to (service) innovation performance (Li et al., 2019; Shanker et al., 2017). However, a holistic view connecting DL to IWB and SIP is still missing.

To address these issues, we pose the following two research questions (RQs): (1) What are the conceptual elements of DL? and (2) what capability dimensions of DL influence SIP? We answer these RQs by employing a multi-method approach with an exploratory research design (Holton and Walsh, 2017; Walsh, 2015), combining qualitative (Corbin and Strauss, 1990; Gioia et al., 2013) and quantitative (Hair et al., 2021) research methodologies. Our study contributes to the innovation literature by providing a qualitatively founded construct of DL (34 semi-structured interviews) and quantitatively (249 survey participants) revealing its influence on IWB and SIP, comparing the results of two structural equation models (multidimensional versus unidimensional view). Overall, these results foster a nuanced understanding of DL as an antecedent to innovation success at different levels, also relevant from a practitioner’s view.

Theoretical Background

Dynamic capabilities view and the microfoundations perspective

We employ the dynamic capabilities view (DCV) (Teece et al., 1997; Teece, 2007) as a conceptual lens to explore relations between the constructs (DL, IWB, SIP) in our research model, taking a microfoundations perspective (Felin et al., 2015). As a form of organisational capabilities, dynamic capabilities target competitive advantage in dynamic environments by purposefully reshaping a firm’s resources, support systems (e.g., governance), strategy, and environment (Schilke et al., 2018). Unlike ad-hoc problem solving, dynamic capabilities are systematic, involving reliable, repeatable actions for strategic change (e.g., Helfat and Winter, 2011). The DCV suggests that firms that excel in developing and deploying dynamic capabilities can establish long-term competitiveness (Helfat and Winter, 2011; Teece, 2007).

The microfoundations perspective denotes a vibrant movement that research interprets differently with respect to the DCV (Felin et al., 2015). We follow an actor-oriented definition from Eisenhardt et al. (2010, p. 1263) conceptualising microfoundations as “the underlying individual-level and group-level actions that shape strategy, organisation, and, more broadly, dynamic capabilities, and lead to the emergence of superior organisation-level performance”. This stance holds significance in DCV research on digital transformation, illuminating the interplay between digital technologies, human elements, and organisational outcomes (Konopik et al., 2022; Vial, 2019; Warner and Wäger, 2019). However, several microfoundations remain underexplored, including the relationships between leadership and dynamic capabilities (e.g., Felin et al., 2015; Schilke et al., 2018), and individual capabilities and organisational innovation (e.g., Scuotto et al., 2021).

We address these research potentials with our work by exploring DL as a lower-order construct with individual capabilities directly and indirectly (i.e., over IWB) linked to dynamic capabilities at the firm-level that drive SIP (e.g., Benitez et al., 2022; Scuotto et al., 2021). Referring to research on the performance effects of service innovation related to dynamic capabilities (e.g., Janssen et al., 2016), we assume that DL and IWB influence a firm’s dynamic capabilities that elevate SIP. Whereas academics widely acknowledges the capability term as a higher-level construct (e.g., organisations, networks) (Felin et al., 2015), the conceptualisation at an individual level is vague (e.g., Scuotto et al., 2021). Following a competency-oriented approach (Liehr and Hauff, 2022; Klarner et al., 2021), we understand a digital leader’s capabilities as leadership behaviours (e.g., visioning), knowledge (e.g., industry regulations), skills (e.g., language skills), abilities (e.g., critical thinking), and other characteristics (e.g., personality characteristics).

Digital leadership (DL)

Academic literature proposes DL as a social influence process in digital environments (e.g., Banks et al., 2022) and strategically as an approach to achieve digital transformational change in organisations (e.g., Benitez et al., 2022; Kane et al., 2019). In the context of differing understandings of digital transformation (Vial, 2019), we follow a DCV-based perspective defining digital transformation as “an ongoing process of strategic renewal that uses advances in digital technologies to build capabilities that refresh or replace an organisation’s business model, collaborative approach, and culture” (Warner and Wäger, 2019, p. 344). We further understand digital technologies “as combinations of information, computing, communication, and connectivity technologies” (Bharadwaj et al., 2013, p. 471), covering technologies such as SMACIT and AI. Digital technologies profoundly alter innovation dynamics (i.e., convergent and generative digital innovations), necessitating transformed microfoundations along digital sensing (e.g., scouting), digital seizing (e.g., prototyping), and digital transforming (e.g., ecosystem navigation) towards a firm’s strategic renewal of its business model, collaborative approach, and culture (Warner and Wäger, 2019).

Leadership is crucial in developing these digital transformation capabilities, such as by encouraging experimentation and the readiness for change (sensing), fostering an innovation culture (seizing), and incentivising entrepreneurial behaviour (transforming) (Konopik et al., 2022). However, the hurdles in digital transformation that firms often face, stemming from a lack of leadership capabilities, indicate the need for an advanced leadership understanding (e.g., Benitez et al., 2022; Cortellazzo et al., 2019; Kane et al., 2019; McCarthy et al., 2021). In this sense, academics find a digital-induced change in leadership parameters (i.e., the who, what, when, where, why of leadership) and advocate the theorisation through new lenses (Banks et al., 2022). Taking the “what” parameter as an example, scholars challenge the effectiveness of traditional leadership styles (e.g., transformational leadership) in digital setups, requiring behavioural complexity (e.g., Banks et al., 2022; Weber et al., 2022). Research groups current theoretical efforts in this emerging and interdisciplinary field under the term “digital leadership”, though also considering it as fuzzy (Banks et al., 2022; Benitez et al., 2022). Table 1 illustrates selected conceptualisations of DL.

The conceptualisations (Table 1) reveal three major aspects. First, DL is a socio-technical phenomenon targeting multiple organisational dimensions (e.g., technology, strategy and business model, culture) that organisations address depending on their digital maturity (e.g., El Sawy et al., 2016; Kane, 2019). Second, DL definitions vary such as related to the unit of analysis (organisational versus individual) with a focus on the individual level (e.g., AlNuaimi et al., 2023; Fisk, 2002; Kane et al., 2019; Tigre et al., 2023). Third, DL conceptualisations often miss a theoretical grounding (e.g., Eberl and Drews, 2021; El Sawy et al., 2016; Fisk, 2002; Kane et al., 2019; Tigre et al., 2023). Research has made some efforts to theoretically back DL, using organisational capability theory (e.g., Benitez et al., 2022), new institutional theory (e.g., AlNuaimi et al., 2022), skill theory (e.g., Sousa and Rocha, 2019), and several leadership theories both at the macro-level (e.g., upper echelons theory (Erhan et al., 2022)) and the micro-level (e.g., transformational leadership (AlNuaimi et al., 2022)). We follow Eberl and Drews (2021) in defining DL as a multidimensional construct spanning the capabilities of a digital leader and their influence on employee leadership and a digital organisation (e.g., vision, culture, structures), to realise a digitally-enabled business model.

DL differentiates itself from traditional leadership approaches in the strategic utilisation of digital technologies, among others realised by digital visions and strategies, and transformation programs (Kane et al., 2019). This human-related perspective on digital transformation requires a blended set of leadership capabilities, covering conventional types that relate to change communication, ownership, employee empowerment, and new capabilities, such as a transformative vision, digital literacy, and adaptability (Kane et al., 2019). Shedding light on the fuzziness of DL (Banks et al., 2022), Brunner et al. (2023b) provide a structured analysis (characteristics, skills, behaviours, functions) of capabilities for digital leaders, to drive technological change through service innovations. Research formulates digital leaders’ capability requirements, especially from a skill and behavioural perspective (Benitez et al., 2022; Erhan et al., 2022; Gilli et al., 2023; Weber et al., 2022). On the skill level, digital leaders require strategic (e.g., Kane et al., 2019), business (e.g., Benitez et al., 2022), digital (e.g., Klus and Müller, 2021), cognitive (e.g., Gilli et al., 2023) and interpersonal skills (e.g., Schwarzmüller et al., 2018). From the behavioural perspective, digital leaders require a high degree of flexibility by contextually combining different, partly competing behaviours, such as transactional and transformational leadership (Fisk, 2002) or task- versus people-oriented behaviours (Weber et al., 2022).

The study of leadership as a key predictor of workplace innovation is burgeoning (Hughes et al., 2018). In this sense, several studies illustrate leadership’s positive influence on IWB (e.g., Sethibe and Steyn, 2017) and innovation performance (e.g., Anning-Dorson, 2016; Benitez et al., 2022). However, in the context of DL, these relations remain under-researched (Benitez et al., 2022; Erhan et al., 2022). Referring to IWB, academic literature underlines IWB’s role in innovation performance and firm competitiveness, requiring leaders to shape work environments and show leadership behaviours that nurture employees’ innovativeness (AlEssa and Durugbo, 2022; de Jong and den Hartog, 2010). This leadership framing of IWB is academically well addressed (Afsar and Umrani, 2020; de Jong and den Hartog, 2007). Yet, few studies on DL reveal its positive influence on IWB in terms of the exploration, generation, championing, and implementation of ideas (Erhan et al., 2022). Service research increasingly stresses the importance that leadership places on the success of service innovations (Furrer et al., 2020; Kurtmollaiev and Pedersen, 2022). This role ties to initial research on DL, revealing its direct and indirect influence on organisational innovation (e.g., Benitez et al., 2022; Wang et al., 2022b). We conclude that the relation between DL and service innovation performance is under-represented in service research, legitimising this study as a new and relevant research perspective.

Innovative work behaviour (IWB)

IWB defines a set of employee behaviours that aim at the creation and implementation of new and valuable ideas, processes, products or procedures within a work role, group, or organisation, enhancing both employee and organisational performance (de Jong and den Hartog, 2010; Farr and Ford, 1990). Research theorising IWB as a multistage process with differing dimensions often links it to stages of a firm’s innovation process (de Jong and den Hartog, 2007; Janssen, 2000; Kanter, 1988; Scott and Bruce, 1994). A contemporary approach that this work follows comprises the phases of idea exploration, generation, championing, and implementation (de Jong and den Hartog, 2010). Scholars suggest that IWB offers capabilities (e.g., creativity and entrepreneurial skills) beneficial for achieving organisational performance and sustaining firm competitiveness in dynamic environments (AlEssa and Durugbo, 2022). Thus, studying influence factors, especially regarding leadership, represents an active and diverse field of research (de Jong and den Hartog, 2010; Scott and Bruce, 1994; Sethibe and Steyn, 2017). In this way, the literature predominantly proposes transformational leadership’s positive influence on IWB (AlEssa and Durugbo, 2022). Further work addresses servant leadership (Cai et al., 2018), empowering leadership (Rao Jada et al., 2019), ethical leadership (Yidong and Xinxin, 2013) or combinations of leadership styles (e.g., Sethibe and Steyn, 2017). However, DL’s role in IWB remains scarce. In their study, following the conceptualisation of de Jong and den Hartog (2010), Erhan et al. (2022) demonstrate that DL significantly influences IWB. Research further acknowledges IWB’s mediating role on (service) innovation performance (Li et al., 2019; Shanker et al., 2017) while also addressing a gap in the understanding of how digital transformation affects IWB and its outcomes (Wissmann, 2021). Consequently, delving into hitherto uncovered leadership approaches that match the digital transformation context calls for further work on aspects such as DL (AlEssa and Durugbo, 2022; Erhan et al., 2022).

Service innovation and service innovation performance (SIP)

In the context of growing convergence between product- and service-based industries (Kowalkowski et al., 2017; Raddats et al., 2022), research highlights service innovation as an important driver for firm competitiveness in service and non-service industries (Feng et al., 2021; Kindström et al., 2013; Plattfault et al., 2015; Storey et al., 2016).

Despite increasing service innovation studies, academic interpretations of the term vary, especially referring to the ambiguous concept of service (Kurtmollaiev and Pedersen, 2022; Soto Setzke et al., 2023). One classification by Kurtmollaiev and Pedersen (2022) distinguishes service understandings into the object (i.e., service as an output, process, or firm’s strategic intent) and the subject (i.e., service firms or generic actors) of innovation. We follow the output view considering a service as “an intangible output offered to satisfy a particular consumer want or need” (Kurtmollaiev and Pedersen, 2022, p. 640), with service-related firms as the actors of service innovation, also including non-service firms that transition to service companies. The DCV-based definition of service innovation from den Hertog et al. (2010) corresponds to this view, characterising it as the development and commercialisation of a new service experience or solution that can take one or several manifestations, such as a new service concept, customer interaction, value system, revenue model or delivery system. This output-related framing of service innovation differs from servitisation, which encompasses the strategic shift of product-centric organisations to a service-centric business and operating model (Kowalkowski et al., 2017; Kurtmollaiev and Pedersen, 2022), and is often linked to radical service innovations (Goduscheit and Faullant, 2018).

Scholars emphasise digital technologies as enablers for service innovation and catalysts for radical service innovations (Johansson et al., 2019; Soto Setzke et al., 2023). This implies that digitalisation fundamentally transforms services (e.g., digital services), service innovation processes (e.g., co-creation), and the strategic impact of service innovations (e.g., new business models) (Matzner et al., 2018; Sklyar et al., 2019; Soto Setzke et al., 2023). Hence, firms require dynamic capabilities and microfoundations (e.g., new revenue mechanisms) to repeatedly deliver service innovations (den Hertog et al., 2010; Kindström et al., 2013). Leadership is essential in cultivating these capabilities in the digital era (Schoemaker et al., 2018), requiring adjusted capabilities such as related to strategic orientation, alignment, collaboration, digital mindsets, and transformational change (Favoretto et al., 2022; Kolagar et al., 2022; Sklyar et al., 2019; Struyf et al., 2021). Research on DL reinforces this perspective, emphasising digital leaders’ role in shaping a transformative vision and strategy, cultivating innovation culture, empowering employees, promoting cross-boundary collaboration, and redesigning organisational structures (Benitez et al., 2022; Kane et al., 2019).

With the rise of the service economy, the relationship between service innovation and firm performance has developed into an essential research topic (Feng et al., 2021). We refer to SIP as the extent to which an organisation achieves a competitive edge based on the implementation of service innovations (e.g., Storey et al., 2016). Reflecting on the characteristics of services (e.g., intangibility and heterogeneity), research emphasises the measurement of SIP as distinct from other innovation types (de Brentani, 1989; Storey and Kelly, 2001), also raising such challenges as the lagging effect of service innovations on financial performance (Aas and Pedersen, 2011; Plattfault et al., 2015). Since the initial contributions (e.g., de Brentani, 1989), scholars have considered various dimensions (such as financial, customer-based and internal measures) and levels (project, program, firm) for measuring SIP (de Brentani, 1989; Storey and Kelly, 2001), with no standard yet established (Plattfault et al., 2015). We follow a common approach in service innovation research, differentiating between financial and non-financial performance. The former includes short-term commercial metrics, such as the service’s profitability, sales and profit goals and market share (Johansson et al., 2019; Mennens et al., 2018). As service innovation’s purpose is to strengthen a firm’s competitive positioning in the long run, research also measures its less immediate but strategically beneficial effects, such as those that relate to new market establishment, customer attraction and retention and operational efficiency (Feng et al., 2021; Storey et al., 2016).

Multi-Method Approach

This study adopts a multi-method research design, following an exploratory design consisting of two subsequent steps (Holton and Walsh, 2017; Walsh, 2015), in which the qualitative research provides the foundation for our quantitative research (see Fig. 1). The application of a multi-method approach to advance leadership theory has received appreciation, proposing that this approach provides significant value for examining leadership phenomena (Stentz et al., 2012). First, we investigate the concept of DL through a qualitative research design and analyse the existing literature in the field of DL, SIP and IWB to design the interview guidelines. The semi-structured expert interviews (Eisenhardt, 1989; Yin, 2011) conceptualised DL as delivering a holistic data structure. Hence, our study’s analysis employs grounded theory (Corbin and Strauss, 1990), particularly following the approach proposed by Gioia et al. (2013). Second, we verify the qualitative results based on a quantitative research design. Accordingly, we derive our hypotheses from the qualitative data and substantiate them from existing literature (Holton and Walsh, 2017; Walsh, 2015). Consequently, we test our conceptual model by applying structural equation modelling (Hair et al., 2021).

Grounded theory analysis

We conducted semi-structured expert interviews with top and middle managers to capture their experiences and knowledge (Bluhm et al., 2011) of the topic of DL and its relation to SIP. By applying a grounded theory approach (Corbin and Strauss, 1990) and based on the work of Gioia et al. (2013), we derived a data structure on which we formulated our hypotheses.

Data collection and sample. Our data set consists of semi-structured expert interviews purposefully collected from 34 organisations in the German service and manufacturing industry over a one-year timeframe (2021–2022) (see Table 2 and Appendix A.1). To create a comparable data sample of organisations (e.g., in terms of SIP), we applied three criteria. First, we determined which industries exceed the threshold of the digitalisation index score of the Federal Ministry for Economic Affairs and Climate Action (Federal Ministry for Economic Affairs and Climate Action, 2022). Therefore, we concentrated on business service providers (finance, insurance, and energy supply services), ICT (internet and software services), electrical and mechanical engineering (manufacturers of machinery and food production), and vehicle construction (manufacturers of motor vehicles), all scoring above the digitalisation index average. Incorporating both service and manufacturing organisations follows current research on servitisation that covers the transition from product-centric to service-centric business models (Kowalkowski et al., 2017). In this sense, we ensured similar levels of digital maturity, a factor frequently linked to an organisation’s likelihood of effectively facilitating prosperous (service) innovation projects (Kane, 2019). Second, building on Warner and Wäger (2019), we targeted incumbent organisations with a decade-long sector presence and publicly accessible business activity information (e.g., websites or annual financial reports). Third, with these criteria in place, we conducted further desk research by searching organisations’ websites for their service offerings and service innovation projects. Where publicly available, we also reviewed organisations’ annual financial reports to evaluate their SIP levels, measured by factors such as profitability, market share, and innovativeness. After applying these criteria, our final sample comprises 20 service and 14 manufacturing organisations. The sample provides a comprehensive research perspective, featuring a balance of 14 top-level (e.g., CEOs and department heads) and 20 middle managers (e.g., (senior) project leaders), with 19 being male and 15 being female (see Table 2). We follow scholars’ calls (e.g., Gilli et al., 2023) and focus on top and middle managers to examine DL capabilities. From a strategic leadership standpoint, adopting a management perspective offers a comprehensive view, considering management’s strategic (top management) and tactical (middle management) responsibilities in defining and executing organisations’ digital agenda (Banks et al., 2022; Benitez et al., 2022).

To find suitable interview partners we based our selection of interview partners on criterion-i sampling (Palinkas et al., 2015), with experts meeting predetermined knowledge and experience criteria (Cresswell and Plano Clark, 2011) concerning DL in service innovation contexts. This allows for the provision of detailed and generalisable information regarding the topic. To sample knowledgeable participants, we adopted three sampling criteria (see Appendix A.1). First, interviewees had to own either top- or middle-management positions allowing for strategic and tactical insights into DL and SIP. Second, participants had to have held leadership responsibility for at least three years. Third, candidates required profound experience in the field of service innovation, for example, by managing service innovation projects or by leading service units.

To recruit the 34 experts, we followed a three-step approach. First, we examined German service and manufacturing organisations to identify potential candidates meeting our sampling criteria. Second, we incorporated recommendations from interviewed experts to expand our pool of interviewees. Third, we scanned our personal business network to acquire additional potential interview candidates. The interview guideline comprises four parts (see Appendix A.2). We designed most questions to be open-ended to grasp the expert’s perspective and uncover DL capabilities and their link to SIP (Gioia et al., 2013; Pratt, 2009); hence to answer RQ1. In this way, the interviews primarily focused on DL and its capabilities. Following Corbin and Strauss (1990), we continuously updated the interview guideline to include pertinent issues. We conducted all interviews via telephone or video call (e.g., MS Teams) in either German or English. Overall, we gathered over 30$$h of audio material during the interview period. On average, the interviews lasted approximately 50$$min, with a minimum length of 39$$min and a maximum length of 69$$min (see Appendix A.1). We transcribed interview data anonymously after each conversation and analysed the files using MAXQDA.

Data analysis. To structure the data analysis, we adopted a grounded theory approach (Corbin and Strauss, 1990). In general, we analysed interview data based on three iteratively executed steps inspired by Corbin and Strauss (1990) and Gioia et al. (2013). In the first step, we analysed interviews using an open coding technique, aligning codes closely with in-vivo quotes of the interviewees, as recommended by Corbin and Strauss (1990) and Gioia et al. (2013). This step produced a substantial 583 open codes. After categorising these initial open codes based on commonalities and distinctions, we reduced the number to 40 first-order codes. In the second step, while building second-order themes via axial coding, we organised and grouped first-order codes, like the first step. From the 40 first-order codes, we derived eight second-order themes. In the third step, using theoretical coding, we categorised the remaining eight second-order themes into three aggregated dimensions (Fig. 2): (i) personal development capabilities (PDC), (ii) employee and team development capabilities (ETDC) and (iii) organisational development capabilities (ODC). To relate the qualitative framework and factors identified in theory we refer to Appendices A.3 and A.4 for an overview. To ensure that our results created rather than duplicated knowledge, we iteratively reflected and compared collected interview data with existing literature (Suddaby, 2006). The data analysis process involved all three researchers (Crosina and Pratt, 2019; Gioia et al., 2013), responsible for independently coding all the data (i.e., first-order, second-order and aggregated dimension coding). Two kept a thematic distance, to challenge the coding process and reach more aggregation throughout the codes. Figure 2 presents the findings of our qualitatively derived data structure. Following our multi-method exploratory research design, as suggested by Holton and Walsh (2017) and Walsh (2015), our qualitative findings contribute to the hypothesis development process. Hence, describing these findings serves as the foundation for deriving our hypotheses in the context of the structural equation model (SEM).

Development of hypotheses

We referred to our data structure (Fig. 2) and literature to derive hypotheses for our SEM (Holton and Walsh, 2017; Walsh, 2015), exploring the relationship between DL and SIP. To establish a rigorous analysis of this relationship, we investigated the construct of DL in two ways. First, we used the aggregated dimensions of the data structure and modelled them as individual constructs (multidimensional view). Second, we applied a unified view by converging aggregated dimensions to a single DL construct (unidimensional view).

Multidimensional view — personal development capabilities (PDC). The qualitative data structure suggests that the personal capabilities of digital leaders have the potential to positively influence SIP. In our case, PDC consists of three second-order themes: (i) self-leadership, (ii) social expertise and (iii) digital literacy.

Self-leadership is not only about leading oneself; it is also a prerequisite for adequately leading others by developing such capabilities as trust, reflection or openness to experience. Tied to that, one interviewed Partner of a service organisation in the field of legal and banking advisory mentioned: “In my opinion, taking on risks and showing courage in the face of resistance and tolerating mistakes while innovation development makes a successful SIP” (Interviewee #12). Similarly, Chatterjee et al. (2020) argue that the courage of organisation management leads to the development of innovations. In a similar vein, our data show that entrepreneurial courage and risk-taking positively influence organisational innovation performance. Managers’ trust is the key component of sharing knowledge, cooperating intra- or inter-organisational or co-creating innovations. Thus, it positively influences organisational innovation (Cabrilo et al., 2020). One interviewed senior project leader from the automotive industry supported the aspect of trust, stating, “[$\dots$] managers no longer have the ability to judge all work steps because specialisation is becoming ever greater. To achieve the most performance out of service innovations, our managers trust their mid-level management in their subjects” (Interviewee #22).

Social expertise is another important theme to positively influence SIP. Following Rahimnia and Molavi (2021), communicating the efficiency of innovative ideas or programmes in an organisation can have a significant positive influence on innovation performance. Moreover, applying networking mechanisms or fostering interorganisational mechanisms of knowledge-sharing as a manager positively impacts an organisation’s innovation performance (Garousi Mokhtarzadeh et al., 2020).

Regarding digital literacy, Benitez et al. (2022, p. 3) provide a broad overview of skills in the DL context, referring to digital literacy as “the competence of leveraging big data analytics and tools, cloud computing and virtualisation, mobile app design and development, complex business systems, web development, IT architecture, ERP systems, social media, security skills, [$\dots$].” Against the background of our qualitative data, we took a refined view of digital literacy by separating it into three subcategories, namely, data literacy, digital collaboration literacy and technology-business translation. One interviewee, the Head of Operations and Sales of a manufacturing company, explained the following regarding data literacy: “One basic task of a manager is to make decisions, thus, I need a selection of possible solutions. Especially nowadays I have access to a lot of information digitally stored in e.g. software. Having the capability to access and to select the right information from them becomes crucial to make strategically correct decisions for digital leaders in service innovations” (Interviewee #29). Digital collaboration literacy refers to leaders effectively using digital processes, tools and working methods. Technology-business translation addresses a leader’s capability of assessing the business value of digital technologies for the organisation, hence, leaders require basic technological knowledge. In line with our findings and backed by literature regarding self-leadership (Chatterjee et al., 2020), social expertise (Rahimnia and Molavi, 2021) and digital literacy (Benitez et al., 2022), we propose the following hypothesis:

Multidimensional view — employee and team development capabilities (ETDC). Like PDC, we assume that ETDC positively influences an organisation’s SIP. Given our data structure, ETDC consists of three second-order themes: (i) structural empowerment, (ii) leadership empowerment and (iii) agility mindset.

Considering structural empowerment, experts often mentioned especially the transfer of responsibility and power. Thus, one Partner from a consulting organisation states: “Back in the days, it was normal that the decision-making competencies were located at the executive’s agenda. It is wrongly assumed that the manager has better decision-making abilities. I understand my leadership role in such a way that I transfer responsibility and power to my specialists to do this” (Interviewee #1). Supporting this argument, Ye et al. (2019) conclude that a leadership style that transfers responsibility and power to employees or teams can support innovation performance. Furthermore, the current academic discourse provides evidence of a positive relationship between freedoms for creativity and development and SIP (Giannopoulou et al., 2014). Academic literature (Barrett et al., 2015; Mustak, 2014; Singh and Sarkar, 2019) also assumes that providing resources, networks and infrastructure positively influences SIP. Regarding the provision of resources, one managing director of corporate development from a manufacturing organisation stated: “Digital leadership is connected to providing resources and creating right conditions for employees to excel in their [service] innovativeness. No matter whether it is about mundane things such as a laptop. In this area, I see digital leaders as very important designers and decision-makers” (Interviewee #33).

We consider leadership empowerment of employees and teams another relevant capability to enhance an organisation’s service-innovation performance. Following our qualitative results, we observe that coaching, as well as participative leadership behaviour, are present concepts. In the context of coaching, experts often referred to digital leaders as “coaches” in interdisciplinary teams, enabling communication at eye level, trust and respect. According to the experts, this role also includes supporting employees developing their professional skills to perform successfully in service innovations. Leaders showing a certain degree of openness, such as involving employees in corporate decision-making (Huang et al., 2021; Wang et al., 2022a), characterise participative leadership behaviour. Accordingly, managers’ participative leadership behaviour has a positive impact on service innovation performance (Ali et al., 2022). Furthermore, Dehghanan et al. (2021) described participatory leadership styles relating to strong communication about aims within an organisation, thus leading to strengthened team innovation capacity, often connected to strong innovation performance (Broadstock et al., 2020; Prajogo and Ahmed, 2006).

To increase SIP, we propose that digital leaders developing an agility mindset as part of their capabilities and, in parallel, enhancing the agility mindset of employees and teams are crucial. Especially, goal orientation and an ownership attitude, as well as a continuous learning attitude, play an important role. Concerning goal orientation, the literature underlines positively influencing SIP. Drawing on goal-oriented theory, Zhou (2021) found that leaders’ and employees’ goal orientation significantly works as a moderator for innovation performance. Referring to ownership attitude, one senior project lead services development, articulated: “The service innovation process is definitely connected to more self-responsibility. To reach the best performance, both leaders and employees must own their process. Think creatively, work independently and be courageous” (Interviewee #18). Furthermore, a continuous learning attitude can positively influence innovation performance. Day (1994) elaborates that a continuous learning attitude consists of asking the right questions, sharing knowledge and acting accordingly. Additionally, a Partner from an audit-advisory background explained the following about continuous learning attitudes: “I enhance cooperation through discussions, external coaching and a lived error culture. So, knowledge is not only created but also passed on. These aspects lead to the creation of a culture of innovation within my team. This creates SIP” (Interviewee #16). Based on our findings, we capture the relationships between the three second-order themes and SIP with the following hypothesis:

Multidimensional view — organisational development capabilities (ODC). To positively impact SIP, digital leaders possess capabilities to strategically guide and adjust organisational work structures towards organisational vision and strategy. Following our data structure, ODC is based on two second-order themes: (i) change leadership and (ii) collaboration redesign.

Change leadership refers to the inspiration and empowerment of employees to join the organisation’s transformation, involving such measures as communicating a vision, developing change strategies and mobilising resources to achieve desired outcomes (Kotter, 2000). Similarly, the qualitative results show that visioning and strategic alignment are crucial for SIP. One Senior Consultant in Digital Transformation explained: “In several projects with service organisations, we are currently about to introduce OKRs [Objective and Key Results] in the service innovation process, meaning that alignment of goals changes. This will be a good step towards reconciling topics, such as vision, value contribution, and strategy, and thus presenting the entire target system of the company” (Interviewee #8). Such visioning and strategic alignment can maximise return on investments or establish a competitive advantage (van de Wetering et al., 2018) for an organisation, hence, increasing the performance of its services. In a digital context, investigations have shown that strategic IT alignment of an organisation leads to better innovation performance (Cui et al., 2015; Héroux and Fortin, 2018). The building of interorganisational guiding coalitions also links to increased organisational innovation performance. Garousi Mokhtarzadeh et al. (2020) argue that innovation performance can be facilitated whenever the interorganisational coalitions are exploited through interorganisational knowledge mechanisms or learning activities.

Another DL theme denotes collaboration redesign, including such aspects as incorporating diverse communities, frictionless information flows or agile structures. Overall, collaboration among internal structures and/or external organisations indicates improvement in innovation performance (Huang and Chen, 2017). Furthermore, experts described the scaled adoption of agile structures within the organisation as leading to more successful SIP. The vice president of a food production manufacturer stated: “In our service teams we incorporate knowledge from each other by pursuing agile working flows. We have a product owner who helps to represent the vision and aligns the requirements of the customer. We have an agility master who is responsible for transforming the team into a high-performance team. To reach the maximum of our service innovation, we leverage synergies” (Interviewee #27). Combined, the introduction of change leadership (Alotaibi et al., 2015) or the redesign of an organisation’s collaborative approach, such as working agile in teams (Denning, 2017), positively links to service innovation and organisational innovation performance. Therefore, we propose our hypothesis as follows:

Unidimensional view — digital leadership capabilities (DLC). As current research has already discussed (e.g., Wang et al., 2022b), DL has the potential to enhance the innovation performance of organisations. As related research (Benitez et al., 2022; Erhan et al., 2022) handles digital leadership unidimensional, we also consider the possibility of collapsing PDC, ETDC and ODC to a single construct, namely, DLC. Consequently, we propose:

Innovative work behaviour (IWB). Academic research denotes a relationship between DL and IWB (Erhan et al., 2022) and emphasises a relationship between IWB and organisational performance (de Jong and den Hartog, 2010). SIP is a major part of organisational performance (Johansson et al., 2019). Thus, we hypothesise:

Furthermore, DL positively influences IWB (Erhan et al., 2022); hence, it is an essential part of describing how leaders can increase their employees’ innovativeness. Nonetheless, the influence of IWB in SIP is still under study, whereas prior research highlighted IWB’s important role as a mediator (Li et al., 2019; Shanker et al., 2017). Thus, we propose the following hypotheses:

Structural equation model analysis

Data collection and sample. We collected our quantitative data by building a questionnaire following the item development process by Brunner et al. (2023a), Howell Smith et al. (2020) and Schneider et al. (2022). Thus, we used our data structure to develop PDC, ETDC, ODC and DLC items (see Fig. 2 for details) and slightly adjusted existing items for IWB and SIP for our final questionnaire (Erhan et al., 2022; Johansson et al., 2019; Moorman and Rust, 1999) (see Appendix A.5). Prior to collecting the final data set, we ran pilot studies with a total of 82 participants. By conducting pilot studies, we leveraged insights to increase the rigour and appropriateness of our study design, iteratively integrated participants’ feedback and refined our questionnaire items. To select fitting participants, we applied several quality thresholds, such as confidence check, educational background, management position, language proficiency and professional years of experience, while we integrated our qualitative criteria to draft an ideal participant’s profile. Subsequently, we launched the final questionnaire with the support of a commercial provider to recruit eligible participants meeting the predefined criteria. Thus, we reached a greater audience to test the results of our qualitative analysis. We collected 249 valid responses, split evenly between male (49.40% and female (50.60%) participants and comprising a diverse set of service and manufacturing organisations (e.g., automotive, construction, consulting, IT, insurance, machine manufacturing) and holding different management responsibilities (e.g., board member, CXO, head of department, team leader). We double-checked participants’ suitability for our study by asking them about their current position and their job title, to identify middle and top managers. Of the 249 participants, 21% were top managers (e.g., board member, CEO, CFO), and 79% were middle managers (e.g., managing director, head of department, team leader). Table 3 provides details of the sample’s characteristics.

Data analysis. We used the statistical tool R to conduct our structural equation modelling, following the approach that Hair et al. (2021) suggested. The participants scored each item on a 7-point Likert scale, ranging from 1, strongly disagree, to 7, strongly agree. To derive the data for the unidimensional view, we converged the three constructs (PDC, ETDC, ODC) to DLC as a single construct. Ensuring statistical rigour, we carefully checked each structural model for several necessary measures and added age, gender, and firm size as control variables. Additionally, we controlled our structural models (multidimensional versus unidimensional view) against the participants’ hierarchical level by separating the data set into two subsamples, top- ($n=52$) and a middle-management ($n=197$), and against the industries (service versus manufacturing), to verify that different settings do not bias our structural models’ results (post-hoc analysis). Accordingly, by verifying that the indicator reliability, the internal consistency reliability and construct convergent validity (Table 4) and the discriminant validity using the Fornell–Larcker criteria and the Heterotrait–Monotrait (HTMT) ratio (with values significantly different from 1) met the statistical threshold requirements, we confirmed the constructs’ appropriateness. We used the Hair et al. (2021) decision-making process to delete or keep indicators. However, none of the indicators satisfied the requirements for deletion (see Appendix A.5; deletion only if: indicator loading and weight not significant and loading below 0.5). Furthermore, we examined the models’ appropriateness by performing collinearity (variance inflation factors (VIF) below 3), redundancy (forming global single items), explanatory and predictive analysis. Since the model fit measures for both the multidimensional view ($\mathrm{\text{CFI}}=0.89$, $\mathrm{\text{RNI}}=0.89$, $\mathrm{\text{GFI}}=0.85$, $\mathrm{\text{RMSEA}}=0.07$, $\mathrm{\text{SRMR}}=0.05$, $\chi 2=588.01$, $\mathrm{\text{df}}=265.00$$p$-$\mathrm{\text{value}}=0.00$) and the unidimensional view ($\mathrm{\text{CFI}}=0.86$, $\mathrm{\text{RNI}}=0.86$, $\mathrm{\text{GFI}}=0.85$, $\mathrm{\text{RMSEA}}=0.07$, $\mathrm{\text{SRMR}}=0.06$, $\chi 2=710.70$, $\mathrm{\text{df}}=296.00$$p$-$\mathrm{\text{value}}=0.00$) corresponded to the respective cut-off values in an exploratory research design, we confirmed adequate model fit (Garver and Mentzer, 1999; Hair et al., 2021; West et al., 2012). Since common method bias could be an issue, we used several steps to avoid it (Podsakoff et al., 2003). We separated the item sections, added a short description before each set of items and reduced the possibility of response bias by ensuring anonymity and appropriate questionnaire design. Additionally, we calculated Harman’s single factor for both structural models, well below the recommended level of 0.5, and conducted the full collinearity assessment that Kock (2015) proposed. To eradicate the possibility of endogeneity, we ran a two-stage least squares regression analysis, using dynamic capabilities (sensing, seizing, transforming) as the instrumental variable (Teece, 2007). The result of the Wu–Hausman test confirms that endogeneity was not an issue.

Quantitative results. Tables 4–6 show the quantitative and statistical results of our study; the multidimensional and unidimensional structural models appear in Figs. 3 and 4, respectively. After assessing the overall quality of the respective models, we estimated the path coefficients and supposed mediation effects using a bootstrapping procedure (with $n=10,000$).

In the multidimensional view (Table 5, Fig. 3), we hypothesise that each construct (PDC, ETDC, ODC) has a positive effect on SIP. In contrast to our expectation, the hypotheses H1, H2 and H3 are not significant (n.s.) and, thus, not supported (H1: $\beta =0.237$, n.s.; H2: $\beta =0.173$, n.s.; H3: $\beta =0.286$, n.s.). However, our results support the hypotheses H6a, H7a and H8a, which describe the direct effects of PDC, ETDC and ODC on IWB, respectively (H6a: $\beta =0.265$, $p<0.05$; H7a: $\beta =0.234$, $p<0.01$; H8a: $\beta =0.393$, $p<0.01$). We also find support for hypothesis H5 that IWB has a positive effect on SIP, which has a direct effect size of $\beta =0.441$ ($p<0.01$). Furthermore, since we suppose that IWB serves as a mediator between PDC, ETDC, ODC and SIP, we carried out a mediation analysis to inspect these effects. Our results show that IWB is indeed a mediator, finding support for H6b, H7b and H8b (H6b: $\beta =0.118$, $p<0.10$; H7b: $\beta =0.103$, $p<0.05$; H8b: $\beta =0.175$, $p<0.01$). Due to the fact the direct effects are not significant, but indirect effects are significant, we conclude that we have indirect-only mediations (Table 5).

In the unidimensional view (Table 5, Fig. 4), we converted PDC, ETDC and ODC into DLC (one construct). Therefore, we propose that DL has a positive effect on SIP. In contrast to the multidimensional view, the effect of DLC on SIP is positive and significantly supports H4 ($\beta =0.594$, $p<0.05$). Interestingly, the effect size of DLC on SIP is not only significant but also has a great value of 0.594, compared to the multidimensional effects of the individual constructs of 0.237 (PDC), 0.173 (ETDC) and 0.286 (ODC) on SIP. Also, the results show that DLC has a positive and significant effect on IWB ($\beta =0.759$, $p<0.01$), thus supporting our hypothesis H9a. Additionally, we proposed that IWB positively influences SIP. Since the direct effect is positive and significant, H5 is supported ($\beta =0.413$, $p<0.01$). Regarding the mediation analysis, we find support for hypothesis H9b ($\beta =0.314$, $p<0.01$), suggesting that IWB mediates the effect of DLC on SIP. We characterise the mediation type as complementary mediation, which occurs when both the direct and indirect effects are significant while pointing in the same direction (product of effects is positive) (Table 5).

Discussion

In our multi-method approach with an exploratory research design, we conceptualised digital leadership and tested its influence on service innovation performance mediated by innovative work behaviour. Based on our qualitative and quantitative results, we conclude that achieving transformational success in terms of innovative work behaviour and service innovation performance requires a digital leader mastering a triad of personal development, employee and team development and organisational development capabilities. We base this conclusion on testing effects of the three qualitatively derived capability dimensions, using two structural equation models. First, by considering the capability dimensions individually (multidimensional view), we find an indirect-only mediation between digital leadership and service innovation performance. This means that a digital leader’s capabilities only affect a firm’s service innovation performance by positively influencing employees’ innovative work behaviour. Second, by treating digital leadership as a collective construct of capabilities (unidimensional view), we identify a complementary mediation, in the sense that digital leadership capabilities directly and indirectly (over innovative work behaviour) impact service innovation performance. To elaborate on our findings, we first discuss the qualitative results (RQ1) followed by the discussion of our quantitative results (RQ2). Further, we propose several contributions to organisations’ digital leadership agenda regarding influencing their service innovation performance.

Theoretical contributions

Conceptualising digital leadership (RQ1). While research underscores leadership’s paramount role in successful digital transformation (e.g., Benitez et al., 2022; Weber et al., 2022), the concept of digital leadership remains fuzzy (Banks et al., 2022) with limited conceptual groundwork (e.g., Cortellazzo et al., 2019), raising questions about its defining elements. To clarify the concept of digital leadership, we introduce a framework (Fig. 5) that, drawing from academic literature and our findings, delineates its conceptual components. We leverage the DCV (Teece, 2007), specifically its microfoundations perspective, to “connect the dots” between the macro- and micro-level theoretical perspectives of leadership. Compared to other conceptual frameworks (e.g., Hensellek, 2022), our work offers a comprehensive and in-depth theoretical exploration of the digital leadership phenomenon. We further introduce a novel perspective on digital leadership by elaborating on the role of strategic leadership.

As Fig. 5 illustrates, our conceptualisation of digital leadership integrates various theoretical domains, which we organised into two overarching categories: “leadership of the organisation” and “leadership of self and others”. This categorisation draws from work on strategic leadership, harmonising micro- and macro-level perspectives on leadership across the dimensions of self, others, and the organisation (Crossan et al., 2008).

In summary, our qualitative research results, characterising the capabilities of a digital leader in the dimensions of personal development, employee and team development, and organisational development, align with the leadership perception of transcending multiple levels (Crossan et al., 2008). First, theories in the “leadership of the organisation” category examine how to steer an organisation towards greater adaptability in dynamic settings, such as digital ecosystems (Benitez et al., 2022; Chatterjee et al., 2020). We ground this perspective in strategic leadership (Boal and Hooijberg, 2000; Crossan et al., 2008; Vera et al., 2022), an area that has received limited attention in the realm of digital leadership (Benitez et al., 2022; Erhan et al., 2022). This field of research explores the functions (e.g., strategic decision-making) and attributes (e.g., managerial capabilities) of strategic leaders (e.g., C-suite roles, board of directors, and top management teams) (Samimi et al., 2022). Strategic leadership, driven by the upper echelons theory (Hambrick and Mason, 1984), has evolved to address topics related to competitive advantage, specifically the role of strategic leaders in digital transformation, empowerment, and innovation. For instance, scholars investigate the competencies and success factors of strategic leaders in the contexts of innovation and digital transformation, considering a democratised understanding of leadership, such as participative decision-making (Singh et al., 2023).

Digital leadership’s objective to strategically leverage digital assets by continually evaluating and adjusting an organisation’s setup (e.g., business model, processes, structures) aligns with the concept of strategic leadership (Benitez et al., 2022). Our results support this connection by providing insights into the strategic nature of digital leadership capabilities, including digital literacy (e.g., technology-business translation), leadership empowerment (e.g., motivation and encouragement), and change leadership (e.g., visioning and strategic alignment). Drawing from interview responses from mid-level managers regarding their strategic orientation in digital transformation processes, and informed by an academic discourse on individuals constituting strategic leaders (e.g., Simsek et al., 2015; Van Doorn et al., 2022), we propose that digital leaders encompass not only a firm’s dominant coalition (i.e., top-level leadership) but also include additional actors, both within and outside of organisational boundaries (e.g., mid-level managers).

Second, theories in the “leadership of self and others” segment delve into effective leadership in digital working environments (Dittes et al., 2019). These environments, which involve alternative mechanisms of interpersonal influence (e.g., shared leadership), also demand a significant element of self-leadership (e.g., self-awareness), which proves beneficial, especially for virtual team performance (Castellano et al., 2021). Within this category, we link leadership skills and characteristics to the skill theory (Mumford et al., 2000), a perspective that digital leadership research has also embraced (e.g., Gilli et al., 2023; Klus and Müller, 2021). Our qualitative research results align with the skills perspective as they shed light on self-leadership characteristics (e.g., courage and risk-taking, openness to experience) and skills (e.g., data literacy, networking skills, communication skills) of a digital leader.

Leadership behaviours play a pivotal role in effective leadership within digital settings (Cortellazzo et al., 2019). While our findings emphasise the focus on behavioural research, they introduce some ambiguity when attempting to align them with leadership theories and styles mentioned in studies on digital leadership (see Fig. 5). In this way, we adhere to research that advocates moving away from broad leadership styles (e.g., transformational leadership) in favour of a combination of nuanced leadership behaviours adaptable to complex and fast-paced scenarios, such as organisational innovation and change (de Jong and den Hartog, 2007; Rosing et al., 2011; Hughes et al., 2018; Weber et al., 2022). The term “digital leadership capabilities”, which encompasses individual sub-factors (e.g. behaviours, skills) across various dimensions (i.e., personal, employee and team, and organisational development), serves as a bridge between micro- and macro-oriented leadership perspectives, providing a deeper understanding of digital leadership. Drawing on the microfoundations perspective of the DCV (Eisenhardt et al., 2010), digital leadership capabilities aim to facilitate an organisation’s digital transformation by contributing to firm-level dynamic capabilities, particularly those related to service innovation performance (Benitez et al., 2022; Plattfault et al., 2015).

Our framework supports the differentiation from related constructs such as e-leadership (Avolio et al., 2014). Research has made some attempts to distinguish e-leadership and digital leadership; however, clarity is still lacking (Eberl and Drews, 2021; Jameson et al., 2022; Tigre et al., 2023). E-leadership considers a form of virtual leadership that explores the dynamic interactions between advanced information technologies (e.g., video conferencing) and leadership. Precisely, e-leadership defines a context-specific social influence process towards organisational goals (e.g., changed behaviours) that advanced information technologies mediate (Avolio et al., 2014). The Adaptive Structuration Theory (DeSanctis and Poole, 1994), which relates to research on organisational behaviour, forms the basis of this theory (Avolio et al., 2014).

From a technological perspective, digital leadership departs from mediation (e-leadership) towards a technology-intrinsic perception that merges digital and physical setups (Cascio and Montealegre, 2016). This view implies that digital leadership encompasses effective leadership (i.e., leadership of self and others) in digital environments (Banks et al., 2022) but also requires the strategic leadership of digitally-transforming organisations (Benitez et al., 2022). In conclusion, digital leadership not only aims to explain organisational phenomena (e.g., using e-leadership) but also seeks to predict and strategically leverage them, which is why we relate it to the field of strategic management (Durand et al., 2017).

Exploring digital leaders’capabilities (RQ1). Current research on digital leadership draws on a diverse body of evidence but lacks a comprehensive capability perspective (Cortellazzo et al., 2019; Jameson et al., 2022; Tigre et al., 2023). Our taxonomy of digital leadership capabilities (Fig. 2), enriching such work as that by Brunner et al. (2023b), Eberl and Drews (2021) or Tigre et al. (2023), addresses this issue by integrating micro- and macro-oriented leadership perspectives. The taxonomy considers an instantiation of the above-described conceptual framework and demonstrates that a digital leader requires capabilities towards the personal, employee and team and organisational dimensions. The personal dimension involves cognitive, social and digital capabilities relevant to a digital leader’s self-development. To grow employees and teams, a digital leader further requires structural and leadership empowerment as well as capabilities to develop an agile mindset. Finally, to strategically develop an organisation, a digital leader requires capabilities towards change leadership and the redesign of an organisation’s collaborative setup.

When contextualising our results with research, advocating a transformed understanding of leadership in digital transformation contexts (e.g., Schwarzmüller et al., 2018), the discussion arises of whether digital leadership denotes a new or varied form of leadership (e.g., Banks et al., 2022; Borah et al., 2022; Kane et al., 2019). In terms of leadership capabilities, research explores such characteristics as adaptability (Bennis, 2013), such behaviours as participative leadership (Schwarzmüller et al., 2018) and skills such as related to innovation and digital skills (Benitez et al., 2022; Sousa and Rocha, 2019). Our taxonomy thoroughly reflects this pattern, predominantly in the form of established leadership capabilities, achieving elevated relevance in the digital leadership context. As an example, a digital leader’s empowering, change leadership, and collaboration redesign behaviours match Yukl’s (2012) taxonomy of task-oriented (e.g., planning), relations-oriented (e.g., supporting and empowering), change-oriented (e.g., advocating and envisioning change), and external-oriented (e.g., networking) leadership behaviours. We hence understand digital leadership as a varied form of leadership whose effectiveness results from flexibly applying multiple leadership capabilities in a new (digital) environment. Based on our results, we consider capabilities related to digital literacy and agility (e.g., agile mindset development, collaboration redesign) as distinctive for digital leaders. Whereas research prominently addresses digital literacy (e.g., Benitez et al., 2022; Gilli et al., 2023; Kane et al., 2019; Klus and Müller, 2021), though with differing terminology and understandings (e.g., Benitez et al., 2022; Fisk, 2002), capabilities to develop corporate agility (Nold and Michel, 2016) remain less reflected academically (e.g., AlNuaimi et al., 2022; Kane, 2019).

Examining digital leadership’s effects on innovative work behaviour and service innovation performance (RQ2). Innovation academics consider leadership a vital factor of organisational innovation (e.g., Benitez et al., 2022), also applying to service innovation research (e.g., Anning-Dorson, 2016; Schneider et al., 2022; Tajeddini et al., 2020). Despite increasing awareness of digital leadership’s influence on organisational innovation, it remains underexplored in innovation research (e.g., Benitez et al., 2022; Erhan et al., 2022). By revealing the multidimensional character of digital leadership and empirically demonstrating its influence on innovative work behaviour and service innovation performance, we contribute to leadership and innovation research in two ways.

First, by following academic calls for multidimensional constructs of digital leadership (Borah et al., 2022), our results enable a nuanced view of its relation to service innovation performance. Previous work confirmed a positive relationship between leadership and service innovation performance (e.g., Anning-Dorson, 2016; Özsungur, 2019; Tajeddini et al., 2020; Wu et al., 2023), also calling for further empirical work in this field (e.g., Divisekera and van Nguyen, 2018; Tseng et al., 2019). Innovation-targeted studies on digital leadership are also sparse and primarily view leadership as a unidimensional construct (e.g., Benitez et al., 2022; Chatterjee et al., 2020; Erhan et al., 2022; Wang et al., 2022b). We infer that research has not sufficiently explored the relationship between digital leadership and service innovation performance, especially covering the multidimensional aspect of digital leadership. Following this, the differing results between the multidimensional (indirect-only mediation) and the unidimensional (complementary mediation) view are of interest for discussion. We conclude that to directly impact service innovation performance, a digital leader must master all three capability dimensions (i.e., personal, employee and team, and organisational development). This aligns with strategic leadership research requiring leadership to be integrally pursued at the levels of self, others, and the organisation (Crossan et al., 2008). To further explain the holistic character of capabilities demanded of a digital leader, we rely on a current academic discourse suggesting that successful organisational change requires the application of complementary leadership behaviours (e.g., Weber et al., 2022). Following this, Weber et al. (2022) show that a digital leader’s people-oriented behaviours foster employees’ trust and innovativeness and reduce their resistance to change while buffering the downsides of mandatory task-oriented behaviours (e.g., digital visioning increases resistance to change). Our capability dimensions mirror this task- and people-oriented pattern, underpinned by our qualitative (data structure) and quantitative (structural equation model) results. Whereas organisational development capabilities primarily consider task-oriented behaviours, personal development as well as employee and team development capabilities largely comprise people-oriented behaviours. This result underlines that to enhance service innovation performance at the firm level, a digital leader’s focus on organisational development capabilities (e.g., visioning and strategic alignment) is insufficient and requires balance with people-oriented capabilities.

Second, another contribution refers to how a digital leader can influence employees’ innovative work behaviour to further enhance service innovation performance. Research widely acknowledges the positive influence of innovative work behaviour on organisational innovation and business performance (e.g. Afsar and Umrani, 2020; AlEssa and Durugbo, 2022; de Jong and den Hartog, 2010). However, we find few studies linking innovative work behaviour to service innovation performance (e.g. Li et al., 2019), which we extend by revealing innovative work behaviour as not only a mediator but also a construct directly related to service innovation performance. From an antecedent perspective, leadership denotes a vibrant field of research, strongly focused on theory-based leadership styles, such as transformational leadership (AlEssa and Durugbo, 2022; Liehr and Hauff, 2022)). We find few studies on digital leadership (Erhan et al., 2022), which we confirm regarding its impact on innovative work behaviour but also further detail by exploring a digital leader’s capabilities, abstracted from generic leadership styles (e.g., Hughes et al., 2018). This view ties to de Jong and den Hartog (2007) identifying leadership behaviours abstracted from theory-based leadership styles that stimulate innovative work behaviour. Our three tested capability dimensions reflect these behaviours (e.g., role-modelling, providing vision, consulting, delegating, providing resources), which underlines the multidimensional character of digital leadership.

Managerial contributions

As leaders take a high stake in driving a firm’s service innovation performance (e.g., Wu et al., 2023), our study also provides important managerial implications. Our findings indicate that organisations should recruit and develop digital leaders exhibiting capabilities across several dimensions (i.e., personal development, employee and team development, and organisational development), both at the top- and middle-management levels. In this way, our taxonomy of digital leadership capabilities serves as a framework to comprehensively assess an organisation’s setup of digital leadership capabilities, identify gaps, and derive such measures as leadership development interventions (Avolio et al., 2010). The taxonomy can further support recruitment purposes, such as checking and enriching job descriptions for digital leaders in certain positions. The literature reflects a lack of empirical work examining the relationship between leadership capabilities and performance in digitalised organisations (e.g., Cortellazzo et al., 2019). Based on our quantitative results, we aim to sensitise decision-makers to the multidimensional character of digital leadership heading towards enhanced service innovation performance. Therefore, our capability dimensions contribute a quality criterion to a balanced investment in enhancing the maturity of digital leaders’ capabilities. As our results suggest that digital leaders can positively influence service innovation performance, this criterion may positively affect an organisation’s return on investment in leadership development (Avolio et al., 2010). Another managerial contribution refers to digital leaders’ influence on employees’ innovative work behaviour positively related to an organisation’s service innovation performance. The knowledge of the multidimensional influenceability of innovative work behaviour, via both strategic and executional leadership capabilities, should raise leaders’ awareness about their role in shaping work environments (AlEssa and Durugbo, 2022; de Jong and den Hartog, 2010) that nurture the development of service innovations. This role is vital as employees’ innovative potential sources an organisation’s competitiveness (Liehr and Hauff, 2022), especially in the service innovation context (e.g., Feng et al., 2021; Plattfault et al., 2015). However, the transition to digital workplaces alters the innovation setting with new ways of interacting and creating value (Erhan et al., 2022). Therefore, to improve service innovation performance, organisations should incorporate leadership capabilities fostering employee innovation (e.g., digital literacy, visioning, participative leadership, fostering self-organisation, promoting agility) as part of their HR programmes to select, evaluate and develop digital leaders.

Limitations and Further Research

Apart from our contributions, this work is not without limitations. Referring to our research design, we encompassed a broad range of German service and manufacturing organisations in our qualitative examination, enriched with an international sample in our quantitative analysis. To deepen the understanding of digital leadership across different cultures, we recommend that future studies include geographic clusters (e.g., European countries). Furthermore, we took an executive perspective in our sampling. As successful service innovations depend on determinants covering multiple stakeholder groups (den Hertog et al., 2010; Hofmeister et al., 2023), future research should also incorporate employees, customers, or business partners. Although we controlled for hierarchy level in our quantitative analysis, further qualitative research may elaborate on the underlying influence mechanisms of digital leaders, such as middle managers in innovation outcomes (e.g., Schneider et al., 2022). In this context, the employee perspective is particularly interesting for further research into the role of digital leadership in influencing service innovation performance. Future studies could incorporate this perspective to verify digital leadership capabilities and assess perceived differences between digital and traditional leadership. Reflecting on our qualitative results, we focused on the comprehensiveness of digital leadership capabilities abstracted from theory-led leadership styles, to contribute to the conceptualisation issue of digital leadership (e.g., Benitez et al., 2022). To sharpen digital leadership’s explanatory power regarding (service) innovation phenomena, we suggest further conceptual work analysing the multiple leadership theories mentioned in this context. Furthermore, studies in this field may target a parsimonious collection of digital leadership capabilities, in the sense of differentiating between necessary and sufficient capabilities (Liehr and Hauff, 2022). Given that current employees will become future digital leaders, it is crucial for both individuals and organisations to understand the specific capabilities required to promote organisation-wide digital leadership (Banks et al., 2022). As our quantitative results are based on cross-sectional survey data, including dependent and independent variables, common method bias and endogeneity risks are inherent, possibly skewing causal inferences between our constructs (Hill et al., 2021). We aimed to mitigate these issues using several established practises (for common method bias, e.g., questionnaire design, Harman’s single-factor test, full collinearity assessment; for endogeneity, control variables, two-stage least squares regression analysis, Wu–Hausmann test). Nonetheless, a residual risk persists, especially as digital leadership and its outcomes have undergone little quantitative study so far. For further mitigation, we advocate longitudinal study designs triangulated with objective performance data (e.g., financials) and addressing the issue of the lagging financial performance effect of service innovations (Aas and Pedersen, 2011). Finally, we recommend the introduction of further outcome variables (e.g., digital mindset) serving as mediators or moderators in our research model. Research considers leaders’ and followers’ digital mindset as fundamental for enhancing competitiveness in the digital era (e.g., Erhan et al., 2022; Klus and Müller, 2021). Cultural variables may reveal insights into the missing direct influence of digital leadership capabilities on service innovation performance in the multidimensional view.

ORCID

Tobias Schuster  https://orcid.org/0009-0006-7318-7031

Timo J. J. Brunner  https://orcid.org/0000-0002-4787-1087

Malte H. G. Schneider  https://orcid.org/0000-0003-0158-4821

Claudia Lehmann  https://orcid.org/0000-0002-2019-3634

Dominik K. Kanbach  https://orcid.org/0000-0003-0956-8009