Are digital platforms a remedy for collaborative challenges? Unpacking the socio-political complexities of digital transformations

Collaboration is both omnipresent across the public sector (Ansell and Gash, 2008, 2017; Emerson et al., 2011; Torfing, 2013) and often seen as a necessary response to wicked problems (Bannink et al., 2024). Such problems are characterized by a requirement for “different, but needed” actors to “gain agreement on what public value is to be co-produced” (Bryson et al., 2015, p. 649). Establishing and maintaining effective collaboration between actors, however, is easier said than done. Where problems are wicked, a collaborative approach is at once necessary and hard to achieve (Bannink et al., 2024). Diverse actor interests, varying perspectives and increasingly high workloads make collaboration difficult to prioritize (Bryson et al., 2015; Vangen and Huxham, 2012). These challenges exist across a wide range of public policy fields, including housing, active and passive labour-market policies, infrastructure, city planning and healthcare. Of these, the field of healthcare provides a particularly relevant setting: to address wicked problems such as an ageing population, multi-morbidity and a shortage of funds and personnel, actors from a wide range of healthcare organizations are required to interact and collaborate. Accordingly, this article focuses on actor collaboration in this domain.

To mitigate collaboration problems, governments and practitioners are increasingly implementing digital communication platforms (e.g. Eines et al., 2023; Norman et al., 2023; Westby et al., 2024) based on the assumption that they facilitate more effective communication and information exchange while simultaneously reducing administrative burdens for the actors involved. Using terms from a framework by Ansell and Gash (2008, p. 550), we understand a digital platform to be part of the “institutional design” of a collaborative effort because its design determines the inclusiveness of participation. Likewise, we view such platforms as relating to the framework's “incentives for and constraints on participation” because they shape how participating stakeholders are empowered, or weakened, through communication. As a result, these platforms affect the “collaborative process”, as defined in the same framework. Emerson et al. (2011), in their “integrative framework for collaborative governance”, locate “collaboration dynamics” in a “collaborative governance regime”. To what extent “drivers” lead the “collaboration dynamics” to actually produce joint action depends on the “regime” actors move in. Belrhiti et al. (2024), in a scoping article reviewing a wide range of articles on “collaborative governance dynamics” in the health care domain, shows the importance of the interaction between the “governance regime”, in Emerson's terminology, and “collaboration dynamics”. They observe (p. 417): “Our scoping review showed that the formation of healthcare networks (HN) is often influenced by contextual factors that motivate the prospect of a new collaborative governance regime”. One of the important elements of such a regime in the health care domain concerns the interoperability of information systems (p. 413; pp. 420–421). Our research centres on this issue: How do information systems (as part of the collaborative governance regime) affect collaboration dynamics?

The implementation of such technologies to improve collaboration is often framed in the literature as a “digital transformation”. “Digitally induced change” has also become a buzzword (Haug et al., 2024; Mergel et al., 2019) to describe the processes through which digital technologies are intended to alter and improve both public service delivery and stakeholder interaction. Nonetheless, the mere deployment of a digital communication system does not magically improve user engagement or collaboration (Pollitt and Hupe, 2011). To enable meaningful communication between actors, these systems must be designed and implemented to meet not only the technical requirements of information transfer but also the collaborating actors' meaning-making processes and political alignments (Carlile, 2004). To date, however, the focus of research on the implementation of digital systems has been on technological issues, leaving the socio-political dynamics of how actors come to use these systems underexplored (Cordella and Bonina, 2012; Gasco et al., 2024). Reviewing the literature on digital change, Haug et al. (2024) conclude that moving from a “technology-focused” approach to an “actor-centred” approach will improve our understanding of digital transformations and their effects on actors' working practices. This is especially relevant when such platforms are implemented to improve collaborative processes, as actors' information needs, working practices and political interests inevitably differ in such cases.

To address this gap, our research therefore follows an interpretive, decentred (Bevir and Rhodes, 2007) approach and focuses on a case from the healthcare field that exemplifies a wider trend in the public sector: the introduction of digital technologies as a means of improving collaborative policy implementation. The research is thus guided by the following question:

• How do collaborating actors in the healthcare sector make sense of the implementation of a digital information system, and what are the implications for collaborative policy implementation?

Specifically, our case study focuses on the implementation of a digital communication platform aimed at improving communication between care professionals in a large city. Collected from a range of semi-structured interviews and meeting observations conducted in 2022, we analysed our data using MaxQDA, paying particular attention to how the design and implementation of the system corresponded with the technical, social and political dimensions at play. Foreshadowing our findings, this analysis revealed a pervasive neglect of actors' underlying social and political dynamics, leading to adverse effects and ultimately causing the system to fail. Our paper contributes to the literature in three interrelated ways by showing (1) the importance of considering socio-political dynamics when pursuing digitally induced change, particularly the alignment of digital platforms to users' working practices and interests; (2) that digital change may exacerbate pre-existing problems and evoke adverse effects, in contrast to the predominantly positive assumptions found in the literature on digital change; and (3) the benefits of adopting an actor-centred perspective, which we show helps clarify why digital systems are not one-way solutions.

In this paper, we will first explain how the concept of boundary objects enhances our understanding of digitally induced change by unravelling the involved socio-political dynamics. We will then describe the methodology of our study and share the findings of our research and, finally, explain how these findings relate to debates surrounding digital transformations in public administration.

Within the field of public administration, the availability and attractiveness of digital technologies has spurred significant interest in digital transformations (Fischer et al., 2021; Haug et al., 2024). Likewise, the continual development of new technological tools, citizen demand (Haug et al., 2024) and increased political emphasis on the potential benefits of such digital tools (Mergel et al., 2019) have all contributed to the omnipresence of digital change across public sectors. Such changes are typically geared towards improving both processes within organizations and communication and relations between actors (Haug et al., 2024) in order to boost organizational efficiency, service delivery and/or collaborative efforts (Fischer et al., 2021). This alluring potential, however, should be taken with a grain of salt. A common criticism of digital change literature is that its purported outcomes are in fact predictions, rather than being grounded in empirical findings (Coursey and Norris, 2008; Mergel et al., 2019). Accordingly, Haug et al. (2024) argue that adopting the term “digitally induced change” will help shift the discussion: from prioritizing the technological tools behind these changes to prioritizing the role of, and impact on, the organizations and actors in which these changes are embedded – better reflecting the reality that the potential effect of digital change depends on the efforts of those involved.

This broad application of and interest in digital change are also reflected in the highly interdisciplinary nature of digital transformation research. However, because this stream of literature originated in the field of computer and information sciences, digital change research initially focused on the “design [of] functional and ideally attractive solutions” (Fischer et al., 2021). When literature on digital change first entered the field of public administration, it was primarily under the heading of e-government, focused on “making public service delivery more efficient and accessible to citizens” (Meijer and Bekkers, 2015) and “changing service delivery from offline to online” rather than on re-evaluating or re-designing service delivery specifically for the public sector (Mergel et al., 2019). Currently, given the rise of New Public Governance (Osborne, 2010) and the corresponding move towards a multiplicity of public and private actors coordinating to govern society, one could argue that “e-governance” would constitute a more fitting term than e-government (Meijer and Bekkers, 2015). Indeed, the field has broadened its scope and nowadays typically uses terms associated with digitalization (Fischer et al., 2021). Current research, for example, also zooms in on collaborative processes and how digital change affects the relationships between internal and external stakeholders (Cordella and Bonina, 2012; Haug et al., 2024) – a logical development, as digital change in the context of New Public Governance inherently implies finding “new ways of working with stakeholders” (Mergel et al., 2019). Moreover, this dynamic underscores the importance of prioritizing the perspectives and practices of the actors who must work with these new digital systems, yet precisely these facets have remained underexplored in the literature (Fischer et al., 2021; Haug et al., 2024; Meijer and Bekkers, 2015).

To address these significant research gaps, recent literature reviews (Fischer et al., 2021; Haug et al., 2024; Meijer and Bekkers, 2015) have pointed to two aspects that merit closer scrutiny. First, how individual actors implement, shape and/or resist digital change must be examined and better understood. To date, research has predominantly focused on the role of technology (Haug et al., 2024) and has explained digital change in a model-like fashion (Meijer and Bekkers, 2015), as a result, largely ignoring the socio-political dynamics such as actors' diverging values, interests or positions (Meijer and Bekkers, 2015). The impact of these dynamics must be considered given that digital tools are not implemented in a vacuum but rather in a social context in which actors already have established routines and working practices. In other words, the specific context and actors involved affect the implementation of digital systems, including whether or not these systems live up to their potential. Second, researchers must also investigate the unintended or negative consequences of digital change and challenge the “normative bias” that currently shapes its discourse in policy and academia – where it is often framed as inherently valuable rather than critically analysed for its actual effects (Meijer and Bekkers, 2015). While the literature has highlighted the many potential benefits of digital transformation, it seemingly “ignore[s] the possibility that barriers … exist” (Coursey and Norris, 2008, p. 532) and thus fails to consider the potential negative or unintended consequences of digital transformation on actors' working experiences. Where these issues have been flagged, they have predominantly focused on problems caused by continuous surveillance or actors being “remotely controlled” (Trittin-Ulbrich et al., 2021; Zuboff, 2019). Recent calls for research into actors' experiences – including for a better understanding of how they navigate the challenges of digital transformation – therefore require an approach that considers the social dimensions surrounding the implementation of a digital system.

Understanding the social dynamics involved in digital platform implementation increases our understanding of how – in terms of Ansell and Gash (2008, p. 550) – the “institutional design” of these platforms, together with the “incentives” they produce, impacts the “collaborative process” and thus actors' collaborative policy implementation. To aid in this analysis, we conceptualize the technology itself as a boundary object. A boundary object (Star and Griesemer, 1989) is an object that is put in place both to combine different social worlds and to help professionals bridge the sectoral, organizational or professional boundaries between them. Boundary objects can take various forms. They can be tangible artefacts, such as diagrams and charts (Star and Griesemer, 1989), posters (Levina and Vaast, 2005), drawings or products (Carlile, 2002). However, they can also be intangible, including metaphors (Fox, 2011), management tools (Allen, 2009) and technologies (Carlile, 2002). As Star (2010, p. 603) reminds us, “its materiality derives from action, not from a sense of prefabricated stuff or ‘thing’-ness:” What demarcates boundary objects is that they are “shared” (Star, 2010, p. 602), meaning that they “inhabit several intersecting social worlds … [and] are both plastic enough to adapt to local needs and the constraints of the several parties employing them, yet robust enough to maintain a common identity across sites” (Star and Griesemer, 1989, p. 393). A digital platform is thus a classic example of a boundary object: it pertains to, and is intended for use by, actors from intersecting organizational worlds. Moreover, it exists to facilitate interaction between these actors and is designed to “satisfy the informational requirements of each of them” (Star and Griesemer, 1989).

However, boundary objects – including digital platforms – are not “magic bullets” for improving interaction (Carlile, 2002). Upon implementation, they tend to be designated boundary objects rather than objects in use (Levina and Vaast, 2005). In other words, the object is intended to facilitate interaction between different organizational worlds but is not necessarily used as such by the collaborating actors. This is sometimes because a designated object makes sense on an abstract level but is not locally practical (Levina and Vaast, 2005). A fitting example of this is underlined in a study by Allen (2009) on the implementation of a care pathway. This pathway was designed as a management tool to help chronologically map out the activities and documents required by various mental health stakeholders, with the ultimate goal of improving the quality of care. As such, the intent of this boundary object was to “[inhabit] several social worlds and [fulfil] a role in structuring relations between them” (Allen, 2009, p. 355). Throughout its implementation, however, the tool proved “unworkable” for the actors involved, prompting the continual use of workarounds. As a result, the process brought “to the surface many of the tensions [the tool was] designed to resolve and like other formal tools, the order embedded in the end product contain[ed] much of the messiness it set out to erase” (Allen, 2009, p. 360). In another example, Levina and Vaast (2005) use their study of headquarters managers who created and mandated the use of posters for their sales teams to highlight the problems caused when powerful actors impose a boundary object without allowing for local negotiations over its meaning, terms of use or cross-field relationships. Because the sales teams ultimately deemed the posters “not suitable to the realities of their trade” (Levina and Vaast, 2005, p. 343), these boundary objects were ignored in practice. In sum, when boundary objects fail to address the involved actors' needs, do not align with their interests or simply prove unusable, they will not progress from designated to being used.

It therefore stands to reason that the implementation of boundary objects must explicitly tie in with actors' common interests and local usefulness. This is because, as alluded to above, boundary objects – especially those designated by powerful actors – have shown themselves at risk of being idealized at the expense of local practice-focused negotiations with their intended users. Such local practices are always influenced by “the context of prior actions and relationships” (Levina and Vaast, 2005, p. 337), making it essential to consider the involved actors' perspectives and positions vis-à-vis the boundary object – both before and throughout implementation – when evaluating its success. In order for actors to manage knowledge in the context of innovation and thus make effective use of a boundary object, Carlile (2002, 2004) argues that three increasingly complex dimensions must be addressed. First, actors must engage in “information transfer” through a syntactic boundary. This dimension involves the creation of a common lexicon (Carlile, 2004) or “shared and stable syntax” to “ensure accurate communication between sender and receiver” (Carlile, 2002, p. 443). In the context of digitally induced change, this could pertain to the technical design of a new platform, including its available options and choice of names. Second, actors must engage in “information translation” through a semantic boundary – a demand driven by the complexity of navigating multiple actor interpretations. This dimension (Carlile, 2004) involves “learning about [the] differences” (Carlile, 2002, p. 444) that underlie actors' different working practices, perspectives and contexts in order to create common meanings. When adopting a new digital platform, this often reflects the need of actors to align these practices or negotiate ambiguities. Third, actors must engage in “information transformation” through a pragmatic boundary. This dimension involves “altering knowledge” and “creating common interests”, which – to mitigate the emerging political conflicts and costs of the boundary object – must be validated by each local actor (Carlile, 2004). It is important to note that, because actors may not hold equal power positions (Carlile, 2004), the ability to create “truly common” interests should not be assumed. In the context of digital platform negotiations, such power imbalances can result in some actors benefitting more from the tool than others. In sum, Carlile argues that boundary objects are only successful when they effectively – and recurringly – address all three dimensions of actor divergence. In this paper, we build on this seminal work to deepen our understanding of how collaborating actors make sense of digital platform implementations, all the while shedding light on their underlying social and political dynamics.

Grounded in the qualitative-interpretive research tradition, our study adopted a decentred approach. With a focus on the social construction of practices (Bevir and Rhodes, 2007), this approach seeks to illuminate and understand the everyday practices actors engage in as they make sense of and (re)appropriate policies in their day-to-day work. It also highlights actors' different perspectives and practices (Ybema et al., 2009) and understands these practices as “historically and culturally relative” (Burr, 2003, p. 14). Indeed, rather than treating actors involved in collaborative processes as a coherent whole, a decentred approach emphasizes the range of meaning-making processes and actions between actors. Its emphasis on questioning taken-for-granted assumptions combined with its sensitivity for how actors navigate dilemmas in their everyday work (Bevir and Rhodes, 2007) meant that, in our study, we were particularly attentive to how actors' actions and ideas regarding the platform's implementation were shaped by their own backgrounds and dilemmas. Moreover, we explicitly recognized that their actions and ideas may in fact collide as opposed to only align, further complicating implementation of the platform. Overall, this approach is particularly well suited to opening up the black box of how actors make sense of, and come to use, digital platforms as they collaborate, hence shedding light both on their social and political considerations and on the unintended consequences that may arise throughout the implementation process.

This paper draws on a decentred study of a digital platform implementation designed to improve collaboration between publicly financed healthcare professionals who provide care for frail older persons at home. The case was situated in the Netherlands, where the healthcare system has functioned as a regulated market since 2006. This implies a mix of public and private elements: healthcare is regulated by the central government and funded through four different laws, but healthcare insurers and healthcare providers are – within strict boundaries – allowed to compete, which influences the prices of healthcare.

Because Dutch healthcare policies also tend to present collaborative governance as a panacea, this context provided a unique opportunity for studying how actors make sense of digital tools implemented to improve collaboration. Over the last decade, multiple healthcare reforms have been implemented specifically to improve collaboration between actors across the medical and social domains, and digital tools have been increasingly suggested as a means of facilitating this process. Nonetheless, the road has been difficult and bumpy.

As already stated, our case study focused on a digital platform implemented to address the collaborative difficulties and fragmented coordination of home care for older people in a large city. Intending to improve the efficacy of care, a project team chose to implement the platform citywide to facilitate communication between general practitioners (GPs), practice nurses (who support GPs), district nurses (who work in home care delivery) and other healthcare providers involved in elderly care. The project was jointly spearheaded by an independent project leader (appointed by the local hospital) and a programme manager from a home care organization. It was funded by the dominant healthcare insurer in the region together with the local hospital. The project team also included administrators from a local GP association and a programme manager from a regional organization that facilitated collaboration among medical professionals. The aim of the project team as a whole was to establish a digital communication platform that would facilitate information exchange among participating care professionals. Collaborating actors were expected to enter all relevant data into the platform and to refrain from using other means of communication (except in emergencies) so that all information would be accessible for all actors. As it turned out, implementation of the system was easier said than done. Some actors resisted its use, and the project team struggled with pinpointing how to improve its efficacy.

Both the specifics of the case and the dynamics that arose throughout its implementation make this case study particularly relevant to our research question. For one, the digital platform used serves as a prime example of a boundary object because it was put in place to improve interaction between different social worlds. In addition, actors' expected use of the digital platform was loosely defined and did not follow, for example, a predetermined care pathway. This created space for more friction and differences in interpretation between the actors involved. These difficulties subsequently underscore the importance of unravelling how actors' social and political considerations shape their attempts to make sense of the implementation of a digital platform designed to improve collaboration.

When designing our study, we paid particular attention to ethical considerations such as protecting participant confidentiality and establishing informed consent, the latter of which we gained through letters and consent forms. We started our study only after gaining approval from our university's ethical board. Respondents were identified through snowball sampling: we asked members of the project team to identify specific practitioners for participation, whom we then also asked to suggest possible participants. A benefit of this approach was that it allowed us to speak to the participants who were actually expected to work together through the platform; a drawback was that it naturally excluded outside actors who could have brought in different experiences. Moreover, because our main aim was to gain insight into the perspectives of each professional group tasked with working with the platform, we chose to interview a variety of actors to obtain an overview of how these different groups perceived and implemented the system.

In the end, we conducted 18 interviews with 14 different actors, including some who were expected to work with the platform – i.e. practice nurses (3), district nurses (3), a GP (1) and a case manager (1) – and the project team members who had initiated and steered the implementation – i.e. the project leader (1), the programme manager of a home care organization (1), a programme manager in digital healthcare (2) and the project leader of a GP association (2). Ten of these interviews were semi-structured, audio-recorded and transcribed, while the other eight were in-depth (repeat) interviews with key informants from the project team, during which we took detailed notes. The semi-structured interviews used a topic list both as guidance and to allow for an in-depth understanding of the topics actors brought up themselves. Developed to explore our key research question, the topic list focused on “participant role and background”, “collaboration processes”, “DP introduction”, “DP implementation and use in practice”, “DP and collaboration”, “DP prospects” and “wrap up and additional points”. The in-depth interviews were geared towards gaining more detailed insights into the situations and problems that had emerged, the backstage frictions that had surfaced, background information on the project and sensitive issues and prospects.

Data collection spanned from 2021 to 2024, the bulk of which took place in 2022. To gain insight into both the interpersonal dynamics between actors and the information supplied to the actors involved, we also observed four multi-actor meetings in 2022: three project team meetings and one “kick-off meeting” with a variety of professionals who were expected to start using the digital platform themselves. All four meetings lasted about an hour. We also collected 18 documents, all chosen for their explanations of the changes introduced by the new digital platform; nine of these were email exchanges and the rest included internal reports, news articles, excerpts from YouTube videos about the project and printouts of selected pages from participating organizations' websites. In addition to these interviews (18), observations (4) and document analyses (18), data were also gathered from project team meetings held in 2021 – which were attended in preparation for the research – and from two additional interviews with key informants conducted in 2024 to learn about possible developments after data collection had finished.

We coded our data in various rounds using MaxQDA. Our analysis followed an abductive approach (Sætre and Van De Ven, 2021; Van Hulst and Visser, 2024), moving back and forth between our findings and the literature. Using Gioia et al.’s method (2013) as guidance, we began by familiarizing ourselves with the data and assigning initial codes to fragments relevant to unravelling actors' perspectives and experiences around the implementation of the digital platform. The aim of this phase was to generate a preliminary code list. To move from first-order codes towards a more conceptual understanding, we used boundary objects, digitalization and collaboration as sensitizing concepts. In particular, we were inspired by Carlile's (2004) framework, which distinguishes between the syntactic (information processing), semantic (interpretative) and pragmatic (political) dimensions – all of which, he argues, must be addressed to enable the effective exchange and creation of knowledge between diverse actors.

We aimed at “theoretical saturation” (Gioia et al., 2013, p. 20; referring to Glaser and Straus, 1967): “Once a workable set of themes and concepts is in hand (and the culmination of the theme and concept development process leads to what Glaser and Straus (1967) termed “theoretical saturation”), we investigate whether it is possible to distill the emergent 2nd-order themes even further into 2nd-order “aggregate dimensions”. “ In the case of our analysis, this process unfolded as follows. Following an abductive approach, first, we noticed that our fragments did not neatly fit the syntactic, semantic and pragmatic dimensions identified by Carlile. Next, we found that more precise clarification of what these dimensions entailed – and which elements of our data were relevant to which dimension – was necessary given our case study's context of digital transformation designed to stimulate collaboration. For example, rather than strictly adhering to Carlile's syntactic dimension, we chose to create a dimension specifically related to the technical design of the boundary object. In the end, our analysis comprised three “aggregate dimensions”: (1) technical design, which we define as the functional aspects of the platform (e.g. its options, functions and which information can be included); (2) social practices, which refer to both how the platform was understood by actors in the context of their daily (collaborative) practices and how it aligned with actors' existing (work) routines, and (3) political considerations, which describe how actors positioned themselves vis-à-vis the platform and its goals. We then grouped the coded fragments under these three “aggregate dimensions” and abductively analysed them in a more fine-grained manner to narrow them down even further. This resulted in the identification of four subcategories (“second-order themes”) for each dimension, all of which were relevant to the involved actors' perspectives and experiences vis-à-vis the digital platform's aim to stimulate collaboration. Table 1 in the Findings section provides an overview of these aggregate dimensions, their subcategories and their effects on the platform's implementation. Doing so, we eventually developed an analytical framework that allows for a better understanding of actor dynamics surrounding the implementation of a digital platform meant to improve collaboration.

Together, this analysis helped us uncover how the platform produced unintended consequences within each dimension. Our observations also revealed an interplay between the dimensions: unaddressed political problems hampered the success of the digital platform and vice versa, and persistent technical issues exacerbated social problems between actors.

Building on the analytical framework introduced above, our case study shows that the effective implementation of a boundary object – such as a digital platform – depends on adequately addressing its technical design and actors' social practices and political considerations. As our case study also reveals, however, this is easier said than done. Accompanied by illustrative quotes from our data, Table 1 provides an overview of our framework's three aggregate dimensions and their related subcategories (second-order themes), along with the observed effects on implementation. In what follows, we will first discuss the key unintended consequences that flow from inadequate consideration of each dimension and then discuss the interplay between them.

Technical design: the unintended consequences of information overload. Despite the technical design of a system typically receiving the most attention in digitalization processes, its features can still cause issues upon implementation – even when these have been carefully selected and developed, as was true in this study. Our analysis shows that the technical design in our case led to information overload, making it difficult for users to navigate the system. No clear choices were made regarding which information the platform would be used to communicate, and actors were unaware of which information their colleagues needed to improve collaboration. This led to “data hoarding”, i.e. the collection of as much potentially relevant information in the system as possible, without it having a clear function. Similarly, the technical design of the digital platform was geared towards offering a wide range of functions, which users did not actually need. The sheer volume of options in the system made it difficult for actors to navigate. Moreover, the interoperability of the platform was low: the digital platform only linked to the GP's home systems. As a result, all other users were required to manually input their information into both their home system and the digital platform, eventually and increasingly inspiring the use of alternative systems to communicate (e.g. phone calls and email). This eventually provoked widespread neglect of the platform itself. The technical design thus overlooked the specific needs of the collaborating actors involved, further contributing to actors' experience of “information overload”. Altogether, these elements not only made it difficult for actors to process and use the information on the digital platform but also caused them to feel overwhelmed and ultimately bypass the system entirely.

Social practices: the unintended consequences of added administrative burdens. Digital platforms are implemented in a context of pre-existing working and collaborative practices, both of which are underpinned by larger social practices. Developers and managers must therefore first understand these social practices and identify a clear position for the digital platform before embarking on its design and implementation. Our analysis reveals that this step was overlooked, leading to extra administrative and communicative burdens. Among the various actors involved, we observed a discrepancy in the interpreted function of the system: for project team members, its goal was to facilitate more effective communication and thus achieve tangible care outcomes (e.g. reducing hospitalizations), while users of the platform expected it to ease their work and time pressure. In practice, the digital platform in fact created extra work. As the platform was only designed to accommodate the working processes of a GP, the other actors experienced its use as a burden. System training, which some actors received upon introduction to the platform, was also geared only towards GPs, leaving other actors to figure it out on their own with little to no training or support. Similarly, actors' collaborative practices – their means of bridging their different worlds – were also overlooked, and the potential uses of the digital platform were not translated into actionable and understandable collaborative working agreements. As a result, the involved actors lacked clarity regarding which information was supposed to be communicated, who was in charge or responsible, what their working relationships entailed and when action was expected. This neglect of the social practices onto which the digital platform was imposed led to a collaborative disconnect and additional administrative and communicative burdens, all of which actors ultimately sought to reduce.

Political considerations: the unintended consequences of increased fragmentation. Our study challenges the assumption that actors will use a digital platform simply because it is made available, showing instead that engagement depends on interest alignment. This means that the political considerations of potential platform users must be addressed prior to the design and implementation stages and that overlooking these considerations risks undermining user support and exacerbating pre-existing fragmentations. In our case, a majority of actors' individual interests clashed with the use of the system due to an uneven distribution of financial or administrative costs, leaving them hesitant or resistant to its adoption. The initial focus of the project team members on one group of users (GPs), coupled with a lack of consultation with the other users on the design of the system or potential improvements, created a feeling of exclusion and deteriorated political traction among the participants more broadly. Ultimately, each participating actor weighed the pros and cons of the system from their own perspective: its advantages (e.g. saving time and more efficient communication) were deemed mostly theoretical, while its disadvantages (e.g. double reporting, lack of interoperability and frustration due to unclear agreements and lack of comprehensive participation) created significant practical burdens. In the end, this misalignment led to abandonment of the platform and even greater fragmentation in communication between actors.

Although our framework distinguishes three dimensions that shape the efficacy of a boundary object, these separations are purely analytical. Observations throughout our case study revealed that the technical design, social practices and political considerations underpinning a digital platform often interconnect, overlap, and reinforce each other in practice: technical design shortcomings can intensify existing social practices, and existing political differences can hinder platform implementation – and vice versa.

Despite the intention of the project team to improve collaboration, our analysis shows that their lack of consideration for actors' underlying social and political dynamics impeded the efficacy and adoption of their initiative. Indeed, imposing a digital platform on a difficult collaboration does not magically resolve its difficulties, especially when the root and dynamics of the problem are not understood or addressed. As seen in our study, when developers and managers focus on the technical design but overlook the needs of its intended users, a system is likely to clash with these users' social practices and interests, eventually causing their political considerations to hamper its support and success.

Moreover, our analysis reveals that the implementation of a digital platform can also exacerbate social and political problems, inadvertently creating more unrest, chaos and inefficiency between actors. For example, as was the case in our study of an already difficult collaboration, platform implementation seemed to cause a reduction in face-to-face meetings, thus creating even more friction and miscommunication between the participating actor groups. It should also be noted that while prioritizing one group of users in the design of a platform may on its face seem beneficial, at least to that select group, the opposite in fact proved true: by overlooking the needs of the other actors, more fragmentation was created, and the system was eventually abandoned, negating its usefulness for the prioritized group (GPs) as well.

Our study has both theoretical and practical implications for the implementation of digitally induced change in collaborative processes.

To start, the literature on digital transformations has thus far focused mainly on technological issues, leaving the socio-political dynamics of digital change underexplored (Cordella and Bonina, 2012; Fischer et al., 2021; Gasco et al., 2024). Framing digital technologies as boundary objects (Star and Griesemer, 1989) has helped us closely analyse not only the technical but also the socio-political dynamics involved in the implementation of digital platforms. Underlined by the seminal work of Carlile (2002, 2004), our study shows the importance of going beyond – and explicitly marks the differences between – (1) the technical dimension (design) to include the dimensions of (2) meaning-making (social practices) and (3) user interests (political considerations). In addition, our study identifies the particular elements that reveal how these three dimensions play out in the context of a digital transformation, specifically when implementing a digital platform. This shift in focus reframes the discussion around digital technologies from what knowledge is needed or relevant to whether and why actors would engage in such a system in the first place. As already illustrated, our study shows that the mere implementation of a digital platform is not a magic bullet. In fact, by overlooking the social and political dimensions, the project team inadvertently caused “the imposition of one world's vision on the rest” (Star and Griesemer, 1989, p. 414) – in our case, by prioritizing the perspectives and priorities of GPs over those of the other actors. Consequently, these other actors were less inclined to use the system, and the platform was effectively rendered useless for all parties. When this happens, a tool remains a designated boundary object only.

Next, due to a “normative bias” in favour of digitization (Meijer and Bekkers, 2015), the negative consequences of digitally induced change have long been neglected (Haug et al., 2024; Trittin-Ulbrich et al., 2021). Our study highlights three unintended and counterproductive consequences of digital platform implementation, all of which may be particularly pronounced when little attention is paid to actors' underlying socio-political dynamics. First, by failing to fully consider the information requirements of each intended user, a platform may not only produce a lack of relevant information but may also create overwhelm and frustration due to information overload. Second, by overlooking the pre-existing working and collaborative practices of its intended users, a platform can create more administrative and communicative burdens rather than less. Third, by neglecting to evaluate whether the involved actors actually adapt their practices to use the system – including, most notably, their underlying political considerations for (not) doing so – a platform may (further) fragment actors' communication efforts. When all three of these counterproductive consequences arise, as was true in our case, the implementation of a digital platform will lead not to greater efficacy and improved outcomes, but rather to more ambiguity, administrative burdens and fragmentation.

Finally, research on digitization in public management has typically taken a “holistic” perspective (Meijer and Bekkers, 2015), studying such tools as wholes defined mostly or only by their technological scope and design – which, contrary to colloquial implications of the word “holistic”, actually neglects the impact on, and practices of, actors (Haug et al., 2024). Practitioners and policymakers instead must acknowledge that the design of such a boundary object or structure affects actors' access to decision-making and perspectives on collaboration. By adopting a decentred approach (Bevir and Rhodes, 2007), we intentionally employed an actor-centred perspective. This has allowed us to shed light on how actors implement, shape and resist digital change. Relatedly, we call for scholars to carefully consider which actors they choose to study. While studying “managerial” experts such as policy advisers, information technology advisers, and consultants (Mergel et al., 2019) does help illuminate digitization trends, triggers, policy evolutions and expected outcomes, digitization users may tell a different story. Our study clearly shows a discrepancy in this regard: the managerial actors in the project team considered the mere existence of the platform a success, pointing to its many potentials, while the users pointed to its problems, including its lack of practical use and feeling unheard by those in charge of the system. This suggests that from a top-down, technical perspective, digital information systems may be seen as a solution to coordination problems, but that from a bottom-up, decentred perspective, these systems can fail to deliver on their promises and may exacerbate pre-existing problems instead. We therefore argue for the importance of analysing the socio-political dynamics at play and adapting the system to the working practices of all involved actors – both before and throughout the design and implementation stages – rather than simply assuming that the availability of a digital system will solve the problems faced by public administrations.

A practical implication, applicable to collaborative governance processes more generally, also starts from this viewpoint. Such conditions of conflict and incongruence in collaborative governance processes are likely in the face of wicked problems, characterized as they are by differences in perspective among involved actors. In such cases, network management needs to work from a decentred perspective. As we have shown, efforts aimed at improving the efficacy of task coordination must be strongly embedded in actors' actual working practices. Prior to the design and implementation stages, discussions should be had with the intended users to flesh out the three key dimensions underpinning a digital platform: What information do parties really need from each other (technical design)? What are the involved actors' current working agreements around the input and use of data (social practices)? And how would the system benefit users (political considerations)? It is especially important to look beyond the potential technical capabilities of the system to first examine how tasks are already coordinated. Only after answering these questions can a digital platform truly be designed and implemented to facilitate more effective communication and reduced administrative burdens.

This article has explored the complexities of digital system implementation from the perspective of its intended users – a lens that both highlights the importance of analysing socio-political dynamics and uncovers several unintended consequences of digital transformations. As an interpretive study, we have not aimed for generalizability or replicability in the technical sense. Rather, interpretive research stresses context sensitivity and multivocality, seeking insight into actors' meaning-making processes and the situated nature of social practices (Schwartz-Shea and Yanow, 2013). Nonetheless, while we acknowledge that our insights are drawn from a single case study in a specific context, we aspired to formulate a generalizable analytical framework that opens up the black box of how actors make sense of and reappropriate policy initiatives in practice. Future research could explore how this framework translates to other contexts, for example, by studying a digital system that is implemented within only one organization or institution or across other public (or private) sectors. Likewise, our case was distinctive in that the information actors were expected to exchange through the platform was only loosely defined. Other studies are needed to examine how, when a platform has been built around a more specific workflow – for example, in healthcare, a strictly defined care pathway – actors' social and political dynamics may be affected differently. Moreover, scholars may want to contrast our study with a more “successful” case, zooming in on the three key dimensions to examine how these were leveraged to achieve actual improvements. Finally, and echoing the call of Haug et al. (2024), longitudinal, process research could be pursued to shed light both on the development of these improvements and on the practices and perspectives of the involved actors over a longer period of time.

Although collaborative ways of working are now commonly pursued across the public sector as a panacea for wicked problems, it is also known to be complicated to put into practice. As the availability and capabilities of digital technologies increase, so does the allure of their implementation to address the complexities of collaborative working. Nonetheless, the mere introduction of a digital tool does not magically solve the problems of collaborating actors, nor does it even ensure its use. To explore both how such technologies are implemented to improve collaborative policy implementation and which complexities emerge throughout the process, this study has examined how collaborating first-line care providers made sense of a digital platform implementation aimed at improving communication and alleviating administrative burdens. Our analysis shows that digital technologies, including digital platforms, are unable to solve collaborative problems in and of themselves. The underlying social and political dynamics of the involved actors must be identified and addressed, and tools must be adapted to their working practices and interests in order for digital change to facilitate collaboration, as well as collaborative governance in itself and live up to its desired potential.

A generalizable implication follows from this last point. While digital platforms can be considered “boundary objects” and the collaborative process starts with building a structure for collaborative interaction, collaborative governance – the benefits of which are implied through the widespread introduction of digital communication platforms – must not be understood as a neutral process or approach, as this risks underestimating the potential for, and significance of, conflict and incongruence between actors' perspectives vis-à-vis such boundary objects. These dynamics have practical and ironic implications for those responsible for managing collaborative processes: a collaborative approach works best when it is least needed. It works when the involved actors already have a shared understanding of the problem at hand. Relating to the dimensions identified in this study, this means that when the political dynamics are resolved, the technical design may easily follow – but not the other way around. Instead, when actors' political dynamics are characterized by conflict or incongruence, a collaborative governance process alone is unlikely to provide a sufficient solution. We consider this insight an important counterpoint to the collaborative governance approach. This counterpoint also has important practical implications for public management. Instead of starting processes of public management from a supposedly shared, albeit wicked, problem and working towards a shared understanding of this problem, public management processes need to start from the different perspectives in the actor constellation. Not the problem is wicked; the actor constellation is.