This study is part of a three-year international Erasmus + project developing educational technology for healthcare practice placements in Spain, Germany, and Poland. It focused on using mobile technology to connect actors and enhance practice-based learning [35].

The project adopted a participatory design approach and co-designed the envisioned solution. A total of 16 people (11 female and 5 male) with different profiles participated in the design team: 3 designers, 4 healthcare professionals, 5 teachers and 4 clinical tutors. They were from 6 different institutions in four European countries: Austria (1), Germany (3), the Netherlands (4), Poland (3), and Spain (5).

The co-design process (June 2022 to May 2023) involved 7 co-design methods across 7 different steps in 10 workshops (Fig. 1). Steps 1, 3–7 were conducted online via MS Teams, using MIRO (https://www.miro.com) as whiteboard to implement the co-design methods. Only Step 2 was face-to-face, using the Value Proposition Canvas, flipcharts and post-its.

The different co-design methods were introduced to the participants before each step:

Note that steps 4–6 (six online workshops, separately conducted in the sub-groups for each trajectory).

Seven interviews were conducted after ethical approval (n. 230672) to explore participants' views on how the co-design process and methods supported the four boundary-crossing learning mechanisms. Participants received information about the study (informed consent, see Supplementary Material 1) before interviews via Zoom. Recordings were transcribed with Sonix.ai using identification codes to protect confidentiality. Interviews lasted 37–101 minutes (M = 62, SD = 0.01), comprising two parts: 1) semi-structured interview [36] reflecting on the co-design process (Supplementary Material 2), and 2) artefact-mediated interview reflecting on co-design methods based on Guliker’s rubik (Supplementary Material 3 and Supplementary Material 4. The artefact in the second part depicted the co-design process and outcomes to support recall (Fig. 2).

Seven participants (6 female, 1 male), aged 3557, came from Spain (3), Germany (2), and Poland (2). Most were educators in nursing or medicine (6), one in pedagogy, with roles in university teaching (3), research (2), and practice placement management (2). Participants were selected via convenience sampling [37] among the co-design team and had no previous experience in co-design. Representation of all profiles and countries was ensured.

A deductive qualitative content analysis was conducted to ensure systematic and theory-based analysis [38]. Literature on the four boundary-crossing learning mechanisms [19, 39, 40, 41, 42, 23, 43, 44, 45, 46, 47] guided the researchers to develop a new codebook with clear inclusion and exclusion criteria to enhance the transparency and rigour of the coding process (Table 1). Two researchers double-coded 10% of the data, achieving high inter-coder reliability (Cohen’s Kappa = 0.80) [48]. After coding all seven interviews, a final reliability check showed strong agreement (Kappa = 0.86). Three researchers reviewed and agreed on the final codes for analysis.

In general, interviewees identified a relationship between the co-design process and the learning mechanisms for boundary-crossing. Most codes related to the co-design process refer to the mechanism of Reflection (46.87%), followed by Coordination (28.13%), Identification (15.62%), and Transformation (9.38%) (Table 3).

Regarding Identification, participants believed the co-design process helped them question the identities and practices of all partners and their own (P02, P05, P06).

P05 confirmed that it helps “to organise all my perspectives and all my insights [on healthcare] in a systematic way. And I suppose I help the [software developers] to, to gain these insights [about their perspectives] and to use them to design the innovation”. They also explained that it allowed them to delineate how participants approached common problems differently (P02, P03, P04, P05, P06). P02 states “I could understand or I could see that we have all the same problems, but we are dealing in a different way” and “it was helpful to see how different countries could manage and use the different solutions”.

Secondly, participants also believed that the co-design process promoted Coordination by allowing them to plan and make decisions together (P05). Furthermore, it reduced boundaries by systemising the way of working between the different project partners (P01, P04, P05). This is reflected in streamlining the cooperation as stated by P01: “the co-design process helps us to focus on the problem, and that was really important in the beginning [...] we didn't know how to advance in how to introduce mobile technology in practice placement”. This coordinative work seems to be more sequential, confirming its cooperative instead of a collaborative nature. P04 mentions it like that: “So we finish our work and we could give the feedback on each stage. So it was also good because we could see overall work and think about the feedback of this work that we agree or something could be better”.

Third, all participants related Reflection to the co-design process. They believe that it helped them to verbalise their point of view and to understand the others’ in different occasions, i.e. perspective making and taking (P02, P04). P04 explains that during the co-design process, they could “share my perspective as a young teacher and also young nurse” and that they “meet a lot of perspectives and also a lot of ideas of different people”. Interestingly, most of the participants (P01, P03, P04, P05) highlighted how the co-design process allows not just making explicit different perspectives but also integrating them to build a new point of view. P05 illustrates it clearly: “so there are other perspectives of other partners and of other ways of doing that can be better or it can be different”. Additionally, P04 explains how integrating these perspectives “was some kind of promoting a new knowledge and practices”. P01 summarises the progress they experienced during the co-design process like this: “I was thinking before [the] start the co-design process that the most important issues, to try to introduce mobile technology in practice placement [for,] was the information and communication [...] but using co-design process, I have changed [...] things”.

Finally, although 3 passages related the co-design process with Transformation (P04, P05), the interviewees addressed generating new knowledge and transformation in general terms. They did not elaborate on these transformations, leaving it unclear whether they represented profound changes in the healthcare learning practices between universities and placements or merely incremental adaptations of the current practice. Thus, the interviewees did not provide enough information to classify these codes as a transformative learning process.

The qualitative analysis of the codes related to co-design methods (n = 50) reveals (see Table 4) that almost half of the mentions were assigned to VPC (46%), Personas and Scenarios (46%), followed by User Journey (30%), and the Innovation Canvas (16%). Altogether, this indicates that the different design methods can act as DBO during the co-design of educational technology.

The following sections present detailed results on how co-design methods support Identification, Reflection, and Coordination. Transformation was not related to any of the design methods by the interviewees.

Based on our exploration of learning at the boundary, we found in line with several authors [49, 18, 30, 31] that design methods can act as design boundary objects (DBO). Participants perceived most of the methods for collaborative design (co-design) as promoting the exchange of perspectives across the boundary via Identification, Coordination, and Reflection: Personas, Scenarios, and Value Proposition Canvas (VPC) emerged as the most prominent ones, followed by User Journey, and finally Mock-ups and Innovation Canvas (IC). This translation of information contributed to understanding differences between the partners, making the co-design methods boundary objects according to Brubaker et al. [29].

Our findings show Personas and Scenarios play a central role in boundary-crossing, aligning with Massanari [50] and Ozkan et al. [51]: the artefacts of both methods served as DBO, helping to recognise participant profiles and roles to overcome discontinuities (e.g., promoting exchange of diverse practices), and leading to high engagement in co-design workshops. Islind et al. [52] stresses that this high level of engagement is decisive for their role as boundary objects. Similarly, Blomkvist et al. [21] illustrate how lacking user engagement can prevent this co-design method from serving as a DBO. The Value Proposition Canvas was also perceived as essential in supporting boundary-crossing, enabling the design team (problem owners and solution providers) to bridge users' needs with intended educational technology solutions. Participants’ active involvement in the VPC established mechanisms for cooperative work across three trajectories: 1) Onboarding, communication, and documentation; 2) Reflective practice and feedback; and 3) Assessment and learning goals. These trajectories significantly impacted the overall project by guiding coordination for subsequent design steps. To our knowledge, this central co-design method has not previously been linked to boundary crossing.

After the equally supportive Personas, Scenarios and VPC methods, User Journeys played a less central role in boundary-crossing. The participants exchanged different views and perspectives on how different actors interact with technology. This is in line with del Olmo et al. [53], who stated that User Journeys can function as boundary objects to share knowledge from different perspectives and to generate new ideas.

Finally, the Innovation Canvas and Mock-ups were the co-design methods which participants attributed the least central role for boundary crossing. We see only in some cases how the different IC instances may contribute to Reflection in line with Dennerlein et al. [7], and how Mock-ups may function as a form of communication between participants allowing them to share their point of view in line with Floyd [54]. The relative novelty and limited engagement of the participants with both methods may have reduced their chance to act as boundary objects.

We analysed the relationships between co-design methods as design boundary objects and the learning mechanisms for boundary crossing [19]. Although each DBO supports almost all learning processes, the facilitation of learning processes is not equally balanced: i.e., Identification is linked strongest to Personas and Scenarios, Coordination to VPC, and Reflection is equally linked to Personas and Scenarios, VPC and User Journeys.

In co-design projects, participants’ different practices from various social worlds may cause conflicts [19] and hinder the co-design process. Identification mitigates this by delineating practices to coexist and avoid conflicts. Our study demonstrated that all applied design methods supported Identification, enabling participants from different social worlds to understand differences and intersections, fostering collaboration in the design process. However, Personas and Scenarios were the DBO that supported the Identification learning process the most.

The learning process of Coordination was supported by all design methods, facilitating smoother cooperation. The VPC stood out, however. It narrowed down design goals and enabled cooperative implementation. This led to the co-creation of three trajectories: 1) Onboarding, communication, and documentation; 2) Reflective practice and feedback; 3) Assessment and learning goals), enabling effective cooperation in parallel teams using the same design methods (IC, Personas and Scenarios, User Journey, and Mock-ups). We observed that these trajectories themselves also acted as boundary objects for coordination, namely an efficient method of communication that promotes the cooperative implementation of co-design tasks in the sub-groups [19]. Dividing work into three trajectories allowed smaller groups to focus deeply, easing boundary-crossing and increasing motivation as well as active involvement. This aligns with Islind et al. [52] highlighting concrete goals as boundary objects, and with the importance of active engagement with design methods for boundary crossing [21, 52].

We identified Reflection as the most prominent learning process. Interviewees stated that they could understand the point of view of others and integrate them into their own perspectives, i.e., learn something new. This became particularly noticeable when stereotypes of real users were formed through the Personas method. The democratic nature of co-design methods, which equally values all stakeholders' perspectives and experiences in the design process, may explain this phenomenon [15].

Transformation was not supported by any of the co-design methods. According to Akkerman and Bakker [19] that draw upon expansive learning [55], a contradiction is the premise of the Transformation learning process. It is the confrontation with a longer lasting and worsening problem that “forces the parties to seriously reconsider their practices and interrelations” (ibid), often resulting in substantially new knowledge, roles and practices. In our case, we noted friction in the design team when framing the problem with the Value Proposition Canvas: the project partners responsible for software development raised issues about integrating their tools, which was resolved by separating the trajectories by technologies and postponing discussions about their integration. The collected interview data does not elaborate on these integration issues and, hence, does not allow us to discern this friction being or evolving into a severe cultural-historical contradiction [55, 56]. For this reason, we did not code the learning process of Transformation in these instances, but only Reflection and Coordination.

During the coding process, we identified a lack of clarity in the conceptualisation of the learning mechanisms [19], which resulted in difficulties when coding them. This is probably due to the nature of each learning process and the complex interrelations and dependencies that exist among them. Consequently, we synthesised literature on the different learning mechanisms and formulated inclusion and exclusion criteria to effectively distinguish between them considering three major aspects:

First, the mechanisms of Identification and Reflection mainly reflect meaning-oriented learning processes [19] and differ in the level of depth of their analysis. While Identification supports the identification and comprehension of diverse perspectives for the recognition of boundaries, Reflection critically analyses these perspectives to enhance understanding or to reinterpret them for crossing boundaries. Thus, passages referring to the core identity of the project partner should be coded as Identification, whereas those addressing diverse perspectives on the topic should be coded as Reflection.

Secondly, some learning mechanisms may be associated with cooperation, whereas others are more closely related to collaboration. The learning process of Coordination facilitates collective efforts towards a shared objective by enabling minimal interaction between the participants and splitting up work for their individual implementation, i.e. to cooperate; it does not necessitate integration of diverse perspectives [19]. In contrast, Reflection and Transformation involve profound reflections that promote meaning making to create something new or to resolve problems synchronously, i.e. collaborative [19, 41]. Thus, to be coded as Coordination, passages should refer to the establishment of mechanisms for cooperation, while those of Reflection and Transformation should be related to collaboration.

Finally, Akkerman and Bakker’s description of Transformation [19] does only recognise the generation of profoundly new practices in terms of expansive learning [55, 56]. Focusing on contradictions and game-changing transformations prevents identifying and analysing smaller incremental changes of practice and respective transformations that might happen more implicitly over a longer term [57, 58]. Thus, to be rigorous in coding, only passages that refer to profound transformations in practices were to be coded as Transformation.

Although Akkerman and Bakker [19] did not relate the four learning mechanisms to each other, we see a certain interdependence. First, boundaries may need to be recognised to overcome them, prioritising identification in the beginning of co-design. Second, boundaries could be overcome by either cooperating and crossing boundaries with minimal interaction, i.e., coordination, or making and taking perspectives in collaboration, i.e., reflection. Transformation may then take collaboration to the next level by increasing the interaction, turning contradictions into development potentials, and forming new shared practices.

The selection of co-design methods is crucial for successfully applying a co-design process that instantiates boundary crossing, especially in initiatives with diverse participants from different countries. Briefly, we suggest it is key to understand how co-design methods can contribute to the different learning processes at the boundary. Subsequently, we provide the following preliminary recommendations for practitioners and researchers:

Recommendation 1: We understand that Identification can initiate boundary-crossing in that participants can express their own identities and understand the identities of the other project partners. In particular, we suggest using Personas and Scenarios in early stages of a co-design process to foster Identification.

Recommendation 2: The learning process related to Coordination should allow efficient communication to ensure that the co-design process progresses as intended. We recommend the Value Proposition Canvas as a viable means as it triggers reflection about the relation between users and educational technology and requires participants to define concrete objectives that can be later used to guide the design process. Trajectories can then allow to distribute work in bigger teams and increase motivation and engagement.

Recommendation 3: Through the Reflection learning process, participants should be able to express their own viewpoints and also listen to viewpoints of others to jointly create a new perspective. We recommend using the VPC, Personas and Scenarios, and the User Journey as they combine common characteristics of the users, the application context and possible solutions.

Recommendation 4: We recommend that researchers clearly delineate and differentiate the learning mechanisms involved in boundary crossing in order to enhance research outcomes. The codebook presented in this article can serve as a guide for understanding and differentiating these concepts, as well as for analysing and interpreting data collected through different methods.