Practices of engagement in innovation platforms: sustainably overcoming complex societal challenge landscapes

Background: South Africa’s public healthcare sector is overburdened, especially its under-resourced primary healthcare delivery system. This burden could be relieved by alleviating the population’s illhealth, focusing on the social determinants of health. These include living conditions and levels of social cohesion. In an attempt to address the aforementioned ‘challenge landscape’, this article considers socioeconomic empowerment of those marginalised members of society living at the base of the pyramid (BOP) to improve factors contributing to poor health. We propose that Innovation Platforms (IPs) offer opportunities to achieve this by drawing diverse stakeholders together, which should include marginalised individuals, to pool resources and knowledge and collaborate around a specific set of challenges. Method: A Grounded Theory approach is utilised to develop the framework comprising concept definition from a systemized literature review. It is evaluated through various progressive stages through three phases of evaluation: 1) the initial framework was subjected to scrutiny in a theoretical case study, 2) a first-pass semi-structured interview and later four more semi-structured interviews with subject matter experts, and 3) an instrumental case study to refine the framework and to understand its application in a particular situation (this included four stakeholder interviews and a workshop and feedback session with the project champion). Results: This article contributes to the extant literature by addressing the lack of guidance on stakeholder engagement practices critical to the proper functioning of IPs in the context of overcoming the complex challenges associated with social determinants of health. The final output of the study is a refined management tool for stakeholder engagement in IPs. The tool provides practical recommendations to support policy makers, researchers and practitioners in 1) establishing IPs, 2) identifying areas for improvement and 3) identifying reasons for an IP’s failure and lessons to learn.


Background
The Sustainable Development Goals (SDGs) have signified a global shift in focus from the outcomes of health to improving the underlying determinants of health (1). These determinants relate to an individual's social and economic environment (e.g. income, education level, relationships with family and friends), their physical environment (e.g. healthy living and working spaces, safe water and air) and their personal characteristics and behaviours (e.g. a balanced diet, exercise, smoking and use of alcohol) (2).
Low-income populations of developing countries tend to have the worst determinants of health due to their challenging physical, social and economic environments (3). These communities are often marginalised and excluded from formal economic systems. Illustrating the complexities of effectively addressing social determinants of health, interventions must be coordinated across improving housing, water and sanitation, access to safe food, help manage high levels of alcohol and substance abuse and help to bridge low levels of social cohesionall of this within the context of an often inadequately performing public healthcare delivery system (4). These are factors that are compounded to create the setting for challenging realities experienced in health and healthcare provisionwhat this research refers to as the 'challenge landscape'.
These challenge landscapes are not merely a sum of their parts, but each part inherently impacts another. Neither are these challenge landscapes inert, rather displaying an inherently dynamic nature. Thus, it becomes important to understand the relationships shared by the parts, requiring the application of the appropriate perspective to analyse the challenges and the context of their existence. Adopting the analogy of an ecosystem has proved advantageous for the purposes of this study, highlighting the dynamic and interconnected nature of the challenge landscapes of healthcare. The innovation ecosystem perspective provides a lens for investigating the relationships between the actors of a challenge landscape and the evolutionary nature of the networks between them (5,6). Approaches to deal with these complex health challenge may emerge from the better understanding of the challenge landscape.
Social and economic policies are structural factors that shape social and economic environmentsor challenge landscapesin which people live (4). Processes and institutions are mechanisms which may allow marginalised people to be integrated into formal systems in ways that do not further marginalize or undermine low income populations' ability to participate (3, 7). Therefore, the role of individuals and institutions with political power and control over decision-making structures cannot be ignored as they influence these policies and structural factors.
It is thus widely acknowledged that efforts to address the social determinants of health should encourage increased dialogue between different sectors, especially when considering the inherent interplay between sectors and stakeholders (4). Relationships that ensure alignment of planning and implementation of actions across different levels of government and between sectors, ministries and departments must be cultivated, while consideration must be given to what these actions may look like at different societal levels (3,4). Emphasis is placed on the active engagement of civil society to expand efforts for health (3,4). This encourages the use of participatory approaches to support efforts to strengthen social health determinants.
The term 'innovation platform' (IP) often refers to a collection of individuals driving learning and change through a collaborative partnership positioned within a specific challenge landscape (19)(20)(21). These individuals come from diverse backgrounds and have different, often competing and diverging, interests (19,22). Despite these differences, IP participants seek to collaborate to better understand and address challenges (19).
The research aimed to address the following main research question: How can sound engagement practices of innovation platforms (IPs) help enhance the collaboration of efforts around social determinants of health in the South African context?
This article contributes to the academic literature through the development and evaluation of a framework to develop and manage IPs from the ecosystem perspective. Based on a grounded theory process, key concepts related to IPs were identified and named through a coding process. The concepts were then synthesized into a coherent theoretical framework. The evaluation and refinement of the framework took place through three stages. First, a theoretical case study was completed on the Safe Water and AIDS Project (SWAP) in Kenya as preliminary evaluation of the relevance of core framework concepts. Second, two rounds of semi-structured interviews were conducted with subject matter experts to assess the credibility and confirmability of the framework. Finally, a qualitative case study was undertaken to test the framework's applicability during which data was gathered through a workshop, interviews and a feedback session with the project champion.
The final contribution of the study is a framework to formulate recommendations and a course of action to address issues that may impact stakeholder engagement in IPs. We thus provide a taxonomy for researchers, policymakers, and network champions to characterise key strategic features of evolving IPs. It provides, amongst others, a basis for generating and guiding policy considerations.
The next section provides an overview of the methodology (Section 2), after which we present the outcome of the literature reviewed for the project through systematised literature review (SLR) (Section 3). Section 4 presents the theoretical framework, Section 5 describes its evaluation, and Section 6 summarizes the contribution made through this article, reflects on the work done and outlines future work.

Methods
The objective of the research was to develop a theoretical framework and a management tool for stakeholder engagement in IPs. The intention of the research to prioritise the empowerment of marginalised communities through the appropriate participation of the necessary stakeholders in the IP, required an especially robust research approach. Jabareen (8) proposes an eight-phase Conceptual Framework Analysis (CFA) procedure for the development and evaluation of conceptual frameworks. His approach relies on the iterative and recursive nature of the Grounded Theory approach, first introduced by Glaser and Strauss (9), whereby data acquisition and data analysis occur in tandem. The CFA procedure provided the necessary structure for a robust research approach while allowing the researchers enough freedom in applying the phases in the manner most appropriate to the research; the phases were approached in a constantly comparative way to refine and organise the data and concepts until the conceptual framework was compiled. A three-part research approach was followed. Figure 1 summarises the three main parts of the research approach and how the CFA phases were completed in each part.

Figure 1: Eight-phases of the Conceptual Framework Analysis (CFA) process
In Part 1 data sources are mapped and selected using a Systematized Literature Review (SLR). Literature is mapped using the Atlas.ti 1 software and a preselected set of codes relating to the type, discipline, year of publication, geographical focus, etc., of the literature sources. Extensive reading of sources result in insights accumulated and presented as a review to present the existing body of knowledge.
Part 2, we analyse each identified concept; understand its context of emergence and its influence on the phenomenon. From the description of the concepts there are some specific 'new ideas' that emerge, under which a range of identified concepts are grouped due to similarities between them. This grouping results in a new set of concepts which form the components of the first pass theoretical framework.
In Part 3, we address the reliability and validity of the research through various steps. Ind-depth interviews with a subject matter expert and stakeholder groups result in greater understanding and reconfiguration of the concepts into an improved framework. In the third step of evaluation, the conceptual framework is applied to an industry case study to evaluate its suitability as a management tool for both researchers and practitioners.
Notice that the research employed a progressive evaluation approach to evaluate and enhance the research output in stages, rather than a post hoc validation. The progressive approach used triangulation to reinforce the validity and reliability.

Part 1: Exploring and identifying core theoretical concepts
Elaborating on the implementation of the CFA phases in Part 1, a Systematized Literature Review (SLR) was used to identify and investigate relevant multidisciplinary literature (see (8)). The review identified studies relating to IPs underpinned by an I4ID philosophy from several disciplines and geographical contexts. The search terms were "Innovation Platform AND [Actors OR Stakeholders OR Participants] AND Innovation for Inclusive Development AND Developing Countries". The search for peer-reviewed literature was performed on the Scopus database for these terms appearing in document titles, abstracts 1 Atlas.ti is a powerful computer software program designed for the analysis of qualitative data (98). 4 or keywords. The search results were screened using a set of inclusion criteria 2 to identify the most appropriate studies for inclusion in the dataset (8). Additional grey literature documents were identified using these search terms to promote a multidisciplinary dataset and to limit publication bias. Figure 2 offers a visual representation of the process. A documentary analysis of the dataset publications followed. The process was Systematized to ensure the rigorous and repeatable identification of the relevant information (10).

Part 2: Framework development: the theoretical framework
Part 2 comprised an in-depth analysis of the content whereby concepts were identified from the literature and named according to their characteristics. The analysis identified 16 fundamental concepts of engagement present in IPs. These have been called 'Practices of Engagement' (PoE).
After identifying the PoE themes, Part 2 continued to deconstruct, analyse and better understanding the respective themes by collecting the engagement mechanisms appropriate to each, as described in the literature. These were compiled into an 'inventory' and became the theoretical foundation upon which the framework would evolve.

Part 3: Framework evaluation: three phases of evaluation
Part three consisted of three phases of evaluation: 1) the PoE themes and the inventory were subjected to scrutiny in a theoretical case study, 2) a first-pass semi-structured interview and later more semistructured interviews with subject matter experts, and 3) an instrumental case study to refine the theory and to understand its application in a particular situation (this included four stakeholder interviews and a workshop and feedback session with the project champion). The pharmaceutical services manager at a not-for-profit organisation (NPO) specialising in clinical care and treatment services, and health and community systems strengthening, the interviewee has over 17 years' experience in South Africa's pharmacy industry. The organisation is a leader in public health innovation and facilitates direct programme implementation and technical assistance for the South African Government and the interviewee manages several of these programmes. They work closely with many stakeholders, including local, regional and national government, and local marginalised communities. Interviewee 1 is co-founder and director of a well-established NPO and research organisation conducting innovative research to strengthen public engagement in many of South Africa's health research projects. The interviewee has a passion to see marginalised communities empowered using innovative participatory approaches, including participatory visual methods and action-orientated approaches. They have many years of experience working with over-researched communities and navigating the dynamics that are associated with the participation of marginalised individuals. The interviewee believes that their approach to research should be accessible to others to learn from, improving engagement practices and policy making.
The interviewee holds a PhD in Immunology and Genetics (University of Cambridge). They have held several research positions both in the United Kingdom and in South Africa.
Interviewee 2 is an expert in community informatics, specialising in collaborative communities. They have experience in both academic and research and development (R&D) contexts, and consult for a variety of communities, organisations and inter-organisational networks in both the developing and developed world. Their services include community visioning and innovation strategy advice, community network mapping, collaborative sense-making and project management. The interviewee adopts an ecosystems perspective coupled with an innovative stakeholder mapping approach to understand the dynamics of stakeholder networks.
The interviewee holds a PhD in Information Management (Tilburg University). They held several research positions in both the academic and private sector before starting their own business; an applied research consultancy on collaborative communities.
Interviewee 3 has experience developing volunteer networks in a diverse range of contexts in both the developed and developing world. They have piloted a volunteer platform in a marginalised community in the Western Cape. They have a novel approach to incentivise volunteering to realise tangible community impact and social development. Their passion for people and for technology is combined in an innovative way to realise transformative social impact in marginalised communities. The insight they have into the importance of the initial roll-out phases of development initiatives, and the stakeholder dynamics associated with the early adoption and dissemination of interventions, proved very attractive to inform this research.
The interviewee holds qualifications in computer science (Massachusetts Institute of Technology) and has founded and co-founded several innovative platforms leveraging technology to relieve social and economic disparity in South Africa.
Interviewee 4 is an independent consultant and founder and managing director of an NPO with a vision to provide holistic support to the poor by facilitating the collaboration of other NPOs and channelling crowd efforts. Their experience as a champion to facilitate the collaborative efforts of NPOs within a single community places them at the centre of a larger stakeholder network. Their experience managing on-the-The interviewee holds a bachelor's degree in Industrial Engineering and has industry experience in management consulting and entrepreneurship. ground issues was attractive to inform the research. Their approach to empower a very marginalised part of society, those living at the BOP and suffering under the realities of homelessness, provides important insight into the dynamics of participatory mechanisms aimed at these members of society.
The researchers re-considered the concepts and their relationships to compile a more robust framework.
Creswell's (11) approach for semi-structured interviews focuses on data collection in the interview, and was adapted for use in this research. In semi-structured interviews, purposively open-ended questions prompted the interviewees to describe their perceptions of social phenomena within their stakeholder networks, and how they make sense of them (12). The researchers were able to learn what the interviewee deemed important when explaining and understanding these social phenomena (13). Though not fully structured, specific focus areas related to the outcomes of the SLR were addressed. This is a key benefit of semi-structured interviews, also having allowed for consistency in the interview style since multiple interviews were conducted (13,14).
Part 4 aimed to present the findings in an appropriate format for use by both industry practitioners and researchers. Part 4 went on to illustrate the use of the framework by applying the research output in the case study of an existing stakeholder network representing an IP. An instrumental case study was used to refine the theory and to understand its application in a particular situation (15). Therefore, the case was not the main focus but played a supportive role in facilitating an improved understanding of the stakeholder engagement phenomenon in the context of the case (16).
In the next section we proceed to discuss the framework evolution process from concept inventory to the final framework and management tool. We begin with the review of the existing IP body of knowledge and present the core concepts. We then focus on key value contributions of the interviews and case studies followed by a presentation and discussion of the enhanced framework and its management tool.

Core concepts: IPs and innovation for inclusive development
We find IPs situated within the body of knowledge of multi-stakeholder approaches. A recent review identified 24 distinct examples of multi-stakeholder approaches in the healthcare sector (see (17)). These include traditional IPs, multi-stakeholder platforms, collaborative research networks, living labs, and more (17). The fundamental practices of the different approaches may be similar, but their purposes and goals are quite different ( (17)(18)(19)). The building blocks of each is influenced by the context, or challenge landscape, within which the approach exists. To account for the variety of approaches, we adopt the term 'innovation platform' (IP) in this research.
Stakeholders participating in IPs depend on the platform's purpose and the level at which it is established (19,20), and may include community members, policymakers, government officials, researchers, field experts and practitioners. Participants are representatives of their home organisations, each having different goals (19,22). Therefore, each actor's perspective is likely influenced by the priorities of their home organisation, which in turn influences problem identification and goal setting within the platform (23). The IP set goals which align in some way with the needs and interests of its participants and the organisations or communities which they represent. It is widely accepted that IPs have significant potential in development contexts (20).

Researchers[GSP[1]
[RE2] have adopted several perspectives when investigating IPs. The innovation systems perspective remains one of the most common of these, especially the agricultural innovation systems perspective, likely because of the popularity of IPs in various smallholder agriculture settings in Africa (see for example (24)(25)(26)). Recent studies have also employed a value chains perspective to investigate the formation and functioning of IPs in healthcare (27)(28)(29). Recently, the innovation ecosystem perspective has gained traction among researchers of IPs (5,(30)(31)(32). This is because it offers unique opportunities for investigating the dynamics of these innovation architectures, particularly with regard to the interconnected nature of stakeholders of innovation and the evolutionary nature of these stakeholder networks (5,6,(33)(34)(35). This is of relevance to the proposed research because of the evolution observed in IPs, which has practical implications regarding platform governance, facilitation, focus and participating actors (19,20,22,36). Research must be sensitive to this evolutionary nature of IPs.
We now consider the idea of marginalised stakeholders -those persons who are 'commonly overlooked'. We acknowledge that its interpretation depends on the context, but in the developing world marginalised stakeholders are often those groups at the 'base of the pyramid' (BOP) (37). The BOP represents society's poorest socio-economic group 3 (37); an estimated 40% of South Africa's population in 2015 (38). That is equivalent to 22,12 million people.
The nature of IPs provides the opportunity for innovation to be co-directed by stakeholders who are commonly overlooked in the traditional innovation architectures of developed markets. These often associate innovation with pioneering technologies, goods and services targeted at high-income consumers (39). Traditional views further associate development with economic growth and social development as merely a by-product (39).
Recent years have seen the view that development should balance economic factors and social factors gain traction (39,40). Development should then consider the social and economic inclusion of marginalised low-income consumers. The result is an innovation philosophy which aims to include economically marginalised groups and individuals in the development of new goods and services to drive development towards technologies that incorporate the needs and interests of these groups (39).
However, innovation for inclusive development (I4ID) is a philosophy that goes beyond seeing the lowincome populations of developing countries as an "accessible mass market" (39) for the sale of goods and services (40,41). The inclusion of marginalised groups and individuals must incorporate their participation in the innovation process, and should thus empower those who usually find themselves on the sidelines of development and decision-making processes (19,40). Inclusion is often erroneously used to describe what is essentially mere representation of the marginalised, but I4ID must consider these stakeholders as participants in the innovation process, sufficiently elevating their social status. Ultimately, innovation for inclusive development seeks to instil a sense of dignity and self-worth in these participants. To this end, appropriate participation techniques must be used, which will allow these participants the necessary level of participation and influence (42,43).
The I4ID philosophy can underpin the formation of various innovation architectures, including IPs (5,44,45). This philosophy allows for the challenge landscape to be effectively identified, considering the context of emergence of the challenge. This is done by including the economically marginalised groups and individuals as participants in the IP because low-income groups and individuals experience challenges that are very difficult to comprehend if one lives outside of their context (20,22). Without considering the context of emergence of the challenge, any solutions proposed by the IP risk dealing with the superficial symptoms of the challenge alone, and not the root cause (46). As a result, the solution may not be effective. After all, novel ideas, improvements and solutions to challenges can only be regarded as innovation if these have been implemented and prove to be beneficial (36).

Concepts from the Systematised Literature Review[GSP[3]
The review set the trajectory of the research as follows (47): The significant portion of publications positioned in the innovation systems (IS) paradigm (with agricultural innovation systems [AIS] being most prominent) encouraged the adoption of the innovation ecosystem perspective as a novel application in IP research. The innovation ecosystem perspective builds on traditional IS perspectives (48) and is useful for investigating stakeholder relations in IPs, emphasising the interconnected nature of stakeholders of innovation processes (5,6). It underscores the evolutionary nature of stakeholder networks (5,6). Since only two publications considered IPs in the context of healthcare, this research contribute to the existing IP body of knowledge by broadening its multidisciplinary scope. Finally, the publications do not focus on the stakeholder engagement practices present in IPs, alluding to a need for a clearer understanding of the concepts related to stakeholder engagement within this context. We this conclude from the review of the primary publications that little to no attention is given to stakeholder engagement in IPs. Yet it may be accepted that dysfunctional engagement between platform stakeholders would result in an ineffective IP at best. However, the review did find that the literature contained key clues, ideas and procedures that may successfully direct stakeholder engagement in IPs.
These have been labelled by the authors as 'practices of engagement' (PoE) and formed the central focus of the review and our efforts to answer the research question.
The 16 PoE themes identified in the systematised literature review are summarised in Table 2. These were expanded into the "inventory of concepts" that makes up the theoretical framework described in Section 4. Using the unique contexts and perspectives of the primary authors, each PoE is shortly introduced below to understand its potential and impact on stakeholder engagement and innovation cocreation. The processes and techniques coupled to the continuous tracking of platform activities, the appraisal of these activities, and reporting the outcomes. Allows for problems to be identified and improvements to be implemented. Participants who are responsible for various platform activities are held accountable.

Participation
The engagement of stakeholders with various platform activities. Stakeholders contribute their knowledge and skill sets towards realising the platform objectives through participation. Participation is required for real inclusion to be realised (40 The development of a "roadmap" (52,53) of what the platform is looking to achieve and how. Visioning is followed by the planning of executable activities towards realising the vision. If visioning and planning are not followed by action, the platform has little to show for its efforts. (20,(50)(51)(52)(53)55,74)

Theoretical framework (Part 2): an inventory list
Considering the desired flexibility for a conceptual framework (8), the screened inventory items were translated into standards or norms to encourage effective stakeholder engagement in the context of an IP. Effective stakeholder engagement accepts that all stakeholders are assigned an appropriate level of participation and that interactions between stakeholders are constructive, so the IP can function well and meet its goals of empowerment.
The framework items are called implementation criteria. The criteria-format lends the framework the desired flexibility. Criteria can be used to inform one's conceptualisation of stakeholder engagement in IPs by introducing those principles that should be in place. A framework of criteria can be useful as an assessment tool to assess the current state of stakeholder engagement in an existing IP by investigating the extent to which the criteria are met. Such a framework may also be a useful guide for establishing a new stakeholder network and developing an IP. Thus, a framework of execution criteria is not only for improved understanding, but may be useful for planning, and monitoring and evaluation.
When translating the inventory items into implementation criteria, careful consideration was given to the purpose of each item to the context of the IP. This was to mitigate the risk of an item's significance being lost in translation. To this end, special attention was given to the language used when developing each criteria-item for the framework.
The first iteration of the stakeholder engagement framework in Table 5 (See Appendix A) does not display the interrelationships shared by the various PoE themes. Rather, these interrelationships become clear when reading the categorised implementation criteria. The implementation criteria, grouped according to the engagement theme which they support, incorporate many nuanced expressions of other engagement themes.
For example, within the category called 'Monitoring, evaluation and feedback', one criterion reads, "feedback on IP activities is used to guide the formulation of next steps and activities." This clearly speaks to the role that feedback has for visioning and planning activities, captured in the 'Visioning and planning' category.
The benefit of the first iteration of the framework is that it displays the content simply and clearly. Evaluating the framework content can follow a structured approach where each criterion can be evaluated as credible and confirmable. This encourages a holistic assessment of the framework to ensure a valid and reliable research output.

Evaluation phase 1: Theoretical case study: the case of the Safe Water and AIDS Project (SWAP)
The aim of the theoretical case study was to investigate an example of how participation in a formalised community [health] network has led to the empowerment of marginalised individuals within a resourceconstrained setting. The case study provided an opportunity to reflect on the inventory of concepts of the theoretical framework, verify the confirmability of the PoE themes by analysing their presence in the case study, and contribute towards the construction of a framework.
The Safe Water and AIDS Project (SWAP) is a Kenyan-based non-governmental organisation (NGO) whose work aims to promote and improve hygiene and primary healthcare practices in local communities (80). One particular initiative targets women living with HIV and who have been ostracised by their communities because of it (81). SWAP has established networks of community health promoters in several communities to educate households about good hygiene practices and provide them with basic health products. This intervention has significantly improved community health. The networks comprise these ostracised women who are trained as community health promoters. Their participation has forced other community members to recognise their contribution to the improvement of the livelihood of the communities, and as a result have been embraced by local society. The sale of health products through this initiative stimulates the economic empowerment of these women. This case was chosen because these networks are well established. Research has found that they have had a significant positive impact on the health of several Kenyan communities (see (82)(83)(84)(85)). Additionally, the case provides an example of formalised networks whose participants experience tangible social and economic empowerment. Finally, the case provided an opportunity to analyse the strategic management of these networks and the role of strategic partnerships in their success.
The investigation focused on the stakeholders present in these community networks, their interactions as described in the case literature and SWAP's strategic management of the network stakeholders. The PoE themes were compared to several instances of engagement described in the case literature to verify whether the PoE themes could appropriately describe them.
It is apparent from this case study that intervention activities require the strategic involvement of key stakeholder groups in a partnership-centred approach. Stakeholders from similar contexts were grouped to better discern the links between them, their purpose and interactions. The interactions are labelled A to H for reference in the discussion.  Table 3 shows how the PoE themes may be used to describe the stakeholder interactions observed in the case literature. The case study found all PoE themes, save for conflict management, to appropriately describe one or several instances of stakeholder engagement present in the SWAP network. This became a departure point for grasping the practical application of the concepts and their implications for stakeholder engagement in IP networks. The investigation revealed that SWAP's focus for its community health network initiative lies in capacity development (including human-resource development), the intervention activities, financial sustainability and research to inform the platform's functions.
The absence of conflict management in the case literature does not disqualify it from the list of PoEs. Rather, it was highlighted to be indicative of bias which excludes descriptions of the challenges SWAP's networks face to engage and manage the behaviour of stakeholders. Never-the-less, the exploratory use of the remaining PoE themes to describe stakeholder interactions in SWAP's community health networks hints at a comprehensive inventory of concepts suited to collaborative networks. All the stakeholder interactions observed in the case could be drawn to at least one PoE theme, while multiple themes were often used to describe a single interaction. This made it apparent that complexity is characteristic to stakeholder interactions in IPs and could not be overlooked as the research continued to avoid an over-simplification of the stakeholder engagement phenomenon.
Literature on stakeholder engagement and participation emphasises the importance of assigning the appropriate level of participation to stakeholders (contrary to the idea that all stakeholders should have the same level of participation) (42).  Figure 3). Their roles are integral to the healthy functioning of the platform, offering support, training and funding, but they remain far removed from the action on the ground (interactions G and H). Consider the Kenyan Ministry of Health which donated land for the building of hubs where community health promoters are trained and can restock their supplies, yet it is not involved in distributing healthcare products to the communities. Thus, stakeholders are operating within their strengths, leveraging what resources and influence they have for the benefit of the network. Force-fitting such diverse stakeholders into participatory roles which are the same will result in an inefficient network and very frustrated participants.
The investigation revealed that the PoE themes require effort and intentionality to implement if they are to contribute to the success of an IP. The SWAP case revealed that this may be facilitated by assigning these themes as identifiable stakeholder roles. A further benefit of which may be that the purpose of stakeholders' presence in the network is understood. In the case study we notice how facilitation is a role taken up solely by actors within SWAP as they mediate the interactions between community actors and non-community actors. SWAP identifies where there is room for improvement and sources the appropriate partners to address this. The research partners provide the resources and skills necessary for effective monitoring and evaluation of the intervention activities, which guides the network's visioning and planning activities. Thus, we see how the network thrives as stakeholders operate within roles that fit them.
While acknowledging that a single case is not comprehensive, several important lessons can be learnt. Firstly, to truly empower people, their inherent potential and human value must be recognised and leveraged. This means that opportunities for growth must be balanced with ownership and the responsibility to improve one's situation. Secondly, empowerment must not be reduced to economic factors alone but includes both social and psychological components. In the case we observe that it is the re-acceptance of the women into their communities and the restoration of their self-worth that is most important, with the economic empowerment of the women positioned as a complementing factor. Finally, it is important to note that the stakeholders in the initiative have different levels of participation according to the roles they play in the network (as opposed to 'blanket-level' participation). Assigning the appropriate level of participation to different stakeholders remains an integral part of an IP's success (see (42,43,(86)(87)(88)(89)(90)).

Evaluation phase 2: Semi-structured interviews
Error! Reference source not found. shows that semi-structured interviews formed part of two stages of the framework's evolution. In Part 3, a single interview complemented the insights gained from the theoretical case study about the inventory of PoE themes, this time considering their appropriateness for the South African health context. After developing a preliminary conceptual framework, another four interviews were conducted with local and international subject matter experts to assess the credibility and confirmability of the framework. It was feasible to use a small sample of interview participants since each is an expert and the analysis process is not a statistical one.

First-pass interview
The preliminary research findings were presented to an expert in public health innovation whose work relies on forming close relationships with local marginalised communities in South Africa. The outcomes of the discussion guided the development of a preliminary framework. Firstly, it became apparent that modifications had to be made to some PoE themes. These are explained in Table 4. Discussing the influence of gender dynamics seemed to inherently lead to discussions of racial dynamics, indicating the importance of including both gender and racial considerations when dealing with stakeholder engagement as both are areas for potential discrimination.
Secondly, considering the participant's use of dialogue when discussing each PoE theme revealed that several other themes were mentioned when discussing a single theme. A new dimension in the complexity of stakeholder engagement had emerged. That the PoE themes should not be regarded as stand-alone became apparent; they are interconnected and any single instance of engagement of any form has several engagement themes subtly at work in the background. A simple example is the important role of communication to manage and prevent conflicts in a network (conflict management). Another example of these interrelationships is shown in Figure 4. The discussion highlighted the importance of clearly understanding a problem before appropriate solutions can be developed. Yet, an initiative will almost never follow the initial plan. Rather, the plan evolves as one learns from implementation and as innovation pathways are adjusted as necessary. Dynamic representation of stakeholders, as new stakeholders join a network and others leave over time, contributes to an evolutionary nature. It remained clear throughout the interview that stakeholder engagement was critical to the success of inclusive development initiatives, yet is often neglected. Early engagement strategies are required to make stakeholders, particularly the beneficiaries, aware of the opportunities available to them. However, these early engagement initiatives are often not implemented. Time constraints for programme implementation and funder demands are the common barriers to early engagement of stakeholders.

Second round of interviews
The analysis of the interviews sought to understand how the PoE themes are perceived in practice by tallying the instances where a theme was confirmed 4 in the interview discussions, but also where the interview participant alluded to important information that was not yet accounted for, revealing a gap in the framework. This was used to evaluate and refine the framework. The results appear in Figure 5. Notice from Figure 5 that Strategic representation has the greatest number of confirmed items. Strategic representation has of the highest number of items in the preliminary framework (second only to Facilitation), alluding to its relative importance as an engagement practice. This may be supported by the interviewees' focus on the theme.
Participation enjoyed a significant amount of attention during the interviews, yet the number of confirmed items in the interviews is relatively low (seven out of 13 items in the framework). Gaps in the framework were highlighted in this category another seven times during the evaluation process, revealing that this may have been where the preliminary framework was weakest and an area needing further investigation in future research.
The attention given to Participation in the interviews and its prominence in the framework may point to the importance and complexity of participation as an engagement practice in stakeholder networks, especially when pursuing the empowerment of BOP stakeholders. The relatively large amount of information not accounted for by the preliminary framework (and thus missing from the literature) may be indicative of the challenges associated with participation. The interviews highlighted that Participation is a process and the end goal must be understood. The interview participants are well rehearsed in the challenges of participation and the consequences of such challenges, and may have ways to address these, but there remains a void in the understanding of what participation is and how to apply it in different contexts.
The interviews brought forth an idea that consideration be given to stakeholders' departure from a network and even the disbanding of the network entirely. If these events are not anticipated and planned appropriately, they may leave a vacuum and undermine the platform's efforts towards empowering marginalised communities. Good relationships need to be maintained with stakeholders who have left a network as these stakeholders may be called upon again in the future. All this needs to be carefully considered by the managers of the network. To this end, Rolldown of participation was added as the seventeenth PoE theme, where 'rolldown' implies a gradual readjustment of stakeholder roles and dependencies as one or several participants plan to leave the network.
Communication received relatively little attention in the interview discussions, despite its prominence in the literature. We do not believe that this would undermine the importance of the engagement practice but it may be related to its inherent role within other engagement practices. It may be that communication was deemed an obvious consideration when dealing with stakeholder engagement and thus not explicitly mentioned by interviewees in the discussion. Indeed, from the literature we understand that "communication is the electricity that powers the platform" (63).
To further understand the conceptualisation of engagement practices in IPs, the prominence of the PoE themes can be represented on a scatter plot using the relative frequencies of their 'appearance' in the IP literature through the SLR, and the interview discussions. Accordingly, Figure 6 represents the frequency by which a PoE theme was coded in the literature to the frequency by which that same theme was coded in the analysis of the interviews. Relative frequencies allow the display of data on the same set of axes.

Figure 6: Comparison of emphasis on engagement practices: literature and interviews
We again acknowledge that additional interviews would certainly increase the confidence in the data displayed above, however, Figure 6 certainly offers insights into the focus areas of both research and practice in the stakeholder engagement landscape. Two groupings are recognisable in the data: those themes which were more prominent in the literature review (represented by a blue trendline) and those more prominent in the interviews (represented by a green trendline). Both trendlines indicate a positive trend, and themes with an increased prominence in the literature are met with a corresponding increased prominence in the interviews.
Resources and capacity (PoE 11 in Figure 6) is most prominent in the literature (relative frequency of 17,4%) but the interviews did not reveal the same focus (relative frequency of 5,3%). Recall that most of the primary publications focused on the agricultural sector, mainly in an African context. Agriculture is a resource-intensive sector and Africa is regarded as a resource-poor setting, and Agricultural IPs are often formed to pool resources to support smallholder farming (see for example (56,67,91)). This may explain the prominence of Resources and capacity as an engagement theme in the literature. However, agriculture was not among the variety of areas of the interviewees' expertise. Thus, this may explain the different emphasis on Resources and capacity, and even the two groupings of the data in Figure 6.
Strategic representation was the most prominent PoE theme in the interviews and enjoyed near equal prominence in the literature, highlighting the importance of identifying and attracting the appropriate stakeholders. Both literature and the interviews place the least emphasis on Shared learning. The benefits of shared learning need to be understood and their potential communicated to stakeholders before these benefits will be realised. Indeed, one interviewee commented that their stakeholders were often unresponsive to efforts to promote shared learning in the network, and the reasons for this were not clear to the interviewee.

Evaluation phase 3: Practical application: an industry case study
The case study considered an IP formed to alleviate vagrancy in the University town of Stellenbosch, in South Africa's Western Cape. The IP goals included coordinating efforts to support and uplift vagrants while raising public awareness around the concept of 'responsible giving'.
For several years, different NPOs' efforts to support and empower vagrants were detached and often duplicated as there was no collaboration around this challenge. This restricted the positive impact envisioned by the NPOs. The emergence of a network champion who recognised the potential of collaboration resulted in the establishment of a network. The network champion understood that individual NPOs offered distinct but complimentary services targeting vagrants, and through collaboration would offer holistic support to the beneficiaries and minimize duplicated effort.
The IP's primary intervention is a coupon system which offers the public an alternative to giving cash, food and clothing to those begging in Stellenbosch. Giving a coupon fosters mutual respect and dignity during interactions between the public and the beneficiaries. Coupon recipients have access to a holistic portfolio of support services from the participating organisations.
Stakeholders identified were the vagrants, several not-for-profit organisations (NPOs) offering them support services, the public, local government (municipality), social service actors, academia, churches and law enforcement. It can be argued that these stakeholders make up an 'ecosystem'. Figure 7 illustrates our conceptualisation of this 'ecosystem' of stakeholders of vagrancy in Stellenbosch. We place the IP at the centre of the ecosystem. The stakeholders participating in the IP are the partnering NPOs and the municipality. The stakeholders in the immediate vicinity of the IP are other organisations and institutions that work with vagrants and these include social service actors, academia, churches and law enforcement. On the periphery of the ecosystem are the stakeholders who interact with the beneficiaries and who are the targets of begging. These include the general public, tourists and local shops and restaurants.

Results from the case study
The enhanced framework was used in the case study to develop recommendations for improved engagement between the network stakeholders. Firstly, potential issues around stakeholder engagement were identified following interviews with the several stakeholders. Secondly, the conceptual framework was applied in a systematized fashion to formulate recommendations for addressing these issues. This served to verify the suitability and relevance of the framework and to guide the development of a management tool. Finally, the recommendations were presented to the IP's network champion for scrutiny and feedback. The case observations are summarised below.

Issue 1: Formal versus informal network processes
One interview participant expressed their need for more formalised network processes while other network stakeholders did not have this need. The network champion had previously acknowledged that a more formalised network may have its benefits. It is recommended that an approach for more formalised engagements be considered while not overlooking the role of informal interactions in stakeholder networks.

Issue 2: Lack of skills necessary to market the IP's interventions
Several interview participants mentioned a lack of public awareness of the IP's interventions. Some attributed the limited awareness to insufficient marketing skills necessary to achieve greater dissemination of information and adoption of the interventions. It is recommended that an additional participant be added to the IP; an appropriate individual/organisation with the skills and resources necessary to champion the marketing of the intervention.

Issue 3: Avoiding the duplication of efforts
Valuable resources are wasted when different stakeholders offer the same service to the same beneficiary. It also creates an opportunity for vagrants to take advantage of the system and access the same services at multiple points. One interviewee acknowledged that the stakeholders are very strict about not having the services duplicated. However, as each organisation evolves, their service offerings evolve, and the risk of duplication increases. It is recommended that opportunities are created for engagement and coordination between the stakeholders to avoid the duplication of efforts.

Issue 4: Stakeholder expectations are mismatched
The case interviews revealed that misalignment of expectations among the stakeholders exists, specifically where certain stakeholders underestimate what other stakeholders are working to achieve. It seems that this has not yet had any visibly negative effects on the collaborative mood; however, it remains a risk to be addressed.

Recommendations for improved engagement developed from the framework
To develop recommendations for improved stakeholder engagement within this IP, the authors drew from elements of the conceptual framework to better understand the root cause of the issues identified and how these might be addressed. The framework was the source of insights and ideas as the case study evaluated its suitability as a stakeholder management tool. Here we discuss our findings for each issue (statements and points taken from the framework are indicated by italics).

Issue 1: Formalising the network without disregarding the value of informal processes
Formalised engagements offer an opportunity for improved alignment and shared learning between stakeholders because a culture of exchanging knowledge and experience can be cultivated. Formalised interactions can be effective if they are properly facilitated. Ideally, a facilitator must not be a stakeholder of the challenge landscape, but rather a neutral and respectful party who is not perceived as a threat by the other stakeholders so that they may be an appropriate mediator of the interactions. A good facilitator might focus on managing aspects of communication to ensure constructive discussions. Conflict management is an important consideration, while formalised engagements enable the environment for addressing misunderstandings and resolving concerns. Some stakeholders may be more prone to competition and less so to collaboration, and formalised engagements must focus on managing the power dynamics of a stakeholder group to prevent power plays and promote a non-threatening environment. These culminate into the development of trust.

Recommendation for issue 2: Identify and integrate a 'marketing champion'
The network might focus on the area of championing; identifying stakeholders who can and will take initiative and leverage their own capacity and resources alongside those of the other network stakeholders. The addition of stakeholders to a network should be a strategic consideration (strategic representation). Several formalised stakeholder identification procedures are available; however, the network's current informal nature may require a similarly informal approach to stakeholder identification. It is common to look to existing relationships with stakeholders from different networks for potential participants.
It remains important to focus on the implementation of interventions; where to focus on the issue to gain the most traction as early as possible. One interview participant alluded to the importance of informing tourists who visit Stellenbosch daily and, being uninformed although well-intentioned, enforce the cycle of irresponsible giving. The tourism industry is likely the most viable audience for the campaign.
Visioning and planning are a strategy to better understand the need at hand. These are appropriate to investigate the resources and skills required and thus help to identify a potential participant with the necessary capabilities.
Stakeholders should be engaged following the appropriate procedures, especially in the absence of a pre-existing relationship, highlighting the importance of communication. It is necessary to establish, build and maintain trust relationships with stakeholders who are to become participants in the network. Credibility and a clear vision play an important role to earn a stakeholder's trust.

Recommendation for issue 3: Improved coordination of activities
This issue identifies a need for improved resource mobilisation and capacity development; coordinating activities (rather than duplicating efforts) strengthens the collective resource position of the stakeholder network. Resources are better dispensed and utilised when they can service coordinated activities.
Through coordinating different activities, stakeholders have greater capacity to focus on other areas for intervention. Visioning and planning and planning activities enable improved coordinating; as stakeholders share their approach to addressing the challenge landscape, other stakeholders would do well to listen and identify potential opportunities for coordination. A space for stakeholders to offer feedback on their performance may assist in identifying areas for improvement. Feedback of coordinated activities allows for improvement of these activities.

Recommendation for issue 4: Greater focus on alignment
The main concern revolves around a lack of alignment of stakeholders

Emergence of engagement-facilitation categories
From the recommendations above the reader should notice three categories emerge from how specific PoE themes were contextualised. Each recommendation focused on a desired outcome; the goal which, if achieved, should address a specific issue or challenge experienced by the platform. From the case study we see such themes as shared learning, alignment, implementation of interventions, participation, and resource mobilisation and capacity development as 'desired outcomes'.
The case study showed that to achieve the desired outcomes the platform must use specific engagement activities for the engagement of stakeholders, including at the interpersonal level. The recommendations made highlight the themes of communication, conflict management, managing power dynamics, transparency and trust building as important 'engagement activities'.
Another very important category of PoE themes can be seen from the recommendations. We refer to these as enablers. Enablers bridge the gap between the engagement activities and desired outcomes. This category contains the PoE themes that, if leveraged correctly, may achieve the IP-specific goals. The recommendations made use such themes as facilitation, strategic representation, monitoring, evaluation and feedback, and visioning and planning as enablers for overcoming the IP's challenges.
The PoE themes are interconnected, and each theme may influence the presence of several other engagement themes in an IP. Visual cues are effective to represent this interconnected nature between PoE themes of different categories. Figure 8 emphasises the visual cues as used in the framework management tool (see section 6.1). The use of two horizontal arrows pointing in opposite directions represents the interplay between engagement activities and enablers. These two categories are placed in a rectangle. Together, they contribute to the desired outcomes, as represented by the horizontal chevron markings pointing to the right.

Concluding remarks: reflections from the case study
Without the intervention of the network champion, the IP may not have been formed. The case interviews made it clear that the ecosystem became more functional once the IP was established formally, and the role of the network champion is well recognised by the participating stakeholders. This is an example of where an overlap in the visions of different stakeholders is not necessarily enough to move them to collaborative action; a champion of the vision is often a necessary component for this.
The need for stakeholders to acknowledge that they are part of an ecosystem is clear from the case; they are not operating alone but are inherently linked with the other stakeholders of the challenge landscape.
If stakeholders fail to acknowledge this, their independent interventions may be detrimental to the efforts of other stakeholders and, ultimately, to the beneficiaries. It is thus worth knowing who the stakeholders are, so that an ecosystem can be understood and potentially managed.
Finally, the IP's collective vision must work towards goals that are attractive to each of the participating stakeholders; it should contribute to their individual visions and goals. Thus, though the finer details of each stakeholder's vision may differ, the IP's interventions must sufficiently capture aspects of all the visions as it works towards a common goal.

Reflecting on the innovation platform
The stakeholder network at the centre of the ecosystem reflects an innovation platform (IP) in that diverse stakeholders participate in a collaborative network around a specific challenge landscape. The network is the 'platform' for value-creating interactions between NPOs and local government, together seeking opportunities to address challenges and achieve a shared objective.
An important characteristic of IPs is that they drive learning and change (19)(20)(21). The network intervention and the distribution of coupons is complemented by posters advertising the intervention and newspaper articles describing how it works, in this way informing the public of the risks of irresponsible aid. The network encourages responsible giving to empower vagrants in Stellenbosch and potentially have a lasting socio-economic impact on their lives. It remains important that IP goals align with the needs of all stakeholders of the challenge landscape (19). The network aligns the visions of its stakeholders to address the beneficiaries' needs through the services offered by the participating NPOs and assists the public in interacting with the beneficiaries through the coupon system.
The I4ID philosophy calls for innovation and intervention that considers the interests and addresses the needs of stakeholders at the BOP (19,40). Indeed, this is true of this IP. The I4ID philosophy encourage the inclusion of BOP stakeholders as participants in the innovation development process (19,40), affording them an appropriate level of influence in the IP's functioning (42,43). The benefits may include a more comprehensive understanding of the challenge landscape, its context of emergence and the development of appropriate solutions (20,22,46). However, the homeless and other persons in need are not participants in the IP and are the primary beneficiaries of its interventions.
The case interviews revealed that for some IP participants, the participation of the beneficiaries in planning and decision-making activities has been a topic of much deliberation. Yet, none have adopted this. The stakeholders believe that until the beneficiary has been successfully reintegrated into a community and no longer relies on the systems of support for vagrants, they may not present an objective contribution to discussions; one where they shed light on the issues contributing to vagrancy and the interventions needed to address these.
It would thus appear that participation is not a 'silver bullet' since it may be appropriate in some contexts but not in others. This echoes the opinions of several researchers in the field of stakeholder participation (see (90,92,93)). However, a good sense of the challenge landscape can be gained by including all other stakeholders in the ecosystem in discussions, to gain exposure to a diverse range of perspectives, experiences and conceptualisations of the issue. It may be appropriate to include representatives of the stakeholders at the periphery of the ecosystem (the public, tourists and local businesses) in discussions around the issue.

Reflecting on the ecosystem
Ecosystem boundaries are important when looking to understand the complexities of an ecosystem. In this case, it is logical to delimit the ecosystem to the stakeholders operating in Stellenbosch and the surrounding areas. In the functioning of this ecosystem, the peripheral stakeholders engage with the vagrants and this greatly impacts the immediate stakeholders. These engagements are perceived to be either constructive, as is the case with responsible giving, or destructive, as is the case with irresponsible giving, and thus either support or undermine the efforts of the immediate stakeholders.
Vagrancy is not an issue unique to Stellenbosch and this ecosystem is not independent, but rather a 'system within a system'. Thus, an intervention that influences this ecosystem will surely influence the larger ecosystem too. Consider for example that houses are given to all homeless persons in Stellenbosch. The case study participants posit that an increase in homelessness would then be seen as homeless people in the larger ecosystem would perceive Stellenbosch as a hub to receive housing, and an influx of 'new' vagrants would likely result.

Reflecting on the significance of stakeholders' visions in an IP
The case study revealed that the issue of individual stakeholders' visions and finding commonality amongst them is complex. The case interviews revealed that stakeholder visions are unlikely to be a perfect match, but they may be complementary. It is in recognising when and how independent visions may contribute to a collective vision that collaboration may become a reality and stakeholder networks or structures, like an IP, may be established.
Interviewees recalled the culture of competitiveness between NPOs operating in Stellenbosch and how a meeting of the stakeholders of vagrancy in Stellenbosch served as the turning point and the birth of a more collaborative mood between them. Though the visions and goals shared by each stakeholder were noticeably different, they managed to identify something common to each; a shared vision for a 'safer Stellenbosch' provided the catalyst for these different stakeholders to acknowledge their role in a system larger than themselves and welcome opportunities for cooperation rather than competition. This may be a display of the benefits of leveraging aspects of commonality of ecosystem stakeholders, rather than focusing on discords.
It was apparent that stakeholders have short-, medium-and long-term visions. For the stakeholders participating in the IP, it seems that their short-term visions all incorporate providing the stakeholders at the periphery of the ecosystem with a means to interact with individuals begging on the streets (the coupons). Another collective vision is fostering a culture of responsible giving in Stellenbosch, which may be a medium-term vision for the stakeholders. These common visions thus contribute to the IP's vision as it aligns with the needs of its participants and the beneficiaries of its interventions.
What is interesting is that the long-term visions of the IP participants were less aligned. From this analysis we might assume that the collaborative mood between stakeholders is more dependent on the alignment of their short-and medium-term visions, and less dependent on the alignment of long-term visions. As visions change and the IP progresses towards the 'long-term' vision, it is likely that the stakeholders represented in the IP will change also.

Enhanced framework and management tool
Guided by the application of the enhanced conceptual framework in the case study, the management tool was compiled. The management tool comprises two elements to be used in conjunction: 1) the framework overview canvas (see Figure 9 below) and 2) the enhanced conceptual framework (see Table  5 in Appendix B). The framework contains 17 PoE themes, now in three categories: 1) Engagement activities, 2) Enablers and 3) Desired outcomes.
A process for using the management tool is proposed in Figure 10. Arrows direct the user's attention from one category to the next, alternating between the framework overview canvas and the framework.
The process begins with the stakeholder engagement issue that the user wishes to address. The user follows a process of considering which 'engagement activities' may address the issue and what the 'desired outcomes' might be. The 'enablers' that would support the achievement of the desired outcomes are then considered. Within each of these categories, guiding questions prompt the user to identify the PoE themes appropriate to address the issue. The implementation criteria associated with the identified PoE themes are then consulted to formulate a course of action.
Users must contextualise the application of the tool to ensure that the recommendations are appropriate to address the issue. The identified issue should be understood properly as the application of the framework requires the user to consider the past, current and future impacts of the issue on aspects of an IP's engagements. The procedure in Figure 10 offers a guide, however, users should move iteratively between the different stages and parts of the framework as they deem necessary to accurately contextualise the application of the management tool.
In the industry case study covered in Section 5.3, the management tool was used to make recommendations for improved stakeholder engagements in the IP. Several diverse engagement issues were addressed, and the tool was appropriate for each. This verifies the transferability of the research output as the framework content can appropriately contribute to the context of the case despite being developed from multidisciplinary sources. The transferability of the research output is another important consideration for valid and reliable research (94,95).
The credibility of the framework and its overview canvas as a tool for managing stakeholder engagements in an IP context was verified in the case study. These components may benefit from exposure to additional case studies and pilot studies to further refine the tool's content and structure (96,97), and possibly adapt it further for use in other contexts. A single use case was enough for the purposes of this research study and this was further motivated by time constraints.
The recommendations were directed at the network champion in this case study; however, the tool maintains its potential to be suitable for a much broader target audience, including those in both research and practice. The tool was applied by the researcher to formulate the recommendations but in practice it may be applied by network champions, facilitators, external consultants, and the like, and may also be useful in group settings. The management tool was used here to formulate recommendations and a course of action when addressing issues impacting stakeholder engagement in IPs. However, other areas of application include the establishment of new stakeholder networks.

Concluding remarks
IPs as spaces for inclusive, context specific innovation creation are well suited to address the complex healthcare challenges faced in resource-limited settings. Proper stakeholder engagement is a necessity for any IP to function properly, and importantly, to appropriately integrate marginalised stakeholders in a manner that empowers them and affects tangible change to their situations.
This study proceeded to contribute a framework of good practices suitable to manage proper stakeholder engagement in IPs, keeping in mind the principles of the I4ID philosophy. The conceptual framework comprises insights developed from a rigorous literature review and field work. The framework is combined with visual 'canvasses' to form the management tool for stakeholder engagement in IPs. The tool offers guidance for the conceptualisation and analysis of stakeholder networks. Its content includes criteria for good practice (called implementation criteria) that can inform the establishment of innovation architectures like IPs. The management tool guides the user in a process of formulating context-specific recommendations when addressing issues of stakeholder engagement in IPs. It may address the needs of researchers and practitioners who want to set up a new IP, who want to identify areas for improvement in existing IPs, or who want to identify reasons for an IP's failure and the lessons to learn. The research also highlights a variety of angles from which social issues that impact the health of a population may be addressed.
Several promising opportunities for further investigation and paths for future research were identified as the research progressed. This research considered mainly the engagement practices and mechanisms associated with stakeholder engagement in IPs.
1. A more thorough investigation of the common barriers to stakeholder engagement, and how to overcome them, would benefit the research domain. This may inform the appropriate additions to the conceptual framework for stakeholder engagement in IPs. 2. The final conceptual framework, and the overall management tool for stakeholder engagement in IPs, may be improved by including recommendations for 'how' its implementation criteria may be addressed. 3. Future research might consider when to engage stakeholders. This may require an investigation into the different IP lifecycle phases and the needs of the IP at these different stages. This may be a complex investigation as IPs are diverse and may have several life cycle phases. 4. The impact of cognitive biases on decision-making and the consequences of unchecked biases on stakeholder engagement need further investigation. 5. The case study did not consider an IP where the marginalised stakeholders and beneficiaries of the intervention are direct participants. Future research should identify case studies where this is the case.

Consent for publication
Not Applicable

Availability of data and material
The data that support the findings of this study are available from the authors, but restrictions apply to the availability of the data, which were used under license for the current study, and so are not publicly available. Data are however available from the authors upon reasonable request. • IP activities are continuously monitored; predefined indicators are monitored • Feedback on IP activities is used to guide the formulation of next steps and activities; new insights and discoveries are implemented, and the IP learns from the mistakes and successes of IP activities • Alignment of IP activities and IP participants with the platform goals is monitored • Feedback is used to identify areas where alignment must be restored • Prioritisation of IP activities is constantly tracked to guard against the effects of power imbalances • Feedback of significant successes is shared with external stakeholders to increase the awareness and interest of stakeholders in the IP; • Feedback of IP activities is used as an opportunity to share insights and experiences and to learn • Participatory monitoring, evaluation and feedback activities are used to promote involvement and transparency • Results of monitoring, evaluation and feedback activities are shared with IP participants and other stakeholders in the ecosystem • Impacts of IP activities are accurately reported by the monitoring and evaluation activities;

Appendix A: Preliminary stakeholder engagement framework
impacts are communicated to the beneficiaries and other stakeholders • Research studies are used to prove the impacts of the IP activities; significant insights are published to disseminate the insights and generate income • Accurate accounts of the reports are disseminated to other stakeholders in the ecosystem through feedback activities • Feedback is used to inform the focus of the IP, adjusting the innovation pathways to address emerging trends and needs Participation • IP participants identify potential activities and select the activities for implementation • IP participants are motivated to participate in the IP and align their resources and capacities • Involvement of IP participants begins early on and continues throughout the life of the IP • IP participants are involved in strategic planning activities which counters their resistance to change • IP activities rely on the involvement of perceived weaker IP participants, promoting their societal status, economic positioning and self-esteem; perceived weaker IP participants are economically active • IP participants have abandoned self-interest for the common good of the IP and its stakeholders • IP participants are actively involved in participatory monitoring, evaluation and feedback activities • IP participants mobilise their individual resources and capacities and direct these into the platform • IP participants have access to the IP's common resource pool • IP participants access opportunities for capacity development; IP participants gain experience, develop insights and gather information; • Commonly marginalised stakeholders have improved social and economic positioning, and self-esteem; commonly marginalised stakeholders are economically active • IP participants share their experiences and insights with other IP participants Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 12 July 2021 doi:10.20944/preprints202107.0250.v1 PoE Implementation criteria • IP participants are involved in the identification of prospective new participants; existing relationships are leveraged to gain new IP participants

Resources and capacity
• IP participants' capabilities are developed to match the needs of the IP activities • IP leverages its capacity to advocate on behalf of its stakeholders (e.g. making policy recommendations) • IP participants mobilise their individual resources and capacities and direct these into the platform • Resource management procedures used in the IP are transparent and trustworthy Shared learning • Capabilities of IP participants are developed because of shared learning • IP participants share information, insights and experiences • Improving sense of collaboration and trust between IP participants • Ideas are translated into executable activities • Improving coordination through joint planning activities Strategic representation • IP participants include representatives of the beneficiaries so that IP activities address the true needs of the beneficiaries • Technologies and innovations introduced by the IP are appropriate to the context of the need • National level stakeholders are represented; IP goals align with the national goals to support government strategies in the specific sector • Stakeholders with the motivation and capacity to serve as innovation champions are represented in the IP • Stakeholders who may be regarded as commonly marginalised are represented in the IP • The presence of unreasonable stakeholders in the IP is mitigated; selecting IP participants considers their influence on power dynamics • Representation in the IP is leveraged to obtain information directly from stakeholders who implement and disseminate innovation • Resource positioning of the IP is positively influenced by the represented stakeholders • Capacity of the IP is positively influenced by the represented stakeholders; IP has increased presence in various economic sectors and at different societal levels; gains legitimacy • Resource potential of stakeholders is established using the appropriate stakeholder analysis techniques • Capabilities of stakeholders are established and areas for improvement (capacity development) are identified using the appropriate stakeholder analysis techniques • Dissemination of IP interventions is positively influenced by the represented stakeholders • IP participants include stakeholders who are willing to contribute to the exchange of knowledge, experiences and insights • IP participants include those stakeholders who are experts in the necessary sectors • Key stakeholders are represented to promote trust in the platform among the wider ecosystem of stakeholders and encourage involvement in IP activities Transparency • Transparency and honesty underlie all communication functions; information is presented completely and accurately and shared within and across IP borders • IP participants share their aspirations, frustrations and self-interest • IP participants are fully aware of the IP activities taking place, decisions being made, and the reasons for these; • IP facilitator operates with neutrality and integrity, sharing all necessary information with the IP participants • Risks are being communicated with the necessary stakeholders • IP operates openly within the innovation ecosystem to raise awareness of platform activities and rally interest from other stakeholders; IP is visible to external stakeholders • All relevant information to guide the visioning and planning activities is made available Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 12 July 2021 doi:10.20944/preprints202107.0250.v1

PoE Implementation criteria
Trust building • Alignment is being strengthened; stakeholders are becoming more willing to compromise and collaborate • Communication channels between stakeholders become more developed; information is shared more easily • Stakeholders become increasingly motivated to contribute from their resource pools • Additional stakeholders are drawn to the IP and willing to contribute from their resource pools • Non-competitive environment allows for shared learning Visioning and planning • IP activities are identified and planned from appropriate planning and visioning activities • IP activities strategically target areas with the greatest potential for innovation adoption and dissemination; appropriate timing for implementation is considered • IP activities identified for implementation align with the goals of the platform and the needs of the beneficiaries • IP activities are planned to incorporate monitorable indicators to allow for monitoring, evaluation and feedback of these activities • Resource requirements (resource types and amounts) are identified using appropriate visioning and planning • Stakeholder capabilities required for the implementation of IP activities are identified using appropriate visioning and planning activities • Strategic representation is guided by the planned scale of the IP activities and the IP's focus • Dynamic representation of IP participants is strategically guided by the visioning and planning of future activities; the change in representation is pre-empted • Initial IP activities are realistically achievable to develop confidence in the IP and its participants

Appendix B: Conceptual framework
PoE theme Implementation criteria Engagement activities Communication