Co-design: from expert- to user-driven ideas in public service design

ABSTRACT

While co-design with users has evolved as a promising approach to service innovation, it remains unclear how it can be used in public service contexts. This article addresses this knowledge gap by applying a co-design framework during the ideation stage of six public service design projects. The findings provide insights into (a) recruiting and sensitizing suitable service users, (b) conditions enabling users to co-design ideas, and (c) requirements for implementation of user-driven ideas. The article contributes an approach that shifts public service design away from an expert-driven process towards enabling users as active and equal idea contributors.

KEYWORDS:

• Co-design
• service design
• user innovation
• public services

Introduction

The active involvement of users in the design and implementation of public policies and services is currently a widely-discussed topic in public management research and practice (Trischler and Charles 2019; Osborne, Radnor, and Strokosch 2016; Verschuere, Steen, and Brandsen 2018). This study focuses on co-design as an approach to collaboratively addressing complex societal problems and driving innovation in the public sector (Bason 2010; Voorberg, Bekkers, and Tummers 2015). Co-design is a specific instance of co-creation practice that allows users to become part of the design team as ‘experts of their experience’ (Sanders and Stappers 2008). It represents a shift away from design as the task of individual experts towards using the collective creativity of a team with members from different backgrounds and interests (Steen 2013). For example, the UK Design Council has adopted co-design as a practice to better understand people’s needs during the early stages of the design process.

The present study investigates the application of co-design during the ideation stage of the service design process. Close collaboration with users (e.g. consumers, customers, citizens) during this stage is considered particularly important because it concerns the translation of user knowledge into new service ideas (Patrício and Fisk 2013). In fact, there is increasing evidence that co-design with users can lead to ideas which are more innovative and which better address user needs (Steen, Manschot, and De Koning 2011; Chang and Taylor 2016; Trischler et al. 2018). However, most studies merely report on the benefits of co-design (outcome focus), while providing little guidance on how co-design activities should be conducted in public sector contexts. This is despite studies suggesting that the effectiveness of co-design is dependent on its process, including which users are involved and how their involvement is facilitated (Trischler, Kristensson, and Scott 2018; Dietrich et al. 2017).

The application of co-design is especially challenging in public service design projects which often deal with sensitive or less engaging topics (e.g. health services, drug and alcohol education, gender equality, etc.) (Engström and Elg 2015), or vulnerable user groups (e.g. children, marginalized user groups, etc.) (Dietrich et al. 2017; Durl, Trischler, and Dietrich 2017). Users in these contexts may either be reluctant to participate or may not perceive themselves as suitable contributors to service ideas. Yet, as emphasized by recent studies, it is important that groups at risk of marginalisation and vulnerability are considered during decision-making processes (Eriksson 2019). This raises the question of how the early stages of the service design process can be opened up to involve relevant users as active and equal contributors?

The present study addresses this question by reporting on insights gained from six public service design projects that actively involved representative users during the ideation stage. These projects applied the six-step co-design framework introduced by Dietrich et al. (2017). In so doing, the requirements, challenges, and benefits of applying co-design could be both identified and cross-compared, along the six steps. The findings of this study contribute a comprehensive understanding of what it takes to involve users in public service design initiatives. This includes a refined framework that specifies, (a) challenges related to recruiting and sensitizing suitable users, (b) activities that need to be undertaken to enable users to contribute their unique knowledge to new service ideas, and (c) requirements for implementing user-driven ideas.

The remainder of this article is structured as follows. Next, the literature on service design and co-design is examined. This is followed by a description of the methodology and the cases underpinning this study. Insights from the cases are then presented and incorporated into a co-design framework for public service design. The article concludes by discussing implications and outlining directions for future research.

Public service design

Service design is an exploratory process that aims to create new kinds of value relation between actors within a socio-material configuration (Kimbell 2011). This definition draws on a key premise of the service-dominant logic; i.e. that value is ‘always uniquely and phenomenologically determined by the beneficiary’ (Vargo 2008, 7). However, public value creation relies on a politically-mediated expression of collectively-determined preferences; i.e. what the citizenry determines to be valuable (Moore 1995) and what the stakeholders involved in implementation see as valuable, jointly having the potential to deliver ‘higher order’ outcomes or benefits (O’Flynn 2007). These extensions highlight the importance of considering the multi-actor nature of value co-creation activities, something which is also key to the recent service ecosystems lens of the service-dominant logic (Vargo 2016).

It is beyond the scope of this article to disentangle the discussions surrounding the value concept in the public sector, and the related applicability of the service-dominant logic (for recent surveys on this topic, see Eriksson 2019; Trischler and Charles 2019). However, in terms of being important for the conceptualization of public service design, the service-dominant logic implies that it is not possible to design services as pre-defined outputs. This is because value is not created until the user integrates and applies the resources of the service provider to other resources in her/his own context (Vargo 2008). The outcome of the service design process is thus best described as a value proposition, consisting of a configuration of resources that users and other actors can transform into value during use (Patrício et al. 2011).

In addition, the service-dominant logic nests both intangible services and tangible goods into an overarching conceptualization of service (Ordanini and Parasuraman 2011). In line with this conceptualization, service design and innovation do not focus on inventing new features, but on realizing or supporting changes in how value is co-created among multiple actors (Michel, Brown, and Gallan 2008; Snyder et al. 2016). This includes changes on the user level (e.g. new ways of thinking, participation, or new capabilities for creating and realizing value) (Michel, Brown, and Gallan 2008), and changes to the wider value constellation (e.g. new roles and influence on the part of politicians, lobby groups, the media, and the broader citizenry) (Trischler and Charles 2019). In this perspective, two aspects seem fundamental to public service design: Firstly, understanding possible future use situations by gaining insights into users’ experiences and needs (Steen, Manschot, and Nicole 2011; Wetter-Edman et al. 2014) and secondly, coordinating resources across multiple actors, including a clear understanding of what it takes to implement a user-driven idea (Svensson and Hartmann 2018). Otherwise, while potentially effective on the end-user level, innovations might not gain the approval or support of those responsible for implementation.

The ideation stage, which the present study focuses on, concerns the transformation of user needs into new service ideas (Patrício and Fisk 2013). This stage is typically ambiguous and chaotic in nature because it requires the framing of, and iteration between, problem and solution spaces (Dorst and Cross 2001). The ambiguous nature of the early service design process stages particularly applies to public services, which are heavily influenced by context-related factors that foster or hinder transformative processes (Dudau, Kominis, and Szocs 2018; Eriksson 2019; Skålén, Aal, and Edvardsson 2015). An iterative and collaborative approach to public service design enables the exploration of problem-conditions-solution combinations across the complex and multilateral interactions of the service system (Trischler and Charles 2019). This collaborative exploration sets the starting point for the generation of new service ideas, ideas which are then conceptualized, prototyped, and finally implemented (Patrício, Gustafsson, and Fisk 2018).

Design-led approaches to innovation

There is increasing interest in the contributions that design-led approaches can make to innovation and creative problem-solving (Brown and Wyatt 2010; Liedtka 2015; Micheli et al. 2019). This field of research is referred to as design thinking and investigates the adoption of design tools, methods, and practices used for innovation purposes (Micheli et al. 2019). While there are numerous frameworks and processes available that describe how design thinking is applied in practice, a number of communalities are noticeable.

Firstly, a fundamental element of design thinking is human-centered design, which is based on the assumption that design solutions have no significant meaning unless they become part of the living experience of people (Krippendorff 2006). The early design stages are thus concerned with applying deep user research, stakeholder engagement, sense-making tools, and empathy design in order to explore the needs of the underlying user (Kouprie and Visser 2009; Liedtka 2015). Many methods are available that help designers in this exploration work, including personas (Grudin and Pruitt 2002), observational methods (Arvola et al. 2012), design probes (Mattelmäki 2008), and ethnography methods (Mariampolski 1999). For example, Stickdorn and Schneider (2010) offer a comprehensive overview of the processes, methods and tools used in service design thinking. Likewise, IDEO has developed a field guide, that includes a human-centred design toolkit, for public sector organizations interested in adapting design thinking to their problem-solving activities.

Secondly, design thinking typically uses a team-based approach involving members with different backgrounds and kinds of expertise. Using multidisciplinary teams has been argued to be an effective way of mitigating a number of cognitive biases during the idea generation process, e.g. projection bias, egocentric empathy gaps, focusing illusions and hot/cold gaps, because it exposes decision-makers to various perspectives (Liedtka 2015). These teams are typically and uniquely assembled on the basis of the specific requirements of the underlying design task, thus differing from work teams which collaborate within an organization on a continuous basis (Cohen and Bailey 1997). Accordingly, these teams also use unique formats, e.g. the scrum methodology, which includes development times divided up into short work cadences or so-called design sprints. For example, Tate et al. (2018) show how innovation sprints can be used during the fuzzy front end of open digital service innovation projects in the public sector. Their study highlights that the methodology and the tools applied during the sprint cycles need to be purpose-built for the underlying context in order to successfully consider related barriers and opportunities.

The present study focuses on co-design applied to the ideation stage of public service design projects dealing with sensitive topics and vulnerable user groups. In doing so, our focus is on enabling and actively involving those affected by the design, a key principle of participatory design (Ehn 2008). While co-design shares with the design thinking literature human-centered design as a core feature, it specifically builds on the assumption that users should have an ‘influence and room for initiative in roles where they provide expertise and participate […] in the early design phases’ (Sanders and Stappers 2008, 5). In contrast, design thinking does not necessarily entail direct user involvement; instead, empathy is commonly referred to as the core of human-centeredness (Micheli et al. 2019). As we discuss next, co-design can bring a number of important benefits to both the innovating organization and the users involved.

Co-design as an innovation driver

Co-design is closely connected with the tradition of participatory design, and the argument that those affected by a design should have a say in its design process (Ehn 2008; Holmlid 2009). This is a response to the critique that designers or developers rarely, if ever, meet any actual users (Holmlid 2009). Co-design uses a collaborative team approach that allows non-designers to become equal members of the design team (Sanders and Stappers 2008). During the process, design tools, e.g. games, visualizations, and probes, are used to engage all the participants and to facilitate a ‘joint inquiry and imagination’ whereby ‘problem and solution coevolve’ (Steen 2013, 18). In so doing, participants who are not normally involved in design activities can directly contribute their knowledge and fresh perspectives to exploring problems and possible solutions.

Co-design has been shown to be especially effective in the exploration of users’ unique knowledge of service usage and latent needs (Steen, Manschot, and Nicole 2011; Trischler and Charles 2019). It helps to overcome the ‘stickiness’ of user knowledge; i.e. user needs cannot be readily articulated and transferred using conventional research techniques, e.g. focus groups, interviews, or surveys (Edvardsson et al. 2013; von Hippel 1994). In addition, co-design also helps to avoid an over-reliance on empathy by means of engaging those who have experienced certain situations first-hand (Vink and Oertzen 2018). Vink and Oertzen (2018) show, in a healthcare context, that using empathy tools alone can risk insights into user experiences being clouded by the interpreter’s own role, identity, and experiences.

Co-design has also been argued to be effective in ‘infrastructuring’ socio-material assemblies where user-driven change can take place (Ehn 2008; Bjögvinsson, Ehn, and Hillgren 2012; Manzini and Rizzo 2011). These studies suggest that collaborative design projects may help to empower individual participants, or the broader community, to ‘design after design’ and to drive transformational processes. These developments point to the possibility that co-design can produce effects reaching beyond immediate design project outcomes, e.g. democratizing social innovation and supporting emergent solutions (Bason 2010; Voorberg, Bekkers, and Tummers 2015; van Eijk, Steen, and Verschuere 2017). Through their active involvement, actors can acquire the skills and confidence to provide input regarding matters of public importance (Jo and Nabatchi 2019; Bingham, Nabatchi, and O’Leary 2005).

While the benefits of co-design are well evidenced, relatively few insights are available into how this collaborative approach can be applied to public sector contexts. This is problematic because, in practice, users are rarely involved in public service design initiatives (Engström and Elg 2015). In an attempt to address this knowledge gap, Dietrich et al. (2017) compare conventional user innovation processes with those involving vulnerable consumer groups. The result is a six-step framework, which includes the stages of resourcing, planning, recruitment, sensitizing, facilitation, and evaluation. The authors argue that the time and costs relating to planning and conducting co-design activities can increase considerably when users do not perceive themselves to be suitable participants, or are not involved in the underlying topic. As a consequence, Dietrich et al. (2017) suggest sensitizing as a key step in preparing and building trust with the user.

However, the framework developed by Dietrich et al. (2017) is limited to one specific study setting, and to a narrowly defined sample. In addition, the users did not need to be recruited per se, instead being selected from the participating schools. Other studies show that public sector organizations can face considerable challenges, or lack knowledge of incentivizing users to participate in co-design activities (Hurley, Trischler, and Dietrich 2018). Thus, a deeper understanding is needed of what the requirements, challenges, and benefits of applying co-design to public service design are. By building on Dietrich’s et al. (2017) framework, this study cross-compares the insights derived from six public service design projects to which co-design was applied during the idea generation phase. This generates a more comprehensive understanding of how the early stages of the service design process can be opened up to involve relevant users as active and equal contributors. The research method and study settings used in this investigation are detailed next.

Research method

This study employs a case study method, which includes the collection of field data from six public service design projects. A case study method recognizes the complexity and dynamic nature of a real-life phenomenon, in the current context concerning the ideation of public service solutions (Voss, Tsikriktsis, and Frohlich 2002). We used multiple cases to build a stronger evidence base and to allow for the cross-comparison of findings (Miles and Huberman 1994; Yin 2009). Table 1 provides an overview of the individual projects and the process stages, including a description of the applied design approaches and the participants involved.

Table 1. Overview of the cases, the co-design process, and the participants.

Project & Aims	Resourcing & Planning	Recruitment	Sensitization	Facilitation
Project A: Alcohol education program development for parents of high-school students in Queensland, Australia.	A systematic literature review (SLR) was undertaken to understand best practices concerning alcohol education for parents. A total of 13 program activities were sourced and featured as activity cards during the co-design sessions.	Parents were recruited via a network contact’s restaurant Facebook page. A total of 24 participants (15 female; aged 38–68) who had at least one child aged between 12 and 16 were recruited.	A brief introduction was provided to the underlying topic and the design task at hand. Researchers assigned people to groups, but then gave them time to individually reflect on each of the activities and to categorize these into likes and dislikes. Next, a moderated group discussion allowed participants to share their thoughts with others.	Two co-design sessions (90–120 minutes) were conducted as part of the ideation stage of program development. Participants collaborated in teams of 4–5 and 3 facilitators ensured that teams stayed on-task. Groups received butcher paper, pens, coloured markers, and post-it notes. The teams pitched their developed programmes to the research team and to other participants.
Project B: Virtual reality (VR) alcohol education content development for high-school students in Queensland, Australia.	An SLR was undertaken to understand current VR use in alcohol programs. None of the identified programs were deemed suitable for inclusion in the co-design session. Different sensitizing stimuli were planned for the co-design sessions.	A convenience sampling technique was used from existing networks. A total of 21 Australian high-school students (9 female; aged 14–16) were recruited. Sixteen indicated that they had already tried/consumed alcohol previously.	Students were asked to fill in a brief survey containing questions on general demographic data, and their perceived peer pressure susceptibility. Following a brief introduction, each group received different sensitizing materials; 1) role play, 2) video exposure, and 3) an online simulation.	One co-design session (60 minutes) was conducted as part of virtual reality script development. Participants collaborated in teams of 5–6. Groups received butcher paper, pens, stickers, coloured markers, and post-it notes. The research team members ensured that the groups remained on-task.
Project C: Social media campaign content development for addressing alcohol-related violence in Brisbane, Australia.	Grey literature was searched through to identify programs that had been implemented previously in Australia to reduce alcohol-related violence. A total of 15 activity cards were created and used during the co-design sessions.	A total of 30 participants were recruited online via the university’s media channels and the commissioning body’s Facebook and Twitter channels. Participants were also recruited through university lectures, the research team’s personal networks and a local bar.	Participants were provided with activity cards summarising existing campaigns and were then asked to individually sort them into piles containing likes and dislikes. A group discussion followed to gain insights into campaign effectiveness, target audience, improvements and what one campaign did well that another did not. Lastly, the participants selected the most- and least-liked campaign.	A total of four co-design sessions were conducted (90 minutes) for the ideation stage of a social media campaign addressing the prevalence of alcohol-related violence. Participants were split into teams of 3–4 and provided with pens, coloured markers, post it notes and butcher paper. Participants were encouraged to express their ideas using words and/or drawings.
Project D: Development of a campaign to reduce household food waste under Redland City Council (RCC), Australia.	An SLR and grey literature search was undertaken to identify previous food waste programmes, including those implemented in Australia. Nine food waste programmes were identified and incorporated into co-design planning.	A convenience sampling technique was used to recruit participants via an invitation email. A total of 21 participants were recruited from a large Australian university and among RCC employees-	Three sensitising activities were undertaken to prepare the participants: 1) 11 words were tested in a word association test. 2) A group discussion guided by 6 semi-structured questions. 3) A discussion about past campaigns to identify what participants liked (and did not like).	A total of two co-design sessions were conducted (45–60 minutes) to develop new ideas for a food waste reduction campaign. Participants were grouped into teams of 3. Teams received pens, coloured markers, post it notes and butcher paper. The teams were asked to pitch their final ideas to the research team.
Project E: Co-design of a weight management program in Queensland, Australia.	Grey literature was searched through to identify programs that had been implemented previously in Australia regarding weight management.	Project partner NGOs recruited overweight and obese respondents in local shopping centres offering them a free health check followed by an invite to a co-design session. A total of 26 participants attended the co-design sessions.	Three sensitising activities were undertaken to prepare the participants: 1) 10 words were tested in a word association test. 2) A 10-sentence reflection task was completed by each participant. 3) A discussion about past campaigns to identify what participants liked (and did not like).	Eight co-design sessions (120 minutes) were conducted. Small groups (1–5) were challenged to develop an idea or a campaign they felt would engage and motivate them as regards losing weight. To spark their creativity, various materials were provided (pens, pictures, glue, scissors, and large blank sheets of paper). Teams were asked to explain their final ideas to the research team.
Project F: Development of a train-the-trainer program to reduce dog and koala interactions under Redland City Council, Queensland Australia.	An SLR was undertaken to identify interventions aimed at reducing wildlife and domestic pet interactions. Expert interviews (n = 14) were additionally used to gain further insights.	A survey was distributed via a council’s media channels (email, social media, and newsletter) and a university’s email list. A total of 635 dog owners responded. Invites to participate in co-designing were sent to participants reporting dog-koala interactions. A total of 41 dog owners attended the co-design sessions.	Participants individually reviewed past campaigns to encourage and reflect on how dog-koala interactions could be reduced. Small groups were formed to further discuss the effectiveness of existing campaigns.	Three co-design sessions (90 minutes) were conducted with dog owners. Teams of 3–4 were formed and challenged to design a programme they felt would engage and motivate them as regards minimising dog-koala interactions. Using materials provided (pens, pictures, glue, scissors, and large blank sheets of paper), teams developed a campaign and were asked to present their final ideas to the research team.

Our study did not employ a grounded theory approach, instead building on the six-step framework proposed by Dietrich at al. (2017). Doing so provided us with a basis for identifying and comparing the requirements, challenges, and benefits of applying co-design across the cases. The cross-comparison of data ultimately led to the development of a refined co-design framework for public service design. This integrative approach of combining both the relevant theories and the insights from application generates a more comprehensive basis for the conceptual development of co-design (Mattelmäki and Visser 2011).

Data analysis

The following data was collected from each of the six cases:

1. Audio recordings of the in-group discussions during the co-design sessions to learn about how the ideas evolved within the groups, the experiences shared by the participants, along with an understanding of how the participants take part and contribute their knowledge.

2. Photos and video recordings of the ideas generated as a result of the co-design sessions in order to understand the ideas, and the underlying needs driving them, from the users’ points of view.

3. Field notes taken independently by each team member to document any peculiarities and challenges noted during the different stages of the projects.

Utilization of these different data collection methods supports the cross-verification of the findings (i.e. triangulation of methods), while the application of methods across the cases increases the validity of the findings (i.e. triangulation of sources) (Patton 1999).

Each project included a team of three to six researchers with backgrounds in service design, marketing, psychology and social innovation. Using the six-step framework of Dietrich et al. (2017) as a basis, each researcher was asked to reflect on each step and to document any peculiarities and challenges observed during the operationalization of the framework. All the researchers were professionally trained in using qualitative research methods, including observational techniques. Within this team, three researchers, who are also the authors of the current paper, analysed the data and developed the framework resulting from the analysis. As such, the same researchers were involved in collecting and analysing the data, and in developing the framework. In order to reduce selective perception and blind interpretive bias, the data was analysed by two researchers independently, and then compared. Any discrepancies that occurred during the coding of the data and the development of the themes were discussed with a third researcher in order to resolve as many disagreements as possible and to arrive at a final version of the co-design framework. An overview of the data analysis procedure, including exemplary evidence, is provided in Table 2.

Table 2. Overview of exemplary data and data analysis procedure.

Raw data example (Project B)			Raw data example (Project C)			Category code	Theme	Framework development
A literature/web search was conducted to identify programs that had been implemented to reduce alcohol-related violence. Eleven programs were identified of which all were expert-driven in their design. An example of a social media campaign aimed at reducing the prevalence of alcohol related violence:			A systematic literature/web search was conducted to understand how virtual reality (VR) is used in alcohol education. Online videos and professionally-developed roleplays, enacting the consequences of alcohol consumption, were used as inspiration when developing a video script for a VR alcohol education program. See below for a screenshot of an interactive video aimed at reducing adolescent drinking:			Expert-driven Lack of user-insights Far from reality Not in context Lack of technology One-sided perspective	Problem definition may be too narrow Resourcing seems to be expert-biased New design possibilities may evolve from technology	Resourcing: Need to understand the issue from a multi-actor perspective Problem space should be kept open to allow for alternatives
Card-sorting activities were prepared for sensitization by using the material from the resourcing stage. See below for examples of two cards (from a total of 15) which were used during the sensitization stage:			Videos and roleplay scripts were prepared for high-school students to both enact in class and reflect on. See below for an excerpt from a roleplay script used on female students:			Many sophisticated design methods Little direction from resourcing Difficult to plan ahead High level of uncertainty Multiple changes to planning Scenario planning	Methods and materials need to be short and adaptable Flexibility is key to planning Every project has its own challenges Unforeseen challenges require iterations	Planning: Background to the topic and the participants involved can affect the planning process Flexibility and agility help when preparing for the unknown
Recruitment using a number of channels to reach potential participants (social media channels of a university and the funding organization, direct emails to students, a media release). Very low response rates led to co-design sessions with few participants:			Recruitment by collaborating with a school to reach suitable students. The school took on the responsibility for recruiting the students. This, however, also meant that the school set the rules, leading to the challenge of facilitating one co-design session with a large number of participants within a 60-minute timeframe:
Session	Date	Total Participants	Session	Date	Total Participants
1	June 2016	6	1	September 2016	22
2	August 2016	4
Participant numbers improved substantially after recruitment was conducted collaboratively with a local restaurant’s social media site:
Session	Date	Total Participants
3	September 2016	10
4	September 2016	10
Participant comment during the group discussion following the card-sorting activity: I do kind of think about I’m angry and I’m just going to knock him out. Like, no way, like, that’s stupid, like, punches aren’t bad, you know, it’s not – it’s not like what you see in, like, a lot of, I guess, movies or media or whatever and it’s just, like, people just knocking each other about. It’s so much – it could be so much more dangerous than that and, like, I hadn’t even really thought that one punch can kill. So, it did make me think. (Male, 21)			Excerpt from a facilitator’s field notes commenting on the sensitizing activity whereby participants were asked to enact a roleplay script and to reflect on drinking scenes during which group pressure led to the overconsumption of alcohol: The roleplay got everyone excited and straight into acting out the peer pressure drinking scene (they were doing this standing up and engaging at a high level – it was difficult for me to stop them and remind them to move to the next step). Transition from roleplay to idea generation and critical reflection on peer pressure scene worked extremely well in this group because they really seemed to get the message. They started to exchange first ideas on how the roleplay script could be improved to make it more engaging for teenagers. (Group 1–4 males – Observation of roleplay activity using beer goggles which simulate drunkenness)			Raising awareness Users take ownership Building trust and confidence Engagement in topic Short but powerful Engaging activities Prepares non-designers	Sensitization triggers reflection Participants become engaged and start to identify with the topic Sensitization seems to help break down barriers Sensitization activities should be interactive and short	Sensitization: Activities should be geared towards triggering reflection and building confidence among participants Sensitization is the first step towards removing the expert from design
Excerpt from a facilitator’s field notes during the co-design session (this group showed high levels of engagement during the card-sorting activity): Group used one existing solution as starting point and developed the idea further by adding additional components. The group found consensus regarding what should be developed early in the process (within the first 5 minutes). The group members became more creative during the process and all members contributed their experiences/skills/knowledge. Little facilitation was required by the facilitators. Group cohesion and collaboration were strong – no disengagement by individuals or intra-team conflicts were evident. It seems that group members allocated specific roles to each other to work more effectively. Everyone seemed to know what to do.			Excerpt from a facilitator’s field notes during the co-design session (this group showed high levels of engagement during the card-sorting activity): Students turned to the script provided and started discussing what they liked and disliked about it. They could not identify any positive aspects despite repeatedly being asked by the facilitator. One participant’s reflection was that ‘the peer pressure’ aspect does not come across well. Within the group, one male participant took on a leadership role in discussing and developing the initial ideas on paper. He started drawing a garden party scene. The other participants provided inputs, including ideas for possible peer-pressure scenarios. Intra-team collaboration was facilitated through the discussion/leadership role taken by the one male participant. It appears that team members took on different roles during the process including questioning, scribing, providing feedback, and analyzing. (Group 2–2 females, 2 males – Observation of intra-team processes)			Experience sharing Participant empowerment Team-building challenges Collective creativity Experts as facilitators Existing solutions as inspiration Activating disengaged users Flexible facilitation approach	Experience sharing breaks barriers Enabling and supporting the novice teams in forming and ideating Aligning facilitation with the needs of the team	Facilitation: Facilitator carries the responsibility of enabling teams to maximize their creative collectivity Facilitating co-design teams means adjusting the facilitation approach to the needs of the diverse teams and their unique dynamics
(Group 2–3 females, 1 male – Observation of intra-team processes):			Additional notes: This project utilized three different sensitizing activities before the co-design activity (watching and reviewing a video, playing an interactive online game, and enacting a roleplay script). The teams more actively engaging in the sensitizing activity required less guidance during facilitation (e.g. Group 1), while the teams that were more passive during sensitization required a lot of encouragement from the facilitators (e.g. Group 3). This finding supports the relevance of applying engaging sensitization activities. The school did not give permission for pictures to be taken during the co-design activity.
Participants describe how their developed idea differs from the expert-driven ones, which were discussed during the sensitization stage: I like the idea of taking a more positive approach and saying, ‘Be logical. Be smart.’ It doesn’t have the negative – this doesn’t have the negative connotation behind most of the other messages. Saying to young adults ‘don’t drink at all’. That’s not going to work. So, I think, yeah, telling them to just, you know, make wise choices and to think about what they do is a better approach at times.” Idea ‘Don’t be a porkchop’A humorous social media campaign which depicts a situation during a night out when Shazza and Bazza are drinking heavily while Dazza is drinking responsibly. Bazza starts acting aggressively when someone accidently bumps into Shazza. Dazza intervenes and says ‘Don’t be a porkchop Bazza, don’t be that guy mate’. Dazza gets Bazza and Shazza some water. Shazza, Bazza and Dazza move to the dance floor and continue having a fun night.			Participants describe how their developed idea differs from the expert-driven ones, which were discussed during the sensitization stage: The content and the storyline were pretty good but the only fail, it was way too, like [pauses], formal, that’s the word it was like someone in their good twenties, thirties wrote the script and they didn’t really have a connection. It wasn’t like they had a connection with our generation so it made it [the script] very bland … like a lot of things, like, I bet ya half the class doesn’t even know what introvert means. Idea ‘House party’A comprehensive guideline regarding how a ‘real’ house party looks. Includes a number of recommendations for a peer pressure scene that is typical of a house party, including atmosphere features, specific background actions, a heterogeneous crowd in terms of age and gender, and the language used by the actors.			Relatable experiences Aligned with the user’s life Close to context Little concern regarding feasibility Improvements to existing solutions New perspectives and ideas regarding problems Too different means rejection Reflection on the original problem	Users contribute context-related insights Ideas generated seem to be more aligned to the users’ needs Users can have very different views of the same problem Ideas are not market-ready solutions Stakeholder involvement may increase support for implementation	Reflection and building for change: Moving user-driven ideas forward requires reflection and open dialogue with those responsible for implementation Implementing user-driven ideas means rethinking how service design was done previously

Findings

The fuzzy and iterative front-end of resourcing, planning and recruitment

Preparation of the co-design activities involved three steps, i.e. resourcing, planning, and recruiting. During the resourcing step, the research team conducted both systematic literature reviews and expert interviews to gain an initial impression of the underlying problem, or the task to be addressed. The insights gained from resourcing informed the planning of the projects, which included specification of the design task and aims, the study setting, the design approach, and communication of the design activities. For example, the systematic literature review provided an overview of existing service solutions related to the topic in question. These insights were summarized on cards and used during the facilitation of co-design. Finally, the recruitment step involved the systematic identification, screening and recruitment of suitable participants based on the underlying problem to be addressed.

Across all the projects, the stages of resourcing, planning and recruitment were highly iterative. One main driver of these iterations was the requirement of exploring and understanding the underlying issue from a multi-actor perspective. For example, across the projects, the initially-framed problem space, as an outcome of the resourcing stage, did not consider the perspectives of the respective end-user because the findings from the systematic literature reviews and interviews only reflected expert-driven insights. In addition, recruitment also required numerous meetings with various stakeholders to ensure their ongoing support within the projects. For example, in Project A, close collaboration with local businesses (e.g. a food and wine bar), public institutions (e.g. schools), and a municipality worker was required to gain access to the right participants. While the engagement of, and close collaboration with, key stakeholders added to the length and the resources required during the planning and recruitment process, gaining their commitment was important because it signalled to the participants that their participation and contribution was valued by the respective public service sector organization.

The iterative nature of planning and resourcing was also visible in the adjustments that were needed, which varied with the participants involved. For example, the participating schools (Project B) required the co-design sessions to be completed in just 60 minutes to avoid students missing classes or becoming fatigued. This added to the challenge of developing an approach that enabled meaningful facilitation within a short timeframe. Subsequently, more sophisticated approaches, e.g. design games or collaborative service mapping activities, had to be replaced by simpler tools such as role play, card sorting activities, and idea templates. In other projects (e.g. Project F), the stakeholders responsible for implementing the ultimate solution were ‘deadlocked’ in a predetermined solution space, which required the research team to slowly move them towards a more open mindset, creating awareness of alternative possibilities and ideas.

Based on the findings, we propose that planning and recruitment are iterative in nature. The participants’ backgrounds can affect the planning process; in turn, the type of task, approach or setting can change the requirements of participant recruitment. In addition, the three steps of resourcing, planning, and recruiting are also suggested to be the ‘fuzzy front-end’ owing to the multiple challenges that need to be overcome during this phase. These challenges seem particularly applicable to public sector contexts addressing sensitive topics, e.g. drug abuse, violence, inequality, poverty, racism, and/or target participants who do not feel any need to change or who do not feel they can contribute to service solutions. The initial resourcing stage should not be seen as an attempt to reduce the fuzziness of the front-end because it might lead to a problem space that is too narrowly defined to start with. Instead, a problem space that is open to alternative solution spaces, as well as extensive networking and the gaining of support from key stakeholders that are well connected within their community, seems essential to the setting up of co-design activities.

Sensitizing in order to prepare

The sensitization step, aimed at engaging potential participants and triggering reflection on the underlying topic prior to co-design facilitation through activities, e.g. playful excises, gameplay, short videos, or thought-provoking questions. Ideally, these activities should be applied during a designated sensitization session prior to the co-design facilitation step to allow the participants, who are typically not involved in the design process, to reflect and prepare themselves. For example, in Project B, the research team introduced both students and responsible school officials to different alcohol education activities (in the form of online games) one week before the actual co-design session. As a consequence, the participants were more prepared, something which helped to overcome the problem of the short duration of the co-design session. The use of a designated sensitization exercise also helped in building trust between the research team and the participants, as well as the school representatives, which was important given the sensitive nature of the underlying topic (e.g. underage binge drinking, which is an illegal activity).

However, in some cases, the same group of participants could not be accessed multiple times (e.g. parents in Project A), which meant that a designated sensitizing session was not feasible. Instead, ‘take home’ activities in the form of digitalized sensitizing packages, alongside thought-provoking questions on the underlying topic, were prepared for any potential participants to reflect on as a part of their preparation for the co-design sessions. These activities formed part of the recruitment process and were distributed via a designated website and via social media platforms. In addition, sensitization was integrated into the beginning of co-design facilitation as an ‘ice-breaking’ activity. During this activity, the participants were asked to share their experiences of the underlying topic in small groups of four to eight. In each group, a researcher participated as a discussant, whose focus was engaging and encouraging insights from all the participants.

The sensitization step was found to be important for a number of reasons. Firstly, it formed a basis for reflecting on topics that the respective users would not readily identify themselves with otherwise, or proactively engage in. Secondly, it triggered an awareness and an interest in the participants regarding the roles they play and the contributions they can make to public service design initiatives. Thirdly, it gave the participants, who may not originally have perceived themselves to be suitable candidates for co-design, the confidence to actively participate and contribute as experts in their own experiences. Sensitization is thus proposed as a key preparation step in the co-designing of user-driven public service solutions.

Facilitating co-design teams

A team approach was used in all the projects, with the team size ranging between three and six participants. Across the six projects, three different facilitation approaches were applied and evaluated: In the first approach (used in Projects A and B), the researchers acted as facilitators, tasked with leading the teams through the various phases of the co-design session. In the second approach (used in Project C), one researcher directly participated in each team as a collaborator. Finally, in the third approach (used in Projects D, E, and F), the researchers encouraged the groups to form, then serving as guides beside for team members taking the initiative to lead the team through the design process.

The third facilitation approach seemed most suitable because it enabled the participating researcher to proactively foster team bonding and cohesion and to thus counteract intra-team processes, e.g. the lack of a team identity, process domination by individuals, or the disengagement of team members. In addition, this approach also changed the role of the facilitator, from someone leading the team to someone empowering and encouraging it to develop solutions that can push beyond what it initially thought possible. In contrast, the overly leading stance adopted by the facilitators led to a lower level of engagement in the participants, entailing that the atmosphere was more comparable with focus groups than co-design teams, where all the participants contributed actively and equally.

A number of different co-design tools were employed during the co-design sessions. For example, during the first facilitation phase, and closely linked to the sensitization step, cards were used in all the projects apart from Project B. The cards depicted existing program components sourced by the research team during the resourcing stage as part of the systematic literature review (see Figure 1 for an example of cards used in Projects A, C and D). These cards provided a useful basis for discussing the key features, the likes, and the dislikes concerning currently existing service solutions. After the group discussion, the teams were asked to collaboratively develop their own overall program design ideas on pre-drafted idea templates, either by building on currently existing solutions (e.g. using the cards) or by developing completely new ideas. The co-design session concluded with each team presenting its design idea to all the participants and facilitators. This concluding phase also served a celebratory purpose whereby everyone’s contribution was recognized and the facilitators praised and thanked each team for their input.

Figure 1. Example of cards used in Projects A, C and D.

Interestingly, no team lacked the skills to put new ideas on paper. For example, one team in Project F drafted an ‘alert App’ for wildlife, including specific details of GPS tracking technology, that would alert a domestic pet owner whenever wildlife was present near his/her home. Teams in Project C sketched out ideas for possible social media campaigns on storyboards. In Projects A, B and E, a number of teams contributed completely new ideas, mostly relating to new ways of using services or better aligning existing solutions with the user’s context. The preceding sensitization step, in combination with the discussion on existing designs, seemed to have provided the participants with the knowledge and confidence required to develop their own ideas.

One important precondition for co-design was finding common ground within the team as regards which ideas to develop. In some cases, such as Projects A and B, the participating researchers played a key role in fostering team bonding (e.g. through assigning roles to group members) and in ensuring progress towards an agreed outcome (e.g. by encouraging the groups to put their ideas on paper). Nonetheless, the research team made sure to de-centre themselves as the main agents or experts, instead focusing on enabling the participants to independently collaborate on ideas addressing their specific needs. During the later phase of the co-design session, in particular, the researchers served merely as guides, leading to greater creativity and engagement among the team members. This highlights the fact that facilitating co-design is about supporting the collective creativity of the participants, who are typically untrained in design.

Reflecting and building for change

Across all the cases, the co-design participants provided insights into their everyday lives, including clear directions regarding what does and does not work as regards the current service solutions. While most of the ideas generated were incremental, with a focus on modifying existing solutions, some of the teams also developed ideas that could not be linked to existing service solutions. The high number of incremental ideas may have been a consequence of reviewing and discussing currently existing solutions during the sensitization step, or introductory step, of co-design facilitation. This points to a trade-off effect of sensitization: while it can help to prepare the participants who do not originally perceive themselves to be suitable candidates for co-design, it might also limit their ability to contribute radically new ideas. Notably, the majority of the ideas developed were not considered to be too narrowly focused on specific user needs, which may be a result of using a team approach and involving users with different needs and experiences.

The outcomes of the co-design sessions were not, in and of themselves, market-ready solutions. Rather, they provided user-driven idea inputs for the research team and service designers to build upon in consultation with other stakeholders (e.g. public authorities or policymakers), who are the ones responsible for creating and implementing the final concept. While the recruitment, sensitization and facilitation steps were aimed at generating user-driven ideas, the last two steps were more focused on analyzing and sharing the insights and the ideas generated with the key stakeholders. In particular, building for change, the last step of the process required an open dialogue between the researchers, designers, partner organizations, and front-line staff in order to assess the feasibility and realization of the ideas. This also made the back-end stages an iterative process because the research team needed to repeatedly reflect on the original design task, or problem definition, alongside the different solution spaces, in addition to providing budgets in order to assess possible ways forward. In fact, across the cases, user-generated insights led to changes to the original task definition because only then could the research team fully understand the complexity surrounding the underlying problem.

In addition, one notable challenge lay in the user-generated ideas being partly perceived as ‘too different’, ‘radical’ or ‘expensive’ by those responsible for implementation. For example, in Project C, the public managers were not actively engaged in the process, ultimately leading to the rejection of the co-designed ideas because their realization was perceived to be too risky. In contrast, in Project F, the public managers were invited to participate as ‘passive observers’ during the co-design sessions, leading to the approval of a user-ideated program that was radically different from the original design brief. It thus appears that the involvement of key stakeholders (e.g. partner organizations) throughout the co-design activity is important when it comes to gaining ongoing support for implementation. Similarly to the front-end, we propose that the back-end of the co-design activity (i.e. reflection and solution building) is iterative in nature, since various stakeholder interests needed to be considered during the development of user-generated ideas into feasible solutions.

Discussion and implications

This study investigated how relevant users can be involved as active and equal contributors to public service ideas. The investigation drew on the co-design framework of Dietrich et al. (2017) to identify requirements, challenges, and benefits that are specific to the application of co-design to public service design. In addition, the findings were compared across six cases to generate a comprehensive understanding of what it takes to involve users in public service design initiatives. We conclude the paper by discussing implications for public management theory and practice.

Based on insights generated using the six cases, we propose that a refined co-design framework is required for public service design. The proposed framework is depicted in Figure 2 and includes the following seven steps: 1) resourcing, 2) planning, 3) recruiting, 4) sensitizing, 5) facilitating, 6) reflecting, and 7) building for change. In contrast to the original framework of Dietrich et al. (2017), we propose that the front- and back-end phases of the process are iterative in nature. During these phases, the needs and interests of multiple actors need to be managed in order to gain support for running the actual co-design activity and implementing the user-driven idea. The close collaboration with partner organizations is particularly important during the recruitment stage in order to overcome the challenge of identifying and sourcing the relevant users. In addition, and in line with Dietrich et al. (2017), sensitization is proposed as a key step preceding co-design facilitation. Sensitization helps to prepare the participants who are typically not involved in design processes to create awareness of the underlying problem and to build trust between users and the facilitation team.

Figure 2. A co-design framework for public service design.

This study also provides insights into the requirements and challenges facing the application of co-design to public service design projects. Along the seven steps, two critical points seem to determine a project’s success or failure; namely, the fuzzy front-end of planning and recruiting, and the back-end of building a platform for change. The fuzzy front-end is widely discussed in the innovation literature and the co-design literature. However, the process is typically described in terms of consecutive steps: Once the design task has been defined, suitable users are identified and involved in the design process (Alam 2006; Steen, Manschot, and Nicole 2011; Visser et al. 2005; Donetto et al. 2015). We show that the front-end can require multiple iterations owing to unforeseen challenges related to participant recruitment, in turn leading to adjustments of the planning of the co-design activity.

The back-end is generally defined as a concluding idea evaluation stage (Dietrich et al. 2017). In contrast, we propose that the stages ‘reflection’ and ‘building for change’ are a collaborative and iterative effort aimed at conceptualizing viable solutions with key stakeholders. Our argument is in line with recent studies highlighting that close collaboration with stakeholders is required, both at the design stage and at the concept testing stage, to ensure innovation success in the public sector (Tate et al. 2018). The reflection stage extends the framework of Dietrich et al. (2017) to highlight the requirement to reevaluate the initial problem definition alongside the solution ideas developed. Since the initial problem is often defined by experts (Holmlid 2009; Vink and Oertzen 2018), it is important for the problem and solution space to be kept open throughout the co-design activity to allow critical evaluation of whether or not the interpretations made by the experts are actually reflected in the user-generated solution ideas.

Few studies are available outlining how co-design can be applied to public sector contexts, including contexts that concern the involvement of vulnerable or marginalized user groups (Hurley, Trischler, and Dietrich 2018). The present study addresses this important knowledge gap by contributing an approach that enables users to contribute actively and directly to the generation of new service ideas. A requirement for facilitating user contribution is de-centring the designer or expert as the main agent during the process. This means shifting public service design away from a traditional approach, where individual designers or experts were solely responsible for design, towards an opening up of the process to other actors (Kimbell 2012). As demonstrated in this study, this shift can be achieved, even in contexts where targeted participants may not see themselves as suitable candidates for co-design, or may be reluctant to participate owing to the sensitive nature of the underlying design topic. Opening up the public service design process is important because ‘the public ha[s] the right to be more than just participants in research, and their involvement can lead to better outcomes’ (Hickey, Richards, and Sheehy 2018, 29).

For practitioners the study provides insights on how those affected can be actively and directly involved in public service design initiatives. Reflecting on the challenges identified across the seven steps, it becomes clear why many public sector organizations still design for rather than with service users. Nonetheless, public managers should be encouraged to move beyond traditional design methods and to apply collaborative approaches enabling those affected to more actively contribute their unique knowledge and experiences, something which, as studies suggest, is key to new service success (Chang and Taylor 2016; Trischler and Charles 2019). The collaborative effort that actively involves users, firms, and public officials, among others, seems particularly important for public sector contexts because bottom-up initiatives are unlikely to be successful when stakeholder support is lacking (Trischler and Charles 2019; Crosby, ‘T Hart, and Torfing 2017). The proposed co-design framework is a first step to motivating managers and scholars into further exploring its application to the public service sector.

Limitations and future research

The study investigated how co-design can be applied in public sector contexts, leading to a seven-step co-design framework for public service design. The framework was developed on the basis of insights generated across six cases covering a broad range of contexts (sustainability, education, health, and environmental protection). Although these cases involved a number of different contextual applications with a total of 163 participants across 20 co-design sessions, the focus is limited to the idea generation stage. Based on this focus, the co-design sessions did not involve professional designers per se, instead enabling representative service users, including vulnerable and marginalized user groups, to become active and equal contributors. This is, in fact, the key purpose of participatory design practices: i.e. to ‘de-centre the designer as the main agent’ and to encourage everyday citizens to contribute their unique experiences and ideas (Ehn 2008, 94).

However, de-centring the designer does not imply that co-design activities should abandon inputs from designers. In our study, we found that, at step seven (building for change) in particular, the involvement of professional designers and other stakeholders is essential when it comes to refining, assessing, and conceptualizing user-generated ideas. Thus, and as highlighted by Michlewski (2008), the intuition, instinct and tacit knowledge that a professional designer brings to the process are important for translating ideas into feasible solutions. One important extension, thus, is research that investigates idea refinement and solution implementation and then compares this process between co-designed and expert-generated ideas.

In addition, this study also followed the same steps across all the cases by adopting the framework of Dietrich et al. (2017). While this enabled the systematic comparison of findings across the six cases, it did not consider the effects that could ensue from changes to the process. For example, we identified unique trade-offs when running sensitization activities prior to the co-design activity. One interesting avenue for future research may be investigating the effects that follow from variations in the framework, e.g. the relationship between sensitizing the participants and idea novelty.

Finally, this study did not systematically evaluate the ideas generated, as has been the case in previous studies (Trischler and Charles 2019; Witell et al. 2011). It is beyond the scope of this study to evaluate the effectiveness of the co-designed ideas, particularly when considering that these typically require a long-term perspective (Bason 2010), and a multi-dimensional approach (Dudau, Kominis, and Szocs 2018). Feedback from the stakeholders involved in the project indicated that the co-design approach not only generated new ideas but also led to an awareness of how user needs and experiences could be more effectively incorporated into future service improvement efforts. Nonetheless, future research should explore ways that would allow the systematic evaluation of ideas generated by co-design teams. In the public sector, this could include criteria that are different to those commonly used in the innovation literature, e.g. transformational value (Blocker and Barrios 2015), wellbeing (Anderson et al. 2013), or social change (Cottam and Leadbeater 2004).

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Notes on contributors

Jakob Trischler

Jakob Trischler has a strong interest in service design and studying the consequences of user involvement in innovation processes. Additional research interests include the public sector and investigating public policy from a service ecosystems lens. A number of research projects are based on collaborations with colleagues at Griffith University, Southern Cross University, and the Management Center Innsbruck. His recent publications appear in Journal of Service Research, Journal of Public Policy and Marketing, and Journal of Service Management.

Timo Dietrich

Timo Dietrich is a behavior change expert. Timo has developed, delivered, and evaluated a range of behavior change programs spanning across social, environmental and health-related contexts. He has published in more than 50 peer-reviewed books, journal papers, and conference papers.

Sharyn Rundle-Thiele

Sharyn Rundle-Thiele is currently working on projects delivering changes to the environment, people’s health and for the greater social good. Selected current projects include changing adolescent attitudes towards drinking alcohol (see www.blurredminds.com.au), farming practice change, preventing koala deaths (see www.leaveit.com.au), increasing healthy eating and many more. Research partners in 2018 include Defence Science and Technology Organisation, Australian Defence Force, Queensland Catholic Education Commission, Redland City Council, Department of Environment and Science and more. Sharyn’s research is published in more than 120 books, book chapters and journal papers.