Evaluating experimentation in the public sector: learning from a Brazilian innovation lab

Abstract This paper presents an initial evaluation of the GNova lab, with the aim of contributing to the policy lab literature and debate on the value generated by public sector innovation labs. GNova is a Brazilian federal government innovation lab dedicated to developing creative solutions to public policy problems through design-led experiments that involve the active participation of members of partner teams. In the context of a political transition, GNova carried out an evaluation process to assess the results of its projects. By using working principles as evaluation criteria, the process was design-led and consisted of two phases. The first was a workshop with project partners, followed by a series of in-depth interviews with participants in six selected projects. The findings were grouped into three types of effects (effects of the process, effects of products and effects of the participation). The evaluation, even though with a limited scope, confirmed the assumption that, in addition to effects from specific products delivered, the lab contributes to the development of competencies in the civil servants who participate in the process, in resonance with a public administration paradigm based on public value.


Introduction
This paper aims to contribute to the policy lab literature with the case of GNovathe Brazilian federal government's innovation lab and to the debate on evaluation and the value generated by government innovation labs. It presents an evaluative process carried out by the lab's team in the second semester of 2018 as part of a larger communication strategy motivated by an upcoming political transition in the federal government of Brazil.
Government labs emerge within the context of developing a public administration paradigm based on public value, which portrays a State that is both post-bureaucratic and post-competitive, overcoming New Public Management (NPM) shortcomings (O'Flynn 2007) and dialoguing with contemporary challenges of organizations. Drawing heavily on Moore's work on public value (1994), this paradigm is built upon collaborative network forms of governance, rejecting one size fits all solutions and adopting a learning-based approach. Here, public value is described as a "multidimensional constructa reflection of collectively expressed, politically mediated preferences consumed by the citizenrycreated not just through 'outcomes' but also through processes which may generate trust or fairness" (O'Flynn 2007). As dominant paradigms influence how public managers and civil servants make sense of their activity, a shift toward a public value model has different implications for teams and individuals, being the development of new competencies an important one. Public innovation labs can be understood as an experimental and site-specific result of this new pragmatism around the world, particularly when it comes to their search for different ways of thinking about and enacting public management practice.
However, being a recent global phenomenon, the actions of public sector innovation labs are not always understood. On the contrary, as stated by Werneck et al. (2020), it is prevalent to find "resistance, suspicions and failed expectations, since the volume of results will not always meet the expectations of managers, who think about gains of scale and short-term political return" (57). The difficulties in understanding the novel and unfamiliar work of labs lead to an even greater need to communicate the goals, methods, and resultswhich can often be intangibleto managers and society. In this sense, it is necessary to select adequate evaluation methods to communicate the public value generated effectively.
There are recent efforts being made to evaluate the work of labs, but there is still little research in the Brazilian and broader Latin American context. According to Ferreira and Botero (2020), labs in Latin America face particular complexities and challenges, such as the needs to work under budgetary constraints, to weave regional networks, and to pay attention to institutional agendas and transitions in public administration, while also being accountable to different levels of society. Additionally, a research conducted on Latin American labs by Acevedo and Dassen (2016) pointed out that a distinctive feature of these labs, in comparison to other regions, is the emphasis given to actions related to transparency and citizen participationdemands arising from democratization processes after periods of dictatorship and less focus on carrying out team impact assessments. This gap makes it even more difficult for labs to demonstrate the effectiveness of their actions and discover new management alternatives.
In this sense, this paper presents an initial evaluation attempt carried out by the GNova team after two years of existence, contributing to the debate on evaluation and value generated by government innovation labs in the Latin American context.
In terms of their contribution to public policies, Olejniczak et al. have presented three central evaluation challenges: establishing what solutions work, explaining why solutions work (or not), and transferring research findings into policy actions (2020). Here, we are looking at what solutions work and why they work (Olejniczak et al. 2020). The main objective was to identify which types of results were produced and why, mainly concerning changing mindsets of civil servants, according to GNova's goals. Additional research is still needed to confirm the findings and develop an initial baseline with indicators, taking the time to conduct a more robust impact assessment.
Our research confirmed the assumption that the lab's approach can generate positive effects related to behaviors and mindsets in civil servants (Maffei, Mortati, and Christiansen 2018). In that sense, there is a potential dialogue with the field of policy labs, policy evaluation and studies about the development of innovation competencies in the public sector to increase the government's capacity to innovate. In this article, we present the results of the evaluation itself, but also the methodological framework used.
This paper consists of five parts. The next section presents GNova's approach and provides the specific context from which this evaluation emerged, aligned with a larger strategy for a moment of political transition in the federal government. Section three presents the scope and methods adopted in the evaluative effort. It includes the evaluation criteria adopted, drawn from GNova's working principles, the evaluation phases and the sensemaking process. The fourth section summarizes the evaluation findings, grouped in the three types of effects identified: effects of the process, effects of the products, and effects of the participation in the process. Section five highlights the main conclusions from the findings, confronting them with our initial assumptions. It also indicates limitations and possibilities for further research.

Background
Initiated in 2016 by the National School of Public Administration (Enap) and the Ministry of Planning (currently the Ministry of Economy), GNova's mission is to promote innovation in the public sector to better respond to society's demands. The lab understands innovation as developing and implementing a new process, service or public policy that generates better results for the public service and public value for society. This definition includes both a significant improvement in an existing process, service, or public policy and the creation of a new approach, service, or public policy that fundamentally changes the organization and its deliveries to society.
GNova is a unit dedicated to the development of creative solutions to public issues through an experimental approach. To enhance the potential of experimentation projects, GNova is also involved in initiatives of prospecting and dissemination through specific actions, such as innovation awards, publications, seminars, workshops, tools, and training. The creation and first years of GNova were supported by MindLab, a lab situated within the Danish government from 2002 to 2018. The collaboration was made possible by an international agreement between the two countries (Danish Ministry of Foreign Affairs 2018; Brandalise, Ferrarezi, and Lemos 2018). During this time there was growing interest and a fertile ground for innovation initiatives in the Brazilian public sector (Cunha and Severo 2017).
GNova's projects are carried out in partnership with federal public institutions. That means that the partners bring specific challenges, and the civil servants responsible for each project work alongside the lab's team, which differs from traditional consultancy work. In general, the lab works on challenges from public institutions that meet the following attributes: there is a clear goal, there is openness to change by the team, there is the possibility of experimenting with new methods and technologies, there is a goal to generate behavior change, and there is also political support and commitment to the continuity and implementation of the project after working with GNova.
As GNova is located in a government school as such, its projects also focus on capacity building (through learning by doing) as a way to demonstrate other ways for working and dealing with wicked problems (Rittel and Webber 1973), such as climate change, public safety, and social inequality. The assumption is that participating in innovation projects is a way of setting a precedent for different ways of thinking and doing in government, rehearsing how things could be done otherwise (Maffei, Mortati, and Christiansen 2018). As a result, teams that participate in GNova's lab can later adopt new behaviors and mindsets in the face of the risks and opportunities envisioned and exercise collaborative practices and narratives that lead to more effective deliveries in their daily life in the public sector.
GNova uses a design-led experimental approach and incorporates methodologies according to the situation, purpose, actors involved, duration, and political conditions. Examples of methods include design ethnography (Halse et al. 2010;Metello 2018) and agile immersion . Additionally, through the collaboration with MindLab, the GNova team developed specific techniques for facilitating groups and applying design and research methods (Brandalise, Ferrarezi, and Lemos 2018).
In GNova, each project is considered an experiment in itself. Experimentation 1 is a series of actions with clear learning objectives that begin from several research sources. The participants of project teams also bring in different sets of backgrounds, interests, and motivations, which make the work process a fertile field which combines different types of information and knowledge. Through experiments, it is possible to test assumptions and pave the way to produce knowledge and valuable solutions before the implementation of policies and services (Williamson 2015).
Underlying the application of methods in a partner's project is incorporating specific working principles, as presented in Table 1. These are ongoing tentative descriptions, acknowledging their complementarity and overlapping nature. Inspired Use of relevant knowledge from both qualitative and quantitative sources to inform processes of decision making, identification of opportunities, and idea generation. 4. Focus on the effects to be achieved to change the current situation Rather than starting with a ready solution in mind, reflect on the theory of change aimed and how the deliveries of the project can contribute to strategic implications and behavior change.

Making ideas tangible through prototypes
Through the materialization of ideas in provisory forms, it is possible to both make exploratory questions and test to confirm or refute initial hypotheses about the problem and the incorporation of learning.
primarily on a design-led approach, the embodiment of such principles in the methodological process of the projects can generate doubt or discomfort, as they differ substantially from the ways that civil servants are used to working in the government, guided mainly by a bureaucratic model of risk aversion, hierarchical and siloed structures, with duties established by rules, norms, and laws to generate predictability and order (Puttick, Baeck, and Colligan 2014;Mcgann, Blomkamp, and Lewis 2018). The actual practice of these principles is what guides the capacity-building goal described above. The assumption is that teams not familiar with the everyday practice of such principles can, after going through the experience of a lab project, later adopt at least some of them in their government units (Table 1). Like other labs of such nature, GNova exists within a conflicting organizational culturethat is, the lab aims to bring in new ways of working in bureaucratic environments which are used to traditional methods of policymaking and are resistant to change. Situated within the Brazilian federal government, the lab is susceptible to political transitions every four years in the country's democratic system. Even though predictable, political transitions are far from a situation of stability and certainty. Especially for innovation labs, which are a relatively recent movement in public administration (Tõnurist, Kattel, and Lember 2017), transitions can be an intense moment of high expectations from sponsors and thus of facing "the challenge of having their methods, objectives, deliveries, and existence validated" (Werneck et al. 2020, 69). Additionally, common challenges among labs, such as "the capacity to mobilize partners, create a narrative around the lab and its activities'' (ibid., 70) tend to intensify in the context of political transitions.
Two years into the lab's existence, GNova was about to face its first political transition. In 2018 there were presidential elections, and the lab decided to gather evidence to be able to communicate to new officers what value the lab had delivered. Combined with its nascent years and a still fragile institutional legitimacy, there was a need for the lab to justify its existence and communicate the results generated by the projects developed. Therefore, in preparation for the transition, GNova's team decided to lead a coordinated effort to document its methods and projects, publish reports and participate in relevant events. The endeavor included, for example, the launch of a newer version of GNova's website with an updated portfolio of projects, the production of the Innovation in Practice (Inovação na Pr atica) book series, and the participation in the organization of the 5th Innovation Week, one of the major Latin American events on public sector innovation.
In such effort, the GNova team decided to carry out an evaluative process, which is the focus of the present article. It aimed to investigate the direct results of projects and the effects of the employment of lab principles and methodologies by partner teams and organizations.
In addition to providing decision-makers with information, the evaluation itself can be critical for labs to obtain legitimacy and consolidate themselves because it allows to determine if the lab delivers the value it claims to offer; to make adjustments in management to adapt to the constraints faced; strengthen or achieve political legitimacy; to make plans to government transition and to be prepared for the changes in political orientation; to make evidence-based decisions; to build a narrative for its team; and to convince managers of the importance of this kind of work in government (Werneck et al. 2020).

Methods
At the outset of the lab's evaluation in August 2018, there were limited resources and the absence of an evaluation framework that fitted the specificities of public sector innovation labs. Moreover, according to Osorio et al. (2019), and based on the literature review, there was no clear understanding of which type of results are the ones to determine whether a lab is successful or not. Given the novelty of the initiative, GNova's evaluation was carried out applying a design-led approach and acknowledging the limitations of being an inaugural effort with few units of analysis.
There was a project team consisting of three people from the labin which two were lab members, and one was a partner consultantdedicated to the conceptualization and execution of the evaluation. A self-conducted assessment meant the evaluation team's direct involvement in the lab's activities could compromise the objectivity of the results. Specific attention was paid to possible conflict of interests. Notwithstanding these limitations, the internal evaluation was adequate because its primary goal was to improve the lab's future interventions. According to Vedung, "there is no need for an external specialist to do it unless there are technical areas of high complexity that demand a scientific expertise" (2014, 88, translation by authors).
There were two evaluation phases: a one-day workshop and a series of interviews. Both are described below. The goal of the evaluation was to assess how GNova's working principles and methods impacted the partner teams in producing innovations and improving the quality of public policies and services delivered to organizations and society in general.
Our underlying assumption informed by GNova's principles was that lab projects go beyond the creation of solutions to the challenges brought, contributing to the development of competencies in the partner's team through their active participation in the process. We decided to use effects instead of results or impact because of the difficulty to isolate variables in such exploratory research, weaving the direct link between projects and consequences. We understand effects as "any behavior or event that can reasonably be said to have been influenced by some aspect of the program or project" (Bond apud Cohen and Franco 2004, 312). Positive effects were understood as value (desired gains or relief from existing pain) or public value, which means offering effective responses to collective needs or demands that are politically desired, which generate benefits for the common good (Moore 2002). The perspective that lab effects can be caused both by the products and by the methodological process since participation in the activities themselves can generate positive effects on the project members, is represented in Figure 1.

Phase 1: workshop
The first phase of the evaluation took place in August 2018, at the lab's anniversary event, and consisted of a workshop to which all project partner teams throughout GNova's history were invited to attend. The purpose was to collect personal reactions from the participation in projects and create a shared sense of belonging as part of GNova's journey. The workshop also served to communicate about the upcoming interviews and helped to elaborate the interview guide.
In the session, participants received a kit with pictures of their projects, stickers and speech bubbles with incomplete sentences, with blank fields regarding the three types of effects previously defined, namely process, products, and participation. Following this, participants collectively built GNova's two-year timeline 3 by incorporating their project's details, identifying milestones, and adding their reactions to the prompts. After the timeline's completion, they shared their experience with the group. The presentations were in a dialogical format, in which people asked questions, shared similarities and differences between projects' circumstances, scope, methods, and results. The materiality of the timeline enabled people to be very specific, pointing to details in the pictures when sharing their favorite moments of projects, for example, or asking clarifying questions. At the end of the event, partners shared their reflections in a space open to the public. 4

Phase 2: project interviews
The second part of the evaluation focused on gaining a deeper perspective and consisted of a series of individual interviews. In the weeks following the workshop, GNova's evaluation team conducted 12 in-depth interviews with partners who had integrated lab projects as active participants of the process. The interviews lasted approximately one hour. The questions followed a semi-structured format, which included specific questions about each type of effect and more open questions about the general experience with GNova.
Out of all the projects displayed in GNova's timelinewhich were around 20, in various formats and timeframes -, we selected six to be further evaluated in the interviews ( Table 2). The selection of such projects was guided by the diversity of topics, partner institutions, 5 and timeframe. They were representative examples of GNova's approach. We selected completed projects and that covered different moments of GNova's history. The evaluation team interviewed two members of each team, one of whom was in a leadership position.
We present a brief description of each selected project below, highlighting the name of the partner institution, duration of the project, context, project goal, the methodology adopted, products, and main results generated (in terms of products).
3.2.1. Central Bank of Brazilhow can we expand the range of possibilities of actions to promote financial citizenship? To understand new ways to promote financial citizenship in the country, the Central Bank of Brazil (Bacen) project was carried out from November 2016 to April 2017, and developed with a design thinking approach. The work started with exploratory research about the subject area, followed by applying design ethnography 6 methods to investigate the following problem: a significant portion of the Brazilian population does not adopt financial planning. The fieldwork included interviews with citizens from five cities and resulted in a set of insights that pointed out possibilities for action in the face of the problem. There were a number of ideas prototyped, including an advertising campaign to raise awareness about the risks of indebtedness by name loans, a gamified savings-investment card to encourage people to meet savings targets in exchange for bonuses, and a competition to induce the creation of financial products suitable for the low-income population.

Ministry of healthwhat are the users' perceptions, needs, and expectations regarding the booking of public medical appointments?
In November 2016, the Ministry of Health was in the initial stages of developing a solution for booking medical appointments online, to reduce long lines and wait times at public health centers in the country. In order to learn from previous experiences, the project with GNova was research-focused and applied a design ethnography methodology in medical centers in four municipalities that had already developed applications (apps) for scheduling appointments. After conducting in-depth interviews with various profiles, from municipal health managers to technicians, doctors, and users, the project's product with GNova was a compilation of insights, which visualized the main findings in the form of a map. At the project's conclusion in February 2017, the Ministry continued to develop the app, incorporating the insights identified in the research into the socio-technical requirements of the service.

Ministry of economydoes SICAF facilitate or hinder commercial relations with the federal government?
The Unified Supplier Registration System (Sistema de Cadastramento Unificado de Fornecedores -SICAF), managed by the then Ministry of Planning, was created to facilitate commercial relations between suppliers of materials and services and the government. To support SICAF's improvement strategy, the project with GNova began with a design ethnography process to develop a more profound knowledge of the users' experience in using the system. The team interviewed four types of users. The interviews were conducted in two municipalities. A map of insights suggested possibilities for action that would improve system management, communication, and technology. After the project, the Ministry started to transition to an entirely digital system, following insights and recommendations from the research with GNova including the exemption of suppliers from delivering physical documents. Consequently, 1,855 registration units ceased to exist, and the 4,000 public servants were assigned to other jobs. 3.2.4. National health surveillance agencywhy does anvisa's call center receive around one thousand calls daily when the information is available on the website? In October 2017, a National Health Surveillance Agency (Agência Nacional de Vigilância Sanit aria -Anvisa) team wanted to understand why people made so many phone calls to the agency's call center instead of searching for information on the website itself. The investigation began with a survey of quantitative data on the types of users and the content of the calls, which subsequently guided a design ethnography process. The team interviewed public agency representatives and citizens in three cities. The product of the project was a series of insights with understanding about users' needs and behaviors. The project also brought specific aspects to be improved on the website, such as the presentation and choice of highlights, terms used, and the search engine.

3.2.5.
National presswhat are the needs and expectations of users in the transition from the printed official gazette to the exclusively digital one?
The Federal Official Gazette (Di ario Oficial da União -DOU), an official advertising vehicle for the government's legal and administrative acts, administered by the National Press, became exclusively digital in November 2017. To implement the transition, it was necessary to understand the needs of users and identify innovation opportunities concerning the search for official information. We applied design ethnography through in-depth interviews with public managers in areas related to information and communication, sporadic users, journalists, private information companies, and the association of those visually impaired. The insights generated by the survey were illustrated by different users' journeys, giving visibility to the details of people's experiences and expectations while interacting with official information. One of the results was creating a permanent committee for continuous improvement of the DOU, as the publication was no longer seen as a newspaper and started to be considered a strategic database. The insights generated were submitted for a hackathon of concepts and prototypes for the digital DOU.
3.2.6. Ministry of economydoes the proposed institutional arrangement ensure effective governance of the PPA??
The Pluriannual Plan (PPA) is a mid-term plan, which establishes objectives and goals to be followed by the country's government over four years. The Ministry of Planning, which was the institution responsible for managing this plan, asked for support from GNova to improve a new governance model for the PPA. The project timeline was limited, so we opted to apply an agile immersion process, 7 carried out between July and August, 2018. After the creation of the concept for a new PPA governance model, the project team designed a few diagrams (prototypes) to represent the proposition. Then they organized interviews with management experts and relevant stakeholders to present the prototypes and collect their reactions and suggestions. Showing early prototypes in the interviews was helpful because of their concreteness and the possibility of quickly gathering information that was dispersed, mitigating risks before the allocation of time and resources, and gaining legitimacy with relevant actors.

Sensemaking: organization of data and insights
Once we had concluded both the workshop and the interviews with partners, we conducted a sensemaking process to identify patterns among the findings. Combining the outcomes of both evaluation phases, we transcribed the audio recordings and organized the information into categories that facilitated the analysis. We made a session to download and share the data collected in a consistent format among the evaluation Table 3. Types of lab effects.

Effects of the process Effects of the products Effects of the participation
Effects of the use of methods and resources in planned activities to reach a specific goal.
Effects of the use of the concrete products delivered as a result of activities developed and resources employed.
Effects of active participation in the process of using methods and resources in planned activities.
team. As each person presented the data, the others took notes of the key elements and stories covered and placed them on a wall to be visually rearranged into common themes. The categorization led to identifying patterns in each group's responses and was the primary input for the analysis. From there, as patterns and relationships between data started to emerge, they were confronted with the lab's working principles, slowly leading to the results identified. From the sensemaking process, we created a typology which is presented in the following section in Table 3.

Evaluation results
This section analyzes both expected and unexpected effects of GNova projects, identified by the partners of the Brazilian federal government, according to the sensemaking process carried out after the interviews and workshop. Below we detail the effects generated. We grouped them into typologies identified within three main categories: effects of the process, effects of the products, and effects of the participation in the process (Table 3). We used direct citations as illustrations of such effects. At the end of this section, we confront the effects presented with the lab's working principles.

Effects of the process
Methodologies matter. Making proper use of methodologies helps one make better decisions during the project journey for several reasons, including: deepen the understanding of the challenge by identifying people's needs and aspects related to the experience, motivations and logic of those who experience a public problem; extend knowledge about the problem, the context, the reality of the actors involved and expand the possibilities for solutions that were not visible; provoke imagination and encourage the elaboration of syntheses. Olejniczak et al. (2020) highlight the dialectical nature of applied policy design and the efforts of labs to integrate diverse activities and perspectives that have traditionally been separated in policy formulation. Specifically, "[p]olicy labs constantly balance between abstraction and fine-grained reality, between research inquiry to understand problems and proactive development of solutions to the problems" ( 101). Using agile methodologies enabled quick answers and a clear project scope definition. Additionally, qualitative methods effectively obtained information. The methodological rigor of the workshops brought reliability to the process and enabled the (re)definition of problems that must be specific, relevant, timely, and well-defined. According to an interviewee: "Coming here [to GNova] was a work of deconstruction, of mindset change to stop proposing solutions without properly understanding the problem. What is the problem? This was the main lesson." (Civil servant at the Central Bank of Brazil) It is essential to listen to the users and the actors involved. The knowledge generated in interactions with users makes it possible to deconstruct initial solutions, and propose new ones that respond to the specific situations.
Interviewees valued the flexibility of the processto be able to take risks and know that you can make mistakes and fix and adjust during the process. Additionally, they highlighted the value of collaboration, recognizing and considering the opinion of all team members regardless of the hierarchy, which has allowed teams to explore different iterations of ideas with freedom for creation, providing gains for the organization as a whole.

Effects of products
GNova projects can generate a variety of products, which are not considered isolated and singular deliveries, but part of a systemic and organizational strategy, with effects in both short and long terms.
The partners highlighted the quality of the products delivered during the projects and the potential for implementation. An important follow-up of projects that had prototypes and insights as products was incorporating these deliveries in the further development of the concept. Prototypes unfold into more refined and relevant products, sometimes encouraging new partnerships according to the related technical qualification.
Another effect identified in the delivery of products is savings for the public administration when delivery contributes to increased efficiency and reducing process, service, and policy costs.
A limitation perceived in terms of the effects of products was that the lab did not follow the development and implementation of the designed solutions. There were several suggestions for GNova to continue to play a role in the post-project, supporting the partner team in improving prototypes and implementation. This point is important because government innovation labs are rarely the "owners of the problem" and are not responsible for implementing the developed solution possibilities. For these labs, which support external teams accountable for the projects, it may be more difficult to know the results obtained by the project (given that the laboratory has no governability over its use).

Effects of the participation in the process
Participating enables capacity building and changing practices. The interviewees' perceptions about the methodological process reveal that competencies related to innovation were developed or acquired new meaning throughout the projects, such as problem analysis, empathy, and incorporation of users' needs.
From the testimonies, it is clear that going through one of these processes expands the perspective of the servants on the possibilities of solutions and challenges the state of affairs: "People here didn't have the habit of listening to the service user. It was very good to demystify what we think we know." (Civil servant at the Ministry of Economy) Some point out that GNova's main contribution was the change of mindset: "The process was very interesting and I think the main thing was the change of mindset. Now, when a demand arrives, we don't start thinking about solutions: wait, let's understand, let's make a prototype, let's go to the field, and listen to people." (Civil servant at the Central Bank of Brazil) In the words of one participant: "I would even give GNova another name, which would be a risk prevention lab. GNova is the space to, before doing anything nonsense, stop and better structure the thoughts, gather experts, go out to the field to talk to users, and come up with a slightly more mature idea to know if it is really worth it." (Civil servant at the Ministry of Economy) Although the process did not always achieve all the objectives initially proposed, in the six cases reported here there was an observable critical analysis and improvement of the model, system, or proposal that was the object of the work with GNova. This is clear from reports indicating the incorporation of insights, and some methods and principles in the partner agency's operation. For some projects, the efficiency gain was evident, and it was monetized.

Analysis of effects in relation to GNova's principles
When confronting the results with GNova's working principles (Table 1), a few observations stand out. The responses focussed on the importance of involving users and relevant actors (principle 1). They considered that the projects expanded or deepened the understanding of the problem and the context experienced by the people involved in the analyzed public service or program, identified the needs of users and actors involved, and designed opportunities for concrete improvement or innovation actions. All partners mentioned how important it was to include users to reach better results (effect of the process). Involvement emphasizing simplicity and agilitycontrasting with and complementing more institutionalized forms that require time and resources brings valuable insights that can reveal opportunities as well as blind spots before implementation.
Regarding principle 2, the framing of concrete and specific problems can create a shared sense of purpose in the team and lead to more effective results. However, the actual framing of a concrete challenge revealed itself to be a difficult task. In the cases analyzed, the actual output of projectsin the form of a map of insights and opportunities of innovationhelped to better frame challenges to be further addressed. That precise observation relates to principle 3, about identifying innovation opportunities based on insights and available evidence. Even though not naming it as such, this principle appeared alongside the application of qualitative research methods to involve users and relevant actors (effect of the process).
When it comes to principle 4focusing on the effects to be achieved to change the current situationparticipants acknowledged the risks of starting a project with a fixed solution rather than focusing on the desired effects to be achieved and allowing for possibilities to emerge throughout the process (effect of the process).
Concerning making ideas tangible through prototypes to test and learn about them, a participant expressed the value of making quick tests with relevant actors in the initial stages of a policy concept. The provisory and concrete nature of the proposition made it easier to get input from relevant actors, who pointed out many gaps in the concept presented (effects from the process). However, this principle was not emphasized by the majority of interviewees, perhaps due to the scope of projects investigated, which were more focused on researching to inform opportunities for action.

Conclusion
The analysis of the results reveals positive outcomes. Also, it highlights some opportunities for improvement and gaps in the work developed by the lab, all of which can guide future efforts. The results exposed that the methodological process had positive effects both in creating high-quality products and changing behaviors and practices adopted by the participants of the experiments. Other research efforts (Osorio et al. 2019) have found similar outcomes. Treating projects as experiments, working dynamically, and systematically documenting the learning process proved itself a value, since it challenges traditional forms of operation and sets precedents for other possibilities.
Analyzing the results, we can say that the vocabulary adopted by the lab team plays a major role in how participants talk about the projects and make sense of their learnings. Looking back to the GNova's history, the high emphasis on user involvement and application of design ethnography methods was very present in its initial projects and discourses around its approach. Labs must become conscious about which methodologies and narratives around innovation they are advocating and their political implications.
The suggestion of project continuity through the development and implementation of prototyped concepts seems to be a question of defining the lab's scope of operations and scalability. It can also be a matter of being clear about the project cycle, being able to exit smoothlywhich can be a very challenging moment for public projects (Bunt and Leadbeater 2012) -, making a transition that points a clear direction of continuity for teams.
In terms of methodology, the debate on principles to be used in work processes is as important as the mastery of tools because they guide the adoption of methods and instruments. It is important to mention that principles are living agreements, and for that reason they should be periodically revisited and reconsidered. Moreover, it is necessary to pay attention to the political dimension that permeates policy labs and their principles, which requires the ability to understand the relationships between different actors and the existing disputes about their views on the project.
The findings reinforce the government's role in developing civil servants' capacities since the methodologies are adaptable and can be helpful in various processes and daily challenges in the public sector, not only in innovation processes. Additionally, the learning by doing approach experienced by civil servants in lab projects resonates with the need to develop new skills to effectively navigate the complexities that come with the paradigmatic change toward the public value model of public administration (O'Flynn 2007), as mentioned in the introduction of this paper.
Even though from a limited research scope, these findings are relevant to expand some perspectives of labs as simply disseminators of tools and techniques that are not typically within the skillsets of many civil servants. Labs can instead represent a shift in how governments think about problems and their capacities to generate knowledge to "solve" them (Bailey and Lloyd 2016). However, it is not clear if this shift will be sustained by civil servants outside the lab and throughout the institution in the longerterm, facing strong organizational and cultural barriers to innovate. Interviewees noted the need to continue to develop skills for innovation after the conclusion of lab projects.
Being it an initial attempt of evaluation, there are limitations in the methods and process employed, especially related to the rigor and expertise of a self-conducted and single-case assessment. We would encourage a further exploration of evaluation instruments with a similar focus.
Finally, this paper contributes to a growing body of literature about labs in Latin America and to the construction of knowledge about how labs work and what effects they have on teams that participate in the design process.