The results are split in the outcomes of Step 1 (literature analysis) followed by Step 2 (insights from practitioners) and the overall analysis of both analyses. The results of Step 1 start with an overview of the publications followed by definitions and characteristics, literature streams, and tools and methods, in line with the sub-questions. Step 2 focuses on validating the definitions and characteristics of CBME based on the practitioner study.
Step 1: Overview of Publications
The literature search led to a total number 17 core journal papers on circular business model experimentation (Appendix A Table 4) plus 22 additional papers to understand the broader sustainable business model experimentation field that pre-dated the circular business model experimentation concept (Appendix B Table 5). The review confirmed that this is an emerging research field as all publications were dated between 2011 and 2020. In the circular economy selection, all except two papers [41, 42] included empirical work, predominantly based on cases. The sustainability selection included only one conceptual paper . Whereas in the older SBME work, fewer tools and processes were developed (4 out of 22 publications), in the newer CBME, 8 out of the 17 publications included a tool or process. This suggests that the circular economy concept lends itself for practical implementation. Furthermore, the empirical work through cases, workshops, and direct experiments with business and other organisations demonstrates the practical focus of the circular economy.
The literature search also identified diverse business contexts, e.g. grassroots organizations , startup , social businesses , and large established businesses . Some of the longstanding and ongoing areas of research in this review focus on (electric) mobility (e.g. [47, 48]) and solar PV in different contexts (e.g. ), suggesting that these might be sectors where experimentation is becoming more commonplace and where business model innovations are gaining traction. Other pockets of research relate to business experimentation in city contexts [47, 50, 51], buildings [52, 53], clothing (e.g. [22, 54]), and farming [44, 55, 56]. Rather specialized, washing machines were represented with three papers [16, 57, 58]. Sector-specific studies included online education , university campuses , travel , water filtration , electric motors , and fast-moving consumer goods . Areas of technical innovation include Internet of Things (IoT) , smart grids , smart cities , and smart collection , suggesting the need to experiment at the intersection of technology and business models (see also [20, 64]).
Finally, different types of circular business models are addressed. For example, Danso et al.  and Yazan et al.  refer to closing the loop through ‘creating value from waste’ in biological resource cycles, and Xue et al.  investigate recycling and smart collection. Others highlight slowing the loop: Bashir et al.  look at refill and reuse; Lieder et al.  compare buy-back, leasing, and pay-per-use; Weissbrod and Bocken  refer to slowing the loop for fashion; and Torrieri et al.  refer to repurposing monastery buildings as a way of direct reuse. Most papers have an environmental focus, whereas Täuscher and Abdelkafi  also consider the economic viability of the business model, and Dobson et al.  integrate social aspects. Specifically, the pay-per-use business model  seems to be popular and is addressed in Lieder et al. , Sousa-Zomer et al. , and [16, 66]). Service models in general, like mobility as a service (e.g. [48, 67]), are popular, which is not surprising as this is one of the more established research fields in the sustainable and circular business model sphere (e.g. [65, 68]).
Defining and Characterising Circular Business Model Experimentation
The literature review results confirm the lack of conceptual clarity on circular business model experimentation. Table 2 shows different literature assertions identified on SBME and CBME.
The literature offers no unifying definition on CBME, but there are clear common characteristics of the concept. We identified ten key characteristics based on analysing dominant characteristics among the articles (Appendix A Table 4):
Designing and testing new circular value propositions
The aim of experimentation is to test propositions (product/service offerings) with customers or other stakeholders to test their viability from a customer perspective and circular economy perspective.
Testing in a real-life context, with stakeholders
Tangible evidence is needed to create evidence to convince stakeholders inside and outside the business about the viability of new propositions. Testing takes place with customers and other stakeholders.
Generating and analysing empirical data
Data are generated and analysed, e.g. through experimentation as a research method or using methods such as Lean Startup .
Iteration and rapid learning
Iteration and rapid learning are recognised in practitioner work (e.g. [24, 35]), and this was referred to in several studies.
Exploration and creating options
Experimentation is about finding out what works under which conditions and identifying and creating options, in the transition from a linear to a circular economy.
Reducing uncertainty, risk, and cost
Experimentation helps to reduce uncertainty and associated risk and cost.
Overcoming organizational inertia (in large established firms)
Organizational inertia hinders circular business model innovation in large established firms, and experimentation can help large firms overcome these.
Vision and purpose and/or goals
Through experimentation, companies work towards a shared vision and goal.
Partnering with others
This involves collaboration with others whether these are businesses, (local) governments, nongovernmental organizations (NGOs), or citizens.
Contribution to a wider transition
The aim is to contribute to a wider transition towards the circular economy and sustainable development.
Nine out of the 17 articles covered all ten criteria, and a further five at least 70% of all criteria (Appendix A Table 4). Criteria 8 and 9 on a shared vision and partnering with others were missing in some of the studies anchored in engineering, namely, the studies by Yazan et al.  which included modelling and computational experiments of the biogas supply chain and Marconi et al.  on a model for effective disassembly times for home appliances.
Literature Steams for Circular Business Model Experimentation: an Evolving Landscape
The literature review identified diverging, yet complimentary perspectives on the concept of circular business model experimentation (Fig. 1, Appendix D Table 7). Figure 1 illustrates how the reviewed literature in this article has evolved since 2011—as this is the date when the first articles explicitly at the intersection of experimentation in a business and sustainability context appeared to have emerged. For example, some authors used the Business Model Canvas in a sustainability context (e.g. [44, 71]), whereas others evolved the canvas (e.g. ) or referred to the Lean Startup (e.g. ). While most studies described business experiments as an innovation activity in companies (e.g. [72, 73]), some studies use experimentation as a research method (e.g. [48, 75]) similar to the earliest work identified . Three studies use both business experimentation as an innovation case and experimentation as a research method [16, 57, 63]. Four dominant research streams emerged from the literature review: literature anchored in business studies, engineering studies, transitions research, and design. Nearly half of the studies (18 out of 37) took a multidisciplinary perspective. Seven studies took a transitions’ and a business perspective, but only three studies integrated three disciplines. The different streams will be discussed next.
Literature Anchored in Business Studies
The first set of articles are the ones anchored in business studies (37 out of 39 papers), mostly include business experiments as an innovation case, followed by experiments as a research method, and three conceptual papers.
Studies typically used the term experimentation in a rather lose fashion, not applying strict ‘scientific rules’ as recommended in work on experimentation as a research method . Only Bashir et al.  and Yoon et al.  used a control group, as typical element of design of experiments as perused in the natural sciences. Aspects like randomization and a control group are typically not referred to; however, in Bocken et al. , the lack of a control group is described as a limitation. The literature review results are aligned with the comparison of the experimentation approach in the natural sciences and the corporate experimentation process by Weissbrod , who describes that control groups are not part to the experimentation process for sustainable innovation due to the complexity of sustainable innovation processes. Thus, most identified literature sources use the term experiment, when in fact ‘quasi experiments’ are described . This means that the term experimentation is used when referring to approaches or processes that lack critical aspects of a ‘real experiment’. It has been argued that a scientific approach to innovation can be beneficial to the success of innovations in organizations, especially through the formulation and testing of hypotheses during establishing new organizations . The question, however, remains to what extent aspects of experimental designs, more common in natural and physical sciences, would apply within the business context and to what extent this is even desirable . For example, some of the reviewed papers  use a more effectual approach building on ‘what is available’ , which seems to contrast the more structured Lean Startup approach, although other cases started off with using the Lean Startup, but ended up taking a more effectual approach .
Using experimentation as an approach to innovation, some articles used an action-oriented research approach with a variation of techniques [12, 44, 54, 69]. Bocken et al.  draw on Ries’  and Blank’s  interpretations of an experimental learning approach. This approach consists of formulating hypotheses, gathering data to test these, and developing conclusions based on the gathered data over a limited time period . Guldmann and Huulgaard  use the interactions with companies to formulate barriers to circular business model innovation. Ramos-Mejía and Balanzo  used a mix of qualitative methods (interviews, focus groups, direct observation, and ethnographic work) within a socio-technical experiment to understand how grassroots ecopreneurs developed sustainable business models in Colombia. Baldassarre et al.  used qualitative and action research techniques, coupled with design science research, to support sustainable startups in prototyping while developing a tool to develop future startups with sustainable business model innovation. Finally, recent work used a straightforward survey as choice experiment to test food provisioning practices in CBMs  or thought experiments to generate insights on future circular business models .
Literature Anchored in Design
The most popular intersection between disciplines lies between design and business studies: Ten studies combine these perspectives (Fig. 1). Studies at the intersection of design thinking, business model innovation, and experimentation already started with the work by Ortega et al. , who were experimenting with social impact innovation. Yet with three studies at the start of 2020, it appears that this is a new and popular research area. Guldmann and Huulgaard  regard experimentations as a solid part of circular business model innovation, making a distinction of internal and external experimentations (e.g. testing a prototype together with a supplier). Baldassarre et al.  tackle the challenge they refer to as the ‘design-implementation gap’ and design an organizational tool (SBM Pilot Canvas) to plan and execute small-scale pilots for implementing sustainable business models. Konietzko et al.  propose a set of principles for circular ecosystem innovation in which experimentation plays one part. Thus, from the viewpoint of the crossroads of design thinking for business model innovation and experimentation, relevant themes for further research include the following: frameworks and processes for innovation and implementation of business models, which is also naturally interconnected with the area of practical tool and method development [66, 81].
Literature Anchored in Engineering Studies
Eight studies are anchored in engineering studies (Fig. 1). Most papers in the field of engineering studies (except Xue et al.  and Torrieri et al. ) used experiments as a method.
Different types of experimental methods are used. Lieder et al.  used a simulation model to run an optimization experiment to find the most cost-effective combination of reused, remanufactured, and recycled components for a business model for a washing machine manufacturer; Torrieri et al.  developed an evaluation model based on multicriteria analysis and a financial model to support the choice of an alternative reuse of an ancient monastery in Italy. Their approach, similar to the tool in Lieder et al. , is suggested as a future decision support tool (Torrieri et al. ). Täuscher and Abdelkafi (, p. 654) developed a simulation based on the ‘learning sector’ by running a simulation for Coursera.
Other researchers developed knowledge more generally to inform business model development for future decision-makers. Yoon et al.  calculated the willingness to pay for solar lanterns in India and found that this is low, despite a trial period and postponed payment to increase sales. Kendel and Lazaric  installed smart meters in consumers’ homes with a control group. In addition to monitoring in this experiment, questionnaires were used. The findings suggest that any interventions that motivate households to change their energy habits would help, which would have implications for future business model development . Xue et al.  also used qualitative research with an open questionnaire and interviews to assess intelligent collection companies. They show the potential for integration of the informal waste collection into the more formal intelligent collection companies using internet and communication technologies . Marconi et al.  developed a method based on datamining to develop disassembly planning for a washing machine and coffee machine. Yazan et al.  developed an enterprise input-output approach to model physical and monetary flows of the manure-based biogas supply chain. Computational experiments highlighted under which conditions; cooperation is beneficial for all actors .
Literature Anchored in Transitions Studies
Seven studies combine a business with a transitions research perspective (Fig. 1). Huijben and Verbong  describe business model innovation in relation to a transition to solar photovoltaics (PV) as part of a wider renewable energy transition, citing Geels [82, 83] on transitions and Geels and Raven  on niche developments in transitions. Jolly et al.  refer to the work by Kemp et al.  on niche developments in their study on business model experiments for off-grid PV solar energy in India. Ramos-Mejía and Balanzo  refer to the work by Kemp et al.  and Smith and Raven  on niches in transitions to sustainability in their work on grassroots ecopreneurs. Xue et al.  do not specifically cite the transitions literature stream but rather refer to the need for wider economic transitions processes in China. Bauwens et al.  refer to the transition to the circular economy and the potential of scenarios to inform strategies and policies.
The ‘most multidisciplinary’ studies are the three studies by Lieder et al. , Xue et al. , and Bocken et al.  (Fig. 1). Lieder et al.  position themselves in both business and engineering studies but also bring in knowledge from design thinking in their study on future business models for washing machine manufacturing. Similarly, Xue et al.  take a business and engineering perspective but also bring in the wider transitions’ lens for sectorial transformation in China. Bocken et al.  make the connection between business models and transitions research, but also design thinking when supporting the ‘design’ of a new pay-per-use business model. The CBME field is diverse in itself, and recent studies also draw on multiple disciplines simultaneously.
Tools and Approaches for Circular Business Model Experimentation
Different tools and processes for circular business model experimentation have been developed. Table 3 provides the overview of tools and processes.
The articles show close ties to the existing practitioner tools such as the Business Model Canvas and Lean Startup method (see also, ). Ramos-Mejía and Balanzo  and Huijben and Verbong  both use the Business Model Canvas as part of their studies. Others like Baldassarre et al.  and Bocken et al.  adapt the Business Model Canvas for sustainability or circularity purposes. Chesbrough  in his work on business model experimentation already mentioned potential approaches to experimentation, including a more structured Lean Startup style approach using hypothesis testing , versus a more emergent effectual approach to business model experimentation, focused on ‘what is available’ . Some articles identify this more effectual approach in addition to the more structured Lean Startup being adopted by business (e.g. [12, 54]), but the dominant sources for inspiration appear to be the Lean Startup, as well as the Business Model Canvas by Osterwalder and Pigneur . Chesbrough  also notes the importance of an iterative approach, in line with design thinking, which is also being adopted in some articles (e.g. [12, 46, 57, 69]). Some articles also build on more traditional new product development innovation funnels (e.g. [22, 72]).
In studies in the engineering research stream, decision support tools have been developed, such as the work by Marconi et al.  that calculates disassembly times in a future circular economy and the model by Yazan et al.  to assess the potential for collaboration. Torrieri et al. () use multicriteria analysis to support the decision on redevelopment of a monastery. Lieder et al.  and Täuscher and Abdelkafi  use forms of simulation to support future business model development.
Recent work embeds insight from psychology to understand possible consumer behaviour shifts  and insights from scenario analysis to develop circular economy pathways .
Step 2: the Practitioner Study
The practitioner study was conducted to investigate how the CBME field emerged compared to literature. Thirty practitioners answered the question: ‘What does “circular business model experimentation” mean to you?’
Appendix C Table 6 shows the coded version of the responses by the practitioners against the criteria in Section 3.1.2 to identify similarities and differences. Two out of the thirty practitioners referred to all ten criteria from our definition in Section 3.1.2. This small number is not surprising as we asked for the practitioners’ viewpoints and not a comprehensive definition. Some points were more prominent in the practitioner responses than in the reviewed studies. In particular, the practitioner data sample showed an emphasis on the customer, using and balancing metrics, systemic change, and scaling up. The focus on the customer and balancing financial, with societal and environmental metrics was expected, seeing the business context of the respondents. The focus on scaling up and systemic change is, however, noteworthy as it shows that the CBME concept is seen as a lever for change by practitioners.
Two out of the thirty practitioners with a science background mentioned a controlled lab type of experimentation environment, although most referred to the rather practical setting of experimentation in collaboration with customers, suppliers, and other stakeholders. As for tools and methods, the Lean Startup approach was named once. One participant, however, mentioned ‘build-test-learn’, and three others refer to hypotheses testing. Design thinking was mentioned once.
Characteristics and Definition
While the literature analysis gave comprehensive insights, the practitioner study gave enriched perspectives to the CBME perspective. These related to the customer and transforming consumer behaviour, using and balancing metrics, systemic change, and scaling up. Based on the practitioner perspectives, we evolved the original 10 CBME characteristics (new in italics):
Designing and testing new circular value propositions. The aim of experimentation is to trial propositions (product/service offerings) with customers and other stakeholders to test their viability from a customer perspective, circular economy, and systemic perspective.
Testing in a real-life context, with stakeholders. Tangible evidence is needed to create evidence to convince stakeholders inside and outside the business about the viability of new propositions. Testing takes place with customers and other stakeholders.
Generating and analysing empirical data. Data are generated and analysed, e.g. through experimentation as a research method or using methods such as Lean Startup.
Iteration, rapid learning, and moving from experiment to scaling. The process is iterative (e.g. build-measure-learn) and the complexity of experiments build up over time, towards scaling up from a small-scale towards expansion within and across markets.
Exploration and creating options. Experimentation is about finding out what works under which conditions and identifying and creating options, in the transition from a linear to a circular economy.
Developing the business case while balancing societal and environmental impact. Experimentation helps to reduce uncertainty and associated risk and cost about future propositions, but various criteria (e.g. business case, circularity, sustainability) need to be balanced during the experimentation process.
Overcoming organizational inertia (in large established firms). Organizational inertia hinders circular business model innovation in large established firms, and experimentation can help large firms overcome these.
Vision and purpose and/or goals. Through experimentation, companies work towards a (shared) vision and goal.
Partnering with others. This involves collaboration with others whether these are businesses, (local) governments, NGOs, or citizens.
Contribution to a wider transition. The aim is to contribute to a wider transition towards the circular economy and transforming consumer behaviours.
Based on these ten characteristics, we assert that CBME is
An iterative approach to develop and test circular value propositions in a real-life context with customers and stakeholders, starting with a shared goal. It involves rapid learning based on empirical data to provide evidence on the viability of circular value propositions. Iterations involve increased complexity of experiments. There is a learning focus on initiating wider transitions, such as transforming consumer behaviours for the circular economy.