A business model aimed at sustainability by means of re-use, remanufacturing and recycling depends on products that are returned either to a manufacturer or spe- cialized third parties. The business model needs to have ownership by the manu- facturer as a starting point to close material loops. Ultimately, customers will not buy new furniture, but they only pay for use, i.e., changing from ownership to performance-based payment models (e.g., Stahel2010; Webster2015; Lovins and Braungart 2013). To investigate a new circular business approach and simulate different circular based service scenarios for different customers and type of products, a dynamic modelling approach has been adopted (Groesser, Chapter
“Complexity Management and System Dynamics Thinking” this book). The SD model supports enhancement of the decision-making process by the Gispen man- agement team to develop, implement, and grow a new business model based on a circular economy (i.e., what kind of business model scenario might be successful within the model boundaries and assumptions). We used the software Vensim© (Ventana Systems, Inc., Harvard, Massachusetts) for the development of the sim- ulation software. Vensim is able to simulate dynamic behaviour of systems that are impossible to analyse without appropriate simulation software, because they are unpredictable due to many influences and feedback interrelations.
2.1.1 Development Process
An iterative approach has been used to quantitatively model Gispens’new business model. In the group model building sessions (Vennix 1996) the following steps were undertaken:
• Define the most important central KPI’s, i.e., business objective variables, for Gispen. A shared definition of the business objective variables was determined to evaluate effects of different tested policies and scenarios. Hence, a common understanding of profit, total turnover, market share, etc. for current and future scenarios was formed.
• Define the relevant variables in the causal-context model (Groesser, Chapter
“Complexity Management and System Dynamics Thinking” this book).
A management science approach was used to structure discussions on input variables and important outcomes.
• Determine and quantify the relationships between central KPI’s and variables in the model. Gispen management was frequently consulted to ensure that the model building proceeds in the right direction. Moreover, the Gispen manage- ment was involved in testing the model and evaluating the benefits for Gispen provided by the model. Gispen employees from sales and thefinancial depart- ment were involved to provide data on relevant business parameters which are used as initial values in the model. Macro-economic predictions at an EU level, existing GDP data, market trends for the office furniture market, standard values for cost and time to implement new business models structures and Gispen specific data such as annual reports and branch reports were incorporated (e.g., Cijfers and Trends Meubelindustrie2013). Not all data required by the model (e.g., the quantitative relationship between product attractiveness and company profit) were known. Expert meetings were used to define best expert estimates for these assumptions in the model (Ford and Sterman 1997). Furthermore, several scenarios in terms of macro-economic conditions were taken into account (i.e., negative, neutral and positive trends) as well as a predefined bandwidth for variables with a high level of uncertainty. Furthermore, the scenario and policy variables with the highest impact on business performance as well as the bounds for the set-up of these variables were defined.
• The model was validated on the level of model structure and model behaviour (Groesser and Schwaninger 2012). The focus was on internal and external validity of the model, for instance, were all relationships correctly modelled and KPI’s calculated in a correct manner, and concurrent validity, i.e., does the model give similar results for the model predictions and Gispen historical data.
A circular business scenario was modelled and evaluated. Within this business scenario, office furniture will be leased to an user (who will pay per month) and will get a financial incentive by Gispen after several years of use. In this model, Gispens’current, i.e. linear, as well as the new circular business model were both included (Fig.2).
Pricing Sales Purchase Price Effecve Sales Price Cost Total Cost Equity Gispen
The costs generate the pricing. Revenues Total revenues Revenue from sales / lease Revenue from government / privatePricing generates the revenues. Investment Investment in markeng Investment in design The revenues allow for the investment.
Investments are added to costs.
Sales of furniture New customers Gispen / Compeon New government customers Gispen / Compeon
The sales acƟvites generate costs.
Lease of furniture New lease customers Gispen / Compeon The lease acƟvites generate costs. The investments support the sales acƟviƟes. Product AƩracƟveness Effect of price, desing, delivery me and set up me
The investment impacts product aƩracƟveness.
Product aƩracƟveness impacts the sales acƟviƟes Recycling Recycling compleon rate Returned furniture Total returned furniture from repurchase Total returned furniture from purchase Not usable returned furniture is recycled.
Orders Total orders Percentage of directly purchased furniture
The lease acƟviƟes generate orders The sales acƟviƟes generate orders. ProducƟon of furniture Rate of remanufacturing, Rate of refurbished, Rate of newly produced furniture ProducƟon capacity Producon capacity Gispen Returned furniture provides input for producƟon.
Orders regulate producƟon
ProducƟon capacity regulates producƟon. Furniture in use Purchase, repurchase and lease customers using furniture Produced furniture is delivered to customers Furniture in business is returned.
Model Overview / Sector Diagram
Name of Sector Name of important variables=
Sector Colors are aligned with Vensim model Fig.2AschematicsimulationmodeloverviewofthefirstversionofGispen’scircularbusinessmodelThe main learnings from this model were summarized and included in a more simplified SD model to facilitate an easier understanding. Moreover, the simplified SD model has a narrower system boundary and focused only on the new business model (Fig.3). Following the conclusions of the first model the new business model was treated as a separate business unit with no influence from running businesses, apart from some initial assumptions such as that Gispen already had a customer base. Different model development steps were undertaken to ensure a consistent model in which all relationships were modelled correctly and all KPI’s were calculated in a correct manner. Moreover, the structure of the model has been discussed extensively in several workshops and the face validity of the behaviour of the model was evaluated (for validation see Chapter“Complexity Management and
Fig. 3 High level overview of the final business simulation model (top) and a more detailed impression of a part of the SD model (bottom)
System Dynamics Thinking”). The initial settings of the models parameters were checked and different scenarios evaluated to see its effects on the most important KPI Gispen was interested in: the break-even point.
2.1.2 Results
In this section, the most important outcomes of the simplified model are described.
In thefinal version of the model, historical data of the current business model were used where reasonable. This model included three major loops, the loop of cash, the loop of customers, and the loop of products, i.e., furniture. These three loops were modelled only considering circular economy furniture and not making a difference between refurbished and remanufactured furniture and different types of customers, i.e., new or existing customers in different market segments. Cost structures were implemented in simple terms.
From the simulation model the following conclusions were drawn:
• The implementation of circular economy of assets with a long usage cycles generates long delays with high negative initial cash flows in a pay per use scenario. This leads to the conclusion that lease models, as we currently know and apply, are less usable to drive more sustainable use of products. Integration of service components and solutions to get through the‘first use’period needs to be considered in more detail as this causes a highly negative cashflow (Fig.4).
A possible option, among others, is the intensification of the use of products, i.e., stimulate multiple or serial use (Webster 2015; Stahel 2010). Another option that might be viable is upgrading existing Gispen products, as a service, at the customer site (i.e., move from production to services).
• The business model made it possible to simulate not only Gispen’s internal processes, but also their interaction with market and competitors. This also allowed to focus on the adaptation of the market, competitors and own orga- nization, pinpointing the uncertainty of the adaptation speed that is a critical point in the model. Figure5shows the accumulated profit for the base run and two alternative scenarios. The base run is simulated with an adoption fraction of 0.008, meaning that 8 contacts out of 1,000 between clients result in a successful client acquisition and an effectiveness of marketing of 0.00025, meaning that 25 out of 100,000 potential clients are attracted every month as new customers. To show the effects of different adaptation speeds the scenarios ‘comblow’ and
‘combhigh’ have been created with the settings of 0.004 for adoption fraction and 0.000125 for effectiveness of marketing in the‘comblow’run and 0.012 and 0.0005 for‘combhigh’respectively.‘Comblow’therefore simulates the effects, if the adoption is low in both marketing and word of mouth while‘combhigh’ simulates when the new business model is embraced more quickly by the customers. In terms of effects the breakeven point for the business model in the scenarios are 109 (comblow), 147 (base) and 150 (combhigh). Low adoption rates have therefore a positive effect on the time to breakeven, mainly due to the
Fig. 4 Example of system dynamics simulation outcome: two scenarios of howfinancial funds develop over time given different assumptions for the product and service margins
Fig. 5 Accumulated profit for Gispen for different adaptation rates
fact that investments are low. The challenge for the new business model is that the costs arise at the beginning of a customer contact (production costs), while the flow of revenues is distributed over the entire duration of the contract (breakeven of an individual contract is between 39 and 40 months on a 60 months contract). This is illustrated by the scenario‘combhigh’where many customers are attracted quickly. Once the new business model is implemented however, the higher rate of turnover of furniture (and thus input for refurbish- ment) also makes thefinancial funds grow the fastest.
• Main added value of dynamic modelling is the deeper understanding of the mechanisms simulated. The method forces the user to provide well-founded reasoning and data to make the model reliable.
2.1.3 Upgrade as a Service
In essence the goal is to waste as little resources as possible. To do so one option is to upgrade or remanufacture existing, client owned furniture. A service is provided by the manufacturer and resources remain at the highest utility value. In practise the process starts with an inventory of the furniture in use and an inventory of the desired requirements for this furniture. These two are matched and additional ser- vices are executed, if possible on the customer site to keep transport to a minimum.
A proven concept so far is to reuse desks and remanufacture the pieces to‘as new’ desks. Visually the remade desk cannot be set apart from newly produced, and warranties are applicable for the remade desk. Since there is no shift of ownership and most of the existing materials are reused, the remade desk is considered a service. Which in turn can be embedded in a pay per use model. Even though the costs are still incurred at one moment in time, whereas the service is payed over a period of time a combination of tools can prevent the extreme dip in cashflow as described above. If at initial delivery a service package, including maintenance and upgrades, is agreed a more stable cashflow can be realized
2.1.4 Benefits of System Dynamics Modelling
The current simulation model concentrates on the objectives of Gispen but could also be used as an illustration of added value of business or process modelling for other companies. The development process of a model itself forces participants to create a shared vision/idea of the new business concept. Moreover, it organizes thoughts, concepts and ideas and how these interrelate. To create commitment of management or stakeholders they should be involved in this development.
Furthermore, this development process leads to a better understanding of all related aspects and their relationships. Afirst simulation of strategies (‘trial and error’) can be done in the model before implementation in the real world takes place. Thereby more successful and durable changes in any business model are supported.
2.1.5 Lessons Learned
The lessons learned of the application of system dynamics modelling in supporting the exploration of alternative business models are summarized below.
• To simulate the relevant aspect of reality in detail, a quite comprehensive and thus complex model was developed. The quantification of the resulting rela- tionships is time demanding and challenging, but results in a detailed under- standing of the mechanisms involved. After the detailed understanding of relevant system, the extensive simulation model was simplified. The second simulation model focusses exclusively on the new circular economy business model of Gispen. By using this model with lower complexity and details, it was possible to provide relevant information to the management team enabling them to obtain the insights for their decision making. In other words, only after the detailed model was developed it was possible to focus on the relevant mecha- nisms in the simplified model which then provided better insights in the relevant developments of the business model with concrete results. It is often, not always, beneficial to develop a larger model first to be able to evaluate what aspects of a situation are actually necessary.
• It was possible to demonstrate the robustness of the model through many extreme condition tests and through the consistency of the units of measure. To have a more practical discussion on the feasibility of circular business model scenarios, it is useful to provide detailed information for decision making.
A dashboard which shows the assumptions in the model and visualize the input and output could be a helpful means to enable even deeper discussions.
• In certain cases, to show to the management a certain trend, the timescale assumption was set to 2050. This was necessary since the delay times (use periods of the furniture products) in the modelled business system are relatively large and hence, changes in the underlying business model can only be seen, for instance, after several iterations of remanufacturing of furniture. A time horizon of 2050 is long, given that the time horizon for decision-making is regularly much shorter. After determining a trend by using the model with the long time horizon, it would be useful to then relate again to a timescale of 7–10 years.
Disparities in business dynamics and decision dynamics are challenges which the SD model could demonstrate. But given the dominant paradigms for deci- sion makers and the strong competition in the furniture industry, the SD model could not influence the decision making processes regarding the time expectations.
• The model is a means to evaluate the business potential. Such simulation models are used a few times during a year when the top management team reflects about its current corporate strategy.
• Group model building turned out to be successful in face-to-face meetings (Groesser, Chapter“Complexity Management and System Dynamics Thinking”).
Misunderstandings or decisions taken were easier to understand in these meetings compared to virtual meetings or discussions.
• Involvement of different stakeholders, among others, management representa- tives during the development process and critical decisions on assumptions of the model will require time, but at least the major and important conclusions of every development step should be evaluated by management. Moreover, the assumptions taken during development process should be shortly described and presented to management.