Introduction

About a decade ago, Zott and Amit (2010, p. 218) suggested understanding business model (BM) design as an activity system, describing it as “purposeful weaving together of interdependent activities performed by the firm itself or by its suppliers, partners and/or customers”. This understanding has been widely adopted in the BM research community today (e.g. Amit & Zott, 2012; Laasch, 2019; Tykkyläinen & Ritala, 2021; Visnjic et al., 2018; Zott & Amit, 2013). There are also initial approaches to the use of activity systems in the broader context of entrepreneurship research (e.g. Alaassar et al., 2021; Audhoe et al., 2018; Pauwels et al., 2016; Peronard & Brix, 2018). However, their potential for understanding entrepreneurial process (i.e. the process of venture creation or venture development based on the recognition or creation of opportunities) beyond the BM has not been discussed thus far. Therefore, considering the impact of entrepreneurship on economic growth (Acs et al., 2012; Carree & Thurik, 2010; Rodrigues & Ac Teixeira, 2021; Wennekers & Thurik, 1999), sustainability (Diepolder et al., 2021; Gil-Doménech & Berbegal-Mirabent, 2018), societal progress (Zahra & Wright, 2016) and individual self-fulfillment (Hitt et al., 2011), a deeper understanding of entrepreneurial processes is urgently needed, yet surprisingly underrepresented in the literature (McKelvie et al., 2020).

Given this need for process-oriented entrepreneurship research and the current lack of respective data on a larger scale (e.g. Davidsson & Wiklund, 2001; Guerrero et al., 2021; Uy et al., 2010), this paper aims to suggest a new way for collecting comparable data on entrepreneurial activities and to demonstrate how data collected that way could be used to address current gaps in the entrepreneurship literature. From this flow the following research objectives for this paper:

  • (1) MODEL: expanding the logic of activity systems to the overall entrepreneurial process,

  • (2) METHOD: introducing a new approach for the structured collection of comparable data,

  • (3) DATASET: demonstrating types of activity data as empirical foundation

  • (4) EXAMPLES: illustrating how this model, method and dataset can contribute to future data-driven and evidence-based entrepreneurship research.

Thus, even if this paper will create benefits for both entrepreneurs and their supporters, the main beneficiaries of this work will be the entrepreneurship research community.

The remainder of this paper is structured as follows: First, we provide a short and concise literature review on the use of activity systems within entrepreneurship. Secondly, we present a proposal for using activity systems in entrepreneurship research, introducing a Systemic Entrepreneurship Activity Model (SEAM) for the whole entrepreneurial development process that goes beyond the conventional understanding of business modelling and including pre- as well as post-business modelling activities. Thirdly, we lay out how this helps enabling an open, non-participant observation approach carried out using a digital platform. This can enable future empirical studies to focus on process-oriented research, which has been addressed as insufficient (e.g. Davidsson & Wiklund, 2001; Uy et al., 2010). Fourthly, we demonstrate the magnitude of activity data gathered thus far, highlighting how SEAM and the observation approach led to the development of a database encompassing 13,927 entrepreneurial projects from 106 countries. On this basis, we exemplify the potential of SEAM for future entrepreneurship research, providing three examples of how this data can and will be used in the future for data-driven and evidence-based entrepreneurship research (Frese et al., 2014). Finally, we draw conclusions on how our work contributes to the future development of the entrepreneurship research field.

Theoretical background

While acknowledging the invaluable contributions of theory related to the business model (Zott & Amit, 2010) as well as any prevailing business model setups like the Business Model Canvas (Osterwalder & Pigneur, 2010) or the Lean Canvas (Maurya, 2012), we want to focus on the holistic use of activity systems in the entrepreneurship development process.

Activity systems originate from general systems theory (GST) introduced by Bertalanffy (1951) as well as the theory related to cybernetics as described by Ashby (1956). Bell and Solomon (2002) firstly used activity systems within entrepreneurship by defining it as a “holistic lens for evaluating business processes”, whilst Halecker and Hartmann (2013) define a system as “a set of interacting elements with interrelationships among them.” According to Capra (2015), systems thinking means that something is placed in the context of a larger whole, which may enable a shift of the analysis from individuals to the interactions between them (Jantsch, 1972). Daniel et al. (2022) combines complex adaptive systems (CAS) and entrepreneurial ecosystems (EE) (Stam & Spigel, 2018) into a framework of dynamic and diverse actors, factors, and interdependencies. Pfeiffer (1971) describe the architecture of an activity system as consisting of “input, human resources, organization, technologies and output” while Zott and Amit (2010) have the most accepted description, including content (What activities should be performed?), structure (How should the activities be linked and sequenced?) and governance (Who should perform the activities, and where?).

Reviewing the current understanding of entrepreneurial activity systems, we identify three main areas of interest in the academic debate, focusing on 1) holistic systems of interrelated activities, 2) ecosystems and finally 3) dynamic and normative systems.

  1. 1.

    Afuah and Tucci (2000), Morris et al. (2005), Zott and Amit (2010) and Daniel et al. (2022) all describe systems that is made up of components, linkages between the components, and the dynamics in which these components interact. Amit and Zott (2001) describe this activity system as “the content, structure, and governance of transactions designed so as to create value through the exploitation of business opportunities”, building on Shane and Venkatamarans definition of entrepreneurship as such (2000). The common denominator between these is the ambition to include as much as possible of the operations of the entrepreneurial ventures in this modelling endeavor. According to Zott and Amit (2010), “activity system design describes how firms do business” This means that the activity systems perspective provides managers and academics with a language and a conceptual toolbox to tackle these challenges and participate in creative design processes and insightful discussions. They argue that there are four advantages of analyzing entrepreneurial development in the context of activity systems: (1) focusing on activities is a natural perspective when deciding on business model design; (2) the activity system perspective encourages the designing of the business model in a systemic and holistic way; (3) the assumptions made in the transaction cost economics literature can be relaxed; and (4) the viewpoint is suitable for further theoretical development and refinement of the BM.

  2. 2.

    There is a widespread agreement among theorists that business modelling, and entrepreneurial business development in general, relates not only to the focal firm but to the entire ecosystem surrounding it. Zott and Amit (2010) describe it as “a system of interdependent activities that transcends the focal firm and spans its boundaries,” and Zott and Amit (2013) refer to it as “a more systemic perspective that emphasizes inter-dependencies and complementarities between a firm and third parties in order to properly understand how value is created.” Visnjic et al. (2018) state that “an activity system may exceed the boundaries of one firm and grow to represent a system of interdependent activities that transcends the firm’s boundaries and spans the ecosystem of firms interconnected by virtue of their value-creation functions.” This fits with the general understanding of the construct of Entrepreneurial Ecosystem (Stam & Spigel, 2018). Lynch et al. (2021) uses this extended perspective to build on the construct of sustainable entrepreneurship.

  3. 3.

    There seems to be a general agreement (Autio, 2017; Teece, 2018; Cosenz & Noto, 2018) that the development of an entrepreneurial project should be done in a dynamic and normative manner—not purely static and descriptive. Logically, this means two things. Firstly, a business should be seen as changing over time regarding the content, structure, and governance of the elements. Secondly, the business model of the venture should be viewed normatively (Autio, 2017). This means that the modular elements of the business are dynamically changed and adapted to improve the creation and capture of value. Aversa et al. (2015) consider the business model a cognitive and analytical tool for playing with alternative scenarios, and modelling various possible outcomes of strategic decisions. This adaptation of the business model elements also seems to be somewhat integrated into the term business model innovation, which is described as a dynamic, normative activity. “Thus, studies should not simply ask the question what is a business model, but rather ask how can a business model be innovated” (Schneider & Spieth, 2013). Halecker and Hartmann (2013) describe business models as “neither local nor static” and conclude that “the evolution and innovation of business models requires a dynamic approach.” Aversa et al. (2015) divide business modeling into three phases: “thinking” is the individual effort of cognitively understanding a business; “articulating” is how the business is represented in a simplified way; and “doing” is how the cognitive model is translated into a set of real-world activities. Another way of interpreting business models in a normative way is to view them as archetypes. Zott and Amit (2010) define four archetypical ways of standing out from the competitors: “novelty” means introducing new content, structure or governance into the business model; “lock-in” means using switching costs or network externalities that derive from the structure, content or governance; “complementarities” means bundling activities within the system; and “efficiencies” refers to reducing transaction costs.

A proposal for using activity systems in entrepreneurship research

Amit and Zott (2001, p. 511) describe the activity system as “the content, structure, and governance of transactions designed so as to create value through the exploitation of business opportunities”, in which the content describes what is being done, the governance by whom it is being done and the structure how these activities relate to each other (Zott & Amit, ). Looking into the academic literature on activity systems in the entrepreneurship research field thus far, we see a usage of the construct in six areas. First and foremost, it has been used in discussing BMs (Amit & Zott, 2012; Aversa et al., 2015; Landoni et al., 2020; Tykkyläinen & Ritala, 2021) and business model innovation (BMI) (Amit & Zott, 2012; Aversa et al., 2015; Landoni et al., 2020; Tykkyläinen & Ritala, 2021), which is reasonable as Zott and Amit (2010) developed it discussing BM design. Beyond that, activity systems are used in the context of open innovation (Peronard & Brix, 2018; Troxler & Wolf, 2017; Visnjic et al., 2018), entrepreneurial ecosystems (Audhoe et al., 2018; Mustafa, 2015), entrepreneurship support (Alaassar et al., 2021; Audhoe et al., 2018; Pauwels et al., 2016) and a wider array of different entrepreneurial activities (Filion, 2004; Hasu & Engeström, 2000; Luce, 2010; Rossignoli et al., 2018; Tidhar & Eisenhardt, 2020).

In this section, we propose using activity systems for a structured description of the entrepreneurial process developing a Systemic Entrepreneurship Activity Model (SEAM). SEAM is composed of seven activities:

  1. 1.

    Developing the purpose of the entrepreneurial project,

  2. 2.

    identifying existing resources,

  3. 3.

    developing and testing a business idea (exploration),

  4. 4.

    developing and testing a business model (exploitation),

  5. 5.

    setting objectives for the entrepreneurial project,

  6. 6.

    setting concrete tasks for reaching these objectives and

  7. 7.

    making and updating a forecast how the project will develop.

The following description of these activities highlights that they are composed of 29 elements describing what exactly entrepreneurs do during each activity. Figure 1 summarizes these activities and elements, serving as an illustration of SEAM.

Fig. 1
figure 1

The Systemic Entrepreneurship Activity Model (SEAM)

Full size image

Purpose

The first activity is to agree on the overall purpose on which the entrepreneur(s) are building their project. This activity is composed of the development of three different elements: core values, vision, and motivation.

The identification of the entrepreneurs’ core values is based on Collins and Porras (1996, p. 66), who define them as a set of guiding principles that are essential and timeless components of an organization. They highlight that “core values require no external justification; they have intrinsic value and importance to those inside the organization”. In entrepreneurship research, core values are investigated to better understand entrepreneurial behavior, decision-making and cultural impact (Kirkley, 2016; Morris & Schindehutte, 2005)

Developing a vision for the entrepreneurial project encompasses its core ideology and its envisioned future. According to Collins and Porras (1996, p. 66) “Core ideology […] defines what we stand for and why we exist. […] Core ideology is unchanging and complements the envisioned future.” Previous research on entrepreneurs’ vision, for instance, found that it is related to and can be explained by different motivations of entrepreneurs (Krueger Jr et al., 2000).

Defining the overall motivation or generalized end goals (Sarasvathy, 2001) helps the entrepreneurs to identify why they chose to develop their project. Typically, this will not be a detailed description of where the entrepreneur wants to be, but a value-based explanation of why the project is started. We want to emphasis that this term is used in the meaning of entrepreneurs conscious understanding of why they want to be entrepreneurs or what they want to get out of their entrepreneurial project, rather than the personality traits more often described as entrepreneurial intent (Ajzen, 1991; Bird & Jelinek, 1989; Carsrud & Brännback, 2011; Krueger & Carsrud, 1993; Thompson, 2009) often associated with the measurement of Achievement motivation (Ach) (Atkinson, 1957, 1964; McClelland, 1961).

Resources

The second activity is analyzing the resources that are available for each project. These resources can be “physical assets, capabilities, organizational processes, firm attributes, information and knowledge” (Barney, 1991, p. 101), with special attention being given to the core competences on which the project is built (Prahalad & Hamel, 1990). There are different types of resources available to the entrepreneurs when starting the development or improvement of their projects. Any project will be founded on the resources at hand, and it will develop as these resources change (Barney, 1991; Kraaijenbrink et al., 2010; Sarasvathy, 2001, 2008; Scott & Venkataraman, 2013; Venkataraman et al., 2012; Wernerfelt, 1984). These resources may be re-used in new and creative ways, and the relevance of the resource may not be obvious from the get-go, but become clear as the project progresses (Baker & Nelson, 2005). Typically, the resources used by entrepreneurs have an underlying typology (Clough et al., 2019). Thus, this activity is composed of four different elements: human, physical, marketable, and financial resources.

The human resources available to the entrepreneurial project (Gimeno et al., 1997; Marvel et al., 2016; Unger et al., 2011) primarily include the founding team or employees but also partners and other important enablers from the ecosystem (e.g. Aldrich & Kim, 2007; Hallen, 2008). The physical resources available in the project may be a factory or office building, any kind of manufacturing machine or another physical asset that can be used in the project. The element of marketable resources includes patents and all other forms of intellectual property in addition to products and services that are developed in the project or that are otherwise available to the project (e.g. Hsu & Ziedonis, 2013; Vissa & Bhagavatula, 2012). Finally, there are the financial resources available to or assigned to the project (e.g. Davidsson & Honig, 2003; Evans & Leighton, 1989). These can be in the form of equity, available credit, signed customer contracts and grants that will generate revenues for the project. These are valuable both because of the monetary value that they may have and because they serve as important references (Moore, 1995) that make it easier to attract new customers and partners.

Business idea

The third activity is the development and testing of a business idea. Here, the entrepreneurs define what their project will achieve, whereas the development of the business model (the next activity) defines how it will be achieved. We highlight the importance of this separation, as entrepreneurs should concern themselves with the elements that matter to their customers before they address the how questions of their project (what products or services should I build and how should I sell and price them?). This would secure a focus on establishing whether “Do I have a problem worth solving?” before trying to figure out “How do I solve this problem?” (Dahle et al., 2014). Consequently, the activity of development and testing a business idea takes a customer-centric perspective, asking who they are, what problems they have and why the entrepreneur is the right person to solve them. It is closely related to the business mission (Klemm et al., 1991), to the unique value proposition (UVP) framework of Moore (1995) UVP and to Osterwalder et al. (2014) value proposition canvas. We propose structuring this activity using four different elements: key contributions, key markets, core competence and problem interview.

By discussing their key contributions, the entrepreneurs are encouraged to identify the problems that they will solve for their target groups or customer segments. This is strongly related to Christensen et al. (2007) discussing the identification of which job a product does for the customer. The identification of key markets helps the entrepreneurs to describe their target group or customer segments. Within entrepreneurship, bottom-up targeting of customers and other influence groups is more relevant than traditional top-down segmentation of markets (Stokes, 2000). As with all other elements, these target groups will often be described more clearly as the project develops, and in many projects the focus will change from one target group to another. The next element relates to the identification of the entrepreneurs’ core competence, defined as “a harmonized combination of multiple resources and skills that distinguish a firm in the marketplace” (Schilling, 2013, p. 117). Identifying their core competences enables the entrepreneurs (and organizations in general) to be competitive in their target market over time (Javidan, 1998; Prahalad & Hamel, 1990; Schilling, 2013). As indicated by Alizadeh and Khormaei (2012), core competences are not resources in the sense of factors of production but are rather based on a competence view and related to the innovativeness of the organization (Forsythe & Khormaei, 2011). This is linked to the dynamic capabilities of an entrepreneurial project, focusing on the dynamic and constantly changing nature of its firm-specific advantages (Teece, 2007; Zahra et al., 2006). Finally, carrying out problem interviews helps the entrepreneurs to test their business ideas and to identify the importance of different contributions prioritization by representatives from different markets. The purpose of these interviews is to determine whether the project is attempting to solve a real problem that is considered important for the target group.

Business model

The fourth activity is the development and testing of the business model. According to Zott et al. (2011), there is no common understanding of business models in academia yet. While there is as wide array of different approaches, concepts and models to BM design (Baden-Fuller, 1995; Baden-Fuller & Morgan, 2010; Johnson et al., 2008; Morris et al., 2005; Ritter & Lettl, 2017), we build our understanding of the BM in SEAM mainly on Amit and Zott’s work (Amit & Zott, 2001; Zott & Amit, 2010, 2013; Zott et al., 2011), the business model canvas (Osterwalder & Pigneur, 2010) and the lean canvas (Maurya, 2012), as these concepts are widely used by practical entrepreneurs and providers of entrepreneurship support (ES). The activity of business modelling is composed of seven elements: co-creators, value proposition, product features, ecosystem, sales model, price model, and solution feedback. These relate to the four elements of developing and testing the business idea.

The identification and addressing of co-creators is described by Moore (1995) in his book Crossing the Chasm. Here, Moore states that everyone has a different attitude toward purchasing new products or services. He uses Everett (2003) diffusion curve to separate all potential customers into the early market and the mass market. Then, he suggests selling to the early market first and using customers as references to be able to enter the mass market. Prahalad and Ramaswamy (2000) offer another view on how particularly motivated early customers can contribute to both the problem definition, the development of the solution and the introduction into the market. By defining their value proposition, the entrepreneurs derive a “single, clear compelling message that states why [they] are different and worth buying” (Maurya, 2012, p. 5). Cooper and Vlaskovits (2013) suggest developing a value proposition by means of a so-called customer–problem–solution. Here, the entrepreneurs enter the name of their customer, the customer’s problem, and a description of the solution to the problem. An entrepreneurial project can have one or a few unique value propositions, always linked to a specific target group.

The definition of the product features leads to a detailed product and service description included in the delivery. These products and services can be developed in a iterative and customer including process as according to the lean startup movement (Blank, 2007; Cooper & Vlaskovits, 2013; Frederiksen & Brem, 2017; Ghezzi, 2018; Ghezzi & Cavallo, 2018; Maurya, 2012; Ries, 2011) and the thoughts of design thinking (Brown & Katz, 2011). The identification of the ecosystem concerns the selection of allies. Stam and Spigel (2018) identify 10 components that make an entrepreneurial ecosystem, explaining the influence of regional, economic, and social factors on the entrepreneurial process (Bahrami & Evans, 1995; Dubini, 1989; Pennings, 1982; Van De Ven, 1993). One of these components is the network available to the entrepreneur. There are several types of potential partners that can create value for them, including distribution partners, partners with complementary products and deliverance partners. Parties influencing the customers’ purchase decisions may be particularly relevant.

Defining their sales model helps the entrepreneurs identifying the approaches to sell their products or services. It can be direct sales or indirect sales via partners and proactive sales or reactive sales involving advertising and selling to respondents. The sales process itself can be organized in several ways. All these different tactics to achieve sales are registered in the business model. Choosing the price model describes how the entrepreneurs get paid. It may receive its payment from the customer directly or through an alternative method. There are also a great number of potential creative payment tactics that can be used. Here, we describe the chosen revenue tactics. In entrepreneurship research, the specifics of entrepreneurial sales and price models are for instance being investigated to derive specific marketing strategies (Flatten et al., 2015; Pitt et al., 1997; Schindehutte & Morris, 2001; Siems et al., 2012). Finally, solution interview feedback is obtained through interviews with customers who have been testing the product or service. This is typically achieved by presenting the potential customer with a type of minimum viable product (Ries, 2011) in the form of a mock-up, prototype or just a general written or graphic description of the product. The problem interview tries to answer the question “Do we have a problem worth solving?” while the solution interview tries to answer the question “Are we solving the problem?”. This corresponds to the test phase in the IDEO inspired design thinking method from Stanford University’s d.school (Brown & Katz, 2011).

Objectives

The fifth activity is concerned with the entrepreneurs translating the general strategies of the business model into tangible objectives for the entrepreneurial project. These objectives are typically the kind of key performance indicators (KPIs) used to manage progress, measuring both financial and non-financial achievements (Kaplan & Norton, 1996). According to both the effectuation and the lean startup view, these objectives will be very dynamic and change throughout the course of the project. We also emphasize that there can be more KPIs than the traditional financial ones. The development of knowledge in the entrepreneurial organization is vital (Prahalad & Hamel, 1990). Thus, it is natural to develop specific KPIs measuring the buildup of strategic knowledge through recruitment and training. We also argue for a specific focus on KPIs regarding product development, sales, and marketing. The financial KPIs will be closely related to the traditional revenue side of a profit and loss analysis, encompassing revenues and grants, and the payments side of a cash flow analysis encompassing new equity and loan funding. Setting objectives is structured by five elements related to skills, sustainability, product development, sales, and marketing, as well as financial objectives. All of them are quantifiable and assigned to a specific time-period. Objectives are described as milestones, as numbers or in monetary terms.

Skills objectives are objectives regarding increasing the competence in the project. In a balanced scorecard, this would be the learning and growth perspective (Kaplan & Norton, 1996). Sustainability objectives describes KPIs concerning positively addressing the environmental, societal, or economic objectives of the United Nations 17 sustainability objectives (Elkington, 1993). Product development objectives could be the finishing of a new prototype, or the improvement of an existing service. In a balanced scorecard, this would be part of the customer and internal process perspectives. Sales and marketing objectives are types of KPIs that are at the core of customer relationship management projects: how many customer contacts have been created, how many sales letters have been distributed or how many likes the project has received on Facebook. In a balanced scorecard, this would also belong to the customer and internal process perspectives. Financial objectives are all objectives that are described in the form of monetary values. In a balanced scorecard, this would be the financial perspective.

Tasks

The sixth activity concerns the management of the activities or tasks needed to fulfill the objectives. This can involve an individual task list for a sole entrepreneur, or it can be a way to structure the collaboration for the whole team. Entrepreneurial tasks planning is a dynamic process of constant changes. Therefor it is not suited for traditional static and linear project management techniques (Gross & McInnis, 2003). In the same way as objectives are related to revenues and incoming payments, tasks are related to costs. According to the concept of activity-based cost setting (Kaplan & Norton, 1996), costs and tasks should be treated in unison.

Forecast

In the seventh and final activity, the revenues and costs related to objectives and tasks are structured in a forecast. The flexibility gained through coordinated treatment of tasks and their cost are described well in the activity-based costing (ABC) literature (Cooper, 1993; Cooper et al., 1988). This activity comprises five elements.

The first element relates to the revenues and the cost of goods sold (COGS) associated with them. Recurring costs are the costs that will return periodically, and they will typically constitute a high percentage of the entrepreneurs’ non-COGS-related costs. Salaries, consultancy fees and rent for buildings and machinery will be stated here. Task costs are the costs that are directly linked to activities, like travel or advertising costs. Penultimately, we describe the financial forecasting elements. These are the equity infusions or loans injected into the project and their related cash flows. Finally, we describe the rules and regulations that apply to the specific project.

Enabling an open, non-participant observation approach

A key challenge for evidence-based entrepreneurship research is the current data situation (Frese et al., 2012, 2014; Martelli & Hayirli, 2018; Roy & Das, 2016). The application of data science methods has great potential within entrepreneurship research, especially to answer new research questions that so far cannot be solved with traditional research methods (Prüfer & Prüfer, 2020). Data used so far are primarily based on qualitative and quantitative surveys via interviews or questionnaires as well as self-assessment text documents such as business plans (e.g. Dana & Dana, 2005; Leunbach et al., 2020; McDonald et al., 2015). The results of these survey methods often show an insufficient number of cases and generalizability in the case of interviews, or a lacking level of question-related detail in the case of standardized questionnaires (Saunders et al., 2019). Big data analyses form the essence of this digital revolution (e.g. Hazen et al., 2018; Obschonka, 2017; Prüfer & Prüfer, 2020), replacing these currently prevalent collection methods.

The possibilities of analysing multidimensional and unstructured datasets (e.g., larger sets of business plans) have been insufficiently exploited (Prüfer & Prüfer, 2020) due to lack of capacity to process large amounts of data (e.g. Obschonka & Audretsch, 2020). Initial approaches to the use of big data in entrepreneurship are mainly represented by the application of machine learning and text mining methods (Prüfer & Prüfer, 2020). Due to the lack of primary data, these methods are mainly used to support the comprehensive review, cataloguing and analysis of existing literature or uncategorized text material from websites. Examples of such use of machine learning is a study of randomly sampled psychological, demographic, and family characteristics of different individuals with the aim of identifying their entrepreneurial spirit (entrepreneurial orientation) (e.g. Antretter et al., 2019; Prüfer & Prüfer, 2020; Sabahi & Parast, 2020) and a text mining approach to analyse 20,188 crowdfunding campaigns (e.g. Kaminski & Hopp, 2020; Rauch, 2019).

Structuring the entrepreneurial process by means of SEAM and using that structure in the course of Entrepreneurship Support (ES) (Ratinho et al., 2020) allows for big data to be collected via a digital project management tool used by the entrepreneurs. We refer to this tool as an Entrepreneurship Management System (EMS) (Dahle, 2021). The EMS is built to accompany the seven activities of SEAM and comprises the 29 elements introduced in section "A proposal for using activity systems in entrepreneurship research". While the entrepreneurs are using the EMS to develop their ventures, time-annotated and anonymized logs are kept of when (structure) which member of an entrepreneurial team (governance) makes which entry (content) in the system within these elements, i.e. creates, modifies, or deletes an entry (Zott & Amit, 2010). At the beginning of any ES program using SEAM, the entrepreneurs are informed that the EMS collects this data for academic research purposes. Thus, we use open, non-participant observation leading to time sensitive qualitative and quantitative data. We refer to this as Digital Observation Method (DOM).

DOM leads to the development of a unique data set, enabling future empirical studies to focus on process-oriented research, which has been addressed as insufficient thus far (e.g. Davidsson & Wiklund, 2001; Uy et al., 2010). It also allows for the use of novel data science methods for the investigation of entrepreneurial activities, which has not been used for this purpose in the entrepreneurship research field so far.

The magnitude of activity data gathered thus far

In the remainder of this section, we set out how SEAM and DOM have been used to build an activity-based dataset comprising 13,927 entrepreneurial projects from 106 countries. At this point, we underline that we do not argue for SEAM to be the only systems-based model describing the business development of entrepreneurs, nor that it necessary is better than other methods. The contribution of SEAM is the distribution to entrepreneurs via ES programs and the subsequent development of the dataset via DOM. This enables a novel research design, addressing a series of entrepreneurship research challenges.

The current version of the EMS used for DOM was introduced on May 7th, 2015. Between that date, and June 22nd, 2022, when this section was written, the EMS has gathered data from 13,927 entrepreneurship projects distributed on 12 groups of ES programs (Table 1):

Table 1 Overview of ES programs providing data
Full size table

The EMS allows the entrepreneurs and their teams (under facilitation by the ES) to enter and update digital Cards containing a heading of up to 56 characters into each of the 29 categories. Within the 13,927 projects, and in the time period between May 7th, 2015 and June 22nd, 2022, 935,745 such cards have been entered into the EMS. They are distributed between the elements as follows (Table 2):

Table 2 Overview of collected data on the seven activities and 29 elements
Full size table

Four Elements have no number of cards due to different technical reasons: Element F1 (Revenues) has no unique cards as they are derived from the revenue cards in O5 (Financial Objectives). Element F3 (Task Cost) has no unique cards as they are derived from T1 (Tasks). Element F4 (Financials) has no unique cards as they are derived from the finance cards in R4 (Financial Resources). Finally, F5 (Rules and regulation) has no registered cards yet, as it has just been added in the most recent version of the database and will be collected in the future.

As of January 28th, 2021, we included a simple demographical query in the onboarding process of the EMS 2225 of 3,067 surveyed entrepreneurs have filled in this query so far, representing a response rate of 72.5%. In this sample, the demographical distribution of the entrepreneurs is as follows (Table 3):

Table 3 Demographics
Full size table

As of January 28th, 2021, we included three extra questions in our onboarding procedure, allowing for description of Vertical Industry, Idea Stage and Ambition Level. 1,130 Entrepreneurs have replied on these questions with the following distribution (Table 4):

Table 4 Vertical Industry, Idea Stage and Ambition Level
Full size table

The potential of SEAM for future entrepreneurship research

This section introduces three exemplary research areas within the entrepreneurship research field to illustrate the potential of SEAM, DOM, and the derived data for addressing current research gaps. The chosen areas are entrepreneurial motivation, entrepreneurial ecosystems, and performance indicators for new ventures. We lay out opportunities for future research within these three areas but furthermore describe how a combination of them can lead to novel and interesting insights.

Using these examples, we highlight how different kinds of data can be used, increasing the complexity in each one. The first example illustrates how a qualitative analysis of one element combined with self-assessment via query can be used to research entrepreneurial motivation. The second example on entrepreneurial ecosystems suggests a qualitative analysis combining data from several elements. Finally, the third example on performance indicators for entrepreneurial projects suggests how big-data analytics and machine learning algorithms can be used to exploratively assess the text data from all elements.

Example 1: entrepreneurial motivation (single element)

The first case exemplifies a mixed-method approach that codes qualitative categories and thus enables us to e.g. provide large-scale descriptive analyses: When ES programs are designed, they often assume that the core motivation of entrepreneurs is to achieve financial wealth (Toscher et al., 2020), taking for granted that most entrepreneurs fit into the model of the homo economicus, being “unswervingly rational, and completely selfish” (Levitt & List, 2008, p. 909). This may be related to most studies attempting to make a normative list of motivations or traits that make an entrepreneur, which has been criticized by Gartner (1988) and Ramoglou et al. (2020). For instance, Shane et al. (2003, p. 257) state that “to provide a road map for researchers interested in this area, we discuss the major motivations that prior researchers have suggested should influence the entrepreneurial process”.

An alternative to this approach could be an analysis of what entrepreneurs themselves claim motivates them. On this basis, we could enable the identification of what individual entrepreneurs want to get out of becoming an entrepreneur instead of trying to find personality traits that correlates with this motivation. We argue that this approach may be more practical to understand how entrepreneurs start a new venture and how ES could help them in the best possible manner.

The SEAM dataset has already been used to do a descriptive pre-study of motivation (Toscher et al., ). This was done by manually decoding 1,513 motivation cards filled in by 776 participants in three different ES programs. It was soon clear that all these cards fit into one out of four categories: 21% of the cards was labeled as GET (an external, extrinsic reward in the form of money or recognition above what is a normal salary), 26% of the cards into GIVE (to GIVE something to society. Rooted in idealism or values), 36% of the cards into MAKE (to experience the fulfillment of making a product, service, or organization) and 16% LIVE (necessity entrepreneurs having little other motivation than having a steady income). A more extensive main study is currently taking place, aiming to combine card data and survey data from 5,000 entrepreneurs.

Example 2: Entrepreneurial ecosystems (multiple elements)

Another analysis approach can be to focus on using that data for network analysis: Entrepreneurial Ecosystem research has gained considerable attention over the past decade (Borgatti et al., 2009; Knoke & Yang, 2020). However, the field is still undertheorized and neither the governing mechanisms, nor typologies for entrepreneurial ecosystems, the composition of particular ecosystem elements, or its influencing factors are well understood yet (Cao & Shi, 2021; Theodoraki et al., 2022; Wurth et al., 2021).

The SEAM dataset enables the empirical exploration of ecosystem network typologies from a start-up level of analysis, thereby focusing on the network as one of the ten ecosystem elements specified by Stam and van de Ven (2021). Using the content dimension of the activity system, the dataset allows for the exploration of which actors in the ecosystem are addressed by entrepreneurs to help them. Figure 2 shows a network graph built with the data mining software Orange (Demsar et al., 2013) that exemplifies which ecosystem actors were identified by 402 entrepreneurs in Norway to help them. It includes all terms that were mentioned at least 10 times by the entrepreneurs. The lines connecting the terms indicate that they have been mentioned jointly in one card. The higher the number of joint appearances, the thicker the line.

Fig. 2
figure 2

Illustration of network partners in the ecosystem identified by 402 Norwegian Entrepreneurs

Full size image

Beyond that, using the structure dimension of the activity system, the dataset allows for the analysis of how different groups of entrepreneurs access their ecosystem. For instance, future research could use the SEAM dataset to assess how network typologies of start-ups with individual entrepreneurial motivations differ regarding their network partners and whether these differences hold across regions.

Example 3: data-driven indicators for entrepreneurial projects (all elements)

Another example of the vast analysis possibilities lies in classical hypothesis testing and more advanced multivariate analysis such as Structure equation models, as the SEAM dataset provided both dependent and independent variables for various research questions: Research on startup development and survival has received much attention in recent decades (Soto-Simeone et al., 2020). In this regard, the positive impact of startups on the economy and society is undisputed (Agarwal et al., 2007; Rodrigues & Ac Teixeira, 2021; Zahra & Wright, 2016), but threatened by high failure rates of startups (Headd, 2003). Therefore, there is great interest in exploring the prerequisites and causes for the survival of these ventures, as this is the necessary condition for their long-term growth and success. However, a profound understanding of the activities that influence the survival and success of startups is – as recent studies show – not sufficiently given (e.g. de Mol et al., 2020; Saura et al., 2019; Soto-Simeone et al., 2020). First, there is a lack of distinction between individual goals of entrepreneurs, which necessitate different measures of success. Moreover, indicators determined by traditional valuation methods, such as net asset value or discounted cash flow (DCF) methods, can only be used to a limited extent to value early-stage startups. This complicates both the selection process and the continuous improvement of the design of startup support measures.

The SEAM dataset enables the empirical exploration of how entrepreneurs define the success of their ventures and which performance indicators for the ventures of different groups of entrepreneurs can be derived from their activities. For this purpose, entrepreneurs’ individual motivations are identified based on unstructured text from the SEAM dataset. This is a task of unsupervised learning using natural language processing (NLP) that can be undertaken with text clustering methods.

Financially motivated ventures from the SEAM dataset can be augmented with financial ratios to capture the survival of the resulting ventures at different points in time. The intersection of the datasets requires entity resolution (Ebraheem et al., 2018; Li et al., 2020), which may need to make use of machine learning methods. Using the SEAM dataset (user activity, free text, and numeric data) and financial metrics, classifiers to predict the survival of new ventures can be trained. For this purpose, indicators can be developed in an iterative process and validated as features together with predictive models. Models with the highest predictive power will be selected and finally the global and local influence power of the individual indicators will be examined. The model prediction and influence power of each indicator will be further investigated in three dimensions: 1) to predict the survival of new ventures after the decided time periods (e.g. 1, 3 or after 5 years), 2) based on this, to be able to identify the potentials in the early stages of new ventures (e.g. the starting point of the indicators, which is crucial for survival) and 3) to derive recommendations for action based on the indicators, which will, for example, serve to improve the chances of survival through ES.

Which research questions can be answered by data from SEAM?

We have briefly discussed three examples of insights generated by the SEAM dataset. When considering the whole model and the amount of data gathered over time, the opportunities for new research are abundant. For instance, we suggest that the following research questions that can be answered by analyses of this data:

  1. (a)

    How do entrepreneurs perform the various activities needed to establish a new venture?

  2. (b)

    Which activities are most important for entrepreneurial success?

  3. (c)

    Which specific characteristics of activities are related to entrepreneurial success (e.g., the characteristics of UVPs, the Ecosystems, or the Financial objectives)?

  4. (d)

    How do the various activities influence each other (e.g., how do the characteristics of the vision and motivation influence the financial objectives or the sustainability objectives)?

  5. (e)

    Which activities do entrepreneurs spend more time on – and which activities do they find more difficult?

  6. (f)

    How do the answers on questions (a)–(e) vary across types of entrepreneurs (e.g., commercial vs. social, BTC vs BTB, high-tech vs. low-tech, immigrant vs. native, etc.)?

This list of research questions is far from exhaustive, but even the answers to this limited set of questions would significantly advance our knowledge of how entrepreneurs operate and why they fail or succeed.

Conclusions, limitations and future research

In this paper, we contribute to the development of data-driven and evidence-based entrepreneurship research by developing a holistic activity MODEL of entrepreneurship and by introducing a new METHOD for the structured collection of comparable data on this basis. We furthermore describe the existing DATASET and illustrate three EXAMPLES how this new data can be used in entrepreneurship research.

We used Systems Theory and the Activity System perspective of Zott and Amit (2010) to develop SEAM, a holistic activity model of entrepreneurship. We suggest that by structuring the entrepreneurial process via SEAM, future research can focus on what entrepreneurs do when planning and conducting the development of their ventures, contributing to process-oriented entrepreneurship research (Davidsson & Wiklund, 2001; Uy et al., 2010). This model informed the construction of a digital entrepreneurial management system (EMS), which assists entrepreneurs in making the decisions necessary to develop a full-fledged plan for their venture. It enables the application of non-participant observation studies on entrepreneurship by means of DOM, allowing for future longitudinal studies of entrepreneurial projects independent of geographical presence and direct contact with the observed entrepreneurs, which is exemplified by three research opportunities in this paper. The DOM uses the EMS to register all decisions and accumulate data, which represent new opportunities for analyses of entrepreneurial activities and new opportunities for development of predictive theory on business modelling and entrepreneurial efficiency. We want to highlight the benefit of using a reconciled model of entrepreneurial activities rather than necessarily arguing the virtues of the SEAM-model specifically. First, it is clear that the individual entrepreneur benefits from handling their business development in a structured, iterative and systemic way (Ratinho et al., 2020). Secondly, the coordination of ES initiatives greatly benefits from the presence of a universal language. Not only does this enable interaction between different programs and between different facilitators in the same programs, it is also key to collect comparable data that is needed on a larger scale (e.g. Davidsson & Wiklund, 2001; Guerrero et al., 2021; Uy et al., 2010). Thus, a reconciled model benefits both research and practice.

The data obtained by means of SEAM, the EMS and DOM are bundled and structured by the SEAM Research Institute, a Norwegian non-profit organization that has just been established. Its mission is to allow entrepreneurship and innovation researchers access to the anonymized and GDPR-secured data, leading to a new foundation for data-driven and evidence-based entrepreneurship research contributing to a better understanding of entrepreneurial activities and ecosystems. Policymakers within the financing and evaluation of ES programs will have access to a uniform dataset to base decisions on, whilst the dataset will enable both the development of best-practices within ES as well as entrepreneurship research in general.

Covering seven years of experience and 13,927 entrepreneurial projects using the EMS, the authors describe the system, give examples from projects, and present a set of research questions that can be answered by analyzing data from such systems. Digital EMS provide unique opportunities for bringing research on entrepreneurial activities and business modelling forward.

Next to the discussed opportunities of our suggested MODEL, METHOD, and DATASET, these also come with various limitations. Regarding SEAM, we want to highlight that the choice of elements significantly impacts the research opportunities that flow from using the model. Although we provided a rational for our suggested elements anchored in the current entrepreneurship literature, this needs continuous reflection and adaption, for instance to include new urgent topics of entrepreneurship research such as sustainability. While the EMS and the DOM yield clear advantages compared to interviews or surveys, the data collection method is limited by the willingness for and level of interaction of entrepreneurs on the platform. As for other, established methods, there cannot be a guarantee that the observed entrepreneurial processes represent the full spectrum of the entrepreneurs’ actual activities. Related to the current state of the SEAM dataset, despite the high magnitude of the collected data thus far, we see a limitation concerning the focus on specific geographical regions. At the moment, the Scandinavian countries make about 70% of the dataset.

Based on our propositions we suggest two areas for future research. The first relates to the wide range of research opportunities using the DATASET we already set out in section "Which research questions can be answered by data from SEAM?". The second relates to the improvement of the suggested MODEL, METHOD, and DATASET. Guerrero et al. (2021) highlight a need for studies on the transition from new ventures towards established ventures for innovative and non-innovative entrepreneurship. While our suggested approach certainly covers both these kinds of entrepreneurship, a common model and data collection method would benefit from lock-in mechanisms, making entrepreneurs use it not only for several weeks or months, but for several years. Thus, future research could explore opportunities to achieve larger scale longitudinal data on this basis. Moreover, several authors highlight the current lack of data on entrepreneurial activity in emerging economies (e.g. Guerrero et al., 2021; Marozau et al., 2021). While our dataset covers South Africa to a large extent, future work should address the accessibility and dissemination of unified models and data collection methods in further emerging economies. Another aspect future work should explore and integrate into a model and method for collecting comparable data is measuring venture survival and performance, a field that needs more research and currently lacks sufficient data (Miner et al., 2012; Soto-Simeone et al., 2020; Ungerer et al., 2021).