Startup decision-making: integrating trial-and-error with the entrepreneur-as-scientist approach

Establishing successful companies has long been a significant challenge throughout human history. Despite the availability of extensive information and a wide range of entrepreneurial methodologies, the global startup failure rate remains around 90% (Kalyanasundaram et al., 2021). This suggests that many entrepreneurial projects continue to fail, not due to a lack of theories or methodologies, but rather because of the difficulty in selecting and appropriately adapting the correct information from the vast resources available to fit the specific project. This process, known as tailoring, involves the mindful selection and adjustment of various project factors—such as processes, tools, methods, and models—to determine which elements are most suitable based on the project's type, size, complexity, and specific context (Project Management Institute, 2021).

In new business creation projects, two approaches have been used in recent decades: the trial-and-error model and the entrepreneur-as-scientist model. The trial-and-error model, epitomized by Lean Startup methodologies (Ries, 2011), advocates for experimentation, rapid prototyping, and iterative learning. On the other hand, the entrepreneur-as-scientist model emphasizes hypothesis-driven processes, careful planning, and the systematic testing of beliefs to minimize risks and optimize outcomes (Zellweger and Zenger, 2023). This dichotomy between empirical flexibility and structured rigor underscores a critical gap in understanding how these methodologies influence entrepreneurial performance metrics such as profitability, market share, and innovation capacity. Yet, the entrepreneur is usually led down one path or another. In this research, we have addressed these two approaches through the lens of organizational ambidexterity, which describes how two contradictory and apparently incompatible processes can occur in organizations, even showing positive and significant relationships in the company's performance (Kassotaki, 2022).

Accordingly, this study explores the impact of trial-and-error and entrepreneur-as-scientist approaches on the performance of newly established companies, particularly in terms of problem identification, solution formulation, and value proposition design. The objective is to support entrepreneurs in effectively selecting and tailoring these approaches to enhance business success. Hence, the Ambidextrous Tailoring Model for Experimental-Structured Approaches is presented, based on the concepts of organizational ambidexterity (O'Reilly and Tushman, 2013) and Popperian epistemology (Popper, 2002). The model integrates structured and experimental strategies, offering a framework to examine how entrepreneurs respond to uncertainty (Andriopoulos and Lewis, 2009; Gibson and Birkinshaw, 2004) and serving as a bridge between two previously distinct schools of thought in entrepreneurship research (Camuffo et al., 2020).

The results of this study suggest several important insights. First, it allows for an understanding of how problem identification, the proposed solution, and the design of value propositions influence business outcomes, which may challenge conventional assumptions about the linear progression from problem recognition to business success. Second, the analysis of the moderating effect of entrepreneurial approaches, such as the entrepreneur-as-scientist and trial-and-error perspectives, on the relationship between the solution/value proposition and business outcomes will enable an assessment of whether a balanced application of empirical and theoretical methodologies is possible, particularly in entrepreneurship education and mentoring. Consequently, this paper contributes to entrepreneurship literature by bridging the theoretical gaps associated with experimental and structured entrepreneurial strategies and provides an extended theoretical framework. It also offers actionable insights for educators, mentors, and policymakers aiming to enhance entrepreneurial outcomes.

The remainder of this paper is structured as follows: Section 2 discusses the theoretical underpinnings and the development of hypotheses. Section 3 details the methodology, including data collection and analysis techniques. Section 4 presents empirical findings, while Section 5 outlines their theoretical and practical implications. The paper concludes with recommendations for future research and policy directions.

In ventures embedded in dynamic business environments, it is essential to continuously renew their competitive advantage to achieve strategic market positioning, contribute to society, generate economic growth, and engage in potential networking with other enterprises. These aspects can be linked to organizational ambidexterity as a pathway to entrepreneurial success (Bertello et al., 2022), being a fundamental mechanism for reconfiguring resources to exploit existing competencies while exploring new opportunities (Kolte et al., 2022a, b; Weiss and Kanbach, 2022). The term organizational ambidexterity, introduced in the 1970s, refers to how two contradictory and seemingly incompatible processes (exploring and exploiting) take place in organizations simultaneously (García-Lillo et al., 2016; Kassotaki, 2022). Exploration involves the creation of new knowledge in management, the development of novel organizational systems, innovation through experimentation with new technologies and processes, and long-term orientation, led by open-minded individuals who continually reevaluate beliefs and decisions. Exploitation focuses on improving and increasing efficiency in each experience, deepening the knowledge base of the business, optimizing routines, structures, and organizational systems, and enhancing processes, competencies, and technologies, with an emphasis on execution efficiency and short-term orientation (Brem, 2017; March, 1991; Pertusa-Ortega et al., 2021; Volery et al., 2015).

Ambidexterity has been analyzed from different perspectives. Hwang et al. (2023) found a relationship between organizational ambidexterity and firm performance moderated by open innovation; Sarmento et al. (2024) analyzed the positive effect of organizational ambidexterity on co-creation. Taha et al. (2024) found that various dimensions of capital and financial performance are interconnected through the mediating influence of organizational ambidexterity. Anokhin et al. (2025) showed that technology sourcing ambidexterity (balancing internal and external innovation inputs) yields strategic benefits in firms with considerable organizational slack. Ambidexterity also extends to areas such as sustainability, as proposed by Thrassou and Festa (2023), who argue that aligning sustainable exploitation with sustainable exploration emerges as one of the most relevant challenges of present and future industries, markets, and businesses. Kolte et al. (2022a, b) mention that implementing the exploitation of existing scenarios (e.g. India's global supremacy in generics) with the exploration of potential opportunities derived from R&D allows for improved healthcare delivery. Ju (2022) and Ju and Elliott (2024) also analyzed how ambidexterity works in companies in emerging economies.

Interestingly, there are companies where, by their nature, ambidexterity is present, such as in institutional venture capital firms, who seek to maximize the efficiency of their predominantly financial resources, and at the same time continually invest in new business opportunities (Rossi et al., 2020b), or in large pharmaceutical companies, which are constantly dedicated to feeding (exploration) and boosting (exploitation) their business pipeline (Festa et al., 2018). Some companies have opted to create separate structures, establishing corporate venture units focused on exploring new business opportunities (Rossi et al., 2021). However, mechanisms that foster ambidexterity in new ventures have also been examined (Dai et al., 2017), like Wang et al. (2018) who found that expertise breadth, external connectivity, and empowering leadership support the practice of temporal ambidexterity in new ventures, and Su et al. (2023), who analyzed how startups achieved high business performance by balancing their technology sourcing within an organizational ambidexterity framework.

When reviewing different studies on organizational ambidexterity, it is remarkable to see how two seemingly opposing approaches have succeeded in different business contexts. Even more, by combining the exploration of new market opportunities with the use of existing core competencies through strategic planning, innovation, and execution, it is possible to develop new competitive advantages (Rossi et al., 2020a; Weiss et al., 2024), which, according to the Resource-Based View (RBV), originate from the possession and strategic management of VRIO resources, those that are valuable, rare, inimitable, and organizationally embedded (Barney, 1991) strategic management of VRIO resources, those that are valuable, rare, inimitable, and organizationally embedded.

Popperian epistemology also reframes organizational ambidexterity, but beyond the traditional exploration-exploitation balance (March, 1991), engaging entrepreneurs in theory-driven and empirically grounded learning to face uncertainty in a systematically critical manner. While intuition, interaction, and adaptation are commonly applied in entrepreneurship theories, it is essential to examine the epistemological foundations of entrepreneurial decision-making and the mechanisms through which knowledge is generated and refined. In this regard, Popperian critical rationalism serves as a fundamental base.

Karl Popper's (2012) philosophy of science provides a conceptual foundation for critical rationalism, linking structured experimentation (Entrepreneur-as-Scientist) with heuristic learning (Trial-and-Error). According to Popper, it is falsification rather than verification that advances scientific knowledge, as hypotheses are systematically subjected to attempts at refutation. This principle is particularly relevant within entrepreneurial ecosystems, where decisions are made under conditions of uncertainty, and assumptions about value creation, customer needs, and business models are continually challenged. In Popperian logic, the entrepreneur is not merely a provider of solutions but rather a problem solver (Harper, 1999; Popper, 2012). The Popperian cycle; conjecture → refutation → new conjecture closely mirrors the entrepreneurial learning process, in which experimentation (with prototypes, markets, and customers), trial, error, and falsification foster continuous improvement and adaptation of entrepreneurial offerings.

Epistemologically, both the Entrepreneur-as-Scientist and Trial-and-Error approaches are grounded in Popperian methodology. In the former, the formulation of a value proposition is operationalized through Minimum Viable Products (MVPs) and market experiments, which are designed to expose weaknesses (Zellweger and Zenger, 2023), using structured reasoning, deductive logic, and iterative empirical testing. In the Trial-and-Error approach, Popper's notion of knowledge growth emerges through repeated failures that enhance environmental understanding and enable the elimination of errors (Popper, 2012; Snihur and Eisenhardt, 2022).

Within the Popperian framework, entrepreneurial opportunity discovery must be analyzed as a unique process involving novel combinations of elements. The inherent novelty shaped by uncertainty, complexity, and ongoing change renders the accurate prediction of future opportunities through large-scale statistical modeling implausible. Therefore, a constant critical approach, exploration of novelty, and contextual monitoring are essential (Murphy, 2009). In this light, it becomes imperative to analyze entrepreneurial ecosystems and their interactions with entrepreneurs from a systems thinking perspective (Bertello et al., 2022).

Popperian epistemology elevates both the Entrepreneur-as-Scientist and Trial-and-Error approaches to knowledge-construction processes. In this context, Popper's critical rationalism conceptualizes entrepreneurial behavior through two complementary lenses: (1) structured (Entrepreneur-as-Scientist) and (2) heuristic (Trial-and-Error), both contributing to knowledge generation via the systematic formulation and refutation of hypotheses. It offers a framework to interpret the entrepreneur's cognitive capacity as grounded in critical experimentation (discovery) and the exploitation of corroborated knowledge (validation), positioning organizational ambidexterity as a strategic expression of Popperian epistemology.

Thus, Organizational ambidexterity and Popperian epistemology offer a robust theoretical perspective as a framework for examining the coexistence of two seemingly contradictory approaches that, together, help entrepreneurs cope with uncertainty: trial-and-error and the entrepreneur-as-scientist approach. This coexistence can be understood through the concept of “tailoring,” advocated by the Project Management Institute (2021) as a means to enhance project success, including early ventures.

Entrepreneurs in the early stages of business development often rely on various methodologies to navigate challenges and uncertainties such as Design Thinking (Braz et al., 2019) and Lean Startup (Ries, 2011), and with the help of tools such as Business Model Canvas (Osterwalder and Pigneur, 2010) and Lean Canvas (Maurya, 2012), it is possible to help startups design their business models, monitor progress and testing hypotheses (Euchner and Blank, 2021). Specifically, the Lean Canvas was designed for entrepreneurs in the early stages of their projects; while all the blocks are necessary, the correct construction of a business model begins when the client's Problem is defined correctly, identifying what the client truly values to create a suitable value proposition, and ensuring that the solution accurately addresses the identified Problem. Figure 1 illustrates the initial steps for building an idea based on the Lean Canvas.

In entrepreneurship, an unsolved problem is an opportunity to create a company. Therefore, identifying a problem can also be called opportunity recognition or opportunity identification. According to Swartz et al. (2022), opportunity identification results from developing the entrepreneur's identity and rejecting the status quo by seeking the transition from a personal life stage and the search for an external opportunity. Opportunity recognition is an essential part of entrepreneurship (Hanohov and Baldacchino, 2018), the first step in the entrepreneurial process (Kuratko et al., 2021a, b), and a key piece at a theoretical level to understand entrepreneurial endeavors (Kreuzer et al., 2022).

When an entrepreneur identifies an opportunity and develops a business project to exploit it, they do so in a context of highly uncertain (Wang et al., 2022), ranging from uncertainty about the existence of market demand for the solution, the entrepreneur's technological capacity to offer it, to uncertainty about the institutional environment (D'Andrea, 2023); thus, despite the high levels of stress generated by this highly uncertain environment and even impacts on organizational creativity, many entrepreneurs are motivated by hope and optimism to achieve their goals, (Darvishmotevali et al., 2020; Zayadin et al., 2022), constantly seeking, adapting and shaping strategies and resources that allow them to boost or attenuate volatile, uncertain, complex and ambiguous (VUCA) environments (Enwereji et al., 2024; Rindova and Courtney, 2020; Scheibel et al., 2025). Thus, entrepreneurs try to solve problems in creative ways, as suggested by Chang and Chen (2020), who found that resource availability generally strengthens the effect of entrepreneurial creativity and opportunity recognition on entrepreneurs' career success.

Problem-solving is one of the entrepreneurs' most prominent personality traits (López-Núñez et al., 2020), helping them ideate possible solutions and move forward with their business (Weinberger et al., 2018). Accordingly, the problem-solving process translates into a market opportunity. Thus, once entrepreneurs deeply understand the problem, they can begin to create solutions for it (Linton and Klinton, 2019). Based on the relevance of the previous processes in company creation, it is expected that the company's outcomes in terms of profitability, sales growth, market share, innovation capacity, and job creation will be influenced by the identification of the Problem and the Solution carried out by the entrepreneur. Hence, the following two hypotheses are proposed:

When a solution is developed, it should ideally offer value to the customer. Several authors have addressed the concept of value proposition. Biloshapka and Osiyevskyy (2018) define the value proposition as the promised set of benefits the company offers its customers. Rydehell et al. (2018) state that the value proposition is reflected in the degree to which a product or service is unique or different from its competition. A common thread among these perspectives is that “the value proposition is the reason why customers turn to one company over another” (Osterwalder and Pigneur, 2010, p. 22), highlighting why this element is key to the success of the company. Payne et al. (2017) also observed that the customer value proposition captures the essence of strategic marketing by helping the company achieve superior performance through the identification of the firm's competitive advantage. Similarly, Sjödin et al. (2019) found that business success is determined by aligning specific value creation and value capture activities throughout the business model innovation process. Based on the above, hypothesis 3 is postulated:

When analyzed through the epistemological lens of Popper (2002), both the Entrepreneur-as-Scientist approach and the Trial-and-Error method reveal a common foundation in the formulation of falsifiable hypotheses. The Entrepreneur-as-Scientist approach applies this principle through a systematic and deductive structure, whereas the Trial-and-Error method does so via iterative learning processes. In both cases, the entrepreneur engages in a cycle of conjectures and refutations, which enables the pursuit of entrepreneurial success.

Although agile methodologies such as Lean Startup adopt scientific logic into the process (Ghezzi and Cavallo, 2020), their weakness lies in the insufficient attention given to the value of carefully crafting a theory that is amenable to efficient testing and the careful interpretation of feedback. To address this, Zellweger and Zenger (2022) offer an integrative, pragmatic perspective of the Entrepreneur-as-Scientist. According to these authors, it is possible to explain how entrepreneurs seek to create value in uncertainty through a sequential process (Zellweger and Zenger, 2023): (1) Belief formation, where a value creation theory is developed and supported by testable hypotheses, (2) Belief testing, which involves testing the proposed hypotheses, and (3) Responding, where an effective response to update beliefs provides the entrepreneur with a means to manage uncertainty and explore the true value of their beliefs. Thus, according to these authors, unlike the Lean Startup, the entrepreneur with a scientist's perspective places greater emphasis on the impact and value of carefully formulating beliefs about opportunities before testing them in the market.

From this perspective, this study hypothesizes that the relationship between business performance and problem identification, solution proposal, and value proposition design will be moderated by the entrepreneur's influence when they develop these processes with the rigor offered by the Entrepreneur-as-Scientist approach. The following three hypotheses are proposed:

Lean Startup methodologies help reduce costs and time by allowing startups to pivot their business model, secure funding, and better understand and satisfy customer needs (Ghezzi and Cavallo, 2020; Lindgren and Münch, 2016). Accordingly, the Agile methodology is widely applied, where Trial-and-Error, the Build-Measure-Learn framework, and iterative development of the minimum viable product are used to refine the final product (Edison et al., 2015). In the dynamic process of business model, the entrepreneur must reconfigure their business into a new business model, and using agile methodologies facilitates faster adaptation.

Zhuge et al. (2023), in a sample of 325 Chinese tech startups, examine the iteration of the proper definition of the Problem, the proposed solution, and the value proposition about market dynamics, finding that Trial-and-Error positively moderate company performance and recommending entrepreneurs strongly utilize the Lean Startup methodology with deep iteration to improve business outcomes. Thus, Trial-and-Error is employed to identify the Problem through customer needs, innovate in the proposed solution, and adjust the value proposition through iterative learning, avoiding market assumptions that could lead to an incorrect definition of demand (Zhuge et al., 2023). Furthermore, Dehghani et al. (2022), in a qualitative study involving 10 startup managers, explore the iterative learning model as a means to adapt to market-driven performance, helping entrepreneurs define the Problem, design the solution, and create a value proposition aligned with customer needs. Hence, based on this, the following hypotheses are formulated:

Studies carried out internationally, such as the Global University Entrepreneurship Spirits Students' Survey (GUESSS) and Global Entrepreneurship Monitor (GEM), enable the analysis of gender in the entrepreneurial community. According to the GEM report (2023) women lead one-third of businesses. Additionally, 25% of businesses worldwide are managed by women, with higher percentages observed in low-income countries and North America. However, a phenomenon observed in micro-commerce among non-university educated individuals is that while women initially start many businesses, they often end up being controlled by men, a trend strongly influenced by the cultural context (Gbandi et al., 2023).

Academic entrepreneurship practices promote a more balanced environment and challenge negative cultural stereotypes of female entrepreneurship that may be prevalent in the given cultural context (Hagg et al., 2023). In this regard, Gonzalez-Tamayo et al. (2024) found that gender equity is more advanced among university entrepreneurs. Likewise, Brixiová et al. (2020) identify a positive effect of tertiary education on the performance of women-led ventures, particularly in developing essential soft skills critical for success. Hence, university education plays a crucial role in promoting gender equity, enabling women to overcome challenges in leadership roles (Maheshwari et al., 2025; Donaldson et al., 2022).

In terms of performance, studies indicate that men benefit more from training, while women with a university education show a positive relationship with sales and greater tenacity (Brixiová et al., 2020). Additionally, Santos-Jaén et al. (2022) found that women place significant value on soft skills, especially during the early stages of their business, as well as on building trust, both interpersonal and professional, essential for success in their ventures. Based on this discussion, the following hypothesis is proposed:

Nationality is “the legal bond between a person and a state, denotes membership to said state, and gives rise to reciprocal rights and duties between individual and state” (van Waas and Recalde Vela, 2023). Culture, by contrast, is a term that has been polysemic, as an example we see that in the classic monograph by Kroeber and Kluckhohn in 1952 there are more than 160 definitions, so it is recommended to specify how the term will be used in each particular context (Jahoda, 2012). In this study, culture is addressed as a multidimensional construct encompassing the values, norms, and shared beliefs of a group, shaped by factors such as religion, ethnicity, and region.

Hofstede's six cultural dimensions theory has been extensively applied across various fields, particularly in entrepreneurship, to examine how cultural values shape behavior. The dimensions include power distance index, individualism, masculinity, uncertainty avoidance index, long-term orientation, and indulgence vs restraint (Hofstede, 2011). National culture also influences entrepreneurial decisions: in countries with low individualism, low power distance, and low uncertainty avoidance, people are more likely to acquire existing businesses. However, entrepreneurship is often preferred in high-power-distance countries, even in non-entrepreneurial cultures (Block and Walter, 2017). The intercultural cognitive model highlights the significant role of context in shaping entrepreneurial development, focusing on how situational influences, beliefs, and behavioral patterns associated with each nationality affect individuals' approach to entrepreneurship.

Therefore, the national context exerts a significant influence on innovation strategies, both at the level of individual entrepreneurs and within multinational firms (Boone et al., 2019). In cross-cultural and international settings, the presence of nationally diverse top management teams tends to enhance firm performance, as such diversity introduces a broader repertoire of perspectives and strategic capabilities (Reckendrees et al., 2022). Álvarez et al. (2025), using country-level data from 99 countries between 2001 and 2017 sourced from the Global Entrepreneurship Monitor (GEM), conclude that for the survival of new ventures, institutional dimensions; regulative, normative, and cultural-cognitive are a determining factor through national public policies aimed at fostering entrepreneurship. In addition, nationality, mediated by contextual factors, influences the opportunity cost of engaging in entrepreneurial activities (Dheer et al., 2019). The national origin of entrepreneurs plays a fundamental role in shaping their perceptions of risk, value systems, beliefs, and prior experiences since “nationality is a marker of social identity, shaping people's beliefs, values, and how they view the world” (Mohan et al., 2024, p. 1). Based on this, hypothesis 7 is proposed, completing the configuration of the conceptual framework model (Figure 2).

The data for this study were sourced from the Global University Entrepreneurial Spirit Students' Survey (GUESSS Project) ^[1], with some authors, as country leaders, having access to the dataset. The GUESSS Project aims to serve as a benchmark for university entrepreneurship, enabling multi-level analysis at a global scale by country and research teams (Sieger et al., 2024). The project underwent rigorous ethical review and was approved by the Ethics Committee of Universidad Católica del Uruguay, with approval documented under resolution number 230901b, available on the project website. GUESSS coordinated the distribution of a pre-validated survey link, translated into various languages, which was sent to country leaders. These leaders collaborated with participating universities to promote the survey and maximize response rates while minimizing potential biases. Participation was voluntary and initiated via email invitations. In compliance with the General Data Protection Regulation, participants signed a consent form before completing the questionnaire.

A non-probabilistic convenience sampling technique was employed, as participation was limited to students from participating universities (Sieger et al., 2016). Given the complex nature of the entrepreneurship discipline, many studies have employed non-probability sampling techniques despite their limitations in generalizability (Maheshwari and Kha, 2022; Nowiński et al., 2019), and hence, for this study, the non-probability sampling technique was found to be suitable. Data was collected globally from September to December 2023. The 10th edition of the study involved 57 countries and resulted in the participation of 226,716 students. For this study, the objective was to analyze how the Trial-and-error and Entrepreneur-as-scientist approaches influence company performance, focusing on identifying problems, proposing solutions, and designing value propositions. To this end, only students who were currently running their own companies and had provided complete responses were included in the analysis. The final sample comprised 20,631 students, of whom 72% were undergraduate students, 14% were at the master's level, 4% were pursuing PhDs, and the remaining 9% were in other categories. The average age of undergraduate students was 24 years. The sample had a nearly balanced gender distribution, with females accounting for 47% and males 53%. Students represented 89 nationalities, predominantly from America and Europe (Table 1).

The dependent variable for this study was Business Outcomes (BO), measured using a five-item Likert scale ranging from 1 (strongly disagree) to 7 (strongly agree). The items for this variable were adapted from Eddleston et al. (2008) and simplified in alignment with the GUESSS framework to enhance response rates among students, as recommended by Dess and Robinson (1984).

There were three independent variables in this study: Problem (Prob), which focuses on customer-oriented entrepreneurial approaches and was assessed with a two-item scale; Solutions (Sol), which addresses ideation, business development, creativity, customer orientation, and a rational approach to problem-solving, assessed using a seven-item scale; and Value Proposition (VP), which captures the articulation of the value proposition and its associated solutions, measured with a two-item. Each item of the independent variables was measured using a Likert scale ranging from 1 (strongly disagree) to 7 (strongly agree).

Additionally, the study included two moderating variables: Entrepreneurs as Scientists (ES), reflecting adaptability, iterative refinement, and a focus on the commercial viability of decisions, assessed with an 11-item scale adapted from Zellweger and Zenger (2022), and Trial-and-Error (TE), measured using a three-item scale inspired by the Lean Startup methodology, which evaluates agile, experimental, and adaptive entrepreneurial practices. The items for both moderating variables were also measured on a Likert scale ranging from 1 (strongly disagree) to 7 (strongly agree). Lastly, two control variables, gender and nationality, were included in the analysis to account for potential demographic influences. The item descriptions for each variable are detailed below, as outlined in Table 2.

The data for this study were analyzed using a three-step hierarchical regression model (see the conceptual model as presented in Figure 2), selected for its suitability in testing moderator effects and its widespread application in entrepreneurship and management research (Yang and Xiangming, 2024; Maheshwari, 2021). This method allows for the stepwise integration of variables across models, facilitating a structured examination of the relationships among predictors, moderators, and dependent variables. Model 1 included only control variables (gender and nationality). In Model 2, the main independent variables (Problem, Solution, and Value Proposition) were added, while Model 3 introduced interaction terms to test moderation effects (e.g. Problem × Entrepreneurs as Scientists).

To minimize multicollinearity, all predictor variables were standardized prior to the creation of interaction terms (Aiken, 1991). Change in explained variance (∆R²) were assessed to determine the contribution of each step, and F-statistics were used to confirm model significance. The analysis was conducted using SPSS version 28 for hierarchical regression and AMOS version 28 to evaluate the goodness-of-fit of the measurement models, particularly using CFI, TLI, and RMSEA indices.

To ensure the robustness and reliability of the model, preliminary checks were performed, including assessments of normality (via skewness and kurtosis), multicollinearity (using VIF), internal consistency (Cronbach's alpha (CA) and composite reliability (CR)), and discriminant validity (square root of AVE versus inter-construct correlations). These steps are reported in detail in the Results section.

To understand the influence of educational level on business outcomes, an additional one-way Analysis of Variance (ANOVA) was conducted to examine whether business outcomes differ across educational levels. The categorical variable on educational levels included four groups: Undergraduate (Bachelor level), Graduate (Master level), PhD, and Other (e.g., MBA). Educational background can shape entrepreneurial outcomes through varying degrees of exposure to entrepreneurship, access to resources, cognitive skill development, and networking opportunities (Van der Sluis et al., 2008; Marvel et al., 2016). Hence, to determine specific group differences, Tukey's HSD post hoc test was employed. This supplementary analysis was conducted to enhance the interpretability of the results and to address potential heterogeneity in entrepreneurial outcomes across different stages of academic progression.

The descriptive statistics in Table 2 highlight key insights across constructs. The Problem, Solution, and Value Proposition constructs have high mean scores of 5.67, 5.58, and 5.53, respectively, with moderate variability, indicating strong positive perceptions. Entrepreneurs as scientists (ES) and trial-and-error (TE) has means of 5.52 and 5.23. In contrast, the Business Outcome construct has the lowest mean of 4.81 and the highest variability (SD = 1.43), suggesting slightly lower ratings and more diverse responses.

The correlations with the dependent variable, Business Outcomes, show significant but moderate positive relationships with all other constructs, as presented in Table 3. The strongest correlation is observed with TE (r = 0.416), followed by Solution (r = 0.426) and Value Proposition (r = 0.408), underscoring their importance in influencing business outcomes. Weaker correlations are noted with Problem (r = 0.339) and ES (r = 0.402), suggesting that while these constructs contribute, their impact is less pronounced compared to others.

The model was first tested for skewness and kurtosis to assess data normality for the dependent, independent, and moderator variables. The skewness and kurtosis values, presented in Table 2, were within the acceptable ranges of −2 to +2 and −7 to +7, respectively, confirming that the data were normally distributed (Hair et al., 2021). To evaluate multicollinearity, the Variation Inflation Factor (VIF) values were calculated, all of which were below the threshold of 5 (Maheshwari et al., 2023) (Table 3), indicating no multicollinearity issues. Next, to assess the model's reliability and validity, the gender and nationality variables were excluded due to their qualitative nature. Internal reliability was measured using Cronbach's alpha (CA) and composite reliability (CR) (as shown in Table 2), while discriminant validity was evaluated through the square root of average variance extracted (AVE) (as shown in Table 3). Cronbach's alpha values ranged from 0.764 to 0.918, except for Value Proposition (VP), which had a lower value of 0.663, falling below the recommended threshold of 0.7.

Nevertheless, the VP items were retained, supported by prior research suggesting that alpha values below 0.7 can be acceptable in exploratory studies (Schmitt, 1996). CR values ranged from 0.856 to 0.933, exceeding the 0.7 threshold and ensuring strong internal reliability. The AVE values for constructs ranged from 0.561 to 0.824, exceeding the threshold value of 0.5 (Bagozzi and Yi, 1989). Discriminant validity was confirmed by comparing the square root of AVE with construct correlations, where the AVE values consistently surpassed their correlations (Table 3). This demonstrated the distinctiveness of each construct and affirmed the model's discriminant validity (Fornell and Larcker, 1981).

After assessing normality, multicollinearity, reliability, and validity, AMOS version 28 was employed to evaluate the fitness of the measurement model (as shown in Table 4). The results indicated a strong model fit, with CFI and TLI values exceeding the recommended threshold of 0.9 and RMSEA values remaining below 0.07 (Hair et al., 2021).

The one-way ANOVA analysis suggested significant differences in business outcomes across the four educational levels (F = 24.304, p < 0.001), indicating that students' level of study is associated with variations in reported business outcomes. To explore these differences further, Tukey's HSD post hoc test was conducted. The results demonstrated that students in the “Other” category (e.g. MBA programs) reported significantly higher business outcomes (Mean = 5.072) compared to those in all other groups: PhD (Mean = 4.642), Graduate/Master's (Mean = 4.641), and Undergraduate (Mean = 4.835). Specifically, the mean differences between MBA students and PhD (mean diff = 0.42934, p < 0.001), MBA and Graduate (mean diff = 0.43024, p < 0.001), and MBA and Undergraduate (mean diff = 0.23717, p < 0.001) were statistically significant. Additionally, undergraduate students also outperformed PhD and graduate students, with significant differences observed between undergraduate and PhD (mean diff = 0.19218, p = 0.003), and undergraduate and graduate students (mean diff = 0.19307, p < 0.001).

These findings suggest that MBA-level education may offer practical and strategic advantages, including enhanced access to entrepreneurial networks, advanced managerial training, and exposure to market-driven problem solving, which translate into improved business performance. While differences may partly reflect unequal group sizes (with undergraduates forming the largest group), the use of Tukey HSD, which adjusts for such imbalances, ensures the robustness of these results.

Before testing the five hypotheses designed in the study, it was essential to address the potential concern of common method bias, which is particularly relevant in self-reported surveys. To examine this, Harman's one-factor test was conducted as a commonly used diagnostic tool. The analysis suggested that a single factor accounted for 40.6% of the total variance explained, which is below the accepted threshold of 50% (Podsakoff et al., 2003). This result confirms that common method bias did not substantially impact the study's findings.

The analysis of our final model (Model 3), as in Table 4, suggests that gender (β = −0.080, p < 0.001) and nationality (β = −0.004, p < 0.001) significantly and negatively influence business outcomes, supporting hypotheses H₆ and H₇. These findings highlight the impact of demographic factors and the need to address systemic disparities. Among the independent variables, Solution (β = 0.198, p < 0.001) and Value Proposition (VP) (β = 0.178, p < 0.001) are positively significant, supporting hypotheses H₂ and H₃. Both variables positively influence business outcomes, with Solution having a slightly stronger effect. This highlights the importance of effective problem-solving and clearly articulating unique value propositions to achieve favorable results. In contrast, the third independent variable, Problem, is not statistically significant (p > 0.1), indicating its limited direct impact on business outcomes and failing to support the first hypothesis (H₁).

Next, we examined interaction effects to evaluate the influence of the moderator variables Entrepreneurs as Scientists (ES) and Trial-and-Error (TE) on business outcomes. The interaction between VP and ES (VP * ES) is significant (β = 0.041, p < 0.05), positively impacting business outcomes and supporting hypothesis H_4c. This indicates that the moderator variable ES enhances the relationship between a well-defined Value Proposition and business outcomes. Additionally, the interactions between Sol and ES (Sol * ES) and VP and TE (VP * TE) are significant at the 10% level (β = 0.031, p < 0.1) and (β = 0.028, p < 0.1), respectively, supporting hypotheses H_4b and H_5c. These results suggest that ES and TE act as moderators (though not very significant), strengthening the relationship between Solution and Value Proposition and business outcomes, though their significance is weaker compared to other effects. Among these interactions, VP * ES demonstrates the strongest impact, followed by Sol * ES and VP * TE.

Conversely, the interactions Prob and ES (Prob * ES), Prob and TE (Prob * TE), and Sol and TE (Sol * TE), do not show statistically significant effects. This indicates that the identified problem and its interactions with the moderator variables ES and TE do not significantly contribute to the predicted outcomes.

The graphs further (Figure 3) illustrate the moderating role of Entrepreneurs as Scientists in the relationship between business outcomes and two key factors: Solution and Value Proposition. The first graph (graph a) shows that for entrepreneurs with a low scientific identity, business outcomes improve marginally as solutions progress. In contrast, those with a strong scientific identity see a more significant improvement, highlighting the benefit of a scientific approach in leveraging solutions for better business outcomes. The second graph (graph b) suggests that entrepreneurs with a low scientific identity experience only a slight increase in business outcomes as their value proposition strengthens, while those with a high scientific identity see a more pronounced improvement, emphasizing their ability to translate a strong value proposition into success. The third graph (graph c) shows that entrepreneurs with low trial-and-error practices experience only a modest improvement in outcomes as their value proposition improves. However, those with high trial-and-error practices show a slight improvement, indicating that entrepreneurs with a scientific orientation are better at converting strong value propositions into measurable success.

The summary of the results of the study's hypotheses is presented in Figure 4 and Table 4.

The findings provide valuable insights into how trial-and-error and entrepreneur-as-scientist approaches impact key business outcomes such as profitability, sales growth, market share, innovation capacity, and job creation. A key finding of this study is that, in contrast to the findings of Chang and Chen (2020), problem identification does not have a direct impact on business performance. The lack of direct significance of problem identification in this context may stem from its role as a foundational step in the early stages of entrepreneurship. At the same time, the solution and value proposition, which are more directly tied to market outcomes, play a more immediate and impactful role in business performance. This suggests that problem identification is important but not sufficient by itself to drive business success.

The results of this study show that both the solution and the value proposition positively impact a company's outcomes, consistent with the findings of Linton and Klinton (2019) and Sjödin et al. (2019), with the solution having a slightly greater effect. This supports the fundamental purpose of a company: to address a customer's unmet need, which is crucial to its success. Additionally, if the company creates a competitive advantage that positions it ahead of its competitors, it contributes further to business success. Therefore, while the solution is the essence of a company's success, the value proposition serves as the driving force behind that success. Based on this, it is recommended that the development of the solution proposal be prioritized as a central aspect of the entrepreneurial process.

When analyzing the impact of the Entrepreneurs as Scientists (ES) and Trial-and-Error (TE) approaches on the relationship between the two key elements, solution and value proposition, that influence business performance, the results provide several insights. The Entrepreneurs-as-Scientists approach significantly strengthens the connection between a well-defined value proposition and business outcomes, supporting the idea that entrepreneurs who methodically formulate their beliefs and hypotheses before testing them in the market tend to achieve better results. This finding aligns with Zellweger and Zenger (2023) and underscores the value of a structured, scientific approach to business development. On the other hand, the Trial-and-Error approach, while influencing the relationship between the value proposition and business performance, shows a weaker effect. The TE methods also do not significantly impact the relationship between solution definition and business performance, suggesting that an unstructured, trial-based approach is less effective in this context. These results suggest that the Entrepreneurs-as-Scientists approach has a stronger and more consistent impact on business success compared to the Trial-and-Error approach. The weaker effect of the TE method may stem from its reliance on an iterative process that may not consistently yield the clear insights needed for long-term strategic decisions.

Our finding that the Entrepreneurs-as-Scientists approach strengthens the link between a well-defined value proposition and business outcomes should be interpreted with caution. While the statistical analysis shows a significant association, the mechanism likely lies in how a methodical, hypothesis-driven approach reduces uncertainty and facilitates more targeted resource allocation. By explicitly articulating assumptions and testing them systematically, entrepreneurs can identify flawed ideas early, pivot more effectively, and refine their value proposition based on validated learning. This is consistent with prior work on hypothesis-driven entrepreneurship (Camuffo et al., 2020) and decision-making under uncertainty (McGrath, 2010), which suggests that structured experimentation improves strategic choices. Nevertheless, we acknowledge that our study cannot establish causality, and future experimental or longitudinal designs could provide stronger evidence for this relationship.

Lastly, the ANOVA analysis by educational level provides deeper insight into the heterogeneity of entrepreneurial performance among students. The significant variation in business outcomes across education levels, particularly the higher outcomes reported by MBA (Other) students, highlights the potential influence of advanced academic exposure on entrepreneurial success. This may be attributed to MBA students' greater access to strategic training, experiential learning, and professional networks that foster opportunity recognition and venture growth. Interestingly, undergraduate students also performed better than PhD and Master's students, suggesting that early-stage entrepreneurial enthusiasm or applied business learning at the bachelor's level could play a role.

This study aimed to examine the impact of trial-and-error and entrepreneur-as-scientist approaches on the performance of newly established companies, particularly in terms of problem identification, solution formulation, and value proposition design. The above is to help the entrepreneur select and tailor trial-and-error and entrepreneur-as-scientist approaches to achieve business success. The results suggested no significant direct relationship between problem identification and business performance (outcomes). However, both the solution and value proposition were found to significantly impact business outcomes, with the solution having a slightly greater effect. After analyzing the relationships between problem identification, solution, value proposition, and business outcomes (in terms of profitability, sales growth, market share, innovation capacity, and job creation), we examined how the selected approaches (trial-and-error and entrepreneur-as-scientist) moderate these relationships. The findings show that both approaches influence the relationship between value proposition definition and business outcomes. However, the entrepreneur-as-scientist approach impacts both the solution-business outcomes and value proposition-business outcomes relationships, while the trial-and-error approach only affects the value proposition-business outcomes relationship. Since no significant direct relationship was found between problem identification and business outcomes, neither approach influences this relationship. The theoretical and practical implications of these findings are discussed next.

This paper arises from the challenge of measuring the effectiveness of experimental models, which are commonly used by nascent entrepreneurs and educators (Balocco et al., 2019) but have lacked strong theoretical foundations for evaluation (Honig and Hopp, 2016). The study's contribution lies in providing an analysis of over 20,000 university students from diverse nationalities who are currently running businesses, offering valuable insights into the experimental (trial-and-error) and structured (entrepreneur-as-scientist) approaches. Focusing on the early stages of the entrepreneurial process, particularly during the gestation of the idea and development of the business model, this research explores how entrepreneurs identify a problem, develop a solution, and offer a value proposition to the market. It examines whether these actions are more aligned with an experimental or structured process, and how these approaches impact business outcomes.

When analyzing the trial-and-error approach, the results indicate that while it is widely used, it only influences the relationship between the value proposition and business outcomes. This suggests that although the experimental approach can help entrepreneurs develop an initial business model, only the value proposition created using this method will significantly affect key business performance indicators such as profitability, sales growth, market share, innovation capacity, and job creation. Notably, the study's sample consisted of university students with an average age of 24, and experimental approaches rooted in experience and rapid learning may yield different results with older entrepreneurs who possess more practical business experience. The entrepreneurs-as-scientists approach, by contrast, has a greater impact on the relationships between solution development and value proposition with business outcomes. This structured approach is well-suited for university students in the early stages of developing their business ideas. However, it is important to note that this approach demands more resources, such as time, knowledge, and financial investment, than the experimental method. Although the Problem identification does not have a direct relationship with business outcomes, it is still relevant since we consider it corresponds to the initial stage of the creation of the business idea, being the basis for the creation of an entrepreneurial opportunity idea and the starting point for an entrepreneurial journey (Selden and Fletcher, 2015). Hence, this variable is still suggested to be included in the theoretical model.

Aligned with organizational ambidexterity, which posits that two contradictory and seemingly incompatible processes can successfully occur within organizations (Kassotaki, 2022), and Popperian epistemology that engages entrepreneurs to face uncertainty in a systematically critical manner (March, 1991), we propose tailoring both entrepreneurial approaches (structured and experimental) depending on the results of the hypotheses. Accordingly, we propose the “Ambidextrous Tailoring Model for Experimental-Structured Approaches” (Figure 5), where the foundation of an entrepreneurial project lies in the accurate identification of the problem (Hanohov and Baldacchino, 2018; Kreuzer et al., 2022; Kuratko et al., 2021a, b). Subsequently, based on the results of the hypotheses, we recommend selecting the entrepreneur-as-scientist approach during the process of creating and validating the business idea's solution. In parallel, the trial-and-error approach should be used to the value proposition design process. While both approaches (entrepreneur-as-scientist and trial-and-error) contribute to business success when developing a value proposition, the trial-and-error approach, by adhering to Lean Startup principles, allows for resource optimization (Ghezzi and Cavallo, 2020; Lindgren and Münch, 2016; Ries, 2011), which is especially helpful for entrepreneurs in the initial stages of their project, where there is usually a lack of resources.

Although these approaches may seem contradictory and incompatible (experimental vs structured), our model proposes that their coexistence is feasible when tailored to the appropriate stages and variables. This aligns with the principles of Organizational ambidexterity and Popperian epistemology, proposing that it is possible to systematically tailor two seemingly opposing approaches in early-stage business development to achieve the desired business performance.

It is worth noting that in this paper, beyond addressing organizational ambidexterity as a mechanism where exploration and exploitation coexist in companies, ambidexterity is analyzed from the understanding of how two contradictory and seemingly incompatible processes occur simultaneously. This latter approach is the basis of the proposed model. Thus, this approach can encourage the analysis of organizational ambidexterity from a more holistic perspective that helps understand the functioning and integration of seemingly incompatible systems in other areas of knowledge.

Given that this paper suggests the integration of two approaches with significantly different methodologies, it is recommended that the process be guided by individuals with extensive knowledge of both approaches. Mentoring is crucial in this context, as extensive research consistently underscores its vital role in enhancing an entrepreneur's likelihood of success (e.g. Fauchald et al., 2022; Kubberø et al., 2018; Kuratko et al., 2021a, b; Nabi et al., 2019). Mentors should provide strategic insights; help navigate uncertainties and ensure that entrepreneurs do not fall into common traps. An important implication of this study is the recognition that mentors, while vital to entrepreneurial success, may themselves require targeted support and development. There are instances where mentors lack specific capabilities or relevant industry experience for the business type they are advising, which can lead to misaligned guidance, conflicts with mentees, and suboptimal business outcomes. Addressing this requires structured mentor training programs and more deliberate mentor–mentee matching processes. Furthermore, the findings are relevant not only for entrepreneurs engaged in creating and innovating but also for those pursuing acquisition entrepreneurship and intrapreneurship, where the business model, risk profile, and capability requirements may differ significantly. Policies and programs should therefore be tailored to support these diverse entrepreneurial pathways, ensuring that mentorship, funding, and capacity-building initiatives are aligned with the unique needs of each type.

For entrepreneurship training institutions, it is important to continue teaching both well-structured business plans and experimental approaches, while emphasizing that results from experiments should be treated with caution and accompanied by a mentor's guidance. Institutions should consider incorporating mentorship programs into their curricula and providing a network of mentors or guiding students towards external mentorship, ensuring that students receive comprehensive support in both experimental and structured methods. Mentoring training programs should offer clear guidelines on when to apply each approach, balancing theory with practical support.

Furthermore, the observed differences in business outcomes across educational levels highlight the distinct learning needs, prior experiences, and developmental capacities of students at various academic stages. Exploring such program-level differences can yield deeper insights into the effectiveness of curriculum design and engagement strategies. Tailoring entrepreneurship education to the specific needs of each educational level may enhance the impact and relevance of university-led entrepreneurship initiatives, fostering more inclusive and effective entrepreneurial ecosystems.

Additionally, taking into consideration the resource-based view (RBV), university incubators should extend the scope of the RBV within innovative ecosystems by positioning themselves and the student-led startups they support as ambidextrous organizations. This entails developing capabilities that enable a more effective utilization of financial, human, and social resources, thereby facilitating rapid adaptation to business and environmental changes (Colombo et al., 2024; Zahra, 2021).

The institutional frameworks (regulative, normative, and cultural-cognitive) define the level of support entrepreneurs may receive and influence the achievement of social and economic development in each country (Álvarez et al., 2025). Educational policies, in particular, are essential for enhancing entrepreneurial skills and empowering individuals to cope with uncertainty.

Finally, for policymakers and investors, the study highlights the importance of mentorship and resource allocation for young entrepreneurs. Policymakers are encouraged to design support programs that priorities mentorship and provide access to resources for entrepreneurs adopting the entrepreneur-as-scientist approach. Investors, too, can guide startups toward more structured methods when appropriate and ensure that entrepreneurs are provided with the right resources and mentorship to increase their chances of success.

One of the main limitations of this study is that, although the sample size is large, it exclusively focuses on university students who own businesses. This limits the scope of the findings, as it primarily represents individuals still in the process of acquiring knowledge, compared to more experienced entrepreneurs who have completed their professional studies and gained additional life experience. This difference in experience may influence how structured or experimental tools are applied. Therefore, it is recommended that future research includes a more diverse sample, comprising individuals with varying backgrounds and experiences, to identify better which approaches are most relevant for different groups and the reasons behind this. Future research could also examine how pedagogical approaches, mentorship structures, and institutional support mechanisms differ between undergraduate, graduate, and professional programs (e.g. MBA), and how these variations influence students' entrepreneurial competencies and outcomes.

While our analysis identifies a significant correlation between the Entrepreneurs-as-Scientists approach and improved business outcomes, the observational nature of our data limits causal inference. Entrepreneurs who adopt a scientific approach may also possess other unobserved characteristics (e.g. higher prior experience, better networks) that contribute to success. Future research could employ experimental or longitudinal designs to isolate the causal impact of Entrepreneurs-as-Scientists.

A further limitation of the study is that this database does not collect information on ventures involving value transfer, acquisition entrepreneurship, or intrapreneurship. However, as future lines of research, we suggest investigating this topic using a mixed-methods approach that would enable an in-depth analysis of such ventures. Another limitation is that the database analyses the relationship between students and entrepreneurship (in particular, in this article the sample considers those who already have their ventures in operation) but does not provide information on how universities deliver entrepreneurship education. As a future line of research, it would be advisable to consider the global trends in project/problem-based learning (PBL) in entrepreneurship education, identifying the credit system used and the current curriculum development policy at participating universities, and comparing the results across the different universities.

Additional future lines of research could explore aspects of ambidexterity through qualitative studies that analyze entrepreneurs' responses within the framework of different types of ecosystems, as proposed by Bertello et al. (2022). An added limitation is the omission of the family business variable. In experimental methodologies, rapid learning plays a key role, and this ability is often developed through experience. Students from entrepreneurial families may already possess this skill due to their exposure to business environments. It would be valuable to explore whether a connection exists between coming from a business family and the choice of method for developing a business model.

Furthermore, this study does not consider the role of a validated business canvas, which is often a requirement for accessing public and private support within entrepreneurial ecosystems. Future research could examine how these approaches are applied across various regional ecosystems. Another potential direction for future research is to analyze small- and medium-sized enterprises or micro-enterprises that are not started by university students, expanding the scope beyond those still pursuing their degrees. Lastly, while this study offers important insights into entrepreneurial approaches across national groups, it has some limitations. First, our analysis of nationality was based on self-reported regional categories and did not account for respondents with multiple national backgrounds or those with migration histories. These groups may engage with entrepreneurship differently due to exposure to diverse cultural norms, institutional environments, and adaptation processes. However, our dataset did not include variables capturing immigration status or dual/multiple national identities, limiting our ability to explore this dimension. Future research could address this gap by incorporating migration-related variables and examining how multicultural experiences influence entrepreneurial strategies and decision-making. Furthermore, in addition to considering the nationality of the participants, it would be very interesting for future studies to address Hofstede's cultural dimensions framework to classify the nations involved, providing insights into how specific cultural traits influence entrepreneurial preferences and behaviors in different contexts.

We thank the GUESSS organizers, the GUESSS partners, the teams responsible for data collection in the different countries, and the students participating in this global survey.