The Theory - Based Approach
In this article we explore the Theory-Based Approach (TBA), a strategic decision-making framework that builds on scientific entrepreneurship by transforming intuitions into testable hypotheses, particularly useful in uncertain, high-impact business contexts. It guides decision-makers to structure problems through key attributes and logical links, enabling them to theorize and adapt strategies in the absence of historical data. TBA incorporates beliefs and confidence levels in uncertain scenarios, using falsifiability to iteratively refine or abandon strategies. The article dives first in the foundations of the framework and then explores the practical advantages of adopting it in high-uncertainty entrepreneurial activities.
In recent years, the Business Strategy scholars have started intensifying their research over the “scientific paradigms” of decision making in organizations. The core idea of scientific entrepreneurship and management is to convert the ideas and intuitions into falsifiable hypotheses, experimenting and testing them to assess what the best course of action might be. This paradigm represents a natural evolution of the “Lean startup” phenomenon that took place in the early 2000s, which advocated similar principles, giving a strong impulse to the research on the matter. The recent concentration of scholars’ efforts on these scientific entrepreneurship frameworks can be explained by the possibility of leveraging the copious amount of big data that are available today and the advancement of AI and its applications to Business Administration, which make implementation of these paradigms in firms easier and more effective.
Among the several streams of research that investigate this subject, the Theory-Based Approach (TBA) has emerged as a promising framework in management science to navigate complex and uncertain business environments. Rooted in the idea that theories and beliefs can guide entrepreneurial decisions, TBA emphasizes proactive engagement with uncertainty, particularly in "low-frequency high-impact" scenarios, where the decision-maker has no previous data available and the problem is very context-specific, as for instance the launch of a new product or an M&A decision (Felin et al., 2024; Camuffo et al., 2024)[1] . The approach recommends framing the problem by identifying its salient characteristics, which are technically called attributes. For instance, for a firm in the automotive industry that wants to switch its business to EVs, some attributes might be the future costs of renewing the factories, the legal environment in which the firm operates, and the future customers’ demand. Once the attributes are recognized, the DM finalizes their theory by connecting them through logical links. In the previous example, the firm will switch to EV production if the demand is high. So, the strategist constructs a theory in which they hypothesize that the cost of renewing the factories increases the final price of autos, decreasing the demand, while a favorable legal environment, like the EU one, might increase it through incentives and bans on carbon fuel vehicles.
However, as the ecological transition has never happened before, the costs, the evolution of the legal environment, and their ultimate relationship with demand cannot be inferred from historical cases. So, the strategist internalizes the uncertainty by developing beliefs on the likelihood of these events happening and they adapt their strategy based on the confidence they have in these beliefs.
The requirement of falsifiability of the hypothesis allows the DM to test it and ultimately update the above-mentioned beliefs until they are confident enough about whether they should pursue it, pivot it to another theory, or drop the problem considered. The framework works best when the strategist compares different very innovative theories, thus the only limitation lies in their creativity.
Generative Rationality vs. Bounded Rationality
One of the core concepts underpinning TBA is generative rationality—a stark contrast to the widely accepted notion of bounded rationality. While bounded rationality, championed by Simon (1956a), accepts the limits of human cognition and promotes "satisficing" under informational constraints, generative rationality advocates for a more active role in perception.
Generative rationality introduces the idea of an “organism-specific environment” where perception is driven by the individual’s goals and mental imagery. In this view, individuals’ perception is directed to certain elements of the environment, which might be even difficult to notice without such focus or even missing [1] (Felin & Koenderink, 2022).
For instance, let us say that we are hungry. Bounded rationality would foresee a statistical machine that wanders through the house, interacting with all the elements of the house, stopping at the first object that is big, smelly, and colored enough to stimulate the individual to say, “this is food, and it would be enough to make me satisfied.” For Generative Rationality, the DM would focus on a cognitively constructed class of objects, rendering irrelevant stimuli nearly invisible. In this case the DM interacts with an element of the environment only if it is somehow connected to its food research, and it would be able to notice whether the food is missing from the table or the fridge, gaining more information and adjusting its search path accordingly.
The implications are profound: decision-making is driven by what the mind chooses to perceive, based on theories it generates, rather than passively adapting to the environment. The adjective “limited” isn’t here seen as a flaw but as an optimization method to grasp some information from the environment that we wouldn’t be able to grasp without restricting our analysis. Ultimately, we can generate a competitive advantage by getting clues that others do not even consider, becoming first movers.
Applying the approach
Practically, does thinking scientifically work? Why would someone waste time and resources in theorizing and testing when you can count on intuition, experience and heuristics?
The strength of TBA consists in avoiding committing too soon to strategies whose cost in terms of time and money would be far higher than experimenting.
Camuffo et al. (2020) ran a randomized controlled trial (RCT) on 116 startups, demonstrating that those trained in scientific hypothesis testing significantly outperformed the control group. These startups pivoted more frequently, showing a greater ability to avoid false positives. Camuffo et al. (2024) and Novelli & Spina (2024) extended this evidence, confirming the value of rigorous experimentation.
Theories ultimately shape the organization’s strategy and structure. A consequence of the above statement is that heterogeneity in market performances between firms wouldn’t be explained only by the different access to resources, like previous models such as the “resource based” view would argue, but it would also depend on the different theories the entrepreneurs envision and the beliefs about them. Thus, the TBA also becomes a tool to understand how the market is evolving and try to infer the underlying ideas and theories on the strategies that our competitors are applying.
[1] This concept is similar to the Bounded Rationality as Simon originally presented it, and pretty distant from all the heuristics-based rationality frameworks that were developed later. However, while the first two paradigms require the environmental objects to have certain physical characteristics to be perceived by the DM, the Generative Rationality foresees the collection of information of “physical” irrelevant or even missing object, as soon as the environment does match the mental image that my need has induced.