The classical rational model assumes an omniscient decision-maker, capable of inventorying all options and calculating their consequences. In practice, we work with partial data, tight deadlines, and contradictory signals. Understanding decision-making models is not about ticking a theoretical box, but about choosing the right cognitive framework according to the level of uncertainty we face.
Deciding under uncertainty: the Bayesian framework as an operational alternative
Most models taught in management start from an implicit assumption: the necessary information exists, it just needs to be collected. The pure rational model, the political model, and even Simon’s bounded rationality presuppose a definable data scope. When data is incomplete or contradictory, these frameworks lose their prescriptive power.
Related reading : The A500 contract: definition, advantages, and complete guide for better understanding
The Bayesian approach treats decision-making as an experiment, not as a fixed arbitration. The principle is to pose an initial hypothesis (a “prior”), to launch a limited action to gather feedback, and then to correct reasoning based on observed results. HEC emphasizes this logic for high-stakes decisions: replacing raw bets with testable hypotheses and gradual learning.
In practical terms, this means breaking a strategic decision into micro-tests. Rather than validating a complete project based on a static market study, we launch a limited pilot, measure feedback, and adjust. Exploring the different decision-making models helps identify which one fits this type of iterative context.
Further reading : Understanding the GARIP LDEF Direct Debit: Explanations and Solutions for Card Payments in Mérignac
This framework is particularly suitable for environments where the cost of delay exceeds that of a correctable mistake. A product launch, a recruitment for a new position, entering an unknown market: in each case, waiting for perfect data costs more than testing an imperfect hypothesis.
Bounded rationality and the garbage can model: when to really use them
The bounded rationality model (Herbert Simon) and the garbage can model (Cohen, March, Olsen) are often presented side by side in textbooks. Their operational use differs radically.
Bounded rationality and satisficing
Simon introduced the concept of satisficing: we do not seek the best option, we retain the first that exceeds an acceptable threshold. This model works when decision criteria are stable and options are comparable. A supplier choice with a clear specification, a candidate selection based on measurable skills.
Its limitation appears when the acceptability threshold itself is vague. If the team cannot define “good enough,” satisficing loops or produces default decisions that no one owns.
The garbage can model in real context
The garbage can model describes situations where problems, solutions, and participants intersect in a nearly random manner. We regularly observe this in matrix organizations or cross-functional committees: a pre-existing solution (a tool, a budget) clings to an emerging problem simply because both are present at the same meeting.
Recognizing that we are in a “garbage can” process is already a decision-making gain. It prevents rationalizing a choice that did not follow a sequential logic. The team can then consciously decide to validate the problem-solution coupling or to reopen the process.
Political decision-making model: arbitrating power dynamics
The political model starts from the observation that decision-makers do not share the same objectives or information. The decision then results from negotiations, coalitions, and compromises. This framework is particularly relevant for decisions involving multiple departments with contradictory performance indicators.
Three conditions signal that a decision-making process has become political:
- Stakeholders defend mutually exclusive objectives (growth vs. cost reduction, innovation vs. compliance)
- Information is asymmetric: some actors hold data that others do not have and do not share spontaneously
- Formal power (organizational chart) does not correspond to real power (influence, expertise, access to the final decision-maker)
In this context, analytical tools like decision matrices do not serve to find the right answer, but to structure the negotiation. A weighted criteria table makes the preferences of each actor explicit. It shifts the discussion from “I am right” to “here are my criteria and their relative weight.”
Choosing the right model according to the nature of the problem
The most common mistake is to apply a decision-making model out of habit rather than suitability to the context. We recommend qualifying the problem before selecting the framework:
- Structured problem, available data, clear criteria: the classical rational process or satisficing is sufficient
- New problem, partial data, high stakes: the Bayesian framework (hypothesis, test, correction) offers controlled progression
- Organizational problem, multiple stakeholders, divergent objectives: the political model structures the negotiation
- Emerging problem in a fluid organization, without clear ownership: identify if we are in a “garbage can” operation to avoid unowned opportunistic decisions
The choice of model is itself a decision that conditions the quality of the outcome. A rational process applied to a political problem will produce a technically correct recommendation that no one will implement. A Bayesian framework applied to a binary and urgent choice will waste time on unnecessary iterations.
The decision-making competence that matters most is not mastery of a single model, but the ability to quickly diagnose the nature of the problem, then select the appropriate process. It is this meta-decision competence that management training rarely addresses, and that high-performing teams develop through experience and critical feedback on their past choices.