What Is a High Quality Decision?

Is it one that led to the best outcome? Or one that integrates all the relevant and available information? Maybe one that is liked by a majority? If decision governance is followed to the letter, will that guarantee a high quality decision?

This text is part of the series on the design of decision governance. Decision governance are guidelines, rules, processes designed to improve how people make decisions. It can help ensure that the right information is used, that that information is correctly analyzed, that participants in decision making understand it, and that they use it before they make a decision. Find all texts on decision governance here.

The quality of a decision depends on the following:

Quality of the outcome
Quality of intended outcomes
Uncertainty of the outcome
Quality of the explanation of outcome from actions
Acceptance of the decision
Clarity of the decision
Integration of relevant information
Feasibility of the option
Fitness to agents
Robustness of planned actions

The reason a decision is made is to unlock actions that lead to some desirable outcome. Certainly then, one criterion for measuring the quality of a decision is the quality of the outcome it led to. The quality of the outcome, in turn, can only really be evaluated when we can experience or observe it, or if it impacted mostly others, hear them out. This first criterion has some unavoidable limitations. One, we cannot evaluate the quality of the decision when it is made, we need to wait to see the outcome. Two, preferences (can) change over time: the decision maker might be convinced in the option they chose at that time, but if they wait long enough to see the outcomes, they may have moved on, wanting something different even when the predicted outcome does occur. This problem of changing preferences is more pronounced when evaluations of outcomes are subjective: the decision maker may, for reasons only they know, dislike an outcome that may correspond to what was predicted at decision time. The issue is less significant when there are agreed upon measures of possible outcomes: if the outcome is to, for example, reduce temperature in a room, there are measurement instruments to determine this unequivocally (provided there’s agreement on how to measure it – which sensors to use, where they are positioned, how long the measurement needs to be done, and so on).

Before the decision is made and actual outcomes observed, the quality of the decision depends on the quality of intended outcomes.

As intended outcomes may not realize, the quality of an option depends on the uncertainty for it to lead to intended outcomes. Uncertainty influences quality in that the less probable an option makes the intended outcome, the lower the interest in that option relative to others.

Outcomes of actions may be uncertain because we don’t know enough about the cause and effect relationships between actions and outcomes (i.e., epistemic uncertainty), and/or they can be uncertain because even with the best knowledge, there are parameters which we cannot control (aleatoric uncertainty). Leaving aside what we cannot explain, we still want the best explanation of how actions lead to outcomes, and so, the quality of the decision will depend on the quality of the explanation (more on explanations here) of the mechanism by which the option we choose leads to the outcomes we want.

Some decisions are more popular than others. Acceptance of a decision will influence quality only to the extent that acceptance of the decision matters at decision time. It does when acceptance is expected to increase the quality of the outcome; for example, acceptance is assumed to increase engagement of those responsible to act as decided.

It should be clear what actions should be taken, how, and by whom when the decision is made. Clarity of a decision matters, in that it cannot lead to desired outcomes if it does not specify to the relevant parties what their role is. You can think of an option as pointing, among others, to the process to deliver desired outcomes, or you can see it as embedding that process (i.e., the process being part of the option). In either case, the option inherits the quality of the process: if the process is deficient, the probability of the option to deliver target outcomes is lowered, and it becomes less attractive.

Integration of all available and relevant information into the options considered at decision time is also a criterion for the quality of a decision. For example, if you have access to information that could be relevant to a decision you need to make, you are aware that this information exists and that you can in fact access it, and you choose not to access and use it, then from the standpoint of the integration criterion, you lowered the quality of the decision. Obviously, your choice to not use some information could be well motivated, e.g., it would delay the decision, but that does not matter for the integration criterion – you made a trade off and prioritized time to the decision over integration of the available information.

The chosen option needs to be feasible, or better, lean, in that it would involve only the necessary and sufficient effort to produce the intended outcome.

A decision leads to actions that someone will be taking, and the option needs to be designed so as to fit their abilities and motivations. The option needs to be such that they will be as successful as possible in executing the actions. This is called the fitness of the option to agents.
Planned actions need to be robust, such that they account for risks anticipated on the path from decision time until the intended outcomes occur.

The Table below summarizes the conditions, and gives simple examples of each.

Table: Conditions for a high quality decision
Condition	Definition	Example
Quality of the outcome	The decision leads to a desirable and measurable result.	A company launches a new product, resulting in increased market share and profitability.
Quality of intended outcomes	The decision targets high-quality outcomes, even if they haven’t yet been realized.	A government policy is implemented to reduce carbon emissions, aiming to achieve environmental goals in the long term.
Uncertainty of the outcome	The decision has a high probability of leading to the desired result, with low uncertainty.	A pharmaceutical firm invests in a drug with a strong likelihood of regulatory approval based on clinical trials.
Quality of the explanation of outcomes from actions	The cause and effect relationships between the actions and outcomes are well understood.	A firm develops a new product which is a variant of a successful one, and addresses well understood additional customer needs.
Acceptance of the decision	The decision is supported by relevant stakeholders, increasing the chances of successful implementation.	A company changes its supply chain strategy with widespread managerial support, leading to smoother execution.
Clarity of the decision	The decision clearly specifies actions, processes, and roles to achieve the intended outcome.	A CEO outlines a detailed restructuring plan with timelines and roles, ensuring all employees know what is expected of them.
Integration of relevant information	All relevant and available data are incorporated into the decision-making process.	A city planner selects a location for a park based on analysis of population density, traffic, and existing green spaces.
Feasibility of the option	The decision involves only the necessary effort and resources to achieve the desired result.	A tech company releases a software update that requires minimal resources but has high potential for improving user experience.
Fitness to agents	The decision aligns with the abilities and motivations of those responsible for its execution.	A hospital implements a patient care protocol that matches the skills and experience of the nursing staff.
Robustness of planned actions	The decision accounts for risks and uncertainties between decision time and outcome realization.	A logistics firm adjusts its shipping strategy to mitigate risks like fuel price volatility and regulatory changes.