A decision tree is a visual representation of a decision-making process. It's like a flowchart that helps you make a choice based on a set of conditions. In this post I will explain how to move from creating simple decision trees to using decision trees in strategic decision making.

Established best practices often recommend to identify and quantify risks and consequences. By using decision trees, this is taken one step further, where the different risks and consequences are connected to strategic decisions, and the impact is seen in relation to this. This is an effective way to identify mitigating actions and capturing upsides.

Constructing a Simple Decision Tree

Here's how you can construct a simple decision tree:

Start with a question that needs to be answered. This could be anything, from "Should I go to the beach today?" to "Which college should I attend?"

Identify the possible outcomes or choices. For example, if the question is "Should I go to the beach today?" the choices might be "Yes" or "No."

Next, identify the factors that will help you make your decision. These are known as "attributes" in decision tree terminology. For example, if the question is "Should I go to the beach today?" the attributes might include "Weather," "Distance to beach," "Time available," and "Company."

For each attribute, identify the possible values. For example, for the "Weather" attribute, the values might be "Sunny," "Cloudy," "Rainy," and "Windy."

Now, for each attribute and value combination, identify the resulting outcome or decision. This is known as a "leaf" in decision tree terminology. For example, if the attribute is "Weather" and the value is "Sunny," the resulting decision might be "Go to the beach."

Once you have identified all the possible attribute-value combinations and their resulting decisions, you can organize them into a tree-like structure. Start with the initial question at the top of the tree, and then draw branches for each attribute, with each value branching off from its corresponding attribute. Finally, draw the resulting decision or outcome at the end of each branch.

Congratulations! You've just constructed a simple decision tree. You can use this tree to guide your decision-making process based on the attributes and values that you encounter.

Decision Trees in Strategic Decision Making

While a simple decision tree can be useful for making straightforward decisions, decision trees can also be used in more complex decision-making scenarios, such as strategic decision making. Here's how to go from simple decision tree sketching to using decision trees to solve complex problems and for more informed and insightful decision making.

Identify the problem or opportunity: Before you can begin building your decision tree, it's important to clearly identify the problem or opportunity that you're addressing. This will help you to focus your analysis and ensure that you're considering all relevant factors. The decision problem might be a complex business problem, such as determining the best product to launch or the best market to target. Strategic decision making often involves several decisions and outcomes that are linked over time, and that cannot be evaluated in isolation.

Define your decision criteria: Once you've identified the problem or opportunity, you'll need to define the decision criteria that you'll use to evaluate different options. This might include things like economic value,  cost, risk, and feasibility, depending on the specific decision you're making.

Identify the different options: Next, you'll need to identify the different options that are available to you. This might include things like different product strategies, marketing approaches, or investment opportunities. Often, a problem or opportunity can be broken into smaller steps. For example, a technology investment can involve early testing at a smaller scale before full-scale implementation.

Identify the relevant factors: Once you've identified the different options, you need to determine the relevant factors that will impact the decisions. These factors might include costs, risks, and financial considerations.

Collect data: Once you've identified the relevant factors, you need to collect data on each of them. This might involve analysing product costs,  market research, financial reports, industry trends, and other relevant data sources.

Create a decision tree: Once you have the data, you can create a decision tree that maps out the possible outcomes based on the relevant factors. This decision tree will be more complex than the simple decision tree created earlier, as it will have multiple layers of branches and leaves.

Evaluate the probabilities and outcomes: Once you've built your decision tree, you'll need to evaluate the probabilities and outcomes associated with each option. Running sensitivities will help you to determine which option is most likely to yield the desired outcome. It will also tell you how robust each decision is. Tweaking the probabilities or adjusting the input data is often an iterative process to better reflecting the realities of the situation or provide additional insight to the situation. The best course of action is often based on calculating the expected value of each possible outcome, taking into account the probabilities of each outcome occurring. The set of decisions and outcomes that give the highest expected value is known as "the optimum path" in decision tree terminology.

Make the decision: Finally, you can use the decision tree to make an informed decision based on the analysis. This decision can be made with greater confidence, as it is based on a rigorous analysis of the relevant factors and potential outcomes.

Overall, decision trees can be a powerful tool for strategic decision making, as they allow you to map out the potential outcomes of complex decisions and make informed choices based on a rigorous analysis of the relevant factors.

Technical description of the different types of nodes in the decision tree

As explained, a decision tree is a graphical representation of a decision-making process. To construct the tree, we use different types of nodes that represent different points in the decision tree where a particular action or event occurs.

Decision nodes: These nodes represent a decision point in the decision-making process. At a decision node, a decision maker must choose one of several possible actions or options. For example, in a decision tree for a job interview, a decision node might represent the choice between accepting or rejecting a job offer.

Event (or chance) nodes: These nodes represent an uncertain event or outcome that is beyond the control of the decision maker. At an event node, the decision maker does not make a decision, but rather must evaluate the likelihood of different possible outcomes. For example, in a decision tree for a medical diagnosis, an event node might represent the possibility of a patient having a particular disease based on test results.

Terminal (or end) nodes: These nodes represent the end of the decision-making process, where a final outcome is determined. Terminal nodes are reached either when a decision has been made or when an uncertain event has occurred. For example, in a decision tree for a job interview, a terminal node might represent the final outcome of getting the job or not getting the job. In a medical diagnosis decision tree, a terminal node might represent the final diagnosis or recommended treatment.

In summary, decision nodes represent decision points, event (or chance) nodes represent uncertain outcomes, and terminal (or end) nodes represent final outcomes or decisions.