Decision tree analysis is one of those project management tools that looks more objective than it actually is. You draw a tree, assign probabilities, calculate expected monetary values, and the math produces a number — which can feel like a decision has been made. In my experience, the number is useful, but it’s only as good as the probability estimates feeding it, and those estimates are almost always guesses dressed up as data. That limitation doesn’t make decision tree analysis less valuable. It just changes how you should use the result. This article covers how to build and calculate a decision tree, with a worked subcontractor example, and ends with the situations where the tool gives you false confidence rather than genuine insight.
Table of Contents
What is a Decision Tree?
The concept decision tree is both used in operations management and machine learning. In machine learning, decision trees are used to create a model that forecasts the value of a target variable based on a set of input variables.This article analyzes the decision tree analysis based on The Expected Monetary Value (EMV) while making operations research.
This technique can be used for many different project management cases. For example: Should we upgrade the software that we are using in our organization? Should we build a prototype for our new project? Should we select the low-price contractor? Should we select the low-budget project? Which contractor best fits our project’s budget and schedule milestones?
What makes decision tree analysis different from a simple pro/con list is the explicit assignment of probabilities and monetary values to each branch. This forces you to quantify your assumptions rather than leave them vague — which is both the strength and the limitation of the method. The strength: it makes your reasoning visible and challengeable. The limitation: if your probabilities are wrong, the math will give you a precise wrong answer.
Expected Monetary Value (EMV) for Decision Tree Analysis
The decision tree analysis provides a visual template to calculate the values of potential outcomes and the possibilities of achieving them. It allows us to select the most suitable choice relying on the existing information and best forecasts.
The PMBOK Guide defines The Expected Monetary Value (EMV) as a statistical concept that calculates the average outcomes when the future includes the scenarios that may happen or may not happen.
Probability x Impact is equal to The Expected Monetary Value (EMV). For the opportunities this value is positive and for the threats it is negative.
To learn how to calculate The The Expected Monetary Value (EMV), please read The The Expected Monetary Value (EMV) article.
One nuance worth noting: EMV is an average across scenarios, not a prediction of what will actually happen. An EMV of $330,000 doesn’t mean you’ll spend exactly $330,000 — it means that if you made this same decision many times under the same probability assumptions, $330,000 would be the average outcome. On a single project, you’ll either pay the penalty or you won’t. EMV is most useful for comparing options against each other, not as a precise cost forecast.
How to Create a Decision Tree in 4 Steps?
After identifying the problem or decision you are going to make, follow these four implementation steps to draw a decision tree diagram;
Step 1. List all the decisions and prepare a decision tree for a project management situation.
Step 2. Assign the probability of occurrence for all the risks.
Step 3. Assign the impact of a risk as a monetary value.
Step 4. Calculate The Expected Monetary Value (EMV) for each decision path.
Finally, evaluate the outcome of each decision and decide which alternative provides you more benefits than others. Sometimes selecting the one with highest score may not be the ideal choice. Because highest expected value means a high risk that the organization may be unwilling to take.
Decision Tree Analysis Example
Suppose you are a project manager of a power plant project and there is a penalty in your contract with the main client for every day you deliver the project late. You need to decide which sub-contractor is appropriate for your projects critical path activities. But while selecting a sub-contractor, you should take into consideration the costs and delivery dates.
• Sub-contractor 1 bids $250,000. You estimate that there is a 30% possibility of completing 60 days late. As per your contract with the client, you must pay a delay penalty of $5,000 per calendar day for every day you deliver late.
• Sub-contractor 2 bids $320,000. You estimate that there is a 10% possibility of completing 20 days late. As per your contract with the client, you must pay a delay penalty of $5,000 per calendar day for every day you deliver late.
You need to determine which sub-contractor is appropriate for your projects critical path activities. Both sub-contractors promise successful delivery and high-quality work.
Following 4 Basic Steps
Step 1: List decisions and prepare a decision tree for a project management situation.
In Figure 1 below, a decision tree is prepared based on the decisions, costs, and rewards of uncertain events.
Step 2: Assign the probability of occurrence for the risks.
In this example, the possibility of being late for Sub-contractor 1 is 30% and for Sub-contractor 2 is 10 %. This means that the possibility of completing on-time for Sub-contractor 1 is 70% and for Sub-contractor 2 is 90 %. In Figure 2 below the probability of occurrence for the risks are assigned.
Step 3: Assign the impact of a risk as a monetary value.
In Step 3 we are calculating the value of the project for each path, beginning on the left-hand side with the first decision and cumulating the values to the final branch tip on the right side as if each of the decisions was taken and each case occurred. Figure 3 below shows the value of each path.
As shown in the figure, path values are calculated by the formulas given below.
Sub-Contractor 1
Path value of completing on-time = Bid Value = $ 250,000
Path value of being late = Bid Value + Penalty = $ 250,000 + 60 x $5,000 = $ 550,000
Sub-Contractor 2
Path value of completing on-time = Bid Value = $ 320,000
Path value of being late = Bid Value + Penalty = $ 320,000 + 20 x $5,000 = $ 420,000
Step 4: Calculate The Expected Monetary Value (EMV) for each decision path.
In Step 4, we are calculating the value of each node – including both possibility nodes and decision nodes. We begin with the path values at the far right-hand end of the tree and then proceeding from the right to the left calculate the value of each node. This calculation is called “folding back” the tree.
The Expected Monetary Value (EMV) of each node will be calculated by multiplying Probability and Impact. Figure 4 below shows The Expected Monetary Value (EMV) of each path.
As shown in the figure, The Expected Monetary Value (EMV) of each path is below.
Sub-Contractor 1
EMV = %30 x $ 550,000 + %70 x $ 250,000 = $ 340,000
Sub-Contractor 2
EMV = %10 x $ 420,000 + %90 x $ 320,000 = $ 330,000
In this simple example Expected Monetary Values (EMV) are very close. Now we are selecting Contractor 2 because of low cost and low possibility of being late.
Notice how close these numbers are — $340,000 versus $330,000, a difference of $10,000 on a contract where the penalty alone could run $300,000. When decision tree analysis produces results this close, the choice isn’t really being driven by the math. It’s being driven by which probability estimate you trust more. If Sub-contractor 1’s late-completion probability is actually 25% instead of 30%, the EMV difference nearly disappears. That sensitivity is worth acknowledging explicitly when presenting the result to stakeholders. The decision tree analysis didn’t make the decision — it structured the reasoning. That’s still valuable. But it’s not the same as certainty.
Decision Tree Analysis and the Probability Problem
The hardest part of decision tree analysis isn’t drawing the tree or calculating the EMV. It’s Step 2: assigning probabilities. In the subcontractor example above, we said there’s a 30% chance Contractor 1 finishes late. But where does that number come from? Historical data on similar subcontractors? The project manager’s judgment? The contractor’s own estimate?
In most real projects, probability estimates come from a combination of past experience, expert judgment, and informed guesswork. That’s not necessarily a problem — but it becomes one when those estimates are treated as facts rather than assumptions. I’ve seen decision trees presented to steering committees with two decimal places of precision on the EMV figures, when the underlying probability estimates were essentially educated guesses rounded to the nearest 10%.
A practical way to handle this: run the calculation with two or three different probability assumptions and see how much the result changes. In the subcontractor example, if Contractor 1’s late probability shifts from 30% to 20%, the EMV drops from $340,000 to $310,000 — now substantially cheaper than Contractor 2. That sensitivity tells you something important: the decision is quite dependent on how accurately you’ve estimated that one probability. If you’re confident in the 30% estimate, proceed. If you’re not, the margin of $10,000 is too thin to be decisive, and other factors — track record, communication, relationship — probably matter more than the math.
prons and Cons of Decision Tree Analysis
Although decision tree analysis can provide you many benefits while evaluating the outcomes of various alternatives, it also has some disadvantages.
Pros
- It is a visual tool that enables to see the whole picture.
- It is easy to implement without deep expertise.
- It is a flexible tool where you can add new alternatives or delete the existing ones.
- An analytical tool relies on quantitative data and and numbers.
Cons
- It may be difficult to manage large and compex inputs.
- If there is a small change in the inputs, the remaining data might be affected.
- It is difficult to capture complex relationships or interactions between variables as effectively as other modeling techniques.
One con that doesn’t get listed often enough: decision trees can create an illusion of rigor when the underlying inputs are soft. The visual structure and precise arithmetic give the analysis a scientific appearance that the probability estimates may not actually support. This isn’t unique to decision trees — it applies to any quantitative risk tool — but it’s worth naming directly because the tree format is particularly persuasive.
When Decision Tree Analysis Gives You the Wrong Answer
Decision tree analysis works best when the decision branches are genuinely discrete, the probabilities are estimable from data or defensible expert judgment, and the outcomes can be expressed meaningfully in monetary terms. When those conditions don’t hold, the tool produces numbers that feel like answers but aren’t.
Three situations where I’ve found it least reliable:
When branches aren’t independent. The subcontractor example treats “on time” and “late” as independent outcomes with fixed probabilities. In reality, if one contractor is late because of a supply chain disruption, the other contractor might face the same disruption. Decision trees don’t naturally capture these correlations. For decisions where risks are likely to move together, Monte Carlo simulation handles the complexity more honestly than a tree.
When the number of branches is large. Decision trees become unwieldy — and therefore less useful as communication tools — when you have more than two or three levels of branching with multiple options at each level. The visual clarity that makes them valuable disappears, and you end up with a diagram that’s technically complete but practically unreadable. At that point, a simpler framework or a different tool serves better.
When the decision involves values that can’t be monetized. EMV works when outcomes can be expressed in dollars. Many real decisions involve factors that resist monetization: reputational impact, relationship value, strategic positioning, team morale. A decision tree that only captures the quantifiable part of a decision may optimize for the wrong thing. The analysis should inform the decision, not replace the judgment required to weigh incommensurable factors.
A Decision Tree Is an Analytical Tool
Decision trees are used as a model that helps in discovering, understanding, and communicating the structure of such decision problems (Ref: Clemen and Reilly (2001) and Waters (2011)).Note that the decision tree analysis is a statistical concept which offers a powerful way of determining, finding out and analyzing uncertainty. For quantitative risk analysis, calculating the Expected Monetary Value (EMV) by using Decision Tree is an extensive technique. It is an efficient tool that helps you to select the most suitable action between several alternatives. Also, this technique enables to present complex data for decision making visually. In order to perform a Decision Tree Analysis correctly, you need to know the Difference Between Quantitative and Qualitative Risk Analysis.
Decision Tree Analysis Alongside Other Risk Tools
Decision tree analysis doesn’t operate in isolation — it fits into a broader risk management toolkit. Knowing which tool to reach for in which situation matters more than mastering any single tool.
For decisions with two or three clearly defined alternatives and estimable probabilities, decision tree analysis is often the right choice. It’s fast to construct, easy to communicate, and produces a clear comparison. For decisions with many interacting variables and correlated risks, Monte Carlo simulation produces more reliable output — though it requires more setup and is harder to explain to non-technical stakeholders.
The project selection methods framework uses decision trees regularly for go/no-go decisions and for comparing investment alternatives. The 5 Whys technique is useful for understanding the root causes that feed your probability estimates — if you’re assigning a 30% probability to a contractor being late, 5 Whys analysis of past late deliveries can give that estimate a firmer foundation than intuition alone.
Used well, decision tree analysis is a structured way to make your reasoning visible, your assumptions explicit, and your comparison between options consistent. Used poorly — with soft probabilities treated as hard data, or with results presented as more precise than the inputs justify — it produces confident-looking conclusions that may not hold up when the assumptions are examined. The tool is reliable. The inputs require honest scrutiny.
See Also
Francois Simosa is the head of training for the Gragados Training Associates, which provides special project management and risk management training programs.
