Decision Tree Analysis for PMP: Expected Value, EMV, and Optimal Path Selection

Decision tree analysis is a structured, quantitative technique for making choices under uncertainty. On the PMP exam, it appears in the context of risk management — specifically, as a tool within the Perform Quantitative Risk Analysis process. Decision trees allow you to evaluate multiple decision alternatives, each with multiple possible outcomes, and select the path that maximizes expected monetary value (EMV). This guide covers how to build a decision tree, calculate EMV at each node, select the optimal path, and navigate the common question patterns PMI uses to test this skill.

Unlike EVM formulas (which you will use heavily on the exam), decision tree analysis typically appears in only one or two questions. However, those questions are often multi-part calculation problems worth significant points. A solid understanding ensures you capture them with confidence.

What Is a Decision Tree?

A decision tree is a graphical representation of a decision problem. It maps out every possible decision alternative, the uncertain events that may follow each decision, and the financial outcome (payoff or cost) of each possible scenario. By assigning probabilities to uncertain events and calculating expected values, a decision tree reveals which alternative has the highest expected payoff or the lowest expected cost.

Decision trees use three types of nodes:

• Decision Nodes (squares ■): Represent points where the decision-maker chooses between alternatives. Each branch from a decision node represents a choice the project manager or organization can make.
• Chance Nodes (circles ●): Represent points where an uncertain event occurs. Each branch from a chance node represents a possible outcome with an associated probability. All probabilities at a chance node must sum to 1.0 (100%).
• End Nodes (triangles ► or terminal leaves): Represent the final payoff or cost if that path is realized. These are the numeric outcomes at the end of each branch.

The standard convention reads the tree from left to right (building the tree) and then solves it from right to left (calculating expected values back to the decision node). This right-to-left calculation process is called rollback or folding back the tree.

The Core Formula: Expected Monetary Value (EMV)

The fundamental calculation at every chance node is the Expected Monetary Value:

EMV = Σ (Probability of Outcome × Monetary Value of Outcome)

Where:

• Probability is expressed as a decimal (e.g., 0.30 for 30%). The probabilities at any chance node must sum to 1.0.
• Monetary Value is the payoff (if positive) or cost (if negative) associated with that outcome. On the PMP exam, outcomes are almost always expressed as profit (positive) or cost (negative).

After calculating the EMV for each chance node, you compare EMVs across all branches at each decision node and select the branch with the highest EMV (if evaluating payoff/profit) or the lowest EMV (if evaluating costs). On the PMP exam, decision tree questions almost always evaluate profit/payoff scenarios, so you will typically be selecting the highest EMV.

Building a Decision Tree: Step-by-Step

Let's walk through the construction and solving of a decision tree from scratch. This is the process you should follow on exam day when you encounter a decision tree question.

Step 1 — Identify the Decision

What choice are you making? Is it "Build or Buy a component"? "Invest in new technology or stick with existing systems"? "Accept a fixed-price contract or negotiate a cost-plus contract"? The decision defines the first decision node (the square at the far left of the tree).

Step 2 — Draw the Alternatives

From the decision node, draw a branch for each alternative. A decision tree must have at least two alternatives (otherwise there is no decision to make). Label each branch clearly with the name of the alternative. If there are upfront costs associated with choosing an alternative (e.g., an investment required to pursue a build option), note those costs on the branch — they will be subtracted from the EMV when comparing alternatives.

Step 3 — Identify Uncertain Events for Each Alternative

For each alternative, determine what uncertain events could occur. These become chance nodes (circles) at the end of each decision branch. Common uncertain events on the PMP exam include market conditions (strong/weak demand), technology success/failure, regulatory approval/denial, and competitive response.

Step 4 — Assign Probabilities and Payoffs

For each chance node, assign probabilities to each outcome (they must sum to 1.0) and attach the monetary payoff or cost to each outcome. The exam will provide these numbers — you do not need to estimate them. Your job is to use them correctly in the EMV calculation.

Step 5 — Calculate EMV at Each Chance Node (Rollback)

Starting from the rightmost chance nodes, calculate EMV = Σ(Probability × Payoff). Write the EMV above or inside each chance node.

Step 6 — Select the Optimal Alternative

At the decision node, compare the EMVs of all alternatives (after accounting for any upfront investment costs). Choose the alternative with the highest EMV (for profit/payoff problems) or lowest EMV (for cost problems).

Worked Example 1 — Simple Two-Alternative Decision

Scenario: A project manager must decide whether to build a component in-house or buy it from an external vendor. The projected profits and probabilities for each decision are as follows:

Build In-House (requires $100,000 upfront investment):

• 70% probability of success → profit of $500,000
• 30% probability of failure → loss (negative profit) of $200,000

Buy from Vendor (no upfront investment):

• 90% probability of on-time, on-spec delivery → profit of $300,000
• 10% probability of vendor issues → profit of $50,000

Step 1 — Calculate EMV for Build:
EMV(Build) = (0.70 × $500,000) + (0.30 × −$200,000)
EMV(Build) = $350,000 + (−$60,000) = $290,000
After upfront investment: $290,000 − $100,000 = $190,000

Step 2 — Calculate EMV for Buy:
EMV(Buy) = (0.90 × $300,000) + (0.10 × $50,000)
EMV(Buy) = $270,000 + $5,000 = $275,000

Step 3 — Compare and select:
EMV(Buy) = $275,000 > EMV(Build) = $190,000
Decision: Buy from the vendor. Even though the build option has a higher upside ($500,000), the 30% chance of a $200,000 loss and the $100,000 upfront investment drag down its expected value below the more reliable buy option.

This example illustrates a critical exam point: high-upside alternatives can have lower EMVs than moderate but reliable alternatives. The exam will tempt you to pick the alternative with the highest best-case scenario — but the correct answer is always based on EMV, not the best-case outcome.

Worked Example 2 — Multi-Branch Decision Tree with Sequential Decisions

Scenario: A company must decide whether to develop a new product. If they develop, they face market uncertainty. If the market is strong, they can choose to expand capacity. Here is the full tree:

Decision 1: Develop the product (upfront cost: $200,000) OR do nothing ($0).

If Develop is chosen — Chance Node:

• 40% probability of strong market → Decision 2 required
• 60% probability of weak market → profit of $150,000

Decision 2 (if strong market): Expand capacity (cost: $150,000) OR maintain current capacity ($0).

If Expand — Chance Node:

• 80% probability of high demand → profit of $1,200,000
• 20% probability of moderate demand → profit of $600,000

If Maintain Current — Chance Node (strong market):

• 80% probability of high demand → profit of $800,000 (limited by capacity)
• 20% probability of moderate demand → profit of $500,000

Solving the Tree — Roll Back from Right to Left:

Solve Decision 2 first:
EMV(Expand) = (0.80 × $1,200,000) + (0.20 × $600,000) = $960,000 + $120,000 = $1,080,000
After expansion cost: $1,080,000 − $150,000 = $930,000

EMV(Maintain Current) = (0.80 × $800,000) + (0.20 × $500,000) = $640,000 + $100,000 = $740,000

Decision 2 result: Expand ($930,000) > Maintain ($740,000). If the market is strong, expand capacity. The EMV of the "strong market" branch is $930,000.

Now solve Decision 1:
EMV(Develop) = (0.40 × $930,000) + (0.60 × $150,000) = $372,000 + $90,000 = $462,000
After development cost: $462,000 − $200,000 = $262,000

EMV(Do Nothing) = $0

Final Decision: Develop the product. EMV of $262,000 > $0. If developed and the market proves strong, expand capacity.

This multi-level tree tests whether you can correctly sequence the rollback. The exam may present a similar tree and ask: "What is the EMV of the 'Develop Product' decision?" or "Should the company expand capacity if the market is strong?" or "What is the EMV of the optimal strategy?" Each of these questions requires you to solve different portions of the tree.

When Is Decision Tree Analysis Used on the PMP Exam?

Decision trees appear in the following contexts on the exam:

• Perform Quantitative Risk Analysis (primary): Decision trees are listed as a tool and technique in the PMBOK guide for this process. The exam frames decision tree questions as risk analysis problems where you must quantify the financial impact of uncertainty to support a decision.
• Plan Risk Responses: After quantifying risks, you may use decision tree results to select the optimal risk response strategy. For example, a decision tree may show that accepting a risk (buy insurance) has a higher EMV than mitigating it (invest in prevention).
• Project Selection / Business Case: Decision trees can support go/no-go decisions on projects or phases. A tree may evaluate whether to proceed to the next phase based on the EMV of outcomes.

PMP Exam Question Patterns for Decision Trees

Decision tree questions follow these patterns:

• Direct EMV Calculation (most common): A simple tree with one decision node and one chance node per alternative. You are given probabilities and payoffs and asked to calculate the EMV of a specific alternative or recommend the best decision. Example: "What is the EMV of Option A?" These are straightforward if you remember EMV = Σ(P × V).
• Choose the Best Decision (high frequency): You must calculate EMVs for all alternatives and select the one with the highest EMV. The exam provides four answer choices, each naming a different alternative (or "do nothing"). This tests whether you correctly computed and compared all EMVs.
• Multi-Level Tree (lower frequency, higher difficulty): A tree with sequential decisions. You must roll back from right to left. These take more time and test whether you understand the rollback logic. The exam may only ask for the final EMV of the first decision, requiring you to solve intermediate nodes correctly along the way.
• Conceptual / "When to Use" (low frequency): A scenario describes a decision under uncertainty and asks which tool or technique should be used. The correct answer is "Decision Tree Analysis" when the scenario involves multiple alternatives with probabilistic outcomes. Distinguish this from "Monte Carlo simulation" (which models overall project risk with thousands of iterations) and "Sensitivity analysis" (which tests how sensitive outcomes are to changes in one variable).
• EMV of Risk Events (emerging pattern): Instead of a full decision tree, you calculate the EMV of a single risk: EMV = Probability × Impact (in $). This is the building block of decision trees. The exam may ask: "A risk has a 30% probability of causing a $200,000 loss. What is the EMV of this risk?" Answer: 0.30 × −$200,000 = −$60,000.

Common Decision Tree Mistakes

• Calculating left to right: You must fold back from right to left. Calculate EMVs at the rightmost chance nodes first, then use those EMVs to evaluate the decision nodes to their left, and continue folding back until you reach the initial decision node. Attempting to calculate from left to right produces incorrect results on multi-level trees.
• Forgetting to account for upfront costs: If an alternative requires an investment (like the $200,000 development cost in the multi-branch example), the investment must be subtracted from the EMV before comparing alternatives. The exam may display the investment on the branch and expect you to incorporate it.
• Selecting by best-case scenario instead of EMV: A common trap: Option A has a small chance of a huge payoff. Option B has a high chance of a moderate payoff. EMV may favor Option B, but candidates dazzled by the huge number pick A. Always compute and compare EMVs — never pick by the maximum possible payoff.
• Not checking that probabilities sum to 1.0: At any chance node, all probabilities must add up to 1.0. If the exam presents probabilities of 0.40 and 0.50, there is a missing 0.10 — likely an unstated "status quo" outcome with $0 impact, which you must account for in the EMV calculation.
• Confusing payoff with profit: The exam may give you "payoff" numbers (gross outcome) and "cost" numbers separately. If the scenario says "payoff of $500,000 with a cost of $300,000," the net profit used in EMV is $200,000. The exam expects you to recognize the difference.
• Rounding intermediate results: On multi-level trees, rounding intermediate EMVs can compound error and lead to an incorrect final answer. Keep intermediate calculations precise and only round the final answer to reasonable precision.

Decision Trees vs. Other Quantitative Risk Tools

The PMP exam tests your ability to distinguish between decision trees and other quantitative risk analysis techniques. Here is how they differ:

Technique	What It Does	When to Use	Output
Decision Tree Analysis	Evaluates discrete decision alternatives under uncertainty	Choosing between specific options with known probabilities	EMV per alternative; optimal choice
Expected Monetary Value (standalone)	Calculates EMV of a single risk event	Quantifying individual risks for the risk register	A single dollar value per risk
Monte Carlo Simulation	Simulates thousands of outcomes using probability distributions	Modeling overall project risk (cost/schedule)	Probability distribution, confidence levels
Sensitivity Analysis (Tornado Diagram)	Determines which variables most affect outcomes	Identifying the most critical risks to focus on	Ranked list of sensitive variables

The exam may describe a scenario where you have three specific alternatives with distinct outcomes and ask which tool to use. If the scenario involves discrete choices with explicit probabilities, the answer is Decision Tree Analysis. If the scenario involves modeling the impact of multiple interacting risks on the overall project, the answer is Monte Carlo Simulation. If the scenario asks which risk has the biggest impact, the answer is Sensitivity Analysis.

Final Exam Day Checklist for Decision Trees

Before you take the exam, make sure you can:

• Calculate EMV from a set of probabilities and payoffs: Σ(P × V)
• Draw and solve a simple two-alternative decision tree
• Roll back a multi-level tree from right to left, solving chance nodes before decision nodes
• Distinguish between payoff (gross) and profit (net of costs) in the EMV calculation
• Explain when to use decision tree analysis versus Monte Carlo simulation or sensitivity analysis
• Interpret the EMV and use it to select the optimal decision alternative

Decision tree analysis rewards methodical, step-by-step problem solving. Do not rush. Draw the tree (or at minimum, write out each EMV calculation on your scratch paper) and work through each node systematically. One or two decision tree questions done correctly can be the difference between a passing and failing score.

← Previous: Communication Channels  |  Next: Procurement Contract Types →