Core concepts
The key terms and mental model behind Decision Tree.
| Concept | Description |
|---|---|
| Decision node | A point where you (or the business) choose a path. No probabilities — you pick the best branch. |
| Chance node | A point where the outcome is uncertain. Each branch has a probability; all branches must sum to 100%. |
| Leaf node | The end of a path. Carries a terminal payoff in USD. |
| Expected value (EV) | The probability-weighted average payoff. Computed live at every node. |
| Payoff | A USD value — positive (revenue, gain) or negative (cost, loss). Any node can carry one. |
| Sensitivity analysis | Drag a slider to see how changing a single probability or payoff affects the EV of every branch. |
How EV blending works
Decision nodes do not blend or average their children's EVs, because a decision means choosing one branch, not averaging across all of them — each child branch of a Decision node shows its own independent EV, and the branch with the highest EV is highlighted in blue to guide you toward the optimal choice.
Chance nodes work the opposite way: because you don't control which branch occurs, a Chance node's EV blends its children. It's the probability-weighted average of every child branch's EV, plus any payoff at that node — in plain language, the sum of each branch's probability multiplied by that branch's EV.
Intermediate payoffs
Any node — not just leaves — can carry a payoff. Intermediate payoffs represent costs or revenues incurred at that point in the path, before the outcome is known.
Example: A Decision node for "Acquire company" might carry a -$500,000 due-diligence cost. That cost is added to the EV of every path that flows through the acquisition branch.
Payoffs on Chance nodes work the same way — the value is added once per path that passes through the node, regardless of which branch is taken.
Sensitivity analysis terms
Sensitivity parameter — A probability or payoff field from any node in your tree, added to the sensitivity panel so you can test how changes to that value affect the overall EV.
Saved values — The current stored values in your tree. Sliders start at the saved value and revert there when released; sensitivity analysis never modifies your saved model.
EV impact — The change in the best-path EV as you drag a slider away from the saved value.
Crossover indicator — A warning shown when moving a parameter causes a different decision branch to become optimal. It identifies the exact threshold at which the decision reverses.