altMBA - Decision Tree
A decision tree maps a choice with branching outcomes, assigns probability and payoff to each branch, then works backwards to the expected value of every option.
Branches fork from a single choice, each carrying a probability and payoff back to one expected-value number.
Reach for this when…
- A decision has several branching outcomes and gut feel gives a different answer each week.
- You need to separate the emotion of a choice from its actual expected value.
- Two options both look reasonable and you can't articulate why you'd pick one.
How to run it
- Map the complex decision with multiple possible outcomes.
- Start at the left with the initial decision point.
- For each option, draw branches for possible outcomes and their probabilities.
- At the end of each branch, assign a value or payoff.
- Work backwards to calculate the expected value of each option.
- Choose the option with the highest expected value, weighted by risk tolerance.
A worked example
Situation. Budi Santoso ran a cold-chain trucking company in Surabaya, Indonesia, and couldn't decide between buying five refrigerated trucks outright or leasing them, with contract uncertainty on both sides.
Applied. He mapped a decision tree: buy or lease, branching into contract-renewed and contract-lost outcomes, assigned rough probabilities and payoffs, then worked backwards.
Result. Leasing won on expected value once he weighted in his real risk tolerance, and he stopped agonising over the binary choice.
The catch
The tree is only as good as the probabilities and payoffs you put in, and those are usually guesses dressed up as numbers. It handles a handful of branching outcomes well; it buckles once a decision has many interacting variables. And expected value ignores what a bad outcome would actually cost you if you can't absorb it.
A decision tree with made-up numbers still gives you a precise-looking wrong answer.
Origin: John F. Magee (Harvard Business Review, 1964)