Frameworks · Choosing strategy
Decision trees
A diagram of the options in a decision and their possible outcomes, with probabilities and payoffs, used to compare options by their expected value.
Section 01: The framework at a glance
Text description
A fictional retailer chooses between two options. Opening a second store costs $200k: success, with probability 0.6, returns $500k; failure, with probability 0.4, returns $150k. Its expected value is 0.6 × 500 + 0.4 × 150 = $360k, a net $160k after the cost. Refurbishing the existing store costs $80k: high demand, probability 0.7, returns $250k; low demand, probability 0.3, returns $100k. Its expected value is $205k, a net $125k. On expected value alone the retailer opens the second store, and the refurbishment branch is marked as rejected. Square nodes are decisions, circles are chance events, and the probabilities are estimates.
Section 02: What it is, and what it is for
A decision tree is drawn from left to right. A square, the decision node, branches into the options, each with its cost. A circle, a chance node, branches into the possible outcomes of an option, each with a probability (the probabilities at a node add up to 1) and a payoff. The tree is then worked from right to left: the expected value at each chance node is the sum of each payoff times its probability, and the net expected value of an option is that figure minus the option's cost. The option with the highest net expected value is preferred on the numbers, and the others are marked as rejected.
The method makes uncertainty explicit, but it is only as good as its estimates, and expected value treats a risky option and a safe one with the same average as equal, which a cautious business may not.
What it is for
To compare options whose outcomes are uncertain, using numbers that can be questioned and tested. The toolkit guide sets out the method in full, with an interaction for building and testing a tree.
Section 03: How it is built
The parts of the framework, and what each one holds.
- Decision nodes (squares)
- Points where the manager chooses between options, each branch carrying the cost of choosing it.
- Chance nodes (circles)
- Points where the outcome is uncertain. The branches out of each one carry probabilities that add up to 1.
- Outcomes and payoffs
- The end of each branch, with the financial result if that outcome happens.
- Expected values
- Worked from right to left: at each chance node, the sum of each probability multiplied by its payoff.
- Net expected values and the choice
- At each decision node, the expected value of each option minus its cost. The option with the highest value is chosen and the others are crossed out.
Section 04: When to use it
The business situations it suits, and where another tool will serve better.
It suits
- Choosing between a few options whose outcomes are uncertain but can be estimated, such as opening a second site or upgrading the first.
- Decisions taken in stages, where a later choice depends on an earlier outcome, such as a trial followed by a full launch.
- Showing how sensitive a choice is to an uncertain probability, by finding the probability at which the decision would change.
Look elsewhere when
- Decisions where the probabilities cannot be estimated with any honesty, such as a new technology's adoption over a decade; scenarios serve better.
- One-off decisions where a single bad outcome would ruin the business, so the average matters less than the worst case.
Section 05: How to use it, step by step
A short method. The toolkit guide gives the full one, with the evidence each step needs.
- Draw the decision node and the options, with the cost of each.
- Draw a chance node for each option's outcomes, with probabilities that add up to 1 and a payoff for each.
- Work from right to left: calculate the expected value at each chance node.
- Subtract each option's cost to find its net expected value, and mark the rejected branches.
- Test how sensitive the choice is to the probabilities, and add the qualitative factors.
Section 06: Applied: Coastline Kayaks chooses where to expand
Coastline Kayaks, a fictional rental business, can open a second beach site for $40,000 or upgrade its current site for $15,000.
- New site: a 60% chance of a busy summer earning $90,000 and a 40% chance of a quiet one earning $30,000. Expected value = 0.6 × $90,000 + 0.4 × $30,000 = $66,000; net of the cost, $26,000.
- Upgrade: a 70% chance of $45,000 and a 30% chance of $25,000. Expected value = 0.7 × $45,000 + 0.3 × $25,000 = $39,000; net of the cost, $24,000.
What it showed
On the numbers the new site is slightly ahead, $26,000 against $24,000, but the gap is small and the new site has the wider range of outcomes: in a quiet summer it would earn $10,000 less than it cost, while the upgrade covers its cost in both. A cautious owner could reasonably choose the upgrade.
Section 07: Reading the result
How to interpret what it shows, and what that does and does not justify deciding.
The expected value is the average result the option would give if the same decision could be taken many times. It ranks options on the numbers, but a decision is usually taken once, so read the spread of outcomes as well: the worst result, and the chance of a loss behind each option.
When two options are close, the ranking is fragile, and small changes in a probability may reverse it; find the probability at which the decision would change and ask how confident anyone can be on either side of it. The figures are only as good as the estimates, so the tree's chief value is often that it makes those estimates explicit and open to challenge.
What it can justify deciding
- Preferring the option with the highest expected value when the gap is clear and the worst outcome is affordable.
- Choosing a lower-value option with a narrower spread when the business cannot bear the worst outcome of the other.
- Investing in information, such as a trial or market research, when the decision turns on a probability no one can estimate well.
What it does not justify on its own
- Treating the expected value as the result the business will receive: no single outcome need equal it.
- Ignoring qualitative factors, such as staff, reputation and ethics, which the tree leaves out.
- Precise conclusions from rough estimates; a gap of a few per cent between options is within the error of the guesses.
Section 08: Strengths and limits
Why it helps
- It makes the options, risks and assumptions visible and open to challenge.
- Expected values give a consistent basis for comparing uncertain options.
- It shows how sensitive a decision is to its probabilities.
Where it falls short
- Probabilities and payoffs are estimates, often guesses, so the precision can be false.
- Expected value ignores attitudes to risk and the spread of outcomes.
- It leaves out qualitative factors such as staff, ethics and reputation.
Section 09: Common mistakes
Errors often made with it, by students and by managers, and how to avoid each.
-
Mistake Probabilities at a chance node that do not add up to 1.
Instead List every outcome at the node and check that the probabilities sum to exactly 1.
-
Mistake Forgetting to subtract the cost of each option.
Instead Subtract the cost at the decision node, after working out the expected value of its chance node.
-
Mistake Working from left to right.
Instead Roll back from the outcomes on the right: expected values first, then costs, then the choice.
-
Mistake Recommending on expected value alone.
Instead Add the spread, the worst case and the qualitative factors, and say how confident the estimates are.
Section 10: Apply it
The application mode: the platform's tools that apply this framework, and a worksheet for your own case.
Frameworks structure judgement; they do not produce answers. Each one organizes evidence so that you can see it, test it and argue from it, and the decision, and the reasons for it, remain yours.
The tool that applies it in full
- Strategy Studio IB toolkit Decision tree builder Build a tree of any shape with decision and chance nodes; it checks that probabilities add up, rolls back the expected values net of costs, marks the best branch and shows the worst result and the chance of a loss behind each option.
For part of it
- Strategy Studio Scenario planner Where the probabilities cannot honestly be estimated, test the options against four scenarios instead of averaging over them.
Your worksheet
The tool is the fuller application. Use the worksheet for a quick pass, or to apply the framework to a second case, part by part.
Your worksheet could not be restored on this page; you can still fill it in and copy your work.
Kept on this device as you type, and nowhere else. Not saved without JavaScript: print the page or copy your answers before you leave.
The Framework Application Lab: choose a framework by the decision you face.
Section 11: Try it
The platform's own exercise. Write your answer, then open the guidance.
A school drama club must choose between staging one big musical, with a 50% chance of a $6,000 profit and a 50% chance of a $1,000 loss, and two smaller plays that will almost certainly make $1,800 each. Draw the decision tree, calculate the expected values and recommend a choice.
Kept on this device as you type, and nowhere else. Your answer is not saved without JavaScript; copy it before you leave.
The draft could not be restored on this page; you can still write and copy your answer.
Guidance: what a strong answer does
- Check the probabilities at each chance node add up to 1.
- Show the working: probability times payoff, added, for each option.
- Compare the expected values, then consider the risk of a loss.
- Add one qualitative factor that might change the recommendation.
Section 12: Related concepts
In the glossary
- Decision tree A diagram that sets out the options in a decision and the possible outcomes of each, with their probabilities and financial results, so that the expected value of each option can be calculated and compared.
- Decision node A square on a decision tree marking a point where the manager must choose between options; each line leaving it is one option.
- Chance node A circle on a decision tree marking a point where the outcome is uncertain; each line leaving it is a possible outcome, labelled with its probability.
- Expected value The average outcome of an option over many repetitions, found by multiplying each outcome's payoff by its probability and adding the results, then subtracting the option's cost.
Read with
- The Ansoff matrix IB toolkit Four routes to growth, by whether a business sells existing or new products in existing or new markets, with risk rising as it moves away from what it knows. Compare side by side: Decision trees and The Ansoff matrix
- The Boston Consulting Group matrix IB toolkit A grid that classifies a business's products by market growth and relative market share, to decide where to invest the cash they generate. Compare side by side: Decision trees and The Boston Consulting Group matrix
- The business plan IB toolkit A document that sets out what a new business or project will do, for whom, how it will operate and what it will cost and earn, to test an idea and win support for it. Compare side by side: Decision trees and The business plan
- Force field analysis IB toolkit A way of weighing the forces driving a change against the forces resisting it, to judge whether it will succeed and how to help it. Compare side by side: Decision trees and Force field analysis
Compare it with any other framework in the library.
Published works behind it
- John F. Magee (1964). Decision Trees for Decision Making . Harvard Business Review.
Cited as bibliographic references in the platform's source register; the explanations on this page are the platform's own.
The explanation, method, example and exercise are the platform's own. Statements about the IB course cite the Business management guide; the business in the example is fictional.
Arjun Agrawal | The Business Brief, https://the-business-brief-dun.vercel.app/frameworks/decision-trees/. Original teaching material, not an IB publication. The organizations in the teaching cases are fictional unless a case says otherwise.