Strategy Studio · Stage 3 of 4: Decide

Decision tree builder

Build a decision tree of any shape, with options, costs, chances, probabilities and payoffs, and see expected values rolled back to the first decision, with the best branch marked.

Back to 1.5 Growth and evolution

IB toolkit tool SL and HL Worked example: fictional

What the tool is for, and its limits

Section 01: Build your analysis

The tool opens on a worked example. Change anything: the analysis follows, and your work is kept on this device.

Scripts are off, so this is the worked example as a finished analysis. Every part of it can be read; to build your own, turn scripts on.

Your work is kept on this device as you type.

Your inputs

m, k or blank.

The tree

A decision node branches into options, each with its cost; a chance node branches into outcomes, each with its probability; an outcome ends a branch with its payoff. Change a node's type to grow or end the tree.

    1. Blank means no cost.

      1. Between 0 and 1.

      2. Between 0 and 1.

        1. Blank means no cost.

          1. Between 0 and 1.

          2. Between 0 and 1.

        2. Blank means no cost.

    2. Blank means no cost.

      1. Between 0 and 1.

      2. Between 0 and 1.

    3. Blank means no cost.

      1. Between 0 and 1.

      2. Between 0 and 1.

    4. Blank means no cost.

15 of 40 nodes; up to 5 levels below the first decision.

The analysis

Worked example Northvale Cycles is fictional and its figures are simulated. Change anything and this becomes your own analysis.

The tree, rolled back

Northvale Cycles. Which growth investment should Northvale make over the next three years? Payoffs are before the costs on the way to them.

Decision tree: Which growth investment should Northvale make over the next three years?Decision tree for Which growth investment should Northvale make over the next three years?, with 15 nodes. Best option: Launch a fleet cargo bike, expected value €2.28m net of its cost. The chosen branch, Launch a fleet cargo bike, is drawn heavier; rejected branches are dashed and crossed with two short strokes.Launch a fleet cargo bikecost €2mEV €4.28mStrong fleet demandp 0.6Weak fleet demandp 0.4EV €1.7mCut the fleet pricecost €0.3mEV €2mFleets respondp 0.5Fleets do not respondp 0.5Sell the bikes to families insteadRelaunch commuter e-bikes onlinecost €1.2mEV €2.25mStrong online salesp 0.5Weak online salesp 0.5Export city bikescost €0.8mEV €1.34mDealers sign upp 0.4Few dealers sign upp 0.6Do nothing new€6m€3m€1m€1.5m€3.5m€1m€2.6m€0.5m€0mEV€2.28m
  • Square: decision
  • Circle: chance
  • Triangle: outcome
  • Heavy blue branch: the choice with the highest expected value
  • Dashed, crossed branch: rejected

Launch a fleet cargo bike has the highest expected value, €2.28m net of its cost, ahead of Relaunch commuter e-bikes online (€1.05m). Following the best choices, its results run from −€1.3m to €4m, with a 0.2 chance of a loss. An expected value is an average over many imagined repeats; this decision happens once, so weigh the spread and the factors the numbers leave out.

The options compared

Each option of the first decision: expected value net of cost and the spread of results under the best later choices
OptionCostExpected value, netWorst resultBest resultChance of a lossVerdict
Launch a fleet cargo bike€2m€2.28m−€1.3m€4m0.2Highest expected value
Relaunch commuter e-bikes online€1.2m€1.05m−€0.2m€2.3m0.5Rejected
Export city bikes€0.8m€0.54m−€0.3m€1.8m0.6Rejected
Do nothing new€0m€0m€0m€0m0.0Rejected

Worst and best results follow the best choices at every later decision, and include all costs on the way.

The working

  1. Cut the fleet price: 0.5 × €3m + 0.5 × €1m = €2m
  2. Weak fleet demand: Cut the fleet price €2m − €0.3m = €1.7m; Sell the bikes to families instead €1.5m. Choose Cut the fleet price (€1.7m).
  3. Launch a fleet cargo bike: 0.6 × €6m + 0.4 × €1.7m = €4.28m
  4. Relaunch commuter e-bikes online: 0.5 × €3.5m + 0.5 × €1m = €2.25m
  5. Export city bikes: 0.4 × €2.6m + 0.6 × €0.5m = €1.34m
  6. Choose a growth investment: Launch a fleet cargo bike €4.28m − €2m = €2.28m; Relaunch commuter e-bikes online €2.25m − €1.2m = €1.05m; Export city bikes €1.34m − €0.8m = €0.54m; Do nothing new €0m. Choose Launch a fleet cargo bike (€2.28m).
Every node, with its value
Every node in the tree with its figures and expected value
NodeTypeProbabilityCostPayoffValue
Choose a growth investmentDecision€2.28m
Launch a fleet cargo bikeChance€2m€4.28m
Strong fleet demandOutcome0.6€6m€6m
Weak fleet demandDecision0.4€1.7m
Cut the fleet priceChance€0.3m€2m
Fleets respondOutcome0.5€3m€3m
Fleets do not respondOutcome0.5€1m€1m
Sell the bikes to families insteadOutcome€1.5m€1.5m
Relaunch commuter e-bikes onlineChance€1.2m€2.25m
Strong online salesOutcome0.5€3.5m€3.5m
Weak online salesOutcome0.5€1m€1m
Export city bikesChance€0.8m€1.34m
Dealers sign upOutcome0.4€2.6m€2.6m
Few dealers sign upOutcome0.6€0.5m€0.5m
Do nothing newOutcome€0m€0m

Checks

No problems found: every chance node has probabilities that add up to 1 and every outcome has a payoff.

Section 02: About this tool

What it is for, how to use it, what it cannot do, and where it stands in the IB course.

Purpose

To compare options whose results are uncertain, by weighting each possible outcome by its probability, and to see how a later decision changes the value of an earlier one.

IB syllabus status

IB toolkit tool SL and HL

Decision trees are one of the guide's toolkit tools for SL and HL, suggested for units 1 and 5 (Business management guide, p. 14), and taught at AO2 and AO4 alongside the syllabus content (p. 13). An option's expected value is the sum of probability times payoff, less its cost; the growth brief (1.5) works one in full.

Serves 1.5 Growth and evolution

How to use it

  1. Name the decision at the root. Add a branch for each option, with what choosing it costs.
  2. Make each option a chance node and give it its outcomes, each with a probability. The probabilities at each chance node must add up to 1, and the tool checks that they do.
  3. End each branch with an outcome and its payoff, before the costs on the way to it. Where a later decision follows an outcome, change that outcome's type to a decision and grow the tree from there.
  4. Read the rolled-back values. A chance node's value is the sum of probability times value; a decision takes the branch with the highest value net of its cost. The chosen branch is drawn heavier and the rejected ones are crossed.
  5. Read the spread of results behind each option, not only its expected value, then test the probabilities you are least sure of.

Limitations

  • The probabilities and payoffs are estimates, and the answer is only as good as they are. Small changes to an uncertain probability can change the choice.
  • An expected value is an average over many imagined repeats. A decision taken once will produce one outcome, which may be the worst.
  • It ignores attitudes to risk. A business that cannot survive the worst outcome may rightly reject the option with the highest expected value.
  • It leaves out what the numbers do not capture, such as control, quality, ethics and the effect on people.
  • Payoffs at different times are treated as equal, unless they have been discounted first.

A model is an aid to judgement, not a universal answer. Use what it shows, then decide, and say why.

Section 03: Using what it shows

Recommended use

The recommended use is to give the expected values and then go beyond them. The fleet cargo bike has the highest expected value, €2.28m, but a 0.2 chance of losing €1.3m; say whether Northvale can bear that, which probability you trust least, and what a paid trial would do to it.

Read alongside