DecisionBranch
Build a tree

Decision trees for consultants and analysts

A recommendation you can defend in the room, with the number that would change it.

Build a decision tree – freeSee the example

The problem

  • The deck says "we think"

    A recommendation without an expected value is an opinion. The client wants to know what it is worth and by how much it beats the alternative.

  • The numbers live in a spreadsheet nobody trusts

    Nested probabilities in cells, a formula someone broke in version 7, and no way to show the structure. A tree shows the logic and the arithmetic in one picture.

  • The client asks "what if"

    What if the win rate is lower? What if the bid costs more? Sensitivity turns that question into a switch point you can answer on the spot.

Worked example

Bid on the Metro tender?

Open in the editor Describe your own

A firm can put in a full bid, subcontract to a prime, or pass. The full bid costs a proposal team for six weeks and has to survive a shortlist before the final decision. Payoffs are the gross margin on the contract.

RecommendedFull bidExpected value 210,000 · 90,000 ahead of Partner with a prime
Bid on the Metro tender?−60k60%50%50%40%−15k45%55%Bid on the Metro tender?EV 210,000→ Full bidFull bidEV 270,000ShortlistedEV 450,000Win900,000Lose0Not shortlisted0Partner with a primeEV 135,000Prime wins300,000Prime loses0No bid0

Switch point. If the chance of Win under Shortlisted falls below 33.3% (it is 50% now), Partner with a prime wins instead.

0%33.3%100%

Risk profile. Under the recommended policy the chance of ending up with a loss is 70%; the worst case is −60,000 and the best is 840,000.

30%−60,00040%−60,00030%840,000

Lead with the recommendation and its margin over the runner-up, then show the switch point: it tells the client exactly how confident they need to be in the win rate. The risk profile goes on the same slide so nobody is surprised by the chance of writing off the bid cost.

The outline behind this example
# Analytical: bid on the Metro signalling tender?
Bid on the Metro tender?
  Full bid (cost 60000) # proposal team for six weeks
    Shortlisted (p 0.6)
      Win (p 0.5): 900000
      Lose (p 0.5): 0
    Not shortlisted (p 0.4): 0
  Partner with a prime (cost 15000) # subcontract, smaller share
    Prime wins (p 0.45): 300000
    Prime loses (p 0.55): 0
  No bid: 0

What you get

  • Expected value on every node

    Rolled back from the outcomes, net of costs, updated as you type.

  • Sensitivity and switch points

    Sweep any probability, payoff or cost and see exactly where the answer flips.

  • Value of information

    The most a study, a pilot or an expert is worth before you commit.

  • Risk profile

    Every outcome under the recommended policy, with its probability, so the chance of a loss is on the page.

  • Unlimited private trees

    Keep client work private; five private trees free, unlimited on Pro.

  • A link that recomputes

    Share the tree; the client can change a number and watch the recommendation move.

How people use it

  1. 01

    Pre-mortem

    Before the workshop, build the tree with your best estimates and list the three inputs whose switch point is closest. Those are the questions to ask in the room.

  2. 02

    Client workshop

    Put the tree on the screen, and let the client argue with the probabilities. Every change recomputes, so the meeting ends with a number everyone touched.

  3. 03

    Board paper

    One page: the tree, the recommended path lit, the expected value, the switch point and the risk profile. Attach the outline as the appendix so anyone can rerun it.

Templates for this work

All templates · How to make a decision tree · Expected value and decision trees

Questions

Can I keep client trees private?
Yes. Free accounts keep up to five private trees; Pro is unlimited. Public trees have their own page and can be embedded.
Can the client change the numbers?
Yes. Share the link and they get the same editor; the tree lives in the link, so their edits do not touch your saved copy unless they save a fork.
Can I put it in a deck?
Every public tree has a server-rendered page you can screenshot or embed, and the outline text copies from any page and pastes back into the editor unchanged.

Start with the example, or your own decision.

The editor is free and needs no account. Describe the decision in a sentence and the AI builder drafts the tree.

Open the exampleDescribe your decision