You will be able to set up your decision journal and use everything from the course to review one decision in full.
Most decisions go wrong in more than one way at once. A plan can rest on a weak assumption, a misleading chart and a bias you did not notice, all in the same choice. Over this course you have collected tools for each of these problems: argument maps, evidence rankings, chart checks, causal questions, a bias checklist, door types, pre-mortems, probabilities and expected value. Used separately, each tool catches a particular kind of mistake. This project puts them together on your own decisions.
You produce three things: a decision journal, a full review of one past decision, and a one-page summary. The journal is meant for you to keep using after the course ends. The whole project takes about an hour, and you can split it across two sittings if you prefer. Nothing more than these three deliverables is required, and nothing less will build the habit.
Start with the journal. It has to live in a tool you actually open every week. A spreadsheet works well, because you can sort entries by review date and, later, count how your predictions turned out at each probability. A notes app or a paper notebook is acceptable if that is what you will open. The best tool is the one you already check every week, and a beautiful system you never open is worse than a plain list you see every morning.
Set up the fields from Lesson 8.2, What goes in a decision journal:
date and decision options choice and reasons expectations with probabilities state of mind review date
Add two extra columns, outcome and process grade, and leave both empty for now. You fill them in when each review date arrives.
Write at least five live entries, for decisions you are making now or will make in the next few weeks. They do not all need to be large, and a mix is recommended. A suggested mix is one career decision, one work project, one purchase, one people decision and one smaller choice.
Each entry needs at least one prediction with a probability and a time frame, written the way you practised in Lesson 7.1, Decisions are bets. When the entries are done, put each review date in your calendar alongside the quarterly reviews from Lesson 8.3, Review your decisions every quarter.
For the review, pick one decision from the past six to eighteen months. It must have a known outcome, be important enough to deserve an hour, and be recent enough that you can reconstruct what you knew at the time. Work through the steps in order and write brief notes for each.
Begin with an argument map from Module 1. Note the main claim, the supporting evidence, the unstated warrant that links them, and the weakest link in the chain.
Next, look for the fallacies from Module 2: ad hominem, false choice, appeal to popularity and sunk cost. Note any that played a part.
For numbers, from Module 3, ask whether a figure or chart influenced the decision. If one did, check whether it survives the tests on axes, averages, base rates and missing data.
For causes, from Module 4, ask whether the decision rested on a belief that one thing causes another, and whether alternative explanations were considered.
For biases, run the checklist from Lesson 5.4 against the decision as it was made, using only what you knew then.
For door type and process, from Module 6, decide whether it was a one-way or two-way door, count how many options you considered, and note whether you did anything like a pre-mortem.
For probabilities, from Module 7, write the realistic outcomes and the rough probabilities you could have given them at the time, then the expected value and the worst case.
Finish with the process grade from Lesson 8.1, Good decisions can have bad outcomes. Grade how the decision was made separately from how it turned out.
Daniel, from Module 4, is an invented example. His team ran an app download campaign last year. Downloads rose, but spending per customer barely changed, so he chooses that campaign for his review.
His argument map shows the weakest link was the warrant that downloading the app causes customers to spend more. At the numbers step he finds that the "twice as much" figure the team relied on compared all app users with all web users, with no before-and-after comparison. At the causes step he names loyalty as the confounder nobody checked.
The bias checklist picks up confirmation bias: the team wanted the app to matter and searched only for data that supported it. On door type, he sees the campaign was a two-way door treated as a one-way door. The team spent the whole budget at once when they could have tested it on a random sample.
Daniel grades the process as poor. He notes that the outcome would not have changed the grade, because even if spending had risen, the team would not have known whether the app caused it.
The change he commits to is that before any campaign over a set budget, the team runs a randomised test on a small sample first, as in Lesson 4.2, and decides on the full campaign from the test result.
The summary fits on one page and should be readable by someone else in five minutes. Cover the decision, the outcome, the process grade, and the two or three review steps that taught you the most.
End it with one change, written as specifically as the changes you wrote in Lesson 8.3. Daniel's change says when it applies (any campaign over a set budget), what the team does (a randomised test on a small sample) and how the final call is made (from the test result).
The course's final project applies the same tools to a decision that is still ahead of you, and you will record that decision in the journal you build here. Set up the journal first, then the activity takes you through the rest.
Set up the journal with five entries, review one past decision in full, and write the one-page summary.
Junxiong-WFG Organisation is an authorised representative of AIA Financial Advisers Private Limited (Reg. No. 201715016G).