Every framework, model, and rule for making better decisions โ sorted by when to use each one. This is the field manual you'll reach for in real life: at work, in negotiations, in business cases, in your own choices.
20+frameworks
4tool families
9cognitive bias defenses
1master chooser
1The meta-question
What kind of decision is this?
The single most underused skill in decision-making is asking what kind of decision you're facing before grabbing a tool. Use the wrong tool and even good logic produces bad outcomes.
Three layers to get right, in order:
1 ยท The decision itselfWhat problem are you actually solving? What's the real question (not the symptom)?
2 ยท The conditionsReversible or one-way? Big or small? Certain or uncertain? Urgent or important?
3 ยท The right toolThen โ and only then โ pick the framework that fits.
The most important distinctionTwo-way doors vs one-way doors (Bezos): Reversible decisions (two-way doors) should be made fast โ you can walk back through. Irreversible ones (one-way doors) deserve slow, careful analysis. Most people treat them the opposite way โ over-thinking reversible choices, rushing the permanent ones.
2The diagnostic step
Classify Before You Decide
Three classification grids that sharpen what tool to reach for. Use whichever one matters most for your situation.
A ยท Reversibility ร Stakes
Low stakes
High stakes
Reversible
JUST DO ITDecide fast, learn from doing. Don't waste analysis here.
Quick experimentRun a small test (MVP). Reversibility = your safety net.
Irreversible
Default to noIf small but permanent, the upside rarely justifies the lock-in.
SLOW DOWNFull analysis, multiple options, premortem, second opinion. The 5% you must get right.
B ยท Certainty: Cynefin (Snowden)
A model for matching how you decide to the kind of system you're in:
Clear (simple)Cause & effect obvious โ sense โ categorize โ respond. Use best practices/SOPs.
ComplicatedCause & effect knowable with effort โ sense โ analyse โ respond. Bring in experts, run the numbers.
ComplexCause & effect only clear in retrospect โ probe โ sense โ respond. Run experiments, learn fast.
ChaoticNo cause-effect link โ act โ sense โ respond. Stabilise first, analyse later.
The classic error โ ๏ธTreating a complex problem (innovation, culture change, market entry) as if it were complicated โ "let's analyse our way to the answer." Complex problems don't yield to upfront analysis; they need probes.
C ยท Urgency ร Importance: the Eisenhower Matrix
Urgent
Not urgent
Important
DOCrises, deadlines. Handle now.
SCHEDULEStrategy, learning, prevention. The zone of long-term wins.
Not important
DELEGATEInterruptions, most meetings. Hand off.
DELETETime-wasters, busywork. Cut.
Memory hook ๐ง Most people live in "urgent & important" (firefighting). Strategic people live in "important but not yet urgent" โ that's where compounding decisions happen.
3Tool family I
Analytical Decision Tools
For when you can quantify the options. These give you numbers to compare.
๐
Decision Matrix (Weighted Scoring)
Use when: choosing between 2โ5 options with multiple criteria (job offers, vendors, products, strategies).
1
List the criteria that matter (cost, fit, risk, returnโฆ).
2
Assign each a weight (must sum to 100%).
3
Score each option per criterion (1โ10).
4
Multiply & sum: highest total wins.
Job offer exampleSalary (40%) ร score + Growth (30%) + Culture (20%) + Location (10%) โ one defensible number per offer. Forces explicit trade-offs, kills "gut" thinking.
๐ณ
Decision Tree
Use when: sequential choices with uncertain outcomes (launch decisions, R&D, court vs settle, build vs buy).
Squares = decisions you control ยท Circles = chance you don't ยท End nodes = payoffs. Multiply probability ร payoff, take the max.
Expected Value
EV = ฮฃ (probability ร outcome value)
Pick the option with the highest EV โ but only if you can stomach the worst-case outcome.
โ๏ธ
Cost-Benefit Analysis & NPV
Use when: evaluating projects/investments with costs & benefits over time.
Simple CBATotal benefits โ total costs. Useful for one-period decisions.
ROI / IRRReturn %, easier to compare differently-sized options โ but trust NPV in disputes.
๐
Pareto Analysis (80/20)
Use when: deciding where to focus effort โ bug fixes, customer issues, time allocation.
~80% of effects come from ~20% of causes. List causes by impact; tackle the top few first โ diminishing returns hit fast after that.
In one lineFind the vital few; ignore the trivial many. Most "to-do" lists have 1โ2 items doing 80% of the real work.
โ
Marginal Analysis
Use when: deciding "should I do one more?" โ hire one more, produce one more, study one more hour.
Compare the marginal benefit of the next unit to its marginal cost. Keep going while MB > MC; stop when they meet. (Same logic as MR = MC from economics.)
๐ฒ
Bayesian Updating
Use when: new evidence arrives and you need to update your belief properly.
Start with your prior belief. When new evidence comes in, ask: how likely is this evidence if I'm right vs wrong? Adjust accordingly. Avoids two errors: ignoring new info (stubbornness) and overreacting to one data point (whiplash).
Memory hook ๐ง "Strong opinions, weakly held." Commit to your current best guess, but update freely when the evidence changes.
4Tool family II
Strategic Decision Tools
For when the problem is qualitative or strategic โ choosing direction, not just options.
๐ฏ
SWOT โ Action
Use when: setting strategic direction. From your Strategy module.
SWOT only matters if it ends in action. Match Strengths to Opportunities (attack), Weaknesses to Threats (defend). If a quadrant has no follow-through, it's just description.
๐
Root Cause: the 5 Whys
Use when: you suspect you're treating a symptom, not the cause.
Ask "why?" five times to drill from symptom to root cause. "Sales fell โ why? Repeat customers dropped โ why? Quality complaints โ why? New supplier batches โ why? No incoming QC โ why? We cut it last quarter." The real fix isn't a sales push โ it's restoring QC.
๐
Fishbone (Ishikawa) Diagram
Use when: a problem has multiple possible causes across categories (quality, ops, customer issues).
Draw the problem as a fish head, then branches for cause categories (People, Process, Equipment, Materials, Environment, Management). Brainstorm causes under each branch. Forces breadth before depth.
๐
Six Thinking Hats (de Bono)
Use when: a group keeps going in circles or needs to look at a decision from all angles.
โช WhiteFacts & data only โ what do we actually know?
๐ด RedGut feelings & emotions โ what's your instinct say?
โซ BlackCaution โ what could go wrong, what are the risks?
๐ก YellowOptimism โ benefits, opportunities, why this works.
๐ข GreenCreativity โ alternatives, new options, possibilities.
๐ต BlueProcess โ what step are we on, are we using time well?
๐ฎ
Scenario Planning
Use when: high uncertainty, long horizons (strategy, geopolitical, tech disruption).
Build 3โ4 distinct futures (e.g. "recession," "boom," "regulation hits"). For each, ask: what would we do? Pick options that work across most scenarios โ robust beats optimal-but-fragile.
Robust โ optimalDon't pick the option that wins in the most likely scenario. Pick the one that doesn't lose badly in any scenario. Survival before optimization.
5Tool family III
Behavioral Defenses (the anti-bias toolkit)
Frameworks don't help if your brain is sabotaging the input. These tools are specifically for catching your own thinking errors.
The biases to watch for
Confirmation
Seeking info that confirms what you already believe.
Anchoring
First number/idea drags everything toward it.
Availability
Recent/vivid examples overweight your estimates.
Overconfidence
You think you're more right than you are.
Loss aversion
Losses sting ~2ร more than gains feel good.
Sunk cost
"We've already spent so much" โ keep losing more.
Groupthink
Pressure for harmony kills dissent & better ideas.
Status quo
Defaulting to "what we already do," unjustified.
Survivorship
Studying only winners; ignoring the silent graveyard of failures.
The defenses
๐
Premortem (Klein)
Use before: any high-stakes decision.
Imagine it's 12 months later and the decision failed catastrophically. Write the story of why. Surfaces risks teams suppress in optimistic forward-planning. The single highest-ROI defense against overconfidence.
๐ก๏ธ
Red Team / Devil's Advocate
Use when: a decision is gaining consensus too quickly.
Formally assign someone (or a small team) to attack the proposal. Their job is to find weaknesses โ so dissent isn't disloyal. Built into how good militaries, intelligence agencies, and Amazon work.
๐
The "Kill the Company" Exercise
Use when: reviewing strategy.
Ask your team: "How would you kill our company if you ran a competitor?" Reveals strategic blind spots faster than any SWOT.
๐
Inversion (Munger)
Use when: stuck on "how do we succeed?"
Flip the question: "What would guarantee failure?" Then avoid those things. Often easier and clearer than identifying paths to success.
โ๏ธ
The 10-10-10 Rule (Welch)
Use when: emotionally charged decisions.
Ask: how will I feel about this in 10 minutes? 10 months? 10 years? Short-term emotion dissolves; long-term consequences come into focus.
๐ง
The Cooling-Off Rule
Use when: you're angry, excited, tired, or pressured to "decide now."
Sleep on it. Almost every "urgent" personal/career decision can wait 24 hours โ and the urgency itself is often a red flag. Pressure to decide fast is a tactic, not a fact.
๐
Decision Journal
Use always: the long-game tool.
When making a non-trivial decision, write down: the decision, your reasoning, what you expect, your confidence. Review months later. Reveals patterns in your own thinking โ your biases, your blind spots. Compounds your judgment over a career.
Memory hook ๐ง "Separate the decision from the outcome." A good decision can have a bad outcome (variance). A bad decision can have a good outcome (luck). The journal lets you grade the process instead of being misled by the result.
6Tool family IV
Process Frameworks
For structuring how a decision gets made โ especially repeatable or organizational decisions.
๐ฏ
OODA Loop (Boyd)
Use when: fast-moving, adversarial environments (competition, crises, sports).
Observe โ Orient โ Decide โ Act. Then loop. Whoever cycles faster wins โ speed of iteration beats depth of analysis when the situation keeps shifting.
โ๏ธ
PDCA Cycle (Deming)
Use when: continuous improvement, quality, repeatable processes.
Plan โ Do โ Check โ Act. Run a small change, measure, learn, standardize. The engine of Lean & Six Sigma. Same DNA as Build-Measure-Learn (entrepreneurship module).
๐งญ
The DECIDE Framework
Use when: you want a structured single-pass process.
D โ Definethe problem precisely
E โ Establishcriteria for a good outcome
C โ Consideralternatives (force at least 3)
I โ Identifybest option
D โ Developaction plan
E โ Evaluatethe outcome & iterate
๐
WRAP (the Heath brothers)
Use when: personal or team decisions where you want to actively counter biases.
W โ Widenyour options (kill "whether or not" โ find a third)
R โ Reality-testassumptions (run a small experiment)
A โ Attaindistance (10-10-10 rule, advise a friend)
P โ Prepareto be wrong (premortem, set tripwires)
๐ช
The MoSCoW Method
Use when: prioritizing scope, features, or tasks under constraint.
Must have ยท Should have ยท Could have ยท Won't have (this round). Forces explicit cuts โ "everything is critical" is the surest sign nothing has been prioritized.
7Tool family V
Group Decisions & Ownership
The trickiest decisions involve multiple people. Without clarity on who decides, even the smartest groups stall.
๐ค
RACI / DACI / RAPID
Use when: a decision has many stakeholders and confusion about roles.
The golden ruleOne decider, multiple advisors. Decisions die when "we" decide everything by consensus. Name the decision-maker explicitly โ others give input, but one person owns the call.
โ๏ธ
Disagree & Commit (Bezos / Grove)
Use when: consensus is impossible but a decision must be made.
Voice your disagreement, hear the decision, then commit fully as if it were your own. Half-hearted execution kills more strategies than wrong choices. Disagreement is the input; commitment is the output.
๐ก
Nominal Group Technique
Use when: brainstorming with strong personalities or hierarchies present.
Each person writes ideas silently first, then shares โ before any discussion. Prevents anchoring on the loudest voice or the boss's opinion. Diversity of ideas without groupthink.
๐ฆ
The Two-Pizza Rule
Use when: deciding meeting/team size.
Teams should be small enough to feed with two pizzas (~6โ8 people). Above that, communication overhead kills decision speed. Bigger meetings = worse decisions, almost always.
8โ The lookup table
The Master Chooser
Face any decision โ match the situation โ grab the tool. This is the heart of the chapter.
it's reversible & low-stakes
โ Just decide. Don't over-analyse. Speed compounds.
it's permanent & high-stakes
โ Premortem + 10-10-10 + decision journal + sleep on it.
โ Six Thinking Hats. Cycle through perspectives.
future is wildly uncertain
โ Scenario planning. Pick the robust option.
consensus is forming too fast
โ Red team / Devil's advocate / Premortem.
emotionally charged
โ 10-10-10 rule + 24-hour cool-off.
stuck on "how do we win?"
โ Inversion โ ask how to fail, then avoid that.
fast-moving / adversarial
โ OODA loop. Out-cycle the opponent.
continuous improvement
โ PDCA cycle / Build-Measure-Learn.
need a structured one-pass
โ DECIDE or WRAP.
prioritising features/tasks
โ MoSCoW (Must/Should/Could/Won't).
many stakeholders confused
โ DACI / RAPID. Name one decider.
consensus impossible, must move
โ Disagree & commit.
complex (novel, no precedent)
โ Probe with small experiments (Cynefin: complex zone). Don't analyse first.
chaotic (system breaking)
โ Act first to stabilise, analyse after.
9The meta-rules
Universal Decision Principles
If you forget every tool above, hold onto these. They're true across every decision you'll ever make.
1 ยท Process > outcomeYou can't control luck. You can control how you decide. Grade your process, not the result.
2 ยท Match speed to reversibilityReversible โ decide fast. Permanent โ take your time. Most people invert this.
3 ยท Diagnosis before prescriptionDefine the real problem before solving. The same lesson as the capstone.
4 ยท Force 3+ options"Yes or no?" is a trap. Whenever you see two options, find a third โ even if to discard it.
5 ยท Quantify what you canNumbers force honesty. Even rough estimates beat pure gut.
6 ยท Name the decider"We" doesn't decide anything. One owner, with advisors.
7 ยท Sleep on itUrgency is usually a tactic, not a fact. 24 hours rarely changes the math; it often changes the mind.
8 ยท Plan to be wrongSet tripwires, run premortems, write down assumptions. The world is uncertain โ design for that.
9 ยท Commit after decidingHalf-hearted execution wastes the decision. Disagree, then commit fully.
10 ยท Learn from every callDecision journal. Your future self is the most important client for your present-self's reasoning.
The one-line summary ๐ง Good decisions = right tool ร clear process ร honest input ร committed execution. Miss any of the four and even smart people get bad results.
๐ฏ Active recall
Cover the answer, say it aloud, then tap to check. Then practice on a real decision this week โ apply the chooser table to something you're actually deciding.
What's the most underused first step in decision-making?
Asking what KIND of decision it is before reaching for a tool. Get the diagnosis right (reversibility, stakes, certainty) and the right tool follows.
tap to reveal
Two-way doors vs one-way doors โ and the common error?
Reversible (two-way) decisions should be made FAST; irreversible (one-way) ones SLOW. The common error is the opposite โ over-thinking small reversible choices, rushing the permanent ones.
tap to reveal
Write the expected value formula and when it's right to use.
EV = ฮฃ (probability ร outcome value). Use for sequential decisions under uncertainty โ but only when you can stomach the worst-case outcome (it averages over many tries you may not get).
tap to reveal
What's a premortem and why is it powerful?
Imagine the decision failed 12 months from now and write the story of why. Surfaces risks teams suppress when looking forward optimistically. Highest-ROI defense against overconfidence.
tap to reveal
What is inversion (Munger)?
Flip the question: instead of "how do we succeed?" ask "how would we guarantee failure?" โ then avoid those things. Often easier to identify than paths to success.
tap to reveal
When should you NOT analyse first (Cynefin)?
In COMPLEX problems (innovation, market entry, culture change) where cause-effect is only visible in retrospect โ probe with small experiments first. Also in CHAOTIC ones, where you act first to stabilise.
tap to reveal
Why "separate decision from outcome"?
A good decision can have a bad outcome (bad luck), and vice versa. Grading by results misleads. Judge the process โ what you knew, your reasoning, what you'd do again โ not the verdict.
tap to reveal
โ State the one-line summary of good decisions.
Right tool ร clear process ร honest input ร committed execution. Miss any of the four and even smart people get bad results.
tap to reveal
What does "disagree and commit" mean?
Voice your disagreement, hear the decision, then commit fully as if it were your own. Half-hearted execution kills more strategies than wrong choices. Disagreement = input; commitment = output.
tap to reveal
In one sentence, when do you reach for each tool family?
Analytical = can quantify the options. Strategic = qualitative direction-setting. Behavioral = countering your own brain. Process = structuring how the call gets made. Group = many stakeholders.
tap to reveal
โ Bonus Chapter ยท Decision-Making Field Manual ยท Pin the Master Chooser. Run a premortem this week on one real decision. ๐ง