Decision quality over outcomes: lessons from Thinking in Bets

Executive overview

Most founders judge decisions by results — win or lose. But outcomes are partly luck, and a bad process can produce a good result and vice versa.

Annie Duke's framework, explored here with serial entrepreneur Jordan Staab, shifts the focus to decision quality: assigning probabilities, stress-testing assumptions, and building feedback loops before outcomes are known.

The core insight: you can never have perfect information, so the goal is to systematically shrink uncertainty — not eliminate it.

Separate outcome from decision

• Business always involves incomplete information; waiting for certainty means losing to faster competitors.
• Evaluate decisions against your original assumptions, not against the final result.
• Build a repeatable process so you can improve it — pure luck is the only alternative.
• Frame tests deliberately: define what you expect to learn, not just whether it "works."
• Failure is a feedback signal; the question is whether you learn and adapt from it.
• Every decision should answer: why are we doing this, what outcomes do we expect, and what is our confidence level?

Think in probabilities, not absolutes

• There are no absolutes in business; assign likelihoods to outcomes before starting a project.
• A practical team exercise: ask "would you hit or exceed this target 80 times out of 100?" — it forces probabilistic thinking and reduces paralysis.
• Map all plausible outcomes before committing; look for decision trees that back you into corners a year from now.
• Evaluate ideas on two axes — opportunity size and effort required — with a modifier for whether your team is suited to solve it.
• A billion-dollar idea with a 9/10 difficulty score and a 1/10 fit score is a poor bet.
• Reconnaissance before commitment: explore multiple future scenarios and assign probabilities to each before locking in a path.

Standardise the decision-making process

• Use templates for every major decision; consistency is what makes improvement possible.
• Avoid complexity in planning documents — dense, simple, and actionable beats comprehensive and elaborate.
• The dopamine hit should come from execution, not from writing an impressive plan.
• A problem well-defined is half solved; forcing yourself to write it down exposes gaps you couldn't see in your head.
• Stack-rank ideas quickly with a simple model (opportunity × effort × fit) before deep-diving on any of them.

Resulting is a trap

• Resulting is judging a decision solely by its outcome — winning on a 7-2 poker hand doesn't make 7-2 a good play.
• If a result falls outside your predicted outcome set, pause — either your model was wrong or you encountered a genuine anomaly; don't scale it.
• Differentiate luck from skill: if the outcome was inside your predicted range and your process was sound, that's skill; if it was wildly outside it, that's luck.
• Anchoring to lucky wins creates dangerous patterns — the romance of the story overrides the math.

Build a truth-seeking inner circle

• A tight advisory circle can sanity-check decision-making and break echo chambers.
• Question everything inside the organisation, regardless of who presents the idea.
• For areas outside your expertise, find the domain expert equivalent of yourself — someone who thinks process-first — and learn their framework.
• Information asymmetry is brutal in unfamiliar territory (the wedding-planner effect): expertise compounds over thousands of repetitions, and overconfidence in adjacent domains destroys value.
• Expanding your circle to other subject matters accelerates learning; different verticals reveal the "other side of the fingerprint."