How to make better decisions by removing bias from the process

Executive overview

Most people judge decisions by their outcomes — a habit called resulting — which means lucky bad decisions look good and unlucky good decisions look bad. Outcome quality and decision quality are separate things.

The fix is to map possible outcomes and assign probabilities before deciding, rather than building a pros/cons list that amplifies whatever conclusion you already want to reach.

Getting independent input from others before any discussion starts is the single highest-leverage change a team can make.

Why resulting corrupts decision review

• Ask your team to name their best and worst decisions of the past year — they will list the best and worst outcomes, not the best and worst choices.
• Driving drunk and arriving safely does not make drunk driving a good decision.
• To evaluate a past decision fairly, reconstruct all the outcomes that could have happened and estimate how likely each was.
• A bad outcome that had a 2% chance of occurring probably means you were unlucky, not wrong.
• A great outcome that had a 2% chance of occurring probably means the decision was poor, not brilliant.

Why pros/cons lists fail

• A flat list has no magnitude: ten pros and one con means nothing if the con is catastrophic.
• It has no probability: most startups fail, so a pros/cons list would always vote against founding one — yet the potential upside can justify the risk.
• By the time you write the list, you already have a preferred answer; the list just confirms it (motivated reasoning).
• A good decision process dampens bias; a pros/cons list amplifies it.

Mapping outcomes and probabilities

• For each option, list the realistic ways it could turn out, then estimate how likely each outcome is.
• Doing this forces you to surface what you know, what you don't know, and what you could find out.
• Example: estimating the weight of an unseen desk — even without seeing it, a reasonable range (50–750 lb) eliminates absurd possibilities and signals your level of certainty to collaborators.
• Expressing a wide range tells others you need help; expressing a narrow range tells them you're confident. Both are useful signals.
• The gaps in your knowledge are the main thing you can actually control; luck is not.
• Filling knowledge gaps before deciding is the primary lever available to improve outcomes.

Collecting better input from teams

• Two heads are only better than one if you prevent them from merging into one head before the conversation starts.
• In group discussions, consensus forms around whoever speaks first — especially if that person is senior, expert, or charismatic.
• The goal of a meeting is not agreement; it is to surface the full spread of independent views so the decision-owner can decide with better information.
• Harvard research (Zechhauser & Levy): students answering by hand-raise formed super-majorities; students answering by clicker showed genuine spread of opinion.
• Once someone states a view, everyone who disagrees is forced into open conflict — which most people avoid, suppressing useful information.

How to elicit independent input

• Share the decision brief and the specific question (forecast, rating, yes/no) before the meeting.
• Have each person write or submit their answer independently, without seeing others' responses.
• Aggregate and share the results so everyone enters the meeting having already seen the spread of views.
• Focus discussion on the disagreements, not the points of consensus — that is where the decision-relevant information lives.
• Participants do not need to reach agreement; the decision-owner synthesises the spread and decides.
• For quick in-room decisions: have people write yes/no on paper simultaneously rather than speaking aloud.
• For structured decisions (e.g. hiring): build a rubric in advance, have each evaluator score independently, then discuss the gaps — not the agreements.

Applying this to one-off decisions

• Recurring decisions (e.g. daily commute route) allow data collection over many iterations; one-off decisions do not.
• High-stakes one-offs — strategic plans, hiring, major investments — are exactly where pre-decision mapping matters most.
• Mapping outcomes and probabilities in advance also gives you a benchmark to evaluate the decision honestly after the fact, separate from the outcome.