Most decision reviews are contaminated by a single variable: the outcome.
When a decision works out, we credit the process. When it doesn't, we question it. This feels rational, but it's backwards. Good decisions can produce bad outcomes, and bad decisions can produce good ones. Evaluating them together is how organizations systematically learn the wrong lessons.
The outcome contamination problem
Imagine two traders make the same decision with the same information. One gets lucky — the market moves their way. The other doesn't. In a standard review, the first trader is praised and the second is questioned.
Nothing about their decision quality was different. Only the outcome was different. But the review process treats them as if one was right and the other was wrong.
When you evaluate decisions based on outcomes, you're training your team to be lucky, not good.
This is outcome bias, and it's embedded in almost every performance review, post-mortem, and retrospective.
The damage compounds over time. The team learns that outcomes matter more than process. So they optimize for outcomes — taking bigger risks when they need a win, hiding bad results, and abandoning sound methods after a string of bad luck. The process degrades precisely because the review process rewards results instead of reasoning.
The scale of the problem
Outcome bias doesn't just affect individual decisions. It systematically distorts organizational learning.
In hiring: A manager who made a great hire is credited with good judgment. A manager who made the same assessment about a different candidate who didn't work out is questioned. Over time, this teaches managers to hire conservatively — choosing candidates who are "safe" rather than optimal — because the cost of a visible bad hire outweighs the invisible cost of a great hire never made.
In product development: A feature that shipped and succeeded validates the team's process. A feature that shipped and failed triggers a retrospective that questions everything — including decisions that were sound given the information available. Teams learn to avoid risk, which means they avoid the decisions most likely to produce outsized returns.
In strategy: A bold strategic bet that paid off becomes a case study in visionary leadership. The same bet, made with the same reasoning and the same information, that didn't pay off becomes a cautionary tale. The lesson the organization takes away isn't about decision quality — it's about outcome luck disguised as strategic wisdom.
How outcome-blind review works
The concept is simple: evaluate the decision before revealing the result.
Present the decision maker's reasoning, their information set, their risk assessment, and their process. Ask the reviewers: "Based on what was known at the time, was this a good decision?"
Only after that judgment is locked in do you reveal the outcome.
The moment between the process evaluation and the outcome reveal is where the real learning happens. When the two align — good decision, good outcome — there's nothing surprising. When they diverge, the divergence itself is the data point that matters most.
What changes
When you separate decision quality from outcome quality, several things happen:
- Good process that led to bad outcomes gets reinforced instead of punished
- Bad process that led to good outcomes gets corrected instead of rewarded
- The team learns to evaluate and improve their decision-making process rather than chasing results
- Risk-taking calibrates to quality rather than results
- People become more honest about their reasoning, because the review is focused on the reasoning — not on whether they got lucky
That last point transforms team culture. In outcome-based review environments, people reverse-engineer their rationale after the fact — constructing a narrative that makes the outcome look inevitable. In outcome-blind environments, people document their actual reasoning in real time, because that's what gets evaluated.
The quality of organizational self-knowledge improves dramatically.
The practical protocol
A simple implementation:
-
Document at decision time — reasoning, alternatives considered, risk assessment, expected outcomes with probabilities. This happens before the result is known, which means it can't be contaminated by outcome knowledge.
-
Archive the documentation — store it where it can be retrieved without the outcome attached. A simple system: log the decision in one column, the outcome in a separate column that's hidden during review.
-
Review the decision blind — at review time, present only the decision context. Evaluate: was this a well-structured decision given the available information? Score it on process quality.
-
Reveal the outcome — only after the process evaluation is locked.
-
Analyze the mismatch — compare the process score with the outcome. The four quadrants:
- Good process, good outcome — reinforce
- Good process, bad outcome — reinforce the process, analyze whether the outcome reveals something the process missed
- Bad process, good outcome — correct the process, don't be seduced by the result
- Bad process, bad outcome — correct the process, note that the system is working as designed (bad process should produce bad outcomes)
Where the learning lives
The most valuable cases are the mismatches:
-
Good decision, bad outcome: Reinforce the process, don't punish the result. This is where organizations most often get it wrong — abandoning sound methods because of short-term variance. The trader who followed all their rules and lost money made the right call. The project manager who ran a disciplined process and had the project derailed by an external event did the right thing.
-
Bad decision, good outcome: Correct the process, don't reward the luck. This is the more dangerous mismatch because it's invisible in standard reviews. The trader who broke their rules and got lucky will be tempted to break them again. The team that shipped without adequate testing and didn't get burned will skip testing next time.
These mismatches are invisible in standard reviews. They're the entire point of outcome-blind review.
The goal isn't to predict outcomes better. The goal is to make decisions better — and trust that better decisions lead to better outcomes over time.
The compounding effect
The value of outcome-blind review compounds. Each review cycle improves the team's decision-making process by a small increment. Over months and years, these increments compound into a significant quality advantage.
Organizations that review outcomes learn to be lucky. Organizations that review process learn to be good. Over a long enough time horizon, good beats lucky — because luck regresses to the mean, and process quality compounds.
This is the same principle behind every durable competitive advantage: it's not about any single decision. It's about the quality of the system that produces decisions. And you can only improve that system if you can see it clearly — which means separating it from the noise of outcomes.
The cultural shift
Implementing outcome-blind review requires a cultural change: the organization has to value process over results. This feels counterintuitive, but it's the only way to build decision quality that compounds over time.
The hardest part is the transition. In the beginning, people will resist. "But the results are what matter." Yes — and the fastest path to better results is better process. You don't get there by evaluating results. You get there by evaluating the machine that produces them.
Results are noisy. Process is signal. Review the signal.