Outcome-Process Separation¶

Why It Matters¶

Outcome bias is pervasive—we judge decisions by their results, not by their quality at decision time. This generates false learning: punishing good decisions that got unlucky, rewarding bad decisions that got lucky. Organizations oscillate between opposite errors based on last outcome rather than improving actual decision processes. Separating outcome from process enables learning what actually improves decisions rather than what luck provided.

The Key Move¶

Before evaluating any decision (especially failed ones), reconstruct the decision process independent of outcomes: what options were considered? What information was gathered? What assumptions were explicit? What analysis methods were used? What alternative viewpoints were sought? What uncertainties were acknowledged? Evaluate this process against reasonable standards: was information that should have been available actually gathered? Were known biases checked? Were probabilistic judgments calibrated? Only after evaluating process do you consider outcome—and outcome informs you about luck and external factors, not primarily about decision quality.

Classic Application¶

Investment performance evaluation. A portfolio manager makes a high-risk bet that pays off spectacularly. Outcome evaluation: "Great decision!" Process evaluation: "Was the risk level appropriate for the client? Was it based on actual analysis or gut feeling? Were alternative investments considered? Was downside protected?" A lucky outcome doesn't validate a reckless process. Conversely, a well-reasoned investment that loses money may still be good process. Sophisticated investors evaluate managers on process, not just returns.

Surprising Application¶

Medical treatment decisions. A patient receives aggressive treatment and survives. Outcome bias: "The aggressive treatment saved them!" Process evaluation: "Given presenting symptoms and patient history, was aggressive treatment the evidence-based choice? Were alternatives considered? Was informed consent obtained? Were risks appropriate to benefits?" Sometimes aggressive treatment happens to work despite being poor process; sometimes conservative treatment is correct process but patient dies anyway. Separating these prevents learning wrong lessons.

Failure Modes¶

Hard to maintain discipline when outcomes are emotionally salient (deaths, major losses). Process evaluation requires detailed documentation of decision-making, which is often absent. "Good process" is debatable—different experts have different process standards. Can excuse all bad outcomes as "process was fine, just unlucky"—requires honest process evaluation, not post-hoc rationalization. Doesn't tell you what the outcome teaches about the world (as opposed to what it teaches about the decision).

Go Deeper¶

Russo & Schoemaker, Decision Traps, Chapter 12; Baron & Hershey, "Outcome Bias in Decision Evaluation," Journal of Personality and Social Psychology 54, 1988; Hammond, Keeney & Raiffa, Smart Choices, Chapter 2.