Counterfactual Bias in PM

Hindsight.

Overview

Counterfactual bias is hindsight reasoning dressed up as analysis. "If only X had been done" sounds reasonable in retrospect because the outcome is known, but X was not obvious in the moment to the engineer who did the work. Catching counterfactual bias is what separates fair postmortems from postmortems that masquerade as fair while implicitly blaming the responder.

• Hindsight makes the cause obvious. Knowing the outcome reshapes what feels obvious. Catching this trap is the foundation of fair analysis.
• "Should have known" implies fault for not predicting the future. The phrase masks blame in analysis language. Reject the framing.
• "If only X" assumes X was visible. X was not obvious in the moment, by definition. Catching the implicit assumption is the work.
• Reasonable-actor test plus system focus. Would another reasonable engineer have done the same? Yes? Then the system needs to change, not the engineer.

The approach

Reasonable-actor test on every finding, system-language by default, peer review specifically for bias, documented language guide, training for new postmortem authors. The discipline scales because the patterns live in writing and review rather than in tribal knowledge.

• Reasonable-actor test. "Would another engineer in the same situation have done the same?" Apply per finding.
• System language by default. "The system allowed" rather than "the engineer should have." Catches bias at the sentence level.
• Peer review for bias. Second reader specifically scans for counterfactual reasoning. Catches what the writer normalised.
• Documented guide plus training. Team PM template flags counterfactual phrases; new authors learn the patterns explicitly.

Why this compounds

Each bias-aware PM compounds the next one. Engineers share what they actually saw because the document is system-focused, not blame-focused. Real root causes surface; psychological safety holds; the team's incident culture matures into the kind that produces honest postmortems by default.

• Better PM quality. Bias-free analysis produces real root causes. Learning is the output.
• Better psychological safety. Fair PMs preserve the team. People share what really happened.
• Cultural reinforcement. Fairness reinforces the team's identity. Counterfactual-free framing becomes the default mode.
• Year-one investment, year-two habit. First PMs establish the discipline; subsequent ones run on the patterns.