Stop grading decisions after you know the answer

A profitable trade can conceal poor judgment, just as a loss can make sound judgment look foolish. Consider two investment managers approving the same kind of position with the same information, reasoning, and exposure. One earns a profit. The other loses money because an uncertain event resolves differently. If the organization praises the first manager's thinking and condemns the second's, it has allowed the result to change its account of the decision.

That can feel like common sense. Investment firms exist to produce results, and losses have real consequences. Yet evaluating a decision requires asking what was reasonably knowable when it was made. A review that forgets this can reward a mistake that happened to pay and punish a sound judgment that encountered an unfavorable outcome.

Jonathan Baron and John Hershey demonstrated the problem in a series of studies published in 1988. Participants evaluated decisions involving medical situations or monetary gambles. Favorable outcomes led to more favorable assessments of the decision maker and the quality of the thinking, even though participants had the relevant information available at the time of the decision. The studies examined hypothetical cases rather than fund management, but they expose a difficulty any review process should address. Outcome bias in decision evaluation ↗

The organizational consequences extend beyond an unfair performance review. Consider a manager who breaks a limit and earns a large return. If the firm celebrates without examining the breach, colleagues receive information about what the organization actually rewards. The written rule says one thing. The promotion says another. A later loss may then be treated as an isolated failure of character, although the incentive was visible earlier.

A review should therefore begin with the material that existed before the result. Examine the rationale, the alternatives, and the uncertainties the decision maker identified. Establish whether the position fit the mandate and whether the manager addressed the evidence against it. Only then bring in the outcome and investigate what it adds.

For some decisions, a reviewer can assess the original materials without seeing the result. That will not always be possible, especially after a prominent event, but the order of review still matters. Making an initial assessment before studying the outcome creates a record of what the process itself warranted. It becomes harder to invent a previously invisible flaw solely because the result was bad.

This approach can sound like an invitation to excuse failure. It would become one if the organization accepted any polished memo as proof of sound judgment. Process assessment must be demanding. A rationale that ignores foreseeable risks is weak even if the trade succeeds. Repeatedly incorrect forecasts need investigation. A manager who invokes bad luck after every loss should be asked to explain the accumulating pattern.

Outcomes also contain information that no initial record can supply. They can reveal that a model omitted an important relationship or that an assumed safeguard did not work. The review should use that information to improve future decisions. What it should resist is treating every realized event as something the manager ought to have predicted with certainty.

The distinction becomes more important under pressure. If employees believe that only visible wins count, admitting uncertainty or identifying a risk can look like preparing an excuse. The organization may then receive less honest analysis precisely when it needs more. A credible review process makes it possible to disclose a weakness before a result forces everyone to acknowledge it.

Results should remain part of compensation and accountability. Over a suitable period, they help show whether a strategy serves its purpose. But they need to sit alongside an assessment of how the organization took its risks, including the profitable ones. A firm that investigates only losses will leave some of its most instructive mistakes untouched.

At the next profitable quarter, review a few winning decisions as rigorously as losing ones. Any limit breach or unsupported assumption deserves an answer while the organization can still afford to learn from it.