The observation
A professional poker player goes all-in holding the statistically correct hand — the maths favour her, the odds are with her — and loses anyway, because a card fell the wrong way on the river. Nothing about that decision was wrong. It was the best available bet given what she knew at the time. It simply lost, because every outcome is a mixture of decision quality and luck, and no single hand separates the two.
Annie Duke — the poker player and decision author whose conversation on Lenny's Podcast grounds this lesson — calls the opposite habit resulting: inferring decision quality from outcome quality. Praise a bad process that happened to pay off; punish a good process that happened to fail. It feels natural, because the outcome is the only thing anyone can actually see. The decision itself — the reasoning at the moment it was made — is invisible after the fact, and easy to rewrite once the result is known.
The counterintuitive core
Resulting is not a poker-table quirk. The wiki's Deciding Under Uncertainty theme traces the identical failure into investing and product, wearing a different name each time. In investing it produces survivorship bias: the manager who loaded up on tech in 1998 and happened to exit before the crash looks prescient; the manager who made the same bet under the same information but held too long is written off as reckless. In product, it corrupts the retrospective: a launch that shipped late but succeeded on market luck teaches the team the wrong lesson for the next one — ship late, it seems to say, and it will work out.
Duke's own example is a parent who feels like they have cracked parenting because the first child turned out easy-going, then discovers with the second and third that it was temperament, not technique. The mechanism is the same everywhere it shows up: a single outcome is not evidence, but it is treated as if it were, because it is the only thing that happened where the decision itself is a counterfactual.
The corrective is identical across every domain it touches: judge the process, not the result. Ask 'given what we knew at the time of the decision, was this the best available bet?' rather than 'did it work out?'. That shift is harder than it sounds, because it demands something concrete — the reasoning has to be written down before the outcome lands, or there is nothing left to judge except the outcome itself, and the whole habit of resulting reasserts itself by default.
What it means in practice
- Ask the process question, not the outcome question. 'Was this the best available bet given what we knew then?' is answerable before the result arrives and stays answerable after. 'Did it work out?' answers itself the moment the result lands — and tells you nothing about whether to repeat the decision.
- Write the reasoning down before the outcome is known. A decision reasoned only in someone's head gets silently rewritten once the outcome is in — a loss retro-fitted with doubts nobody voiced beforehand, a win retro-fitted with confidence nobody actually had. Putting the reasoning in writing at the moment of decision is what makes it available to judge later, unedited by hindsight.
- Pre-commit the kill criteria, not just the warning signs. Duke's research finds that teams routinely run a pre-mortem — imagine the project has already failed, list the early signals — and then proceed largely unchanged when those signals appear. A sales team's kill criterion: if a prospect will discuss price but not agree to a demo, the lead is dead. Deciding that in advance removes the in-the-moment rationalising ('maybe they just need more time') that shows up exactly when a decision is hardest to make well.
- The pre-mortem is not the active ingredient — the pre-commitment is. Listing the ways a project could fail costs nothing and changes nothing on its own. What changes behaviour is attaching a specific, decided-in-advance action to each signal, so that when it appears there is a policy to execute rather than a judgement call to agonise over under pressure.
The honest limit
None of this means outcomes are noise to be ignored. Over a large enough sample, they do carry signal — separating luck from skill is exactly what many trials are for, and a single hand, a single launch, a single exit proves nothing precisely because the sample is one. Duke's own First Round Capital experiment makes the point directly: five years of structured investment ratings, scored one to seven against explicit forecasts, produces a dataset large enough to show which partners' confident judgement actually predicts outcomes and which partners' confidence is correlated with nothing at all. That distinction is invisible after a single deal and only becomes visible once outcomes accumulate.
So 'ignore outcomes' overstates the corrective. The discipline is to weight process over any single outcome, not to discard outcomes altogether — a good process that loses once is not yet refuted, but a process that loses consistently, across enough trials to rule out bad luck, is telling you something real. The skill is knowing which situation you are in: judging the last hand by its process, and judging the last hundred hands by their results.
Go deeper
The single best source for this idea in full is Annie Duke on Lenny's Podcast — watch it here, or read the episode.