Pre-mortems and confidence scoring — the two checks that separate sober decisions from optimistic ones

The two failure modes a normal pros-and-cons list won't catch

The pros-and-cons list is the default decision tool of working adults. It is also the wrong tool for any decision where two specific failure modes show up: the unimagined disaster (a way the plan can fail that you simply did not think of), and the missing input (a piece of context you didn't know you needed). Pre-mortems and confidence scoring exist to catch exactly those two failures.

Neither is intuitive. Both feel slightly uncomfortable in the moment because they undermine the certainty you wanted from the analysis. That discomfort is the point.

What a pre-mortem actually is

The technique is from Gary Klein, who studied how experts make decisions under uncertainty. The standard post-mortem looks at a failure after it happened and tries to identify the causes. A pre-mortem inverts the timeline: imagine it is one year from now, the decision was made, and it failed badly. Write the story of how that happened.

The psychological mechanism Klein identified is prospective hindsight. People are dramatically better at generating plausible failure causes when they pretend the failure already occurred than when they're asked "what could go wrong?" in the abstract. The first framing produces vague risks. The second produces specific, narrated, causally connected failure scenarios — the kind you can actually do something about.

The move that makes pre-mortems work is forcing yourself past the easy answers. "The market changed" is not a pre-mortem. "We launched the pricing tier in March; enterprise prospects pushed back on the per-seat model in week three; sales took until June to admit the objection was structural rather than seasonal; by then the cash runway was nine weeks" is a pre-mortem. The specificity is what makes the preventative measures actionable.

Why pre-mortems beat post-mortems for prevention

Post-mortems are valuable for institutional learning. They are nearly useless for preventing the specific failure you are about to walk into, because by the time you write a post-mortem the failure has already occurred. The pre-mortem moves the analysis to the only point where prevention is still possible: before commitment.

The other underrated benefit is dissent legitimization. Teams suppress doubts in pre-decision meetings because raising concerns reads as obstruction. The pre-mortem reframes that: now everyone is required to imagine the failure, so saying "here is how this could go wrong" is the assigned task, not a political move. Concerns surface that would otherwise have stayed buried until they showed up as real problems.

What confidence scoring is actually measuring

A confidence score is not a measure of how good the decision is. It is a measure of how complete the inputs to the analysis are. Two decisions can have identical pros-and-cons lists; the one made with full context is materially better than the one made with three critical unknowns.

Reloadium Decisions outputs a 0–100 confidence score alongside every analysis, calibrated against what information the decision actually required and what was provided. When the score is low, the system doesn't just tell you the score — it tells you exactly which inputs are missing: the success criteria are undefined, the time horizon weights are inconsistent with the stated priorities, no constraint information was provided.

This matters because the alternative is false precision. A pros-and-cons list with five well-articulated factors feels like a thorough analysis, even when the five factors were chosen because they were easy to think of rather than because they were the most important ones. The confidence score is the system's honest report on its own work.

Common decision fallacies these two checks prevent

• Confirmation bias — the pre-mortem forces engagement with failure modes you weren't looking for
• Planning fallacy — confidence scoring penalizes missing constraint and timeline information that the planning fallacy systematically ignores
• Survivorship bias — pre-mortems generated against the specific decision, not against the abstract category, surface the failure paths you don't see in case studies
• Sunk-cost reasoning — running the pre-mortem before committing means you have not yet invested anything that would distort the assessment

How Decisions runs both automatically

The pre-mortem and confidence score run on every analysis without you opting in, because the decisions you most need them for are the ones where you would have skipped them. The pre-mortem appears as a narrated failure scenario with the warning signals you would have missed and the specific preventative measures available now. The confidence score appears at the top of the analysis, with a list of missing inputs you can address before deciding rather than after.

You can act on either output independently. Address the missing-info warnings, then re-run the analysis. Implement two of the four preventative measures from the pre-mortem, then re-classify the decision. The tool isn't trying to make the choice for you. It is making sure the choice you make has been tested against the two failure modes that ordinarily get skipped.