JUDGMENT DESIGN

We Said We Saw It Coming. Nobody Had Written It Down.

The postmortem ran ninety minutes and ended in unusual harmony. The launch had missed by a wide margin, and around the table, one after another, people said versions of the same sentence. We always knew the timeline was aggressive. The integration was never going to hold. Someone had flagged the vendor risk back in the spring. By the time the facilitator reached the action items, the room had settled on a story in which a failure that surprised everyone had, in fact, been foreseen by nearly everyone.

Nobody lied. That is what makes the scene worth looking at. Every person in that room believed what they were saying, and several of them could even produce a moment in which they had said something like it, a raised eyebrow in a planning call, a half-sentence in a hallway. What none of them could produce was a document. There was no page, dated before the launch, that said what the team expected to happen, how confident it was, and what would make it change its mind. The foresight lived entirely in memory, and memory, it turns out, is the least reliable witness an organization ever calls.

The mind edits the tape.

Once an outcome is known, it stops feeling like one of several possible outcomes and starts feeling like the only one that was ever coming. This is not a flaw of the dim or the careless. It is how human recall works. The brain does not store what you believed on a given Tuesday and retrieve it intact; it rebuilds the belief on demand, using everything you know now, including how the story ended. The ending quietly rewrites the middle. What was a vague unease becomes a clear warning. What was a confident presentation becomes, in retrospect, a performance you always half-doubted.

So the postmortem does not examine the decision as it was made. It examines a cleaned-up copy, assembled after the fact from memories that have already been revised to fit the result. The room is not reviewing its judgment. It is reviewing a better version of its judgment that never existed, and then, with a perfectly straight face, congratulating the real one for its insight.

This has a cost that compounds quietly. If everyone saw it coming, then the failure was not a failure of foresight. It was a failure of something else: execution, bad luck, a vendor, a dependency, the other department. The postmortem sorts itself toward causes that leave the original judgment intact. The organization comes away with a long list of fixes and no revision to the one thing that actually produced the miss, which was the quality of the call made in the room on the day it was made.

Why the confident feel more right than they were.

Run this for a few years and a second distortion appears, quieter than the first. Because every outcome gets retold as expected, every senior person accumulates a flattering record. The leader who backed the winner remembers having been sure. The leader who backed the loser remembers having had reservations. Both memories are generated after the fact, and both are slightly generous to the person doing the remembering. Over time the organization stops being able to tell who is actually good at calling things and who is merely good at narrating them.

That matters more than it sounds, because promotion, trust, and the weight of a voice in the room are all built on perceived judgment. If the record of judgment is edited by the people it flatters, then perceived judgment drifts away from real judgment, and nobody can detect the drift. You end up giving the biggest decisions to the people with the most persuasive memory, which is a different talent from the people with the best calibration, and a far more common one.

There is also a human cost on the other side of the table. The person who genuinely did raise the concern in the spring, who said it clearly and was overruled, finds their warning flattened into a thing everyone supposedly knew. Their foresight is not credited because it is not distinguishable from the foresight everyone now claims. In a room where all hindsight counts as insight, the one person who actually had it cannot prove it, and learns, correctly, that raising a flag is a bet with no payout.

The uncomfortable part.

Here is the reframe. The problem is not that people misremember. Misremembering is the default setting of every mind, including yours. The problem is that the organization has built a process that depends on memory at the exact moment memory is least trustworthy. It asks people to evaluate a decision after the answer is known, using only recollection, and it treats the resulting consensus as learning. You cannot learn from a decision you have not preserved. What you have preserved is a story about it, and the story was written by the outcome.

If nothing was written down before the result, the review is not an audit. It is a eulogy that has been told so many times it feels like evidence.

Notice how cheap the alternative is. It is a single page, written before the decision is final, and it carries three things: what we expect to happen, how sure we are, and what we would need to see to be wrong. Not a business case, which exists to win the approval. A forecast, which exists to be checked. When the outcome arrives, the page does not argue. It simply sits there, dated, showing what the room believed before it knew. Some of those pages will embarrass the people who wrote them. Others will vindicate the person who was overruled. All of them will be more useful than ninety minutes of consensus recollection.

What rebuilding it looks like.

The fix is not a better postmortem. It is moving the real work upstream, to the hour before the decision hardens, and making the room commit to a prediction in a form that can be graded later. Probabilities rather than adjectives. A named person who owns each assumption. A date on which the page gets opened again. Do this for a year and something unusual happens: the organization develops a track record it can actually read. It learns who is calibrated, which kinds of calls it systematically gets wrong, and where its confidence has been running ahead of its results.

Most teams will not do this on their own, because the discomfort is built in. Writing down a forecast means giving the future the chance to contradict you in public, and no one volunteers for that when memory offers such a gentle alternative. That is part of why we build the commitment step into the crisis simulations we run at SSUNDAR. Leaders record their read before the scenario resolves, and afterward the replay sets what they wrote beside what they later remember writing. The gap between the two is usually the most useful number in the room, and it is almost never small. People are rarely wrong about what happened. They are wrong about what they used to think.

An organization that cannot find its own old predictions has not learned from its mistakes. It has simply become very good at having seen them coming.

FOR GCC, BPO AND SHARED-SERVICES LEADERS

Most of what goes wrong in a global capability centre, a BPO operation or a shared-services team is not a skills gap. It is a capacity gap nobody forecast: the seats that will be needed in March against the people who will actually be delivery-ready.

See how Capability Forward forecasts bench strength and time-to-proficiency, or start the March Bench Diagnosis.

TEST YOUR OWN JUDGMENT

Theory is interesting. Data is better.

Five cascading crises. AI-generated. Your decisions compound. Get your personalized Leadership Architecture Report in under 4 minutes.

Run the Simulation.

WEEKLY INTELLIGENCE

One insight on leadership systems. Every Monday.

No fluff. No spam. Unsubscribe anytime.