The Wrong Lesson Gets Learned

Category: Incentives, metrics, and consequences Organisations frequently learn the opposite of what they should.

A project ships late and over budget, and it is painful enough that the organisation decides to learn from it. There is a review. The review finds that a senior engineer made an early architectural call that turned out to be wrong, and that the wrong call was what set the overrun in motion. The lesson is drawn, written up, and circulated: we need more oversight of early technical decisions. From now on, big calls go through a review board.

It sounds sensible. It is also the wrong lesson, and it will quietly make things worse.

What actually happened was subtler. The engineer made a reasonable bet with the information she had, flagged her uncertainty clearly, and asked for the time to validate it before committing. She was told there was no time. The bet was forced early, by a deadline nobody would move, and it went the way unvalidated bets sometimes go. The real lesson was about how the organisation manufactures false urgency and punishes people for wanting to check things. But that lesson implicates the leadership that ran the review, so it is not the one that gets learned.

Instead, the organisation adds a review board. Decisions now take longer, the deadline pressure that caused the problem is untouched, and the engineer — who behaved exactly as you would want — learns that flagging uncertainty gets you blamed. The next person in her position will not flag it. The organisation has held a careful post-mortem and come away more fragile than before.

The Principle

Organisations do not automatically learn the right lesson from a failure. They learn the lesson that is easiest to see, cheapest to act on, and least threatening to the people drawing it — which is frequently the wrong one.

A failure is raw material; the lesson is an interpretation, and the interpretation is chosen under all the usual pressures. The version that survives is rarely the truest one. It is the one that fits an existing belief, blames a convenient party, or can be solved with a visible action — regardless of whether it addresses what actually went wrong.

Why It Is Inevitable

The real cause of a failure is usually diffuse, systemic, and uncomfortable — it implicates conditions, incentives, and often the people running the review. The wrong lesson is usually specific, local, and flattering — it points at an individual’s mistake or a missing process, neither of which asks anyone in power to change. Given a choice between a true lesson that indicts the system and a false one that indicts a person or a gap, the false one is easier to adopt and far easier to live with.

There is also a strong pull toward the actionable. A review is expected to produce a recommendation, and “add a control” is a recommendation you can announce, whereas “stop creating the conditions that caused this” is vague, slow, and hard to be seen doing. So failures reliably generate more process, because process is the lesson shaped like an action, even when process was not the problem.

And hindsight distorts everything. After the fact, the decision that led to the failure looks obviously wrong, so the lesson becomes “that decision was bad” rather than “that decision was reasonable given what was knowable, and the failure was in the conditions around it.” The clarity of hindsight makes the wrong lesson feel like the obvious one.

How It Shows Up

• A failure produces a new control, sign-off, or process — but the underlying pressure that caused it is never named.
• The lesson lands on a person (“X made a bad call”) rather than the conditions that forced or shaped the call.
• The fix prevents the exact previous failure while doing nothing about the family of failures it belonged to.
• Reviews that conveniently exonerate the people running them.
• The same kind of failure recurs in a new form, having simply routed around the specific control added last time.

Why It Causes Damage

The wrong lesson is worse than no lesson, because it comes with the full confidence of having learned. The organisation believes it has dealt with the problem, lowers its guard, and is genuinely surprised when the same thing happens again wearing different clothes. Meanwhile the new controls accumulate, each one slowing the system down in exchange for protection it does not actually provide.

It also teaches the people involved exactly the wrong behaviour. When the engineer who flagged her uncertainty is the one the lesson blames, everyone watching learns that honesty about risk is dangerous. The organisation does not just fail to fix the real cause; it actively trains people to hide the very signals that might have prevented the next one.

How To Counter It

• Ask why the reasonable people behaved as they did, not who made the mistake. If the answer is “the conditions made this the sensible local choice,” the lesson is about the conditions.
• Distrust any lesson that conveniently blames the absent, the junior, or the departed — and any review whose conclusions ask nothing of the people who ran it.
• Be suspicious of “add a control” as a reflex. Ask whether the proposed fix addresses the cause or just the most recent symptom of it.
• Separate the decision from the outcome. Judge whether the call was reasonable given what was knowable at the time, not by how it turned out.
• Check the lesson against the family, not the instance: would this fix have prevented the near-misses too, or only this one exact event?

What Good Looks Like

Organisations that treat a failure as a question rather than a verdict — that resist the first, most comfortable explanation and keep asking why until the answer points at conditions they can actually change. Where reviews are willing to indict the system that ran them, and where the people who surfaced risk honestly are protected, not blamed.

They learn slowly and uncomfortably, and they are far harder to surprise twice, because the lesson they took was the one that was actually true.

A Reflective Question

Think of a “lesson learned” your organisation is proud of. Did it change the conditions that produced the failure — or did it add a control, blame a person, and leave the real cause exactly where it was?