Learning from what goes wrong, why most team reflection doesn't change anything, and what does

Good morning. Most teams have a version of the debrief, the retrospective, the post-mortem, the wash-up after a project or a launch or a hard week. People gather, name what went well and what didn't, and in most organisations the conversation is genuine. People are honest. The insights are real. Then the next project begins, the same patterns resurface, and the debrief document sits somewhere no one opens again.

The gap isn't a lack of willingness to learn. Learning from experience takes more than reflecting on it. It takes changing something as a result, and that second step is where most teams quietly stall.

Amy Edmondson's research on how organisations learn draws a distinction that most retrospectives blur. Preventable failures happen when a known process wasn't followed; they call for better systems or clearer standards. Complex failures happen when familiar things combine in an unfamiliar way; they call for understanding what the team couldn't have seen coming. Intelligent failures happen at the frontier, a reasoned attempt at something genuinely new that didn't work; these are worth treating as the cost of a useful experiment, not as an error to stamp out.

When a team treats all three the same way, one thing reliably follows: people learn to manage the appearance of performance rather than its reality, because from where they're standing every failure looks equally punishable. A team that can't tell an intelligent failure from a careless one teaches its best people to stop taking the risks that produce learning.

The second structural problem is timing. A retrospective held weeks after the fact leans on memory the outcome has already reshaped: once you know how it ended, you misremember how clear it was at the time. Teams that genuinely improve tend to run shorter, more frequent learning moments inside the work, rather than one long inquest bolted onto the end.

The shift worth making is from reflection to revision. A debrief that ends in insight has done half the job. The half that changes anything is the decision that follows: what will we do differently, starting when, and how will we know it worked. Insight is the easy part to feel good about. Revision is the part that moves the team.

These patterns are common, and worth naming plainly, because they mean the team is paying the full price of reflection, the time, the honesty, the discomfort, without collecting the benefit.

They separate the failure from the system that produced it. The question moves from "who got this wrong?" to "what about how we work made this outcome more likely?" and that scrutiny includes the leader. Exempting yourself from it tells the team that reflection is for everyone except the person running the meeting.

They keep the loops short. Not everything needs a formal retrospective. A standing question in a weekly check-in, what's moving, what's stuck, what surprised us, surfaces learning while it's still fresh enough to act on.

They close the loop with a commitment, not a conclusion. Every learning conversation ends the same way: one specific change, an owner, and a signal that will tell you whether it's working. One real change beats ten insights.

The teams that improve fastest have made honest examination of what went wrong an ordinary part of how they work, and they never leave the room without deciding what changes.