Retrospectives should be the engine of continuous improvement. In practice, they’re often the meeting everyone tolerates but nobody values, a ritual performed out of obligation rather than conviction, producing sticky notes that get photographed and forgotten.

I’ve run hundreds of retrospectives over the years, and I’ve wasted more of them than I’d like to admit. The ones that worked had a few things in common. The ones that didn’t were all failing in the same ways.

Why They Fail

The failure modes are predictable:

They get skipped. When the team is busy, which is always, the retrospective is the first meeting to be cancelled. It feels optional in a way that planning and standups don’t. But skipping retrospectives is like skipping maintenance on a machine. It runs fine for a while, and then it doesn’t, and by then the problems have compounded.

They get shortened. “We only have 15 minutes, so let’s keep it quick.” A rushed retrospective produces surface-level observations rather than genuine insights. The important things, the uncomfortable truths, the systemic issues, the patterns that nobody wants to name, need time and space to emerge.

They produce action items nobody tracks. This is the most common and most corrosive failure. The team identifies problems, agrees on actions, and then nothing happens. After a few cycles of this, people stop bothering to raise issues because they’ve learned that raising them doesn’t lead to change. The retrospective becomes performative.

The same issues come up every time. When the team raises the same problems sprint after sprint with no resolution, the retrospective stops being a tool for improvement and becomes a venue for venting. Venting has some value, but it’s not the same as improving.

People don’t feel safe being honest. If the retrospective is attended by managers who might judge, or if previous honesty has been met with defensiveness, people will stick to safe, superficial observations. “The sprint went well” is the retrospective equivalent of “fine” when someone asks how you are.

What Works

The retrospectives that produced real change in my teams shared these characteristics:

Psychological safety is non-negotiable. People need to feel safe saying “this didn’t work” without fear of blame. That means the leader’s response to criticism matters enormously. If someone raises a problem and you get defensive, you’ve just taught the whole team not to raise problems. If you thank them and dig into it with curiosity, you’ve taught them that honesty is valued.

Action items have owners and deadlines. “We should improve our test coverage” is not an action item. “Sarah will create a proposal for improving test coverage in the authentication module by next Friday” is an action item. The specificity matters because it creates accountability.

Follow-up happens. Start every retrospective by reviewing the action items from the last one. What was completed? What wasn’t? Why? This simple practice transforms retrospectives from wish-listing sessions into accountability mechanisms. When people see that actions from previous retrospectives actually get done, they invest more in the process.

The format varies. Using the same format every time, “what went well, what didn’t, what should we change”, gets stale. Rotate formats. Try timeline retrospectives, sailboat metaphors, start/stop/continue, or focused retrospectives on specific topics. The variety keeps people engaged and surfaces different kinds of insights.

There’s enough time. A good retrospective for a two-week sprint needs at least 60 minutes. For longer projects or significant milestones, allow 90 minutes or more. The depth of insight is proportional to the time invested.

The Retrospective on the Retrospective

Marcus Blankenship suggests treating every fourth one-on-one as a retrospective on the one-on-ones themselves. I think the same principle applies to retrospectives. Periodically, maybe quarterly, ask the team: “Are our retrospectives useful? What would make them better? Should we change the format, the frequency, the facilitation?”

This meta-retrospective prevents the process from becoming stale and gives the team ownership over how they improve. If the team decides that fortnightly retrospectives are too frequent and monthly ones would be more valuable, that’s a legitimate conclusion. The goal is continuous improvement, not rigid adherence to a ceremony.

Self-Improving Teams

Cate Huston describes the goal well: self-improving teams critique, iterate, and change things as part of their process. They’re not afraid to discuss what worked and what didn’t. They make suggestions and try changes knowing that some will fail. These teams get better over time with less and less intervention from the manager.

Getting to that point requires investment. The first few retrospectives might be rough, people are cautious, the insights are shallow, the follow-through is inconsistent. But if you persist, if you demonstrate that the process is valued by actually acting on what it produces, the team will gradually open up and the retrospectives will become genuinely valuable.

The retrospective is the team’s primary mechanism for self-correction. Treat it with the seriousness it deserves, and it will repay that investment many times over. Treat it as a checkbox, and you’ll wonder why your team keeps making the same mistakes.