The retrospective keeps producing the same three actions and nothing changes.

What to do when the retrospective keeps producing the same actions

Fort Irwin, California, the nineteen-nineties. Inside the US Army's National Training Center — the desert-based training facility where US units run live-fire exercises against an opposing force trained to fight the way an opponent would actually fight — a specific institutional discipline has been running for several years.

It is called the After Action Review.

The discipline emerged in response to a specific problem the Army had with its own post-mortems. Traditional military post-exercise debriefs had become demoralising rituals in which junior officers apologised, senior officers identified who was to blame, and nothing about the underlying system changed. Same exercise next month, same failures, same blame, same lack of change.

The After Action Review is structured to prevent all of that.

Four questions. Nothing else. What was supposed to happen. What actually happened. Why was there a difference. What will we do differently next time.

Rank is explicitly suspended for the duration of the review. A lieutenant can contradict a colonel, and if the contradiction is correct, the colonel is expected to acknowledge it. The reviews happen immediately after the event, not weeks later. And — this is the part most corporate adaptations later drop — the output of the AAR is a documented change to the standard procedures, not a personal commitment.

The person whose behaviour needs to change isn't the one being fixed. The system that produced the behaviour is.

The format spreads into corporate training in the nineteen-nineties and is imitated with varying success. The places where it works are the places that keep the four-question discipline and the system-level outputs. The places where it fails are the ones that keep the vocabulary and turn the output back into personal action items that go on a list nobody reads.

Why the room, not the format, is usually broken

When a retrospective keeps producing the same three actions and nothing changes, it is rarely because the retrospective format is wrong. It is because the room around the retrospective hasn't been redesigned to support the discipline the format is supposed to enable. The format is the easy part. The room is the structural one.

So let's go to the office and work through it.

Start by finding what is really keeping the change stuck

"The retrospective keeps producing the same three actions and nothing changes."

The feeling is cynical.

The team turns up. The post-its get written. The same patterns surface. The same actions get committed to. And in the next retrospective, the same patterns surface again, and everyone in the room has stopped expecting the actions to land.

Two choices. They look like the same broken retrospective. Different fundamental causes.

When the team is venting out of habit

Choice one: the team is venting out of habit. The retrospective has become a ritual, and the ritual is more comfortable than the change. Everyone has accepted that the meeting is where complaints are stored, not where action happens.

If that's the read, ban any discussion of fixing anything for the first twenty minutes. Force the team to dwell on what the recurring problem actually means, at the level of why the system keeps producing it, before anyone is allowed to propose a solution.

Chris Argyris, working on what he called double-loop learning, observed that the same three actions keep coming up because they are second-loop actions attached to first-loop thinking. The team keeps trying to fix the output without ever examining the rules that produce it. The twenty minutes of forced non-solving is what creates space for the second loop to actually happen.

When the team lacks the authority to fix it

Choice two: the team lacks the authority. The team keeps raising the same complaints because the fixes sit above their pay grade. Nothing changes because nothing can change from inside the team.

If that's the read, take the top recurring complaint and escalate it yourself, this afternoon, directly to whoever has the power to act on it. Not through a process. Not through the next management review. Directly, and visibly, in a way the team can see happening.

W. Edwards Deming's insight is the structural framing here. Most quality problems are system problems, and system problems can only be solved from above the level they are happening at. The retrospective stops being a complaint box the moment the team sees that raising something in it leads to a named person doing something about it. And if escalation doesn't work — if the complaint disappears into the upstream system and nothing comes back — then you know the retrospective was theatre and you can plan accordingly.

Venting, or no-authority. Same broken retrospective. Two different first moves.

How to make the retrospective produce more than a list

Three tools. The discipline is to make the retrospective produce something other than a list. All three tools were unpacked earlier in the toolkit; here, they combine.

The first tool is

Hold the team in meaning before action

What, So What, Now What.

We unpacked the framework at scenario six — Borton's nineteen-seventy reflective-writing model, formalised into Liberating Structures by Lipmanowicz and McCandless. Three sequential phases. What happened. What does it mean. What do we do.

The reason the framework matters here is that retrospectives that produce the same three actions every time are ones where the team jumps straight to the Now What without doing the other two. The actions keep being the same because nobody has understood the pattern underneath them.

The discipline for this scenario is the order. The team is not allowed to propose actions until they have agreed on what happened and what it means. The order is the whole point. Making the team complete the What and So What phases — properly, against the same three patterns that keep coming up — is what produces a different Now What than the team has been producing on its own.

If the same three patterns surface for the third retrospective running, the So What phase has been telling you the team is in second-loop territory and the Now What hasn't followed. Hold the team in the So What phase deliberately; the actions that emerge will be different from the ones that have been failing.

The second tool is

Push past the first explanation that sounds right

Digging Deeper.

We unpacked it at scenario twenty-seven — Teresa Torres and the continuous-discovery tradition's discipline of refusing to accept the first explanation that sounds like an explanation, and pushing past the comfortable stopping points until the actual mechanism is exposed.

Most retrospectives stop one or two whys short of the real answer, which is why the real answer never gets fixed. The team writes we missed the deadline because the spec was unclear; the team commits to write clearer specs; the spec stays unclear-shaped because the cause of the unclear spec was upstream of the team's authority and was never named.

The discipline here is to apply the Digging Deeper move during the retrospective, not after it. We missed the deadline because the spec was unclear. Why was the spec unclear? Because the customer hadn't validated it. Why hadn't the customer validated it? Because the salesperson promised features the customer hadn't asked for. Why? Because the sales compensation rewards features-promised, not features-validated. The structural cause is two layers above where the retrospective stopped, and the action at the structural cause is different from write clearer specs.

The framework's value, applied in retrospect, is that it produces a structurally different action set than the team has been producing.

The third tool is

Close the loop visibly so complaints lead to change

Feedback Loops.

We unpacked it at scenario eighteen — the cybernetic-and-Deming tradition of flag, respond, visibly close, with the structural insight that the loop closes at the visibility moment, not at the action moment.

The reason Feedback Loops matter here is that retrospectives that produce the same three actions every time are loops whose closing-the-loop step has been failing. The complaint is raised. Maybe the action happens. Maybe it doesn't. The team has no visibility either way. The next retrospective therefore raises the same complaint, because as far as the team can see, nothing happened to the last one.

The discipline is to shorten the loop between the complaint being raised and the adjustment being made — and to do the closing visibly. Three weeks ago at the retro you raised X. Here is what we did about it. Here is the result. Said in the room, in writing, in front of the team.

The longer the gap between flag and visible-close, the more the team learns that complaints don't lead to change, and the more the retrospective becomes a place where people go to say the same things they said last time.

That's the toolkit. One more story before we close.

The After Action Review we opened with showed an institution that built a different room around the post-mortem so the post-mortem could produce different outputs. The story we close with shows the same move at the smallest scale — a founder who didn't redesign the meeting format but redesigned what it cost to say certain things out loud inside it.

A precedent: changing the cost of naming failure

Atlanta, the early two-thousands onwards. When Sara Blakely founds Spanx in two thousand with roughly five thousand dollars of her own savings, the ritual she brings into the company has been formed at her family dinner table.

Her father — a personal-injury lawyer — had asked each of his children at dinner every week, since childhood, what they had failed at that week. Blakely tells NPR's How I Built This in two thousand and sixteen that if the answer was nothing, her father was disappointed: the absence of a failure meant the absence of an attempt.

At Spanx, she formalises the ritual into what the firm calls the Oops Meeting — a recurring company meeting whose only agenda item is for employees to name failures in front of their colleagues and be applauded for them.

The usefulness of the ritual isn't the applause. It is the structural consequence.

The retrospective meetings that had been producing the same three safe actions begin producing different ones, because the failures the team is willing to name have changed. A retrospective inside a company that penalises naming a failure — even implicitly, even through a raised eyebrow in a senior stakeholder — will reliably produce the same three actions indefinitely, because those are the three actions that can be named without carrying cost. The actions that would actually change things are the ones the team can't yet afford to put on the list.

So.

The After Action Review built a different room around the post-mortem so the post-mortem could produce different outputs. Sara Blakely built a different cost around naming things, so the things that got named changed, and the actions changed with them. Two different cuts at the same move — change what the room can afford to say, not the meeting format.

When the retrospective keeps producing the same three actions and nothing changes, the mistake is usually to redesign the retrospective. The retrospective isn't broken. The room around it is. Blakely's move wasn't to change the format of the review. It was to change what it cost to say certain things out loud, so that the review had different things to work with.

The agenda was fine. The invisible tax on the agenda was the problem.

So. Your Next Move from this playbook.

Which action item from your last three retrospectives has actually been done — and if the answer is none, what is the retrospective even for?