Retros Don’t Fail Because Teams Don’t Care
- Sarah Gruneisen
- Feb 8
- 4 min read
They fail because trust gets murdered in small, “well-meaning” ways.
When I interviewed Igor Podoprigora, co-founder of Codigy, I expected a product conversation.
Retrospectives. Catalogs. Cross-team collaboration. Templates. AI summaries. Jira syncing.
And yes, we touched all of that.
But the real conversation wasn’t about software.
It was about something that quietly determines whether your organization can ever scale without breaking:
trust.
“Teams are organisms.”
So why do we keep treating them like machines?
One line stuck with me:
Teams are the most important unit.
Not the roadmap.
Not the quarterly plan.
Not the strategy deck.
The team.
And engineering managers matter here because they live in the intersection: hands-on enough to know what’s real, connected enough to see the system, close enough to notice what’s breaking before leadership sees the smoke.
Here’s the provocation:
If engineering managers are mediators, what happens when the organization turns them into messengers of fear?
When their job becomes translating leadership anxiety into pressure?
When they’re asked to “get buy-in” for decisions already made?
When they’re forced to measure people with metrics never designed for human systems?
At 6-10 people, collaboration often works “smoothly enough.”
At 10-40 teams, it starts to look like 60 chefs in one kitchen.
And that’s where trust becomes the real bottleneck.
The most common feature request is actually a confession
People rarely name the real problem directly.
They hide it behind requests.
“Give me team comparison.”
“Show me productivity.”
“Let me see who is underperforming.”
“Make it visible.”
But under the request is usually a confession:
“I don’t trust what I can’t see.”
And under that:
“I’m afraid.”
Here’s the controversial part:
A lot of “metrics” are just fear wearing a suit.
And some measurement requests don’t just fail to solve the problem, they create new ones.
Lines of code.
Individual ranking.
Traffic-light comparisons between teams doing entirely different work.
These metrics don’t generate clarity.
They generate gaming, blame, and silence.
If you want to kill a retro, “just call out who wrote it”
I have never seen a room shut down faster than this moment:
Someone writes:
“I don’t feel safe in how we communicate.”
And the facilitator says:
“Who wrote that?”
Sometimes it’s framed as problem-solving. Sometimes as accountability.
But what it really is, is public exposure.
And the nervous system learns fast:
“If I’m honest, I get targeted.”
So people stop being honest.
Then leaders conclude:
“Retros don’t work.”
No.
Your facilitation murdered trust.
And then you blamed the ritual.
One design choice in the interview hit home:
protect psychological safety by separating what needs to be seen (patterns, trends) from who said it (individual identity).
Because vulnerability with a name tag attached is not vulnerability.
It’s risk, and most people will not take it twice.
Double loop learning: the thing most companies pretend to do
We talked about double loop learning, the difference between optimizing steps and questioning the whole system.
Most companies claim they do it.
Most teams don’t even have the conditions required to attempt it.
Why?
Because double loop learning requires one thing first:
permission to question assumptions without punishment.
If your metrics are used as verdicts, people won’t question the system.
They’ll protect themselves inside it.
That’s why “zoom in / zoom out” matters.
Zooming out is a leadership function, but it’s also a safety function.
Because without shared visibility, the organization operates like heart surgery through a peephole:
each team sees their tiny part, leadership guesses the whole, and complexity grows in the blind spots.
Retros become rituals when people stop believing anyone is listening
This line was brutally simple:
Retros become meaningless when trust in the process is lost.
And how do people lose trust?
When their ideas disappear.
When actions aren’t tracked.
When nothing changes.
Follow-through isn’t an efficiency detail.
It’s a trust contract.
And yes, syncing action items into the systems people actually use can help. Not because any specific tool is sacred, but because action without follow-through trains teams into cynicism.
Still, here’s the hard truth:
No tool can save a culture that punishes honesty.
A tool can only amplify what leadership does with it.
That’s why I respected hearing a product leader say, in effect:
We want to make it hard for leadership to weaponize information.
That’s not a feature.
That’s a stance.
Measuring without fear: the only measurement that matters
Metrics exist whether you like it or not.
The question is what you do with them.
Metrics should not be instructions.
They should be invitations to curiosity.
The moment metrics become moral judgment, people game them.
The moment metrics become curiosity, people learn.
And yes, teams can compare to their past selves. That’s reflection.
But comparing teams to each other in complex systems?
Usually harmful.
Because you end up comparing apples to infrastructure.
And then leaders wonder why their best people leave.
The “glue” problem: how organizations remove their own nervous system
There’s a pattern every experienced leader has seen:
The unofficial team lead.
The stabilizer.
The mentor.
The quiet connector.
The one who prevents outages by noticing invisible risks.
The one who makes other people better.
In metrics, they can look “underperforming.”
And then someone removes them.
And the team collapses.
Because the glue wasn’t extra.
It was the nervous system.
This is why “compare teams” is such a toxic request.
It doesn’t just misjudge performance.
It destroys the invisible work that keeps teams alive.
The dragon question wasn’t cute, it was the whole point
At the end, I asked:
If this product had an inner dragon, what would it protect teams from, and what courage would it help them reclaim?
The answer wasn’t dashboards.
It was autonomy and ownership:
Helping people contribute more than completing tasks, exploring options, shaping outcomes, and building real results.
If you listen closely, that’s also the mission of leadership itself.
A closing question for you
If retros in your org feel like rituals, ask yourself:
Is the problem really the retro?
Or is it that people no longer feel safe enough to tell the truth
and no longer trust enough that truth will matter?
Because tools don’t create trust.
People do.
And leaders either protect it…
or kill it with “well-meaning” control.
🐉 Dragon Insight
A dragon doesn’t guard the kingdom by watching every villager.
It guards the kingdom by making the realm safe enough that people dare to build.
Comments