Why Organizations Struggle to Learn
Why Organizations Struggle to Learn
Almost every organization says it “learns from failures.” And almost every organization repeats the same mistakes.
That gap isn’t accidental. The knowledge exists - but it doesn’t move.
The Illusion: If We Analyze, We’ll Learn
After a failure, the following usually happens: data gets collected. A presentation gets written. Conclusions get drawn.
On paper - learning. In practice - documentation.
Real learning isn’t writing down what happened, it’s changing what happens next time.
Why Postmortems Don’t Change Behavior
A postmortem is a process in which an organization analyzes a failure that already happened: what went wrong, how it was detected, and what actions were taken to prevent recurrence. Its stated goal is learning - not blame.
Most postmortems fail for one simple reason: they focus on the event, not on the system that created it.
They answer:
- What broke
- Who was involved
- And what specific change we’ll add
But they don’t touch:
- Incentives
- The structure of responsibility
- Or processes that keep producing the same failure
The result is a good document - without real change.
Local Knowledge vs. Systemic Learning
In every failure, there are people who learn a lot. The problem is that this learning stays local.
It gets stuck:
- In one team
- In one domain
- Or with two or three people “who know”
The organization, as a system, doesn’t change.
Next time - a similar failure will appear elsewhere, with different names, but for the same reason.
A Practical Example: A Failure That Was Learned - and Forgotten
In a certain organization, a failure was caused by a combination of a short timeout and momentary load.
The postmortem was excellent:
- They understood the root cause
- Improved the timeout in one service
- And added a check
Six months later - an almost identical failure, in a different service, in a different team.
Why? Because:
- The incentive for speed remained
- Assumptions about load didn’t change
- And the learning wasn’t translated into an organizational principle
The knowledge was there. It simply didn’t move.
How an Organization “Forgets” Over and Over
Organizations forget not because of a short memory, but because of structure.
This is what it looks like:
- There’s no ownership of cross-cutting learning
- And systemic changes are politically expensive
It’s easier to:
- Fix locally
- Move on
- And hope it doesn’t happen again
And it always does.
When Real Learning Actually Happens
Learning happens when you change one of these:
- Incentives
- Boundaries of responsibility
- Or a process that produces the same pattern
Not when you add another test. And not when you write another document.
Learning is a structural change - not an intellectual one.
The Bottom Line
Organizations don’t fail at learning because they lack smart people.
They fail because their structure doesn’t let knowledge move.
Looking Ahead
Even when an organization understands its problems, it doesn’t always have the capacity to improve.
In the next post we’ll address that moment - when an organization simply has no more capacity for change, and why adding people or processes only makes things worse.
📚 More in this Series: Human-Scale Engineering
- Part 1 The Organization Is Part of the System (Even If You Never Wrote a Line About It)
- Part 2 Incentives Are Stronger Than Architecture
- Part 3 Ownership as a Double-Edged Sword
- Part 4 Processes Born for Scale, But That Kill Responsiveness
- Part 5 Why "Alignment" Is a Dangerous Concept
- Part 6 Leadership That Allows Mistakes, Rather Than Preventing Them
- Part 8 When an Organization Has No More Capacity to Improve
- Part 9 When You Need to Break an Organizational Structure
- Part 10 Engineering Is a Social System
- Part 11 What We Actually Learned About Human-Scale Engineering