Mistake Learning Bank
Convert failed decisions into reusable organizational knowledge
- Difficulty
- Moderate
- Time to result
- ~ongoing to results
- Steps
- 5
- Confidence
- 98%
Create a culture in which reversible mistakes are expected but unexamined mistakes are not. When data shows that a decision failed, name it promptly instead of finding a convenient statistic to defend it. Reconstruct why the team chose that path, identify which assumption or process needs to change, and turn the finding into a durable rule, check, or piece of shared knowledge. Then stop the harmful action, preserve what the experiment taught, and run a better test. Over time, the organization accumulates a bank of both failed and successful experiments that improves future decisions. The approach depends on honest measurement and psychological permission to admit error; it does not apply unchanged where a single mistake could cause severe harm.
Origin
Kogan used the company's excessive inventory after the pandemic-era boom as a major example. His team acknowledged the call was wrong, cleared inventory, reduced holdings, and shifted more of the business toward subscription and platform revenue that did not depend on owned stock.
Core principles
- 01Mistakes are useful when they produce retained learning
- 02Call a failed decision a mistake as soon as the data supports it
- 03Investigate the decision process rather than hiding the result
- 04Repeat good experiments and stop repeating disproven actions
How to run it
- 1
Read the result honestly
Compare the observed outcome with what the decision was meant to achieve, using relevant data rather than a convenient supporting chart.
Pro tip Agree on success and failure signals before running the experiment.
Watch out Selective statistics can make almost any decision appear justified.
- 2
Call the mistake
State plainly that the decision was wrong once the evidence is sufficient, without treating the admission as personal defeat.
Pro tip Separate the quality of the person from the outcome of the decision.
Watch out Delay increases the cost and reduces the time available to change course.
- 3
Reconstruct the decision
Ask what assumptions, information, and incentives led to the choice and which of them proved unreliable.
Pro tip Include what looked reasonable at the time so the lesson remains fair and usable.
Watch out Hindsight alone does not explain how to decide better next time.
- 4
Bank the learning
Turn the finding into a change in process, data, controls, or shared organizational knowledge that future decisions can access.
Pro tip Prefer a concrete check or changed operating method over a vague reminder.
Watch out A lesson that remains only in one person's memory will be lost.
- 5
Change course and iterate
Stop doing the disproven action, preserve the useful evidence, and choose the next experiment based on what was learned.
Pro tip Keep doing more of what the data shows is working.
Watch out Experimentation is not a defence for repeating the same preventable mistake.
In the wild
After online demand surged, Kogan increased inventory and expanded from roughly 13 to 38 warehouses. Demand then fell as restrictions eased, leaving stock at ports and forcing discounting. The company called the decision wrong, cleared inventory, and redesigned the business around leaner stock plus more subscription and platform revenue.
→ Kogan said inventory fell from well over $200 million at the pandemic peak to around $40 million, while more sales became independent of held inventory.
Common mistakes
Defending the failed decision
Searching for a favourable graph after the result is known prevents the organization from learning and changing course.
Using mistakes as a slogan
Permission to fail must be paired with diagnosis, retained learning, and a concrete change that reduces recurrence.
Is it for you?
Best for
It is best for reversible business decisions and experiments where evidence can reveal what worked and what failed.
Not ideal for
It is not ideal as permission to relax controls in safety-critical environments such as aircraft or hospital life-support systems.
From the transcript
“why did I make that decision what can I change to make sure that decision doesn't happen again”
“collect the data stop doing the bad stuff keep doing more of the good stuff”
From the episode
536: He Made $450M Selling TV's for $0.01
Ruslan Kogan