One Measurable Innovation
Add one testable improvement to every new iteration
- Difficulty
- Easy
- Time to result
- ~ongoing to results
- Steps
- 6
- Confidence
- 98%
Pabrai describes IKEA's rule that every new store must contain an innovation not present in earlier stores. Bartlett generalizes it to recurring work such as podcast episodes, and Pabrai adds the key constraint: the change must be measurable or it is not an experiment. The operator chooses one meaningful improvement, states the expected effect, defines the metric in advance, and runs the next iteration while limiting unrelated changes where practical. The result is then recorded and used to keep, revise, or retire the change. One bounded experiment reduces attribution confusion and creates a steady learning cadence without requiring every iteration to be radically new. The method should not delay urgent fixes, and a measured result still needs enough observations before being treated as a stable conclusion.
Origin
Pabrai says IKEA required every newly opened store to include an innovation absent from prior stores. Bartlett proposed applying one new experiment to each podcast, and Pabrai insisted it be measurable.
Core principles
- 01Repeated outputs should not be exact copies of the last version
- 02One bounded change keeps improvement manageable
- 03A change is an experiment only when its result can be measured
- 04Learning compounds across successive iterations
How to run it
- 1
Select one iteration
Choose the next recurring output, location, campaign, or operating cycle where a change can be observed.
Pro tip Use a repeated unit with a reasonable comparison point.
Watch out Do not force experimentation into a one-off event with no observable outcome.
- 2
Choose one improvement
Identify a specific change that could improve customer value, cost, quality, or learning.
Pro tip Keep the experiment narrow enough to explain clearly.
Watch out Bundling many changes makes the source of the result unclear.
- 3
Define the prediction
State what should change if the idea works and why.
Pro tip Write the prediction before seeing the outcome.
Watch out A vague hope cannot be falsified.
- 4
Choose the measure
Define the metric, observation window, and comparison that will distinguish improvement from no improvement.
Pro tip Prefer a customer or operating outcome over activity volume.
Watch out Measurement chosen afterward invites confirmation bias.
- 5
Run and record
Launch the iteration, collect the defined evidence, and record what happened.
Pro tip Note important contextual differences from the comparison iteration.
Watch out One result may be noisy or confounded.
- 6
Decide the inheritance
Keep, revise, or remove the change, then choose the next measurable innovation.
Pro tip Carry successful improvements into the new baseline.
Watch out Do not keep a change merely because it was novel.
In the wild
Pabrai says IKEA's founder required each new store to introduce an innovation not present in previous stores because failing to keep innovating would eventually weaken the company.
→ Successive stores became opportunities for incremental learning rather than identical rollouts.
Common mistakes
Calling an unmeasured change an experiment
Pabrai explicitly says measurement is what distinguishes an experiment from an untested change.
Changing everything at once
Multiple simultaneous changes make it difficult to learn which one affected the result.
Is it for you?
Best for
It is best for recurring stores, products, episodes, campaigns, or operating cycles that can be compared over time.
Not ideal for
It is not ideal when urgent safety or reliability problems require several immediate changes or when outcomes cannot be observed.
From the transcript
“There has to be some innovation that is going into that store that does not exist in our previous stores.”
“You have to make it measurable, right? Or else it's not a experiment.”
From the episode
Mohnish Pabrai (Billionaire Investor): The $100 Investment Hack That's Disappearing Fast! The Fastest Way To Financial Freedom!