Failure as Discovery Loop
Turn each failed attempt into evidence for the next one
- Difficulty
- Easy
- Time to result
- ~days to results
- Steps
- 3
- Confidence
- 95%
The Failure as Discovery Loop reframes an unsuccessful outcome as information rather than identity. First, distinguish the project failing from you being a failure. Then ask what the result revealed about the approach, product, or role. Convert that finding into a specific change and apply it to the next attempt. The mechanism is outcome separation, followed by evidence extraction and another informed trial. Finkelstein applies this lens both to Shopify's large volume of product releases and to his own move from chief operating officer to a role better matched to storytelling. The model does not claim that failure is costless; it makes the learning explicit so the cost can improve a subsequent decision.
Origin
Finkelstein credits Shopify founder Tobi Lütke with the line that failure is the discovery of something that did not work.
Core principles
- 01Separate a failed project from your identity
- 02Treat an unsuccessful result as evidence about what did not work
- 03Carry the learning into the next attempt
- 04Expect a portfolio of attempts rather than universal success
How to run it
- 1
Separate self from outcome
Describe the project, decision, or role that failed without labeling yourself a failure. This creates enough distance to examine the result.
Pro tip Use language such as 'the attempt failed' rather than 'I am a failure.'
- 2
Extract the discovery
Identify the assumption, method, or fit that the result showed did not work. Make the finding concrete enough to affect a future decision.
Pro tip Ask what you know now that you did not know before the attempt.
Watch out A vague lesson such as 'try harder' does not change the next experiment.
- 3
Change the next attempt
Translate the discovery into a different product choice, role, or execution method. Test that change rather than repeating the same approach.
In the wild
Finkelstein concluded that chief operating officer was not the role in which he could be best in the world. After discussing the mismatch, he moved into Shopify's more externally focused president role and concentrated on storytelling, which he says made his work more meaningful and valuable to the company.
→ The role changed to fit his strongest contribution instead of preserving status or headcount.
Common mistakes
Turning a result into an identity
Calling yourself a failure makes it harder to inspect the specific project or decision that did not work.
Repeating without extracting evidence
Another attempt adds little value if nothing learned from the previous result changes it.
Is it for you?
Best for
Builders making repeated product, role, or business experiments where learning can improve the next attempt.
Not ideal for
High-stakes failures that require immediate safety, legal, or financial remediation before another experiment.
From the transcript
“separating like you being a failure versus the project or the particular thing being a failure is very very different”
“failure is the discovery of something that didn't work”
From the episode
Shopify President: How To Become A Millionaire For The Price Of A Starbucks Coffee!