The Introspection Probe (Decoupling Decisions from Outcomes)
Ask for a win that worked for the wrong reason, or a good call that failed — outcome-blind introspection tests.
- Difficulty
- Moderate
- Time to result
- ~days to results
- Steps
- 4
- Confidence
- 90%
Most interview questions collapse decision quality into outcome, so candidates simply narrate wins. This probe deliberately decouples the two by asking for cases where the result and the reasoning came apart: something that worked but not for the reason you expected, or a good decision that still failed. It is paired with a limitation question that tests whether, at their edges, the person is introspective or blame-shifting.
Origin
Assembled from two guests on Lenny Rachitsky's compilation. Iyo Omojola (Chief Product Officer at Carbon Health, former Square product lead and founder) supplies the outcome-decoupling questions; Scott Belsky (Chief Strategy Officer at Adobe, founder of Behance) supplies the self-limitation question.
Core principles
- 01Outcome does not certify reasoning — a win can be lucky and a good decision can lose.
- 02The people who improve are those who update their model from why things worked, not just that they worked.
- 03A question that presupposes a gap between reason and result blocks the standard highlight-reel answer.
- 04How someone talks about hitting their own limits predicts how they will behave when they hit them on your team.
- 05Blame and finger-pointing at the edges is the failure signal; open self-diagnosis is the pass signal.
How to run it
- 1
Ask for a win with the wrong cause
Ask: tell me something you did that worked out, but not for the reason you thought it would work. This forces them to separate the result they got from the mechanism they predicted.
Pro tip Candidates who cannot produce an example either have never examined their wins or have never had one they didn't fully control. Both are informative.
Watch out Expect a long pause. Do not rescue them from it.
- 2
Ask for a good decision that failed
Ask: tell me something you did that was a good decision that didn't work out. This is the mirror image — it tests whether they can defend process quality independent of a bad result.
Pro tip The strongest answers explain precisely why the decision was right given the information available at the time, and still own the miss.
Watch out Watch for the candidate who converts this into a disguised win ('it failed but actually taught me so much and then I succeeded'). That is evasion.
- 3
Check whether the model actually updated
The purpose of the probe is to see whether they incorporate the finding into their model so they make different decisions next time. Ask what they now do differently as a direct result.
Pro tip A concrete behaviour change ('I now always X before Y') beats a philosophical lesson every time.
- 4
Probe the self-known limitation
Ask what they have learned about themselves that reveals a limitation in how they work. This tests whether, when they hit their limits or struggle, they can be open and introspective — or whether they blame and point fingers.
Pro tip Follow with: what do you now put in place to compensate for that? Real self-knowledge comes with a coping mechanism attached.
Watch out A limitation that is secretly a strength (perfectionism, caring too much) is a non-answer and should be pushed past, not accepted.
In the wild
Omojola describes his whole interview process as teasing out introspection: is this a person who is reflective about the decisions they have made, why they worked and why they did not, and who incorporates that into their model so they make different decisions next time. The two decoupling questions are the instrument for that.
→ Rather than a highlight reel, the interviewer gets a demonstration of the candidate's actual learning loop — which is what predicts their next decision, not their last one.
Belsky asks about something people have learned about themselves that reveals a limitation in how they work, explicitly as a way to test introspection: when this person hits their limits or struggles, can they be open and introspective, or are they going to blame and point fingers?
→ The answer functions as a live rehearsal of how the candidate will behave in their first crisis on the job.
Common mistakes
Accepting a failure story that is really a success story
Candidates are coached to convert failure questions into resilience narratives. If the answer ends with the failure being redeemed, you have learned nothing about their reasoning — push for the case where it simply did not work and they still stand by the call.
Skipping the 'what changed' follow-up
Reflection without a model update is just eloquence. Without asking what they now do differently, you cannot distinguish someone who learns from someone who narrates.
Grading the outcome anyway
Interviewers instinctively reward the story with the bigger result. That defeats the entire purpose of a question designed to decouple result from reasoning.
Is it for you?
Best for
Interviewers hiring for senior judgement roles where the candidate will make consequential calls under uncertainty and must learn from them fast
Not ideal for
Entry-level candidates with too few decisions in their history, or interviews where you only need to verify technical skill
From the transcript
“you did that worked out but not for the reason that you thought it would work or tell me something you did that was a…”
“I like asking about something people have learned about themselves that reveal the limitation in how they work”
From the episode
My favorite interview questions from 100+ guests