Should we abandon data and just go with our guts? Obviously not. The honest position is a middle ground, and the rules for finding it are not complicated.
It is okay to share observations without a bibliography. Not every conversation needs to be a thesis defense. Anecdotes are not the plural of data, but they are often the starting point for the questions that matter most. The hypothesis came from somewhere, usually from someone noticing something.[1]
Acknowledge your limitations. "No stats whatsoever" is a form of intellectual honesty, not a failure of rigor. You are not pretending to evidence you do not have, and that small act of transparency raises the trust level of the entire exchange. Stay open to exceptions. Whatever you believe, hold it loosely enough to accommodate the inevitable edge cases.[2]
Then know when data actually matters. Medical decisions, policy choices, hiring at scale, anything where the cost of being wrong is asymmetric — those deserve rigorous evidence. Casual conversations about life, fast product calls, design judgment in domains you know well? The rigor tax is rarely worth paying. When data does matter, the better question is how to measure data-drivenness at all, rather than performing it.
I've argued elsewhere that the demand for receipts on every observation has become reflexive and that intuition is compressed experience worth taking seriously. The balance is not a clever framework. It is mostly a willingness to match the level of proof to the actual stakes — and to admit, out loud, when you are operating without much of either.
Related Essays
The Tyranny of Show Me the Data
Every observation now demands a citation. Somewhere along the way we lost the ability to simply notice things, and the conversation got worse for it.
The Case for Intuition
Pattern recognition, experience, and inherited wisdom are real forms of knowledge. They are not a substitute for data, but they are not noise either.
Always Exceptions, and That Is Fine
Edge cases do not invalidate general observations. They enrich them. The "well actually" reflex misunderstands what most claims are even trying to do.
Key takeaways
- Match the rigor to the stakes. Medical and policy decisions deserve evidence; casual observations do not.
- Anecdotes are not the plural of data, but they are often the start of the right question.
- Intellectual honesty about what you do and do not have is the cheapest form of credibility.
FAQ
When does data clearly win?
High-stakes, repeatable, slow-feedback decisions. Medicine, safety, policy, hiring at scale. Anywhere the cost of being wrong is asymmetrically large.
When does intuition clearly win?
Fast-feedback, low-cost, reversible decisions in domains where you have real reps. Most product, design, and conversational calls live here.