Product Sense and Judgment
Product sense is the competency nobody can fully define and everybody can recognize. It is why one PM looks at the same research as everyone else and proposes the thing that works, while another proposes something reasonable that lands flat. For a long time the field treated it as innate, as if you either have taste or you do not. I think that is both wrong and lazy. Product sense is trainable, it is one of the most valuable skills a PM can build, and the PMs who improve fastest are the ones who stop treating it as magic.
#What it actually is
The cleanest definition I have found comes from Lenny Rachitsky, who ranks the core PM skills in order: communication first, execution second, and product sense third. So this is the third most important skill on his list, and his definition is the one I use. Product sense is the ability to consistently come up with product solutions that meet the needs of the people you are building for. Two words in that sentence carry the weight: consistently and needs.
- Consistently is the operative word. Anyone can be right once, and that is luck. Product sense is being right more often than chance, and being able to say why before the outcome proves you so.
- Needs, not wants. Real product sense reads the deep need under the stated want, which is why it is downstream of Jobs To Be Done and Voice of the Customer, not a substitute for them.
I would add a third element Lenny emphasizes: product sense is empathy plus creativity. Empathy to genuinely inhabit the user's situation, creativity to generate solutions that fit it. Neither alone is enough.
#Taste, and the thing beyond data
There is a part of product sense that data cannot give you, and I will not pretend otherwise. That part is taste. It is the judgment that this interaction feels right and that one feels cheap, that this is the elegant version and that is the over-engineered one. Taste is what is left to decide after the data has narrowed the field but before the answer is obvious, and in my experience that gap is where most product decisions actually live. theScore's principle of balancing game sense with data names this directly: quant and qual and intuition together, never analysis paralysis, never gut alone. Data tells you what happened. Product sense tells you what to do about it.
- Gut-only. "I just know what's good." Sometimes true, usually ego, and unfalsifiable, which makes it uncoachable and dangerous at scale. If you cannot articulate why, you do not have product sense. You have an opinion.
- Data-only. Refuses to move without a number, so it only ever optimizes what already exists and never invents. Data cannot tell you what to build next when there is no data on a thing that does not exist yet. That is a judgment call, and dodging it is not rigor. It is abdication.
Real product sense lives between these, using data to inform a judgment it is still willing to own.
#How to train it
Because it is trainable, here is how I actually build it, in myself and the PMs I coach:
- Make predictions and write them down. Before a launch or experiment, predict the outcome and your reasoning, then check yourself after. This is the fastest loop, because it turns vague intuition into a calibrated, falsifiable skill. (Hypothesis-Driven Development is the formal version.)
- Do teardowns. Take a product you admire and one you don't, and articulate why down to the decision. Borrowed taste becomes your own once you can name the reasons.
- Increase your at-bats with users. Product sense is pattern recognition, and patterns come from volume. Lenny's advice is blunt: talk to more users, see more products, ship more things.
- Seek the disconfirming case. Hunt for why you might be wrong. The best-judgment PMs are the least attached to being right and the most attached to being calibrated.
#A note on AI
Product sense becomes more valuable as AI commoditizes the mechanical parts of the job. When AI can draft the spec, run the analysis, and generate ten options, the scarce skill is knowing which option is right. As Reforge puts it, the premium on judgment and taste has never been higher. For probabilistic models specifically, product sense extends into a new dimension covered in The PM's AI Literacy and The AI Product Manager: knowing what a model can reliably do, which is its own kind of taste.
#Continue Reading
- Sound Product Decision-Making for the disciplined process that surrounds the judgment call.
- Understanding Trade-offs for where product sense does its hardest work.
- Jobs To Be Done for the framework that reads the need under the want.
- Hypothesis-Driven Development for turning intuition into a calibrated, testable skill.
- Character, Competency, and Craft for why judgment sits among the hard-to-teach competencies.