The Most Valuable Thing a Product Team Does Is Say No

The most valuable thing a product team does isn't shipping features. It's refusing them.

Every feature you build is a liability you own forever. Not a one-time cost — a permanent one. It has to be tested on every release, documented, supported when it breaks, explained to every new user, and carried in the mental weight of every future decision. You pay for a feature not when you build it, but every single day it exists afterward.

Most roadmaps don't fail because they were too empty. They fail because they said yes to everything — a pile of half-used features that each made sense in isolation and together turned the product into something nobody can describe in a sentence. Complexity doesn't arrive in one bad decision. It accumulates, one reasonable yes at a time.

Spend twenty years in credit and this looks familiar. A product is a book of exposures. Every feature is a position you've taken, with an ongoing carrying cost and a tail risk of blowing up at the worst moment. A lender who says yes to every borrower doesn't have a growth strategy — they have a concentration problem waiting to detonate. Same with a product that says yes to every request. The discipline is identical: the quality of what you own is decided by what you refuse.

And refusing is genuinely hard, because every “yes” has a face. A big customer asked for it. A competitor has it. A smart colleague is passionate about it. Saying no means disappointing someone real, now, for a benefit — a product that stays simple and coherent — that is diffuse and shows up later. The incentives all point toward yes. That's exactly why saying no is the job, not an obstacle to it.

The best product people I've seen aren't the ones with the longest shipped list. They're the ones who can explain, crisply, the things they deliberately didn't build and why. Their product does a few things extremely well instead of many things adequately. When you ask what's on the roadmap, the more revealing answer is what they took off it.

Here's the test I'd apply before any new feature: not “would this be useful?” — almost everything clears that bar, which is the whole trap. The test is “is this worth carrying forever?” Useful is cheap. Worth maintaining for the life of the product is rare.

A product, like a loan book, is defined by the deals it turns down.

If you build anything: what's the best feature you ever decided not to ship?