Notes RSS feed

Shipping a feature is not the finish line; it is the start of an ongoing bill. Every toggle and flow adds maintenance, tests, docs, support load, and cognitive weight for new users. A product with a hundred surface features is rarely twice as valuable as one with fifty—it is often harder to learn, harder to change, and easier to break. Ruthless PMs count the carrying cost before they argue for another checkbox. Growth that ignores liability compounds into fragility. The best products feel simple because someone said no a thousand times.

Alignment means people understand what was decided, why it was decided, and what they are expected to do next. Consensus means everyone likes the answer, which is a different and usually impossible bar. You need shared clarity; you almost never need unanimous enthusiasm. Committees that chase agreement optimize for comfort and delay the hard tradeoff. Document the decision, the reasoning, and the dissent if there is any—then move. Product velocity dies in rooms that mistake silence for alignment and politeness for strategy.

Support queues are raw product research: unsolicited, unfiltered, and tied to real failure modes. A user who bothers to write in is telling you where the product broke the promise your marketing made. Industry reports summarize trends; tickets show you the edge cases that become churn. A PM who reads twenty tickets a week builds pattern recognition that no dashboard replicates. Skip the queue and you optimize for the persona in the deck, not the person stuck on step three. Make the inbox part of your ritual.

“We already built half of it” is an emotional argument dressed up as strategy. The hours and dollars already spent are gone; they do not increase the return on the next hour. The only rational question is whether finishing this feature is the best use of what you still have—time, focus, and opportunity cost. Most teams cannot kill work mid-flight because they confuse past investment with future obligation. Kill the feature if the bet no longer clears the bar. The sunk cost is not a reason to continue; it is a lesson you already paid for.

Analytics excel at describing behavior at scale. They are nearly useless at explaining motive. A retention chart that moves eight points is a signal that something changed, not a story about what broke or delighted users. Diagnosis still lives in conversation: interviews, support, session replay with context, the messy qualitative layer dashboards cannot replace. Teams that optimize only on metrics get very good at moving numbers and very bad at understanding people. Pair the chart with a human sentence, or you are treating symptoms while the underlying problem mutates.