Stewart Butterfield: Why thinking is what kills your product

Features aren’t binary; they sit on an S-shaped curve where early effort yields little value, then suddenly a lot, then diminishing returns. Teams should ask whether a feature is still in the ‘junk’ zone (under-invested) or at the plateau (done enough) before killing or doubling down on it.

Most product problems are not about too many clicks but about users not understanding what’s happening or what to do next. Design flows so people can act almost without thinking—clear hierarchy, obvious next steps, and avoiding decisions that make users feel stupid.

Taste can be learned through practice and critique; few teams invest in it. Slack leaned into micro-details (e.g., magic links, the “shouty rooster” warning, smart Do Not Disturb rollout) that signaled care and built an emotional bond, driving word-of-mouth adoption.

Adding thoughtful speed bumps—like the @everyone ‘shouty rooster’ warning—prevents abuse, educates users, and nudges healthy norms. The goal is not zero friction; it’s the right friction in the right places to make usage sane and self-explanatory.

As companies grow, people run analyses, A/B tests, meetings, and decks whose upside is tiny compared to the cost (e.g., weeks of work to maybe extend thread length by 0.03 messages). Leaders must ensure a steady supply of clearly valuable work and aggressively cull busywork.