Win — with a lesson attached.

On the member platform I kept a running list of “must-haves” before launch: richer profiles, more moderation controls, a nicer map, better emails. It was tempting to keep polishing.

I shipped the core loop instead — apply → approve → publish → appear in the directory/map — with a small first group of real members, and left the rest on the backlog.

What happened: the priorities I’d have guessed at were mostly wrong. Real usage surfaced the actual friction (a required-photo rule that silently held profiles in draft, form fields people misread) faster and more cheaply than any planning session. The “nice” features I almost built first barely mattered; the boring clarity fixes mattered a lot.

Lesson: with a small first group of users, shipping the MVP isn’t the compromise — it is the research. Real use shows you what to build next. Now I ask “what would let one real user finish today?” and let that order the backlog.