What we ran
Community engineering first. Not member acquisition — engineering. A channel with a reason to come back that was not price, moderation that held up under the scam pressure any token attracts, and enough informed regulars that a newcomer's question got answered by someone who did not work for the project.
Creator seeding, sized to the stage. Mid-tier creators with genuine audience overlap rather than the largest accounts available. Early on, the premium for tier-one reach is steep and the conversion rarely justifies it.
Press attached to real moments. Coverage timed to things that had actually happened, which at that stage is a short list — and that is fine. Two placements where the buyers read beat forty syndicated reposts, every time.
The measure we watched
The number we watched was not member count. It was whether a member could explain the project to someone else without a script.
That sounds soft. It is the hardest thing to fake and the most reliable predictor of whether a community survives its first difficult month.
The outcome
The programme raised $780K — from a starting position of effectively zero market visibility.
