What we learned finding 301 allies for a launch

For one of our launches we set out to find 50 people who already cared about our cause. We found 301 in two days. Here is how, and what we’d do differently.

For QBTC, a Bitcoin project built by our team, the first goal wasn’t followers or sign-ups. It was a list: 50 people already defending the same cause, rated by how well they fit, by Thursday. By the deadline the list had 301.

Start from the cause, not the product

We stopped describing the product and wrote the cause in one line: “we love Bitcoin and want it to survive quantum risk.” That changed who we looked for. Instead of potential customers, we searched for people saying the same thing in public, whatever their reason.

Classify everyone, keep the allies

We pulled people from public posts and from contact lists we already had, and classified each by stance: ally, neutral, skeptic, competitor or unrelated. By 5 November, 1,570 accounts were classified and 301 rated as allies. Neutral isn’t useless, but allies are who you talk to first.

Rate fit with reasons

A score alone is hard to trust. Every rating came with a one-line reason and the post it was based on. That made it fast to scan the top of the list and spot mistakes.

Watch their posts daily

A list is only useful if you act on it. We check the top allies’ new posts every day and draft an affirming reply that adds something specific to what they said. A person reads it and decides whether to post it.

What went wrong

Our first drafts pitched. They opened with “I’m a dev at…”. People don’t want a project introduced in their replies. We changed the voice so drafts never name the project or claim affiliation, and that rule now applies to everything.
Duplicates. The same person appeared under several names. We merged 18 duplicate rows before the list was usable.
The monitor checked nobody for a night. A parsing bug read every post’s author as “i”. A daily check of “how many allies did we actually look at?” would have caught it sooner.

What we’d do again

Write the cause in one line before searching.
Classify broadly, then talk only to the top of the list.
Keep a human on every reply and DM.
Measure conversations started, not likes.