How We Decide What to Build Next With 82 Paying Users

We build Ziva, an AI assistant for the Godot game engine. We have 82 paying users across Germany (67), Austria (10), and Switzerland (5). Here's how we decide what to build next at that scale.

The feedback loop is absurdly tight

With 82 users, you can read every support message. Every feature request. Every bug report. We don't need analytics dashboards or NPS surveys. We just read the conversations.

Three things we learned from this:

1. The feature users asked for wasn't the feature they needed.

Users kept asking for "better code generation." They wanted the AI to write more complex scripts. But when we watched what they actually did, most were pasting error messages into the chat. They didn't need better code generation. They needed auto-debugging.

So we built a feature that reads Godot's output panel and suggests fixes automatically. It became our most-used feature, and nobody had asked for it by name.

2. Pricing conversations happen in public.

With 82 users, word travels fast. When we set our paid plan at $20/month, three users posted in the Godot Discord about whether it was worth it. Two said yes, one said no. That one "no" taught us more than any pricing framework.

The objection wasn't the price. It was that the free tier didn't give enough credits to evaluate the tool properly. We increased free credits from $1/month to $3/month. Conversion improved because people could actually test the tool before paying.

3. Your roadmap is a conversation, not a document.

We don't have a product roadmap in the traditional sense. We have a list of things users told us they want, sorted by how many people asked. The top item becomes the next sprint.

This sounds chaotic. It isn't. When your market is 82 people, you can talk to all of them. The signal-to-noise ratio is perfect.

The niche advantage nobody talks about

Building for a small market has one advantage that gets overlooked: your cost of acquisition approaches zero.

We spend almost nothing on paid marketing. Our growth comes from the Godot community itself. When someone builds something cool with Ziva, they post it in the Godot subreddit or Discord. Other developers see it and try the tool. The community is the marketing channel.

This only works because the market is small enough to be a community, not an audience.

What changes at 500 users?

We don't know yet. But the feedback loop will probably break. We won't be able to read every message. We'll need to build systems to replace the intimacy.

The founders who've been through this scaling transition tell me the hardest part is admitting that what worked at 82 users won't work at 500. The tight feedback loop, the Discord conversations, the personal support emails. All of it has to evolve or die.

For now, 82 users is plenty. They tell us exactly what to build. We build it. They tell us if it's right.

ziva.sh