We're burning thousands on AI tokens every month and getting nothing back. That's the plan.

We spent $20K last month on AI tokens for users who will never convert.

It's the highest-ROI line item on our P&L.

We call them AI tourists. People who sign up out of curiosity, not necessity. They explore the product, burn our infrastructure, and most of them will never pull out a credit card. So why do we let them?

Because the people least likely to pay are the most likely to talk.

That's not optimism. That's the mechanism most growth teams have completely backwards.

Elena Verna, head of growth at Lovable, framed this exactly right: freemium token cost should be a marketing line item. Not COGS. Her logic: "why would we prevent a person who wants to do all the marketing for us from using us?"

Every token they burn is cheaper than any ad you'll ever run.

If you want to go deeper on this, Elena breaks down the full thinking behind this strategy in this conversation, it's worth every minute.

But most companies don't see it that way. They see non-converting users as a cost problem. They add friction. They cap free usage. They optimize the funnel to filter out people who won't pay.

And in doing so, they filter out their own distribution.

Here's the mechanism. The social currency principle. People don't share things because they're generous. They share things because sharing makes them look good. When someone discovers an AI agent that automates something that used to take hours, they can't stay quiet about it. Sharing it is how they tell the world: look how smart I am. Look how ahead of the curve I am.

That's free distribution. Earned entirely through product.

But it only works under one condition.

Your product has to actually impress them.

If your product is average, tourists try it and forget it. You burned tokens for nothing. If your product genuinely impresses them, they can't not share it. Because the sharing was never about you. It was about them.

Two types of products exist in this world. Products people use. And products people talk about. The gap between them isn't features. It's the moment of genuine surprise.

The $20K wasn't a cost problem. It was a distribution bet on our own product.

We made it because we believe we're building the second kind.

Find your tourist moment

• Ask yourself: if someone with zero context tried your product today, what's the single thing most likely to make them tell someone tomorrow? If the answer comes fast, you're building the right thing. If it doesn't, that's the work.

• Pull your last 30 days of non-converting users. Don't ask why they didn't pay. Ask whether they actually reached the moment you're building toward, the one you just identified. Did they get there?

• Before you add any paywall or usage cap to your free tier, ask two questions: can you enable users to experience that moment before hitting a wall? And how are you actively activating that moment inside your trial?

The best distribution strategy you'll ever build is a product people can't stop talking about.

–Amos

Amos Bar-Joseph

Founder & CEO of Swan AI. Building the first autonomous business scaling to $10M ARR per employee with AI agents. Writing about the future of GTM and leadership.