Balatro
LocalThunk / Playstack

An anonymous developer, a poker deck-builder, and a demo strategy executed to perfection over eight months. 208,000 wishlists on launch day, 1 million sales in under a month.
The demo as a wishlist machine
Balatro is the most recent and most repeatable of the cases. No pre-existing community, no known studio, no significant marketing budget. LocalThunk, anonymous, first game. Playstack publishing but with no Sting on the cast. What worked is entirely documented by LocalThunk himself on his blog: a playable demo, put in the hands of the right creators at the right moment, for eight months before release.
A first demo came out in June 2023, with no campaign, and had almost no impact. It's the next one that changed everything: Dan Gheesling, an established roguelite creator, pitched the game to Northernlion. Northernlion played the demo fifty rounds live, with LocalThunk watching in the chat. At the end of the stream, LocalThunk sent him a full beta key. Direct, measured consequence: the game's wishlists rose to 28,661 by late July 2023, from a base near zero.
From there, the mechanism runs by itself. Each Northernlion video ends with the same natural call-to-action: the demo is free, the game is on wishlist. Other deck-builder creators followed — Retromation, Dan Gheesling himself, specialised channels. The demo acts as a converter: you can't play it without wishlisting, because nobody plays a demo they've already forgotten. The day before launch, 208,401 wishlists.
At launch on 20 February 2024: 50,000 sales in two hours, 119,000 in the day. The accumulated velocity triggered Discovery Queue, Popular Upcoming, global Top Sellers. The algorithm didn't discover Balatro — it cashed in a signal methodically built over eight months. Steam did its job: push what moves. What was moving had been built beforehand.