Why a 400-player game should out-rank a 300k incumbent
Quality-first ranking weighs retention, session depth, and word-of-mouth over raw concurrency — so a loved 400-player game can out-rank a 300k incumbent. Here are the signals we weigh, and why concurrency is the weakest of them.
Quality-first ranking means ordering games by how much players value them — retention, session depth, and word-of-mouth — rather than by how many are playing right now. Under that lens a loved 400-player game can, and should, out-rank a 300,000-player incumbent that churns its audience. Concurrency measures reach, not quality, and it is the weakest signal we track.
The problem with ranking by size
Most discovery surfaces sort by concurrent players. That creates a feedback loop: big games get shown because they are big, which keeps them big, which keeps them shown. New and niche games — often the best-crafted ones — never break in. Players see the same handful of titles and conclude there is nothing else worth playing. There is; it is just buried.
The signals we weigh instead
- →Retention — do players come back the next day and the next week? A game with 71% D1 retention at 400 players is healthier than one with flat retention at 300k.
- →Session depth — how long and how engaged is a typical session, versus a bounce.
- →Word-of-mouth and organic growth — is the game spreading because people recommend it?
- →Recommendation lift — when we surface it next to similar games, do players actually take it up?
- →Concurrency — included, but weighted low. It is a tiebreaker, not the ranking.
What this unlocks
Signal-over-size ranking surfaces hidden gems while they are still small, gives original creators a fair shot against incumbents, and gives players a reason to open discovery again — because it shows them something new that is actually good. It is the whole thesis of RIIP: discovery quality is the product.
Concurrency tells you a game is popular. Retention tells you it's good. We rank on good.
This only works if it can't be gamed. That is why quality-first ranking is paired with anti-clone dampening — so a copycat can't ride a great game's signals — and why the model always falls back to transparent heuristics you can inspect.