Player tags
What players on Steam call a game, with vote counts. "Story Rich" with 10,011 votes says more than a genre label.
Oruvai doesn't guess from a cover or a genre. It reads what players say about a game, scores the reasons behind it and matches those to yours. Every number on this page is one you can follow into a recommendation.
Every recommendation is a short chain you can follow back to what players said and to the reasons you named.
How strongly a reason shows up in a game, and how players feel about it. Plenty of games have dialogue. Few have dialogue people rave about. Pick a game and inspect it.

904,681 Steam reviews · 96% positive
Presence comes from player tags and how often reviews mention the reason. Appeal compares how reviewers who liked the game and reviewers who didn't talk about it. Both are shrunk towards neutral when there is little to go on.
The engine never reads marketing copy and no publisher can edit what it knows.
What players on Steam call a game, with vote counts. "Story Rich" with 10,011 votes says more than a genre label.
A balanced sample of positive and negative reviews per game. Plain counting shows which reasons the people who loved it mention, and which the people who didn't do.
Release, platforms and artwork come from IGDB. They decide what can be recommended, not why it fits.
A game with few reviews or no tags is marked as a preview, never dressed up as a sure thing. Confidence grows with every player who weighs in.
Pick what you loved and why. For every reason you keep, a game earns points for how present it is, weighted by how much players like it. What you skip counts against it.
Playing a tour. Touch anything to take over.
The math for Mass Effect LE
0.9 × 0.92 × (0.5 + 0.5 × +0.46)0.61Choices that matter0.9 × 0.87 × (0.5 + 0.5 × +0.64)0.64Skip shooters- 0.6 × 0.61-0.37Quality nudge0.15 × (0.91 - 0.75)+0.02Score0.91Fits because of choices with real consequences and standout dialogue. Players praise the choices.
Every sentence in a pick is assembled from the two numbers above. No language model chooses or writes your recommendations.
New games with few reviews and games without player tags are marked as a preview, with the catch stated next to the pick.
Say you can't stand hectic shooters and they drop down the list, however popular they are.
Publishers and stores can't buy a better rank. The full rules are in our recommendation transparency.
The engine on this page and on the homepage runs on 24 games. We are widening it step by step, and we say so instead of pretending it already covers everything.
Real player tags and review language, scored and explained. Free for everyone.
270K+ games are listed today; about half link to Steam player data the engine can read. Until a game has been read, Oruvai recommends it with the existing system.
Dozens of reasons instead of six, variety in the top results, and learning from every "not for me" you tap.