Most people never finish the games they buy. Here are the numbers.

Steam publishes one brutally honest statistic and almost nobody reads it: for every achievement in every game, the share of owners who have unlocked it. Find the achievement that marks the end of a game and you know how many buyers ever got there.

This site's pipeline does that detection across the catalog. Right now it holds a finish signal for 8,640 games, read from Valve's global achievement data. The numbers are rougher than a lab study and they are the best public evidence there is. Here is what they say.

Finishing is the exception

Across the 8,640 measured games, seven in ten show a finish rate under 10 percent. Push the line to 20 percent and you have covered more than four out of five games. The typical Steam game is finished by roughly one buyer in twenty.

Not obscure games, either. ELDEN RING, with over a million reviews, shows about a 10 percent finish rate. Cyberpunk 2077 sits near 6. HELLDIVERS 2 lands around 11. Black Myth: Wukong, a straight story-driven action game, manages about 13, and that is one of the better numbers among the giants.

Some famous games barely register, and those cases are honest ones: Terraria shows 0.4 percent, Stardew Valley 1.8, PUBG under 1. Those are not failures. They are games without a finish line, and lumping them into finish-rate talk is how this statistic usually gets abused. The score here treats those genres on population instead, and this guide's numbers should be read the same way: finish rate is a story-game statistic.

Genre sets the baseline

Average finish rates by cohort, from the same data:

• Short indie games: about 23 percent
• Narrative games: about 19 percent
• Strategy: about 11 percent
• RPGs: about 9 percent
• Shooters: about 9 percent
• Multiplayer and roguelike games: about 5 percent, and the number means little there

The pattern is length in disguise. Players finish what respects their time. A four-hour narrative game gets finished at four to five times the rate of an eighty-hour RPG. When someone tells you a game "has 80 hours of content," the data's answer is that almost nobody meets that content.

Good reviews do not mean people finish

Here is the one that surprised me. Group the measured games by review score and the finish rate barely moves: games rated in the 70s finish at about 13 percent, and games rated in the 90s finish at about 13 percent. Loving a game and completing it are close to independent.

People buy a game, respect it, play a weekend, drift off, and still leave a glowing review for the part they saw. The review score measures the quality of the experience. It says nothing about how much of the experience anyone consumes.

What this does to the price math

This is why dollars-per-advertised-hour flatters long games. If a game advertises 40 hours and 6 percent of players finish it, the honest denominator for most buyers is not 40. You are paying for shelf, not for play.

The BANG Score uses this signal directly: the finish rate, where one exists, reserves the very top of the scale for games people actually see through, and a long game nobody finishes cannot ride its hour count to a high score. Every game page with achievement data draws the full drop-off curve, so you can see the exact point where players stop.

The method, plainly: the finish signal is the global unlock rate of the achievement that best marks a game's ending, from Valve's public data. It is noisy for games with buggy or missable end achievements, absent for games without achievements, and meaningless for endless genres, which the scoring treats separately. Figures measured August 4, 2026 from this site's own pipeline; they move a little as the catalog grows. The shape does not.