The games people actually finish

Ask a maker how long their game is and you get a number in hours. Ask a player and you get a shrug, because most people did not get to the end and have quietly decided that is their fault.

It is almost never their fault. Finishing is something a game either affords or does not, and the games on this site with the highest completion rates have very little in common except how they are shaped.

Finishing is a shape, not a length

The games people finish are the ones where the end is visible from the beginning. Twelve levels. Five rooms. One board. A boss you can see on the map from level one.

This does not mean short. A twenty-hour idle game with a stated ending gets finished more often than a ninety-minute platformer that does not tell you how many levels there are. Visible structure is doing the work: the player is not deciding whether to continue, they are watching a bar fill.

The corollary is brutal for procedural games: infinite content removes the reason to keep going as reliably as it removes the reason to stop. If your game generates levels forever, invent an ending anyway. Ten waves. A distance. A boss at floor twenty. Something that turns playing into completing.

The middle is where you lose them

Nearly every game that people abandon, they abandon between a quarter and a half of the way through. Not the first five minutes (those either work or you never had them). The middle.

The reason is nearly always the same: the middle is where the maker stopped designing and started producing. The first levels are hand-tuned, the last levels are the showpiece, and the middle is nine variations of level three built on a Sunday. Players cannot articulate this but they feel it exactly, and what they feel is that the game stopped being about anything.

The fix is unglamorous. Take your middle third and cut half of it. Not the difficulty. The quantity. A game that goes from the tutorial to the good stuff in ten minutes is better than the same game with twenty minutes of competent filler in the way, and it is a much smaller game to finish building.

Save systems break more runs than difficulty does

For browser games specifically, the single most common reason a person does not finish is that they closed the tab and their progress was gone.

Local storage is fine. Local storage that only writes at the end of a level, or only when you click a save button, or that silently fails in a private window, is where the losses happen. The rules that seem to work:

• Save after every meaningful action, not every level. Writes are cheap.
• Save the state you would need to reconstruct the current moment, not the current milestone.
• Show that you saved. One line of text in a corner. People do not trust a browser game with their progress, and they are right not to, so tell them.
• Handle the private-window case. If storage throws, say so on the title screen instead of losing an hour of somebody's evening in silence.

A game that survives a closed tab gets played across three sessions. A game that does not gets played once, for however long that sitting lasted, and then it is over whether or not the player enjoyed it.

Difficulty is fine. Punishment is not.

Hard games get finished. Games that make failure expensive do not.

The distinction is what happens after you die. If you are back in the action in half a second on the same input that killed you, difficulty reads as a challenge. If you sit through an animation, a results screen and a menu, the same difficulty reads as an insult, and the third time it happens somebody closes the tab.

Roguelites work in a browser for exactly this reason: death is the loop, not an interruption to it. Games that borrowed the difficulty of a roguelite without borrowing its attitude to death are the ones that go quiet after launch day.

Endings people talk about

The last thing worth saying is that an ending is a marketing asset and almost nobody treats it as one.

The games that get shared after launch week are disproportionately the ones with an ending somebody wanted to describe. A final level that recontextualises the mechanic. A results screen with a number worth screenshotting. A last line that lands. Twenty seconds of design, and it is the twenty seconds people quote.

If you have cut something to ship, do not cut the ending. Cut the middle.