I expected "play the short ones first" to work. Data from 129 real game backlogs says short games get finished at a third of the rate of mid-length ones.

Every guide to clearing a game backlog says the same thing: start with the short games, get quick wins, build momentum. I built a game backlog tracker, so I could check whether people actually do that.

I joined each game with its HowLongToBeat main-story time and looked at what share ends up in "Done". The data is 129 accounts and 25,276 games, taken on 2026-11-10. Only games with a known length are counted.

Finish rate by game length.

Main story

Games

Marked Done

Finish rate

up to 5 hours

11,667

706

6.1%

5 to 15 hours

5,593

1,072

19.2%

15 to 30 hours

3,012

607

20.2%

over 30 hours

2,445

385

15.7%

The shortest games are the largest group by far (more than half of games with a known length), and they are the ones least likely to be finished. Mid-length games, 5 to 30 hours, are finished about three times as often.

Why I think this happens (hypotheses, not findings).

  1. Bundles and giveaways. Short games are cheap, free, or arrive in bundles, so they land in the backlog without anyone deciding to play them.
  2. No ceremony. A 3-hour game gets played in one evening, and nobody comes back to update the status.
  3. Weak pull. A game you paid little for has little to make you start it.

I can't separate these from the data. What the data does show is that "short" is not enough of a reason to pick a game.

Finish rate by genre (genres with at least 800 games)

Genre

Games

Finish rate

Platformer

988

19.2%

Shooter

2,622

17.8%

Horror

3,511

15.4%

Action

12,005

15.0%

Adventure

7,775

12.9%

Strategy

4,297

7.9%

Simulation

3,428

8.0%

Casual

3,159

7.6%

Indie

9,358

8.2%

Games with a clear end (platformers, shooters, horror) get finished more often than open-ended ones (strategy, simulation). That fits: if a game never really ends, "Done" is a decision, not an event.

What I'd change in a backlog tracker because of this

  • Don't present length as the main way to pick. Length plus genre plus what you finished before is a better signal, which is also why recommendations here use your own finished and rated games.
  • Treat "Done" for open-ended games as optional. A second status for "played enough" would be more honest.
  • Make marking a short game as Done a single click at the moment you stop playing it.

Caveats: Small, self-selected sample. Many people imported whole Steam libraries, so "not done" includes games they never meant to play. Genres overlap (one game counts in several rows). The finish rate is a measure of what people recorded, not what they played.

If you track your game backlog somewhere, I'd be curious whether your short games get finished.