Which metrics matter most for a horror game?
Completion, session length, quits after scares, creator reach and refunds. Many horror games are short and finite, so a player who finishes and never returns is a success. Daily retention targets from mobile games do not fit them.
| Metric | Event to send | Why it matters |
|---|---|---|
| Completion rate | chapter_complete, game_complete | Shows how many players reach the ending you built |
| Session length | session_start, session_end | Horror sessions are short by design; long gaps can mean fear or boredom |
| Quits after a scare | scare_triggered, with the scare name | Separates a scare that works from one that drives players away |
| Deaths and retries per area | player_died, with the area | Finds chase or puzzle spots that turn fear into frustration |
| Refund rate | Steamworks refunds, by week | Short games end inside the two hour refund window |
| Streamer and creator reach | Tagged links and keys per creator | Horror spreads through videos of people reacting |
Sources: How To Market A Game (2 October 2023); our own metric list.
How long should a horror session be?
Short, and that is fine. Zukowski says people can only stand being scared for 30 to 60 minutes. Don't Scream had a 39 minute median playtime, GameDiscoverCo reports. So a good horror game may show short sessions and few returns.
Read session length next to completion. Short sessions with high completion mean a tight game. Short sessions with low completion mean players quit early. Look at where the last session ended to tell the two apart.
Sources: How To Market A Game (2 October 2023); GameDiscoverCo (13 November 2023).
How do I find the scares that make players quit?
Send an event for each scripted scare, then measure quits right after it. A good scare makes players jump and keep going. A bad one makes them close the game. The difference is visible in a funnel from the scare event to session end.
- Name each scripted scare and send scare_triggered with the scare name, the area and the build.
- Send session_end with the last area, and game_complete at the ending.
- Build a funnel from each scare to session_end within 2 minutes. The 2 minute window is our rule of thumb.
- Compare each scare with a calm moment, such as a save point, in the same build.
- Change one scare per build, then compare the quit rate and the completion rate again.
Sources: Manalens product documentation (funnels and event properties).
How do I measure streamer and YouTube reach?
Give each creator a tagged link and a key, and log the date of each video. Horror spreads by reaction videos. GameDiscoverCo reports that xQc got jumpscared by the Don't Scream trailer live, and that the game reached 130,000 wishlists from zero in two weeks. Zukowski reports that Choo Choo Charles got over 90,000 wishlists in 2 weeks after it went viral.
Effects can come late. Zukowski shows that Mouthwashing launched on 27 September 2024 and that a Vargskelethor video came out on 27 October 2024. The game's player peak came on 17 November, with a second peak on 26 December. A video can lift a horror game weeks or months after launch.
Many viewers search Steam instead of clicking a link, so also compare daily wishlists before and after each video.
Sources: GameDiscoverCo (13 November 2023); How To Market A Game (2 October 2023; 8 September 2026).
What refund rate should a short horror game expect?
Higher than the median, if the game ends before two hours. Valve refunds any game bought in the last 14 days with less than two hours of playtime. GameDiscoverCo's 2025 survey found a median refund rate of 9.5% for all games, and 12.6% for games with under 2 hours of average playtime. Don't Scream refunded 29.1% by units.
Watch refunds weekly, and read them next to completion. If many refunds come from players who finished the game, the problem is price or length, not quality. If they come from players who quit early, look at the first scare and the first chase.
Sources: Steam Refunds policy; GameDiscoverCo (4 February 2025; 13 November 2023).
How do I see scares, completion and refunds together?
Manalens is the analytics and growth engine built for games. Send scare, death and completion events from the Unity, Godot or web SDK, sync refunds from Steamworks (the automatic Steam sync is in early access), and read both by build and by creator link.
Sources: Manalens product documentation.
Frequently asked questions
Is low retention bad for a horror game?
Not by itself. Many horror games are short and finite, and How To Market A Game notes that people can only stand being scared for 30 to 60 minutes. A player who finishes and does not return is a success. Read completion rate next to retention before you call it a problem.
What is a good completion rate for a horror game?
We found no public benchmark for horror completion rates. Track game_complete and compare your own builds instead. If completion rises after you change a chase or a puzzle, the change worked. Steam achievement percentages for similar games can give a rough outside reference for the ending.
How do I know which streamer helped my horror game?
Give each streamer a tagged Steam link and a key, and write down the date and time of each video. Then compare daily wishlists and sales before and after each video. Many viewers search Steam directly, so the lift in daily wishlists is often larger than the clicks.
Why does my short horror game have so many refunds?
Steam refunds games bought in the last 14 days with less than two hours of playtime. A game that ends before two hours fits that rule. GameDiscoverCo found 12.6% refunds for games under 2 hours of average playtime, and Don't Scream had 29.1%. State the length clearly on the store page.
Sources and implementation
- How To Market A Game: 14 reasons why every indie game developer should make at least one horror game (Chris Zukowski, 2 October 2023, updated 26 February 2026) ↗
- How To Market A Game: You probably shouldn't launch a horror game in October (Chris Zukowski, 8 September 2026) ↗
- GameDiscoverCo: Don't Scream: jumpscaring a giant 'hook' to >100k sales in 2 weeks! (Simon Carless, 13 November 2023) ↗
- GameDiscoverCo: Steam refunds, how many should you expect? (Simon Carless, 4 February 2025) ↗
- Steam Refunds policy (Valve, read 18 November 2026) ↗