Indie Horror Game Mistakes: What 9,566 Steam Reviews Punish

The most common complaint about indie psychological horror games on Steam is that they are too short, and it is the complaint that does the least damage. The theme The game is very short was raised by 406 reviewers across 38 of 47 games, and 79% of those complaints sit inside positive reviews.

The complaints that arrive with a thumbs down are smaller and nastier. The game is not worth its price comes from 71 reviewers, and only 14% of them recommend the game. For The game is a walking simulator with minimal interaction or gameplay it is 68 reviewers and 31%. The game crashes frequently (72 reviewers, 36%) has the highest severity of the 25 ranked complaint themes, 4.0 out of 5.

The sample behind every number below: 9,566 reviews of the 47 indie psychological horror games released between 2026-03-20 and 2026-09-20 that reached 100 or more reviews, plus post-mortems of 10 older flawed hits (500+ reviews, 40-79% positive, released 2013 to 2026) built from the 142 most helpful reviews of each. Counts are reviewers, not players.

Two conventions. Text in italics is a theme name from the analysis, a label for a cluster of similar statements. A quotation with a game’s name next to it is a reviewer’s own words, copied character for character. 54% of the niche’s 1.9M stored reviews are not in English (Russian and Simplified Chinese lead), and this post quotes only the English ones.

Jokes, one-liners and copy-paste duplicates were dropped first. Then, for each game, the most helpful negative reviews, the positive reviews with a but in them, and a sample of plain positive reviews were read by a language model that sorted complaints, praise and wishes into a fixed category list. Similar statements were clustered into themes.

The rank is a score that blends how many of the 47 games show the theme, how many reviewers raise it, and how severe it is. That is why the rows are not in order of reviewer count.

Rows with a short category name (Story and writing, Pacing) collect the statements in that category that did not group into one repeated sentence.

The column I would read first is the fifth one. A complaint inside a positive review comes from a customer who still recommends you. The same complaint inside a negative review is usually why they did not.

On bugs, the checklist built from this data says they are what most clearly separates the two ends of the niche: 14.2% of complaints in games under 80% positive against 3.6% in games at 90% or more. It adds that one large game weighs on that gap, and that bugs also show in 6 of the 10 post-mortems. From Dreamcore: Rabbit Hole (not recommended, 0.5 hours played): “the game is a buggy mess and crashed several times”.

Length complaints come mostly from happy players, and they are more common in well-rated games. A reviewer who recommends Fears to Fathom: Scratch Creek (2.3 hours played) wrote “I wished it was longer”. That is a compliment with a request attached.

Price changes the tone. The share of reviewers complaining about length or price climbs with each price band:

Five of the ten flawed hits were marked too short for the price, and all five cost $13-25 for 1.5-5 hours. The Park was 1-2 hours for $13. Lust for Darkness was 2-3 hours for $15, and one of its verified review quotes is three words long: “Is that it?”

The checklist turns this into a working guide in its item on price and honest playtime: about 1.5-3 hours at $8 or less, and above $12 only with 5 or more hours of non-padded play. The median price of the 917 games on the niche watchlist is $6.99. That is the checklist’s guide from this sample, not a law of Steam pricing. Its measurement matters more than its numbers: take the median completion time from blind playtests, then subtract any padding you found. The Steam pricing strategy guide covers comparables and regional tiers.

Price is a store page decision, and so is the rest of the promise. The checklist’s line on this: players forgive a small game, not a misdescribed one. Broken expectations showed up in 6 of the 10 flawed hits. If your title says simulator or your tags say exploration, the page is making a claim the build has to keep.

The game is a walking simulator with minimal interaction or gameplay sits outside the top ten, with 68 reviewers across 15 games. It gets a section because of who writes it. 72% of the reviewers who raised it had under two hours played, inside Steam’s refund window, and 0% had more than ten. Only 31% of them recommended the game.

Across the whole gameplay-agency category, the checklist counts only 40% of complaints coming from players who still recommend the game. What reviewers describe is an absence:

• •“there is absolutely zero reason to do this as there’s nothing to do in the apartment” (Interdimensional Vending Machine, not recommended, 6.4 hours played)
• •“it was just straight to the water and then straight back. No exploring. Nothing” (Fears to Fathom: Scratch Creek, not recommended, 3.2 hours played)

The post-mortems found no agency in 6 of the 10 flawed hits, and the verified quotes read like the recent ones. Amnesia: A Machine for Pigs: “Each level is a single, straight, path with no choices on how to proceed”. Autopsy Simulator: “You click when and where you are told to click”.

Padding is the neighbouring failure, judged present in 8 of the 10 flawed hits: forced slow walking, fetch loops, running disabled during narration. In the recent sample, pacing complaints come from 8.3% of reviewers across 43 games. The checklist’s test is cheap. For each 10-minute block of a playthrough, write down what the player decided or learned. Empty blocks are padding.

Jumpscares are not the problem. Jumpscares are effective and well-executed is a praise theme raised by 310 reviewers across 24 games, and The game is quite scary and frightening by 295 reviewers across 31 games, 99% of them recommending. A reviewer of Tung Tung Tung Survival (recommended, 0.6 hours played): “when I heard that unsettling sound and turned to my right, my heart left my body”.

The complaint is repetition. Jumpscares are overused and become predictable comes from 174 reviewers across 23 games, and 67% of them had under two hours played. From Fears to Fathom: Scratch Creek (not recommended, 2.1 hours played): “the jump scares are pretty cheap and predictable”.

Scare quality is a first-session complaint. As a category it makes up about 10% of complaints from reviewers with under two hours played, against about 1% from reviewers with more than ten.

The post-mortems judged horror that does not scare to be a failure in 7 of the 10 flawed hits, and they name three ways a scare dies:

1. 01The threat is telegraphed. From Amnesia: A Machine for Pigs: “The lamp also begins to flicker everytime a monster is near”. A warning that always fires first removes the fear.
2. 02The same scare repeats. Fear Therapy ran jumpscares every minute or two. One of its reviewers: “I don’t think that putting a jump scare moments on every minute increases the fear element, on the contrary, it decreases it”.
3. 03The scare is volume. Painfully loud audio is its own complaint in the Fear Therapy post-mortem.

The checklist’s checks: count jumpscares per hour and list any that repeat, compare the peak loudness of scare sounds with normal play on headphones, and note the timestamp of the first real scare in a blind playtest. It wants something frightening inside the first 30-60 minutes.

This pattern comes mostly from the post-mortems, which say the recent data only hinted at it. Trial-and-error deaths with distant checkpoints were judged a failure in 4 of the 10 flawed hits (Scratch Creek, Layers of Fear 2, Kholat, Remothered), usually in chase sequences.

The Scratch Creek post-mortem describes a final chase with invisible walls, darkness, one valid route and no checkpoints. A reviewer: “I got stuck there for like 40 min”. Layers of Fear 2: “the sheer amount of deaths you go through from cheap chases”. In Kholat, one-hit deaths undid 15 minutes of walking.

What the checklist asks for:

• •Checkpoints right before lethal sections, so a retry costs a minute or two, not fifteen.
• •Chases with more than one working route, readable from light or sound, and no invisible walls.
• •A blind playtest of every chase, with median attempts in low single digits.
• •Skippable dialogue on a retry (a cutscene skip was requested across 12 games).

Difficulty is the complaint of people who stayed. As a category it makes up 12% of complaints from reviewers with more than ten hours against 5% from those with under two.

Story is the axis this niche is judged on. It is the most praised category in the sample (2,831 reviewers, all 47 games) and the most complained about (16.8% of reviewers).

On the complaint side, The game lacks a coherent story or narrative comes from 123 reviewers across 28 games, with only 34% recommending and 51% under two hours played:

• •“story is extremely boring and leads to no where... felt so disconnected” (Fears to Fathom: Scratch Creek, not recommended, 2.5 hours played)
• •“we don’t get a deeper understanding of the events of the game beyond the mission itself” (Static Dread: The Submarine, not recommended, 2.9 hours played)
• •“it leads nowhere” (Interdimensional Vending Machine, recommended, 22.8 hours played)

The third reviewer played 22.8 hours, recommended the game, and still wrote that.

Story was judged a failure in 9 of the 10 flawed hits, every one except Fear Therapy. Layers of Fear 2 had a story nobody could follow without a video explainer. Amnesia: A Machine for Pigs buried its plot in metaphor. Observation ended by explaining little, and one reviewer called it “far shallower than it thinks it is”.

Two checklist items do most of the work here. The main plot has to be understandable without collectibles, so a player who reads no notes still knows what happened. And the ending has to pay off: ending complaints come from 3.8% of reviewers across 39 games. Both have the same test. After the credits, ask a playtester what happened and why. Ambiguity is fine when it is a choice. The checklist notes that “pretentious” and “explains nothing” are the words players use when it is not earned.

The checklist names six combinations of failures. Each matches at least one documented flawed hit, and its advice is to fix a match before anything else. The free readiness check tests for all six, and the flawed hits study has the post-mortems.

“Refund bait” describes how reviewers read the game, whatever the developer intended. Scratch Creek reviewers wrote “Over half of the game is packing boxes” and “Game pads itself over two hours”, and the post-mortem notes they read the chores and driving as padding past the refund window. This dataset has no refund numbers. For how often Steam purchases are refunded in general, see the refund rates data post.

The second row is the one I find hardest to read, because those hooks demonstrably sold. Observation was bought for its premise (you are the AI of a stricken space station) and lost reviewers on movement: “the flying controls are absolutely atrocious”. Remothered is stalker horror whose audio does not tell you where the stalker is: “whether he’s right next to you or upstairs”. Autopsy Simulator impressed even pathologists with its anatomy and still got this: “contains neither simulation and barely any autopsies”.

Wishes are mostly written inside positive reviews, which makes them the closest thing to a backlog from paying fans. The top five wish themes:

The checklist calls more of it the top wish by far. “Only wish there was more” (Dread Fields, recommended, 1.4 hours played) is the whole request in five words.

The rest are specific. Players want more to do: “More meaningful environmental interactions would help too” (Nightmare Frontier, recommended, 3.0 hours played). They want to know where to go: “adding more environmental clues of which path you must take to progress” (Fears to Fathom: Scratch Creek, not recommended, 2.9 hours played).

They want saves they can trust: “please add more save spots or a manual save” (Witchblood, not recommended, 2.7 hours played). They want comfort toggles: a depth-of-field toggle was requested across 13 games. In Layers of Fear 2, a black-and-white filter over most of the game produced this: “80% of the time game is black and white old movie which gave me headache”.

Treat co-op requests with care. 104 of the 116 in the sample came from two titles, and a Scratch Creek reviewer names the cost: “multiplayer element kind of ruins the immersion”.

• •It cannot tell you how to get noticed. Every game in both sources found an audience. The checklist says nothing in it explains why the other roughly 90% of releases in this niche never reach 100 reviews, and calls that a visibility and hook problem it does not cover.
• •Counts are reviewers, not players. Reviewers skew engaged and opinionated. Trust the ranks more than the absolute percentages.
• •It is a small launch-window sample. 47 recent games plus 10 older ones, in one niche. Launch-window reviews over-represent bugs and price and under-represent long-term critique. The full watchlist of 917 games is collected but not analysed yet, and the percentages will move when it is.
• •Post-mortem frequencies are judgments. Every count out of 10 above is a reader’s assessment of 142 reviews per game, not a tally. Of 63 quotes checked against the stored reviews, 48 (76%) were verbatim, and only those verified quotes appear in the study or here.
• •It has no refund data. A review written with under two hours played is not a refund.
• •One game is quoted a lot. Fears to Fathom: Scratch Creek is in both sources, and this post quotes only English reviews.
• •It makes no forecast. Nothing here predicts what your game will sell or score.

The full checklist has 56 items, 20 of them critical, and its scoring rule is blunt: any critical item at zero means not ready, whatever the total. It also says that is a working rule, not yet calibrated against real scores.

Three things to do, in order:

1. 01Run the free check. The psychological horror readiness check walks your game through these findings in about five minutes. The questions ask what playtesters did, not what you hope is true. A full PDF genre report is sold from that same page.
2. 02Get the evidence. Most items need a blind playtest, a timer or a crash log. The playtest or demo tool helps you pick how to get a build in front of strangers, and the Steam Playtest setup guide covers the mechanics.
3. 03Then check the promise. Run your store page through the analyzer and read the tags, title and trailer against what the game is. The horror store page guide covers the page side.

Working in another genre? The Genre Readiness Check hub lists the others, and the same method is written up for two of them in what players hate in souls-likes and indie action RPG mistakes.

What are the most common indie horror game design mistakes?

In 9,566 Steam reviews of 47 indie psychological horror games released 2026-03-20 to 2026-09-20, the top-ranked complaint themes were The game is very short (406 reviewers, 38 games), The game has numerous bugs (203 reviewers, 26 games), Story and writing (154 reviewers, 32 games) and Gameplay and player agency (134 reviewers, 32 games). The ones most tied to negative reviews were price (14% from positive reviews), walking-simulator complaints (31%) and an incoherent story (34%). Counts are reviewers, not players.

Why do horror games get refunded?

This dataset has no refund data. What it has is reviews written with under two hours played, which is inside Steam’s refund window. Scare quality makes up about 10% of complaints from those reviewers against about 1% from reviewers with more than ten hours. The walking-simulator theme is 72% under two hours and the jumpscare-overuse theme is 67%.

Is it bad for a horror game to be short?

Not on its own. 79% of the complaints under the theme The game is very short sit inside positive reviews. The share of reviewers complaining about length or price rises with price, from 10.9% for free games to 24.2% at $12 or more, and all five flawed hits marked too short for the price cost $13-25 for 1.5-5 hours. The checklist’s working guide is about 1.5-3 hours at $8 or less, and above $12 only with 5 or more hours of non-padded play.

What do players mean by walking simulator?

In these reviews it means nothing to do besides walk: a single corridor, puzzles that solve themselves, clicking where told. The theme was raised by 68 reviewers across 15 of 47 games, only 31% of them recommending. The post-mortems judged 6 of 10 flawed hits to have no agency.

Are jumpscares bad?

No. Jumpscares are effective and well-executed is a praise theme from 310 reviewers across 24 games. The complaint theme is about overuse and predictability (174 reviewers, 23 games). The checklist asks for a small budget tied to events, never the same scare twice, and scare audio that does not hurt.

Is there an indie horror game design checklist I can use?

Yes. The free psychological horror readiness check is built from the 56-item checklist behind this post and takes about five minutes.