Reviewers who had played under two hours complain about how a game handles and whether it delivers its promise straight away. Reviewers past ten hours complain about balance, builds and endings. In 12,327 reviews of 30 recent souls-like games, controls and input made up 14% of the complaints from reviewers with under 2 hours played, against 5% for reviewers with over 10 hours. In 9,566 reviews of 47 indie psychological horror games, scare quality was 10% of the under-2-hour group’s complaints and 1% of the over-10-hour group’s.
Two hours is the line because of Steam’s refund window. Valve’s refund policy covers requests made within two weeks of purchase with less than two hours of playtime. Our Steam refund rates post covers the policy and what refunds cost.
One warning before any numbers. A reviewer who posted with under two hours played is not a refunder, and none of the three studies contains refund data. They show what people who reviewed early complained about. That is the closest public signal I know of for a bad first session, and no more.
This post puts the three genre studies behind our Genre Readiness Check side by side. It matters this month because Steam Next Fest runs October 19 to 26, 2026, and a demo is a first session and nothing else. If you would rather test your build than read tables, the quick check asks exactly these first-session questions for three genres.
The three samples and what the two-hour figures measure
| Study | Reviews analysed | Games | Games released between |
|---|---|---|---|
| Souls-like | 12,327 | 30 | 2026-03-21 and 2026-09-21 |
| Indie action RPG | 4,490 | 12 | 2026-03-20 and 2026-09-20 |
| Indie psychological horror | 9,566 | 47 | 2026-03-20 and 2026-09-20 |
Every game reached 100+ reviews. Together that is 26,383 reviews, but the studies were run separately and nothing below pools a percentage across genres. The action RPG sample is small, and the source says so itself: 12 games, where one game can move a number.
The method is the same in all three. After jokes, one-liners and duplicates were dropped, the most helpful negative reviews, positive reviews containing a “but”, and a sample of plain positive reviews per game were read by an LLM, which sorted complaints, praise and wishes into a fixed category list. A complaint here can come from a review that recommends the game.
Two different figures carry a playtime split.
Category shares. The source describes its bounce table as the share of each group’s complaints by category. The 14% above means that of all the complaints extracted from reviewers with under 2 hours played, 14% were about controls and input. It does not mean 14% of early reviewers complained about controls.
Theme splits. Each ranked complaint theme also carries a column labelled under 2h and over 10h. The source does not define it beyond that label. Our published reports describe it as the playtime split of the reviewers raising the theme, and the numbers behave that way. The souls-like rebinding theme, raised by 241 reviewers, shows 41% and 21%: 41% of those reviewers had under 2 hours played and 21% had over 10. The remainder is not shown.
The source calls the first group players who bounced and the second invested players. I use the longer wording on purpose: the counts are reviewers, not players, and a review at under two hours does not prove its author quit.
Souls-likes: controls and combat feel come first
From 12,327 reviews of 30 souls-like games. Each figure is that category’s share of the group’s complaints. The left and right halves are separate lists, so rows do not pair up.
| More common under 2h | Under 2h | Over 10h | More common over 10h | Under 2h | Over 10h |
|---|---|---|---|---|---|
| Controls and input | 14% | 5% | Difficulty balance | 5% | 11% |
| Combat feel | 13% | 8% | Build variety | 2% | 6% |
| Visuals and art | 6% | 2% | Boss design | 1% | 4% |
| Performance | 7% | 4% | Progression and grind | 2% | 4% |
| Interface and menus (UI/UX) | 9% | 7% | World and level design | 2% | 4% |
| Price and value | 3% | 1% | Platforming and traversal | 1% | 3% |
Only three of the 25 ranked complaint themes have more reviewers under 2 hours than over 10, and all three are about handling:
- “Controls feel unresponsive and clunky.” 120 reviewers across 25 games, split 43% under 2h and 28% over 10h.
- “Keybinds cannot be remapped on keyboard, mouse, or controller.” 241 reviewers across 26 games, split 41% and 21%.
- “Combat feels clunky and unsatisfying.” 261 reviewers across 26 games, split 34% and 29%.
From the rebinding theme, a Duskfade reviewer at 0.1 hours played:
“There is no key rebinding”
The camera needs a caveat. The largest camera theme, “The camera is problematic during boss fights.” (269 reviewers across 19 games), splits 17% under 2h and 34% over 10h, and the checklist notes that 64% of it comes from one game. That fits: you can hate the keybinds in minute one, but you have to reach a boss before its camera can annoy you. Some get there fast. From Crimson Moon, at 0.9 hours:
“your camera whips around wildly, making it both extremely disorienting”
At the other end, the build variety theme (123 reviewers) splits 4% under 2h and 74% over 10h, and progression and grind (178 reviewers) splits 7% and 73%. The full ranked table is on the souls-like checklist page, and what players hate in souls-likes covers the whole study.
Indie action RPGs: combat feel inside ten minutes (small sample)
From 4,490 reviews of 12 indie action RPGs. With 12 games, read it as direction only.
| More common under 2h | Under 2h | Over 10h | More common over 10h | Under 2h | Over 10h |
|---|---|---|---|---|---|
| Combat feel | 16% | 7% | Online economy integrity | 1% | 8% |
| Controls and input | 9% | 2% | Loot and itemization | 1% | 5% |
| Visuals and art | 5% | 1% | Build variety | 2% | 6% |
| Length and amount of content | 8% | 5% | Difficulty balance | 5% | 8% |
| Price and value | 4% | 2% | Developer support and updates | 4% | 7% |
| Audio and voice | 3% | 1% | Bugs and crashes | 5% | 7% |
A source caveat on one row: 91% of the online economy complaints come from one game, SpiritVale.
In every ranked theme in this sample but one, more of the reviewers had over 10 hours played than under 2. The exception: “Combat lacks satisfying impact and feedback.” (48 reviewers across 11 games) splits 38% under 2h and 27% over 10h. Next by early share come price and value (57 reviewers, 33% and 47%) and “Combat feels unresponsive and clunky.” (83 reviewers across 10 games, 28% and 35%).
The broader combat feel theme (107 reviewers across 11 games) splits 25% and 45%. From it, a Kernel Hearts reviewer at 1.0 hours played:
“the movement and attacks feel like stop-motion animations. There’s barely any impact, hit feedback and animation are non-existent”
Compare the loot and itemization theme (78 reviewers), which splits 7% under 2h and 79% over 10h. Few reviewers judge a loot system in their first session. I expected price to rank higher on the early list than it does. It is there, at 4% against 2%, well behind combat feel. The action RPG checklist page has the ranked table, and indie action RPG mistakes covers the long game.
Indie psychological horror: the first scare, and something to do
From 9,566 reviews of 47 indie psychological horror games.
| More common under 2h | Under 2h | Over 10h | More common over 10h | Under 2h | Over 10h |
|---|---|---|---|---|---|
| Scare quality | 10% | 1% | Difficulty balance | 5% | 12% |
| Multiplayer and co-op | 2% | 0% | Story and writing | 13% | 16% |
| Bugs and crashes | 6% | 5% | Interface and menus (UI/UX) | 3% | 6% |
| Gameplay and player agency | 7% | 5% | Pacing | 5% | 7% |
| Darkness and visibility | 1% | 0% | Puzzle design | 2% | 4% |
| Performance | 2% | 1% | The ending | 2% | 4% |
Horror turns the theme splits upside down. In 24 of the 25 ranked complaint themes, more of the reviewers had under 2 hours than over 10. The exception is difficulty balance, tied at 23% each.
Be careful with what under 2 hours means here. The checklist’s pricing guide talks about games of about 1.5-3 hours, so a horror reviewer with under two hours played may have finished the game, and the source does not separate the two. The biggest theme, “The game is very short.” (406 reviewers across 38 games, split 60% and 2%), is raised inside positive reviews 79% of the time. Those are mostly satisfied players, and I would not count them as bounces.
The scare themes read differently. “Jumpscares are overused and become predictable.” (174 reviewers across 23 games) splits 67% under 2h and 0% over 10h, and the walking simulator theme (68 reviewers across 15 games) splits 72% and 0%.
A Fears to Fathom: Scratch Creek reviewer at 2.1 hours played:
“the jump scares are pretty cheap and predictable”
Some of this is about the promise as much as the game. The checklist’s first honesty item is that the store page, tags and title describe the game you made. From Lust for Darkness, one of the older flawed hits:
“it’s just not a horror at all”
Your store page makes the promise that the first session then has to keep. Before you tune the opening, check what the page tells people they are buying:
The free analyzer reads your live page and names the biggest issue on it. The ranked horror table is on the psychological horror checklist page, and indie horror game mistakes covers the full study.
What repeats across all three genres
Each study ends in a checklist, and section B of each is called “The first two hours (the refund window)”. Four items come back in all three.
| What section B asks for | Souls-like | Indie action RPG | Indie psychological horror |
|---|---|---|---|
| Controls and camera right on the first try | B1, critical | B2, critical | B5, major: within five minutes |
| It runs and does not break in the first session | B3, critical | B4, critical | B4, critical |
| The core of the game lands early | B2, critical: first fights feel good | B1, critical: combat feels right within ten minutes | B1 and B2, critical: a scare in the first 30-60 minutes, no chores first |
| The player knows what to do next | B4, major | B5, major | B3, critical |
Start with controls and camera. The souls-like checklist lists Dark Devotion, Bleak Faith and Crimson Moon as flawed hits that lost players here. The action RPG checklist cites “The controls feel clunky” from 42 reviewers across 10 games and a too-close camera from 40 reviewers across 6 games. From the Dark Devotion post-mortem:
“Not rebindable in any way, shape, or form.”
Next, it runs and does not break. The wording is identical in all three checklists. Performance is 7% of souls-like complaints under 2 hours against 4% past 10. In horror, “The game crashes frequently.” (72 reviewers across 27 games) splits 42% under 2h and 7% over 10h. A Mahou Arms reviewer at 1.1 hours, from the action RPG sample:
“managed to soft-lock myself twice within the first 30 minutes”
Third, the core of the game lands early. For the two combat genres that means the starting kit, before any upgrade. For horror it means an opening that is not chores, and the checklist notes that players read early busywork as deliberate padding past the refund window. From the Scratch Creek post-mortem:
“Over half of the game is packing boxes”
Knowing what to do next is the weakest of the four as a two-hour finding. Onboarding complaints come from 6.4% of souls-like reviewers in 30 of 30 games and 5.5% of action RPG reviewers in 12 of 12 games, but the onboarding and guidance theme splits 16% under 2h and 59% over 10h in souls-likes, and 19% and 54% in action RPGs. Only horror leans early, at 23% and 16%. The checklists still file it under the first two hours: mechanics explained when first needed.
What veterans criticise instead, and why launch reviews hide it
Now the right half of each table. Souls-like reviewers past 10 hours complain about difficulty balance (11% of their complaints against 5% under 2 hours), builds and bosses. Action RPG veterans complain about loot, builds and, where there is one, the shared economy. Horror veterans complain about difficulty balance (12% against 5%), story and the ending.
What the samples miss matters more. All three are launch-window samples, games released in the six months before the pull, and each checklist states the bias in its known limits:
- Souls-like: launch-window reviews over-represent controls, performance and bugs, and under-represent runbacks, late-game balance and endings.
- Indie action RPG: over-represent bugs, controls and price; under-represent endgame, itemization and long-term balance.
- Indie psychological horror: over-represent bugs and price; under-represent long-term critique.
Runback and checkpoint complaints come from only 2.6% of reviewers in the souls-like sample, yet the way back after death is the second most common failure in the ten older flawed hits, present in 7 of 10. The source’s explanation is that players who quit early never reach the long runback. From the Ashen post-mortem:
“atrocious boss runs, what an awful and frustrating decision”
Action RPGs repeat the pattern. Endgame complaints come from 1.6% of reviewers in the recent sample, while content that runs out or repeats is the most common failure of the flawed hits, in 9 of 10. In horror, ending complaints come from 3.8% of reviewers across 39 games, and story failed in 9 of 10 flawed hits. The flawed-hit counts are hand-made judgments from ten post-mortems per genre, not review counts. The flawed hits study has the games.
A clean first two hours removes the class of complaint this kind of sample sees best. It tells you nothing about hour twenty.
What to watch in a blind first session before Next Fest
The checklists say to answer from evidence, not intention: a blind playtest, a timer, a crash log, recorded boss attempts. Each item below is something to observe in a demo build, taken from their items and checks.
- Do they go looking for rebinding or the options menu, and do they find what they wanted? Test keyboard-and-mouse and controller separately.
- After the first three encounters, do attacking, dodging and blocking feel reliable? The action RPG version asks the same after ten minutes and adds a rule for you: do not answer “it gets better once you unlock X”.
- Where does the tester first fail five times in a row, and what options did they have at that point?
- For horror, what is the timestamp of the first real scare? The checklist’s target is the first 30-60 minutes.
- How many times does the tester stop for more than about 3 minutes not knowing the goal?
- Did it crash, soft-lock or drop frames on ordinary hardware? For a souls-like, also test camera and lock-on against the largest enemy in the tightest room.
- For horror, what did the player decide or learn in each 10-minute block? Empty blocks are padding.
If you have not settled the format, the playtest or demo tool walks through that choice and how to run a Steam playtest covers setup. For the fest itself there is the Next Fest prep tool, and how long a Steam demo should be covers length. Then run the Genre Readiness Check against what you observed, not what you intended.
Limits of this data
- Reviewers are not players, and early reviewers are not refunders. The studies contain no refund data and no data on players who left without reviewing. The source adds that reviewers skew engaged and opinionated, so ranks are more trustworthy than absolute percentages.
- Playtime is not defined. The source does not say whether hours played means playtime when the review was written or total playtime since.
- The action RPG sample is 12 games. Six sit under 80% positive and two are at 90% or higher. One game can move a figure.
- The tags are broad. The souls-like sample includes boss rushes, roguelite hybrids and one extraction PvPvE game, and the action RPG sample mixes top-down loot games, boss rushes and an online grinder.
- An LLM sorted the complaints into a fixed category list.
- Survivorship: every game in the three samples found an audience, and a game that does everything on this page can still go unnoticed.
Frequently asked questions
What is Steam’s refund window?
Valve’s refund policy covers requests made within two weeks of purchase with less than two hours of playtime. Valve also says you can ask for a refund outside those rules and it will take a look. Our Steam refund rates post covers the policy in more detail.
Do players who review a game with under two hours played refund it?
This data cannot tell you. The three studies count reviewers by hours played and contain no refund data. A reviewer with under two hours played may have refunded, kept the game, or, in a short horror game, finished it.
What do players complain about in the first two hours of a Steam game?
It depends on the genre. In 12,327 reviews of 30 souls-like games, controls and input are 14% of the complaints from reviewers with under 2 hours played and combat feel is 13%. In 4,490 reviews of 12 indie action RPGs, a small sample, combat feel is 16% and controls 9%. In 9,566 reviews of 47 indie psychological horror games, story and writing is 13% and scare quality is 10%, and scare quality is the category most over-represented against reviewers with over 10 hours, where it is 1%.
What do players with 10+ hours complain about instead?
In the souls-like sample, difficulty balance (11% of their complaints), build variety (6%) and boss design (4%). In the action RPG sample, online economy integrity (8%), loot and itemization (5%) and build variety (6%). In the horror sample, story and writing (16%), difficulty balance (12%) and pacing (7%).
Does a good first two hours mean good reviews?
No. Launch-window reviews under-represent late problems. Runback and checkpoint complaints come from 2.6% of reviewers in the souls-like sample, but the way back after death is a failure in 7 of the 10 older souls-like flawed hits. Endgame complaints come from 1.6% of action RPG reviewers, and content that runs out or repeats appears in 9 of 10 action RPG flawed hits.