Indie Action RPG Mistakes, According to 4,490 Steam Reviews

In 4,490 Steam reviews of 12 recent indie action RPGs, the highest-ranked complaint theme is progression and grind (133 reviewers, all 12 games), followed by difficulty balance (110) and build variety (113). The largest single specific complaint is that combat feels unresponsive and clunky: 83 reviewers across 10 games.

The second half of the study looks past launch. We also read post-mortems of ten action RPGs that sold well and ended up with Mixed or Mostly Positive scores. Nine of the ten ran out of content or repeated it, six had builds that collapsed into one or two viable options, and five had loot that never felt like progress. Those problems barely register in launch-window reviews. They show up at hour twenty.

Before any of the numbers: this is a small sample. The recent set is 12 games released between 2026-03-20 and 2026-09-20 that reached 100+ reviews. Six of them sit under 80% positive and only two are at 90% or higher, so one game can move a figure. Where one game drives a number, the caveat sits next to it. The ten post-mortems (142 most helpful reviews each) carry much of the weight in this post, and their “n of 10” frequencies are judgments counted by hand from ten written reports, not review counts.

To test your own game against these findings, the free action RPG readiness check takes about five minutes.

An LLM read the reviews, pulled complaints, praise and wishes into a fixed list of categories, and clustered similar statements into themes. 19 games passed the tag rule for the niche and 7 were left out as off-genre. The 12 that remained produced 19,007 extracted points.

Themes are ranked by a score that combines how many of the 12 games show the theme, what share of all reviewers raise it, and how severe the complaints are. That is why interface complaints, with 127 reviewers, land ninth: their severity is 2.3 out of 5.

In nine of the ten rows, 40% to 54% of the complaints sit inside a review that still recommends the game. Developer support is the exception: only 11% come with a recommendation, and developers replied to 0% of those reviews.

A theme row is narrower than its category. Combat feel fills two rows here and two more further down the list, so category totals are larger than any one row:

Combat is both the most complained-about category (17.2% of reviewers, 774 of them, no single game above 21% of those) and the most praised (1,011 reviewers, 12 of 12 games).

Counts are reviewers, not players, and reviewers skew engaged and opinionated, so trust ranks more than absolute percentages. 68% of the analysed reviews are not in English. The quotes below are English ones, copied character for character and attributed by game only.

Steam’s refund window covers the first two hours of play, and players who complain inside it talk about different things than players who stayed. Combat feel is 16.1% of complaints from reviewers under 2 hours, against 7.5% past 10 hours. Controls are 9.2% against 2.4%. Price is 4.1% against 1.7%.

“Combat lacks satisfying impact and feedback” comes from 48 reviewers across 11 of the 12 games. Unreliable dodges: 30 reviewers across 6 games. Parries that do not register: 32 across 4. Clunky controls: 42 across 10. A camera too close to see attacks coming: 40 across 6.

“the movement and attacks feel like stop-motion animations. There’s barely any impact, hit feedback and animation are non-existent” (Kernel Hearts, not recommended, 1.0h played)

One figure needs its caveat. Inconsistent hitboxes are raised by 80 reviewers across 6 games, but one game is 82% of those. Another 29 reviewers raise hitboxes under enemy design, across 6 games. The post-mortems back the theme up independently. From Eldest Souls, a boss-focused game at 76% positive:

“A boss can be facing one direction and swinging straight ahead, yet still hit you when you’re clearly outside the attack animation.”

Combat or controls that feel wrong appear in 4 of the 10 flawed hits. The checklist built from this data turns it into a test you can run this week: a blind playtester, the starting kit, ten minutes, then ask whether attacking and dodging feel good. Answering “it gets better once you unlock X” is not a pass.

A broken promise is the second most common failure in the post-mortems: 7 of 10. The checklist puts it in one line: players forgive a small game, not a misdescribed one.

Eldest Souls was bought as a pixel-art soulslike. Its reviewers describe something else:

“This is a boss rush game, where you travel through an empty world from one boss fight to the next”

Zengeon sold roguelite runs “with hundreds of combinations” and shipped items that do not change how a character plays, on three fixed maps. Van Helsing III was the finale of a trilogy whose fans expected to bring their character along:

“Between VH1 and 2 you could import characters. They removed that from VH3.”

The checklist’s rule is blunt: a boss rush is not a soulslike, and fixed maps with stat items are not a roguelite. Marketing numbers have to be true in play, meaning that many ways to play and not that many stat permutations. If your game leans toward the soulslike label, our soulslike review study covers what that audience complains about.

This is the one finding here that lives on your store page and not in your build. Read your tags, short description and first screenshots as a stranger would, and ask what genre they promise.

The free analyzer reads the live page and names the biggest issue on it. For tags, the tag optimizer compares yours with comparable games, and our pull of 764 store pages’ tags shows how the first five slots get spent.

This is what invested players judge. Past 10 hours, build variety is 5.8% of complaints (1.8% for early quitters) and loot is 5.4% (1.3%).

Build variety is praised by 516 reviewers and complained about by 425 (9.5%), both in all 12 games. The niche wants builds, and it notices when they are not real. The “many builds, few viable” pattern shows up in 6 of the 10 flawed hits. From Wolcen, 55% positive across 61,837 reviews:

“they’re just nerfing the the few viable builds on endgame”

(The doubled word is the reviewer’s. Quotes here are not tidied.)

The checklist’s test: keep a spreadsheet of builds per class that cleared the top difficulty, played by testers who are not the designer. If the community guide would list one build, that is a fail. The post-mortems add a balancing habit, which is to buff the weak options before nerfing the strong one. Cheap or free respec is praised by name in SpiritVale, Eldest Souls and Dragonkin, even by unhappy players.

Loot is the quieter problem. Loot complaints come from 7.7% of reviewers at severity 3.4, and only 41% come from players who still recommend the game, so it costs recommendations. Loot and levelling that do not feel like progress appear in 5 of the 10 flawed hits. Wolcen again:

“every item in the game is a BORING stat-stick”

Dragonkin: The Banished has no uniques and locks rarity behind level:

“There are no heirloom items, unique, or items that you could feasibly use late game that you find early.”

A caveat on drop rates: the “drop rates are extremely low” theme is 92% one game, so it is not a niche-wide finding.

Levelling belongs in the same bucket: Dungeons of Sundaria’s reviewers describe a “level-scaling system that actually makes you WEAKER”.

Difficulty is the second biggest complaint category (15.9% of reviewers, severity 3.5) and enemy design is the third (13.5%). Read that with care: the two best-rated games in the sample are hard, boss-focused games, and their complaints are mostly about exactly this. Players forgive difficulty in a game they love, and still write about it.

What the flawed hits were punished for is difficulty made of numbers. In the recent sample, “Enemies have too much health” comes from 60 reviewers across 5 games, 77% of them from one game, so the post-mortems carry the weight here: unfair difficulty appears in 5 of 10.

“Higher difficulty means enemies take longer to kill” (Dragonkin: The Banished)

“the bosses have 100000 hp and you stab them with a toothpick” (Eldest Souls)

Eldest Souls also lost players at its second boss, when they had one skill point spent. The checklist’s rule is that the first hard gate comes after the player has tools and choices.

The checklist’s checks: measure time-to-kill for standard enemies at each difficulty, list what each tier adds besides numbers, and after every playtest death ask what killed the player.

“Enemy variety is limited” comes from 45 reviewers across 9 games. Difficulty is a wish from 56 reviewers in 10 games, and a step between story and normal is asked for across 6.

Content that runs out or repeats is the most common failure in the post-mortems: 9 of 10, every game except Hero Siege.

The recent sample under-reports this: endgame complaints come from 1.6% of reviewers there, because most of them have not reached the end yet. It already shows as a wish: 59 reviewers across 9 games. The launch-window versions of the complaint are “The game is quite short” (47 reviewers, 8 games), “lacks sufficient content” (27, 7 games) and “boring over time due to repetitive content” (25, 8 games).

“The gameplay gets very repetitive really quick” (Zengeon)

The post-mortems also describe what fills the gap. Grind or time-gates used as content appear in 3 of the 10 flawed hits, and levels that waste time in 3. In the recent sample, “requires a lot of grinding” comes from 30 reviewers across 6 games. Dragonkin’s reviews name both:

“Upgrading your city to the maximum level requires an absurd number of hours spent repeating the same activities.”

“escort objectives, defend the npc objectives, hold an area and backtracking”

And from the recent sample, the grind theme in one sentence:

“The exp required to level up while balanced at the beginning is insufferable once you get into the 60’s” (ZENONIA 1, not recommended, 20.6h played)

The checklist offers two ways out. Either there is something worth doing after the credits, and it exists at launch, or the game is sold as a finite campaign and priced like one.

Online-economy complaints have the highest severity of any category: 4.3 out of 5. Only 12% come from players who still recommend the game, and they belong to invested players: 7.5% of complaints past 10 hours, 0.7% under 2 hours.

The caveat is large. In this sample 91% of those complaints come from one game, SpiritVale, 48% positive across 4,763 reviews. The reason to take it seriously anyway is that the pattern repeats in the post-mortems of Hero Siege and Wolcen: 3 of the 10.

SpiritVale failed almost entirely on operations. Players describe bans they say hit legitimate accounts with no real appeal, bots in the economy, and then:

“Servers are down multiple times per day for multiple hours”

“Player trading has been disabled with no ETA”

“It took two whole weeks just to fix a broken Auction House”

Hero Siege adds overloaded season launches (“Season launches ALWAYS suck.”) and paid extras inside a paid game:

“Stash tabs cost money, inventory tabs cost money, in-game housing requires a subscription.”

Wolcen adds wipes: “Every time they do a big update, you lose your progress.”

Developer support has the second-highest severity of any category (3.9 out of 5). It is 8.2% of weak games’ complaints against 0.2% in strong ones, but that compares 6 weak games with 2 strong ones, and 59% of those complaints come from one game.

The SpiritVale post-mortem’s “Done right” note reads, in the study’s words and not a reviewer’s:

an online economy is an operations commitment: server-side bot detection before launch, a human review and a real appeal for every permanent ban, load testing at playtest scale, and never removing trading as a fix. If the team cannot staff that, make the game self-found or co-op without a shared economy.

The checklist says it in fewer words: if a small team cannot staff the online section, “the documented safe choice is no shared economy.”

The checklist names six combinations of failures, each matched to documented flawed hits. If a game fits one, that is the thing to fix before anything else.

Every game in that table sold. Wolcen alone has 61,837 reviews. All ten were praised for their look or their first hours, and failed anyway. The post-mortems are on the flawed hits study page, and the readiness check for action RPGs matches your answers against these six patterns.

Wishes are mostly written inside positive reviews, which makes them the closest thing to a backlog from paying fans. The top wish themes are build variety (45 reviewers, 10 games, 87% inside positive reviews), combat feel (40 reviewers, 9 games, 77%) and interface (52 reviewers, 9 games, 85%). At category level, interface is the top wish (114 reviewers, 11 games) and build variety the third (87 reviewers, 11 games).

The specific requests repeat: animation cancelling (4 games), New Game Plus and a boss rush mode (3 games each), an endless mode, skipping an animation that replays (15 reviewers, 4 games), and more and better localisation (9 games).

One wish reads as an offer:

“These can be patched though, and once they are, I’ll happily update my review” (Rubinite, not recommended, 11.3h played)

If you ship that patch, say so where the reviewer will see it. Our review management guide covers replying to negative reviews.

• •The recent sample is 12 games. The weak-versus-strong comparisons rest on 6 weak and 2 strong games, and both strong games are hard boss-focused ones. Use them for direction only. The full 239-game watchlist is collected but not analysed yet, and percentages will move when it is.
• •“Action RPG” is a broad tag. Top-down loot games, soulslikes, boss rushes, roguelites and an online grinder sit in the same sample. Loot and endgame findings matter less for a six-hour boss rush.
• •Launch-window reviews are skewed. They over-represent bugs, controls and price, and under-represent endgame, itemization and long-term balance. The post-mortems cover that gap with judgments, not counts.
• •Every game in the data found an audience. Of 2,192 relevant games in this niche, 239 reached 100 reviews, and the median relevant released game has 10. Nothing here explains the rest, which is a visibility and hook problem.
• •No outcomes. There are no sales or refund figures here.

The findings are packed into a 59-item checklist with 22 items marked Critical. An offline game skips the online section: 55 items, 19 Critical. Its scoring rule has not been calibrated against real scores yet, but its first line is easy to apply: any Critical item at zero means not ready, whatever the total.

1. 01Run the free action RPG readiness check. It takes about five minutes. Answer from playtest evidence, not intention.
2. 02If you match one of the six red-flag combinations, fix that first.
3. 03Building something else? The genre readiness hub lists the other genres studied so far.

A full PDF genre report is sold from the genre page. The check is free.

What do players complain about most in indie action RPGs? In 4,490 Steam reviews of 12 recent indie action RPGs, the highest-ranked complaint themes are progression and grind (133 reviewers), difficulty balance (110), build variety (113) and combat feel (107). By category, combat feel is the most complained about (17.2% of reviewers) and also the most praised (1,011 reviewers). The sample is small, and counts are reviewers, not players.

Why do action RPGs end up with Mixed reviews on Steam after selling well? In post-mortems of ten action RPGs with 500+ reviews and 40-79% positive scores, the most common failure was content that runs out or repeats (9 of 10), followed by bugs and lost progress (7 of 10), a broken promise (7 of 10), build variety that collapses to a few viable options (6 of 10) and loot that does not feel like progress (5 of 10). Those frequencies are hand-counted judgments, not review counts. The review score thresholds post explains where the Mixed label starts.

Should a small team build an online action RPG with trading? The study’s checklist says that if a small team cannot staff server stability, human-reviewed bans with appeals and economy protection, the documented safe choice is no shared economy. Online-economy complaints have the highest severity of any category (4.3 out of 5), though in the recent sample 91% of them come from one game, SpiritVale. The same pattern appears in the Hero Siege and Wolcen post-mortems.

How many viable builds does an action RPG need? The data gives no number. The checklist’s test is that several clearly different builds can finish the hardest content, verified by testers who are not the designer. If a community guide would list one build, it fails. The “many builds, few viable” pattern appeared in 6 of the 10 flawed hits.

How big is the sample behind these numbers? 4,490 reviews (19,007 extracted points) of 12 action RPGs released between 2026-03-20 and 2026-09-20 that reached 100+ reviews, plus the 142 most helpful reviews of each of 10 older flawed hits from 2014 to 2026. Six of the 12 recent games are under 80% positive and only two are at 90% or higher, so single games can move a figure.