INDEPENDENT INDIE INTELLIGENCE

Steam Discovery
5,197catalog apps
60digested stories
2026-09-28source check

MARKET PULSE · AS OF 2 SEP 2026

Demand exists. Discoverability is the scarce resource.

Q2 coverage is incomplete. June-only records use overlapping tags and September review observations—not the HTMAG annual methodology.

17,480Steam releases in 2026 YTDSteamDB snapshot, 2 Sep 2026SteamDB game data
~26Ksimple annualized paceOwn arithmetic: 17,480 / 245 × 365. Ignores seasonality.SteamDB game data
2.99%2025 launches with 1,000+ reviews608 / 20,282 in HTMAG/VGI cohortWhat the hell happened in 2025?
38 : 10–9 review games per 1K+ review gameQ1 2026 young-cohort snapshot2026 Q1 Games
DEMAND / SUPPLYQ2 2026: coverage and available evidence
Method/source · /data/steam-june-2026-cohort.json
FULL QUARTER NOT VERIFIED · JUNE SUBSET BELOW

Available Q2 evidence, with its limits visible.

No publicly accessible, complete Q2 2026 genre-supply and 1,000-review outcome table was verified. A separate June-only public-store snapshot is available below. April–June totals and exact quarterly hit rates remain unknown.

Checked HTMAG archive · https://howtomarketagame.com/2026/

Different methodology: anonymous US catalogue, displayed June release dates, up to seven overlapping Steam tags, review-score counts observed 2 September. This is not Chris’s single-genre cohort, not indie-only, and not first-launch or equal-age performance.

2,102Included June-dated apps
≥45Confirmed 1K+ score-count games
1,380Threshold status unresolved
—Exact hit rate: not available

Missing numerical tooltips are not treated as zero reviews. Language-specific counts below 1,000 are also unresolved. A displayed count already above 1,000 confirms a minimum, not all-language total sales or reach. Lifetime reviews can predate a June 1.0 release.

Overlapping Steam tagIncluded June supplyConfirmed 1K+ minimumUnresolvedExact hit rate
Open World Survival Craft5≥21—
Farming Sim21≥114—
Roguelike Deckbuilder18≥110—
Simulation411≥9236—
Management74≥231—
Horror143≥383—
Psychological Horror90≥355—
Idler130≥164—
Tower Defense600 confirmed34—
Puzzle287≥6208—
2D Platformer1610 confirmed130—
City Builder220 confirmed13—
Life Sim59≥133—
Job Simulator120 confirmed3—
Party Game57≥142—
Online Co-Op63≥723—
Cozy490 confirmed17—
Metroidvania140 confirmed6—
Action Roguelike147≥3103—
Bullet Hell133≥293—
Gambling210 confirmed16—
Automation29≥210—
RPG243≥8154—
Fishing90 confirmed6—
Show example games and lower-review comparators
Coverage, reproducibility and source limitations

Captured 30 contiguous public search pages on 2 September 2026, 17:40 UTC. Their displayed dates bracket June, but complete Steam coverage is not claimed. Two package rows were excluded; no duplicate June app IDs were found. Retrieval stopped after an HTTP error; April and most of May were not collected.

  • Not a full-Q2 table. June cannot be scaled to April–June and should not be shown as a quarter-over-quarter trend against HTMAG 2025 or Q1 2026.
  • June is bracketed within the saved search sequence, but anonymous US visibility, removed/restricted products, content preferences and live pagination can omit games. Complete Steam coverage is not claimed.
  • Displayed release dates may reflect full release after earlier Early Access. Lifetime reviews can predate June. This is not a cohort of first-ever commercial releases or reviews earned during June.
  • Reviews were observed September 2, 64–93 days after June dates. HTMAG Q1 used an April 2 snapshot; the observation windows and maturity differ.
  • Review-score counts are not all review posts, units, revenue, wishlists, or organic reach. Steam excludes nonqualifying purchase types and some off-topic activity from the score. Marketing spend is not observed.
  • No numerical tooltip is not proven to mean zero reviews. These rows stay in the denominator and their review-threshold status remains unresolved; consequently the bounds can be too wide to rank genres usefully.
  • Tag assignment uses only the up-to-seven returned IDs, not all visible tags or a single best-fit genre. Rows overlap and must not be added together. Idler is not the entire incremental category; Bullet Hell is not synonymous with bullet heaven; Online Co-Op is not friendslop.
  • Paid and free games, indies and large studios are mixed. A low count is not proof of commercial failure, and a high count is not evidence of profit, affordable scope, or causal organic marketing.
  • Exact full-quarter data needs a legitimately available complete export plus consistent release/EA, purchase-type, language, tag and snapshot rules. No paywall, authentication, CAPTCHA or rate-limit workaround was used.
What would complete the quarterly comparison?

A legitimate complete export with app IDs, first Steam release dates, consistent genre membership and review totals—including zero-review games. Recalculate both quarters using matching definitions and an explicit observation date. Do not combine a hand-classified hit count with an overlapping-tag supply denominator.

Gamalytic export source · https://gamalytic.com/game-listValve review definitions · User Reviews — Steamworks Documentation
METRIC HYGIENEWhat does “100–200K organic reach” actually mean?
100–200K social viewsAwareness test

Good for hook validation only. Pair with tagged Steam visits and same-day wishlist delta.

100–200K wishlistsBreakout intent

A very strong prelaunch outcome—not a realistic three-week baseline.

100–200K paid unitsCommercial outcome

Requires a developer/publisher disclosure. Reviews and owner models cannot confirm this threshold.

$100–200K grossRevenue, not reach

Do not multiply owners by list price. Discounts, regions, refunds, tax and platform share change the economics.

100–200K Steam impressionsPrivate discovery metric

Only the developer can read exact impressions in Steamworks traffic reports.

Public-data ceiling: exact Steam organic impressions and source-level conversions live inside the developer’s Steamworks Traffic Breakdown. SteamDB, VG Insights and review multipliers cannot reconstruct them. Steamworks documentation.

GENRE → NICHE → TROPE → TEST

Eight opportunity hypotheses.
None is an empty market.

A genre is the ruleset. A niche is the buyer + use case. A trope is the instantly recognizable promise. The opening is a differentiated testable combination—not a high tag count.

01 · Simulation × horror / absurdityHigh demo fit

One-room jobs, under pressure

A familiar task with an unacceptable second objective.

DEMAND EVIDENCE

The Cabin Factory reported 20K wishlists in ten days; Iron Nest disclosed >365K copies by day five. Different scales, both concentrated around readable physical actions.

Evidence 1 · How Iron Nest sold >250k copies in less than 3 days...Evidence 2 · https://newsletter.gamediscover.co/p/revealed-new-data-on-steam-wishlist
Potential opening
Hypothesis: returns inspection with morally absurd approval rules. A fantasy and rule twist—not proof the niche is empty.
Three-week slice
One counter, 10–12 objects, one shift, authored surprises. Reuse the room; vary procedures.
Channel test
Test close-up tactile shorts, then small simulation/horror creators with a complete shift.

A satisfying interaction needs sound and animation. Full store management and NPC simulation are outside the slice.

02 · Incremental × spectacleHigh demo fit

Finite, physical incrementals

Make one small action become impossibly large.

DEMAND EVIDENCE

Horripilant reported 50K copies in month one; its retrieved owner models range from 115.6K to 162.2K. Outhold reported 50K in week one; current models span 193.3K–220.6K.

Evidence 1 · Steam community / developer announcement — HorripilantEvidence 2 · SteamDB game data
Potential opening
Hypothesis: absurd QA goblins visibly multiply while solving the wrong problem. The opportunity is a fresh readable fantasy, not another upgrade menu.
Three-week slice
One arena, one repeated action, three unlocks, a visible before/after within two minutes.
Channel test
Before/after shorts → playable loop → incremental specialists. Make the clip show progress, not a spreadsheet.

Cheap entry does not imply cheap tuning. Repetition without a satisfying end state feels like a prototype.

03 · Roguelite × chance × tactile UIMedium demo fit

A machine that makes luck tangible

The player can see the stakes, probabilities and disastrous reversal.

DEMAND EVIDENCE

Slots & Daggers reported 100K copies in ten days. Balatro’s developer documented 208,401 launch wishlists and 119K Steam units on launch day; that is a long iteration story, not instant genre validation.

Evidence 1 · Steam community / developer announcement — Slots & DaggersEvidence 2 · https://localthunk.com/blog/balatro-timeline-3aarh
Potential opening
Hypothesis: a visibly rigged office raffle or haunted vending machine whose rules can be bent. Test the decision, not just the visual skin.
Three-week slice
One machine, one round structure, five modifiers, clear readable probabilities. Fictional stakes only.
Channel test
Clips for the reversal; long-form strategy creators for the decisions and replayability.

Balancing synergies can exceed the entire sprint. Avoid real-money gambling, paid random rewards and large content pools.

04 · Social co-op × physical comedyMedium demo fit

Friends + one shared burden

Everyone wants the same thing; everyone can ruin it.

DEMAND EVIDENCE

The references show shared vehicles, chains, rescues and footage turning coordination into comedy. Content Warning’s 6.6M free claims and established Landfall distribution are major confounds.

Evidence 1 · https://landfall.se/content-warning-press-kitEvidence 2 · https://app.pressengine.net/newsdesk/bastion/stories/news-chained-together-links-up-with-secret-mode-for-future-development
Potential opening
Hypothesis: carry one enormous wedding cake or fragile moving-day object. Replace quota extraction with a very different shared goal.
Three-week slice
One hero object, one room/route, private two-player lobby. Validate remote interaction by day four.
Channel test
Real group voice clips with a complete setup → failure → reaction, then small co-op creator groups.

Without a proven networking stack, fit is Low. R.E.P.O.-scale object physics and Content Warning-grade recording are not three-week features.

05 · Horror × observation / inspectionHigh demo fit

Ordinary places, one wrong detail

The viewer spots the danger one second before the player.

DEMAND EVIDENCE

The Exit 8’s publisher reported >500K PC downloads in April 2024; its developer described deliberate scope reduction. Later inspection games show continuing demand but increasingly crowded supply.

Evidence 1 · https://playism.com/en/news/2024/0417/1559/Evidence 2 · Steam store page — The Exit 8
Potential opening
Hypothesis: a work chat where a non-human colleague is trying to get invited into the room. The interaction grammar must be distinct.
Three-week slice
One room, an authored message-state machine, six escalating rules, one finale. No generative dialogue dependency.
Channel test
Horror creators and short puzzle-like anomalies; show the rule early enough for viewers to participate.

Content consumption is fast. A few jump scares are not replayability, and clones compete with familiar incumbents.

06 · Cozy × puzzle / productivityHigh demo fit

Small rituals, specific audiences

Care, tidy, arrange or accompany—not generic “cozy”.

DEMAND EVIDENCE

My Little Life’s developer reported 30K copies after launch; current owner models span 68.6K–136K. The productivity study found nine of 34 selected games above 500 reviews, with exclusions.

Evidence 1 · https://www.reddit.com/r/LifeSimulators/comments/1inp6uc/my_mini_lifesim_that_lives_on_the_bottom_of_your/Evidence 2 · The Rise of Productivity Games on Steam
Potential opening
Hypothesis: an oddly comforting overnight ritual for remote workers. Validate a specific use context and emotional payoff.
Three-week slice
One ritual, three characters or objects, one progression session. Focus on feel and a single polished vignette.
Channel test
Cozy curators, lifestyle/focus creators, themed festivals and customization sharing.

Art taste, writing and comfort are the product. Tiny Bookshop-scale narrative content does not fit three weeks.

07 · Autobattler × PvE / rogueliteMedium demo fit

A requested mode inside a familiar genre

“I love X, but wish it had Y” from an identifiable community.

DEMAND EVIDENCE

Guildrun disclosed >275K demo players and 3h55 median playtime. Specialist long-form creators worked; short-form did not. Paid creators and ads also contributed.

Evidence 1 · How Guildrun genre-mashed its way to ~300k demo players in 10 daysEvidence 2 · Indie Dev Timeline for How Many Dudes
Potential opening
Hypothesis: a tiny PvE drafting system for a community asking for that mode. Confirm the complaint before inventing a new genre mashup.
Three-week slice
One encounter chain, six units, a single decision loop. Balance telemetry matters more than cinematic presentation.
Channel test
Specialist long-form videos, repeated community playtests and clear decision examples.

These successes took months or years of iteration. A complex synergy system is usually a poor first three-week bet.

08 · Physics × puzzle / partyHigh demo fit

A joke that is also a toy

One ridiculous verb with visible consequences.

DEMAND EVIDENCE

Horse Magnifier’s retrieved owner models span 20.3K–39K—useful demand without a megahit assumption. Super Battle Golf reported 100K copies in 48 hours after 4.5 months of development.

Evidence 1 · SteamDB game data — Horse Magnifier: The Full HorseEvidence 2 · https://www.pcgamer.com/games/sports/that-golf-game-with-orbital-death-lasers-sold-100k-copies-in-two-days-we-made-super-battle-golf-together-in-4-5-months-and-are-so-happy/
Potential opening
Hypothesis: one absurd manipulation tool with at least three genuinely different uses. The caption should be understandable without current meme knowledge.
Three-week slice
One room, one verb, five authored challenges. Solo first; no online competition in the first slice.
Channel test
Raw clips with a visible before/after. Watch whether people ask to play, not merely react to the joke.

A one-post joke can exhaust demand. Prove that a second session is different before expanding production.