SHYPIARD RESEARCH · Launch outcomes
A complete Steam page can still launch into silence.
We compared two very different groups of Steam games. The biggest lesson is how little a checklist can explain on its own.
A trailer is there. The description is there. The screenshots are there. Yet almost nobody leaves a review. For a small studio, that is a harder launch problem to diagnose than a visibly unfinished store page: the work is present, but the audience is not.
Our low-review and high-review groups each contain 100 games. Every game in both groups has a trailer and a long description at the observation date. Those assets matter to the page, but their presence alone cannot distinguish these two outcomes. The useful question is what differs beyond the basic checklist, and which differences a studio can actually investigate.
The comparison reveals a large localization gap, smaller differences in page features, and a different distribution of release days. It also exposes blind spots: most wishlist, festival and creator histories are missing. Those gaps belong in the explanation, because they limit how much of the launch we can reconstruct.
Page completeness is a starting point. It does not tell you whether enough of the right players will ever see the page.
THE OBSERVATION
What the two cohorts look like
| Measure | Low-review · n=100 | High-review · n=100 |
|---|---|---|
| Median supported languagesWhat to check & budget | 1 | 11 |
| At least five languagesWhat to check & budget | 20/100 | 87/100 |
| At least eight screenshotsWhat to check & budget | 59/100 | 82/100 |
| Achievements presentWhat to check & budget | 68/100 | 93/100 |
| Trailer and long description presentWhat to check & budget | 100/100 | 100/100 |
01 · THE ANALYSIS
The clearest gap concerns who can read the game.
The median supported-language count is 1 in the low-review group and 11 in the high-review group. 55 low-review games have one language, compared with 9 high-review games. This is a much larger separation than the screenshot difference. It gives a studio a concrete question: does the game's language support fit the audience it is trying to reach?
That question is more useful than a blanket instruction to buy ten translations. The CSV does not separate interface, subtitles and full audio, measure translation quality, or show the language of each customer. A well-funded game can also add languages after finding an audience. The snapshot cannot isolate the direction of that relationship.
Dwarf Fortress, Selaco and ULTRAKILL appear in the high-review group with one supported language. They are reminders that broad language support is neither necessary nor sufficient for visibility. A localization decision still needs evidence about the particular game's players, terminology and cost of maintaining translated content.
02 · THE ANALYSIS
A richer page is visible. Its conversion rate is not.
82 high-review games have at least eight screenshots, compared with 59 low-review games. Achievements appear on 93 and 68 pages respectively. These differences can guide a page audit, but they do not show which asset persuaded a visitor to buy. We have no page impressions, click-through rates or controlled before-and-after measurements.
Counting assets is therefore the first pass. The next pass is editorial: can a visitor understand the activity, stakes and appeal from the opening trailer and gallery? That requires looking at the assets themselves. An eighth screenshot that repeats the same scene adds to the count without necessarily adding a reason to care.
The same caution applies to tags. Both cohorts have a median of 20 in a source that caps the recorded list at 20. The interesting tail is the ten low-review games with six or fewer tags, rather than a supposed optimum at twenty. The dataset supports inspecting that sparse tail; it does not establish a tag-count target.
03 · THE ANALYSIS
The calendar differs, but the games differ too.
73 high-review games released on Tuesday, Wednesday or Thursday, compared with 48 low-review games. Friday accounts for 15 and 29; the weekend for 3 and 10. The pattern is worth checking against an unannounced date, especially alongside competing releases and the team's availability.
It does not justify moving a promised launch purely to follow a weekday statistic. Among Us is a high-review Friday release in this corpus. The cohorts also differ in release year, audience, publisher support and time available to accumulate reviews. None of those factors is controlled here, so the weekday is a planning question rather than a measured advantage.
04 · THE ANALYSIS
The missing launch history matters more than another score.
Next Fest history is unobserved for 89 low-review games and 91 high-review games. Creator coverage is unobserved for 95 and 95. Only 3 rows contain a wishlist declaration. Those columns cannot tell us whether the high-review group had a stronger launch audience.
Post-launch features also need their time dimension. DLC is present on 75 high-review pages and 16 low-review pages; updates are observed on 99 and 81. These are later states of the product. They can follow commercial traction as well as contribute to ongoing activity. Neither is a day-one requirement established by this study.
For a studio using this research, the next step is to connect a page audit to its own evidence: incoming traffic, audience sources, wishlist movement and the response to a demo or update. That is where an actionable diagnosis begins. The public comparison helps choose questions; the studio's timeline helps answer them.
WHAT THE DATA HELPS YOU DO
The problem to check. The next action. The budget to scope.
Recommendations below are a first investigation, not a claim that every gap needs fixing. Game examples describe the dated evidence; their intervention costs remain unknown unless disclosed.
Language coverage: choose a reachable audience before adding languages
What could be wrong
A language barrier may keep suitable players from understanding the promise or playing comfortably. A low language count alone does not show that your game has this problem.
How to investigate and improve it
- Export the page text and game strings separately. Count source words and identify interface, subtitles and audio needs.
- Choose one priority language from player requests and the audience you can reach. Translate a small playable slice and the page consistently.
- Have a fluent tester check the build, fonts and UI. Scope the remaining game text and future updates before commissioning more languages.
What other games tend to show
Five or more languages appear on 87/100 high-review pages versus 20/100 low-review pages. Broader support is common in that selected group; the single-language exceptions keep it from becoming a requirement.
15 supported languages in the 17 August 2026 row.
Spending on this specific intervention is not disclosed in the corpus.
One supported language and 31,400 recorded reviews: a boundary case for mandatory localization.
Spending on this specific intervention is not disclosed in the corpus.
Budget · Translation scenario: $0.10 per source word / language
A published Loten Labs rate provides a planning reference. At that assumed per-language rate, 1,000 words in one language = $100; 10,000 words in four = $4,000. Confirm the language pair and service scope with a quote.
Initial internal labour: €300–€6001–2 person-days × your chosen rate. Editorial first-pass estimate: String inventory and a pilot review, excluding full implementation. External purchases are separate.
Translation-only planning calculator
$100 translation scenario
Rate is editable. No currency conversion or addition to the EUR labour line; integration, QA, voice and tax are excluded.
How to check whether it helped
Record the localized build/date, language-specific feedback and qualified visits. Compare equivalent campaign periods; a before/after change is not automatically a translation effect.
These are supplier-based scenarios, not the budgets of the example games. Integration, linguistic QA, audio, tax and future text changes require their own scope.
Screenshots: make the gallery explain the experience
What could be wrong
A short or repetitive gallery may leave progression, interface or the main activity unclear. Eight images are a comparison threshold, not evidence that a visitor understands them.
How to investigate and improve it
- List the distinct things a player should understand: core loop, progression, environments and supported social play.
- Capture representative moments from the current build. Replace repeated scenes before adding filler to reach a number.
- Ask unfamiliar target players to describe the game from the gallery. Revise the scenes they misread.
What other games tend to show
At least eight screenshots appear on 82/100 high-review pages, 59/100 low-review pages and 89/100 survival pages. Their quality and effect on conversion were not measured.
Seven screenshots in the snapshot despite 264,694 recorded reviews.
Spending on this specific intervention is not disclosed in the corpus.
Twelve screenshots coexist with 32 reviews. Asset count did not establish reach.
Spending on this specific intervention is not disclosed in the corpus.
Budget · Internal first pass
No new vendor purchase is required for an internal recapture. The team's time is still a cost. New environments or gameplay features are outside this first pass.
Initial internal labour: €150–€3000.5–1 person-days × your chosen rate. Editorial first-pass estimate: Capture and gallery review using an existing playable build. External purchases are separate.
How to check whether it helped
Use a comprehension test first, then review qualified store visits and wishlist activity over comparable periods. Log which images changed.
Do not build new game content just to fill the gallery. A count cannot explain sales.
Achievements: support the loop when the implementation earns its cost
What could be wrong
An absent feature is a completeness question, not proof of a launch weakness. Generic achievements can consume development and QA time without clarifying the appeal.
How to investigate and improve it
- Decide whether milestones fit the game's progression and the audience's expectations.
- Scope a small implementation spike: unlock rules, API integration, save migration and regression checks.
- Ship only milestones that can be tested reliably. Keep this behind missing gameplay clarity or measured reach issues.
What other games tend to show
Achievements appear on 93/100 high-review pages and 68/100 low-review pages. They are frequent, but several highly visible games omit them.
No achievements recorded, with 31,400 reviews.
Spending on this specific intervention is not disclosed in the corpus.
No achievements recorded in the August snapshot, with 540,159 reviews.
Spending on this specific intervention is not disclosed in the corpus.
Budget · Internal first pass
Start with internal discovery. Budget the actual unlock systems, platform integration and QA from the game's architecture instead of borrowing an industry flat price.
Initial internal labour: €150–€3000.5–1 person-days × your chosen rate. Editorial first-pass estimate: Design and integration spike; shipping scope still needs an engineering estimate. External purchases are separate.
How to check whether it helped
Check unlock correctness and player feedback after release. Do not infer extra purchases from the feature's presence.
The first-pass labour estimate excludes full engineering and regression work. Absence can be a deliberate product choice.
Trailer and description: demonstrate the promise, then explain it
What could be wrong
Assets can exist without making the game legible. Presence is identical across the launch cohorts, so these observations cannot assess the quality of your opening footage or writing.
How to investigate and improve it
- Show the activity that defines the game near the beginning, using representative gameplay from the current build.
- Make the short description state what the player does and what makes that loop worth trying. Use the long description to answer the questions the footage leaves open.
- Test the opening with unfamiliar players before commissioning a complete recut. Confirm music, font and footage rights.
What other games tend to show
All 100 games on each launch side have a trailer and a long description. The useful review concerns the assets themselves, rather than buying another asset because its checkbox is empty.
A trailer and a detailed page were present in a low-review launch.
Spending on this specific intervention is not disclosed in the corpus.
Gameplay video and a detailed page are present in the high-review snapshot; their conversion effect is not isolated.
Spending on this specific intervention is not disclosed in the corpus.
Budget · Internal first pass
Use an internal capture/edit to test the premise first. For outsourcing, request a current quote. Derek Lieu's historical 2020 pricing discussion gives $3,000–$6,000 for a solo trailer maker; it is not a current market quote.
Initial internal labour: €300–€6001–2 person-days × your chosen rate. Editorial first-pass estimate: Opening-footage and copy prototype, not a finished external trailer. External purchases are separate.
Derek Lieu: historical trailer-budget discussion (figures written in 2020)How to check whether it helped
Check whether players can name the core activity and differentiator after the opening. Compare visitor-to-wishlist activity only with traffic-source context.
Full capture, animation, sound, music licences, cutdowns and revision rounds change the quote. No trailer spend is disclosed for these examples.
Release timing: remove practical conflicts before chasing a weekday
What could be wrong
A date may clash with competing releases, campaigns or support capacity. A weekday association cannot establish that your announced date is wrong.
How to investigate and improve it
- Check direct audience competition and announced events in the proposed window.
- Confirm build-review deadlines, team availability, creator access and launch support.
- If the date is flexible, compare realistic windows. Preserve commitments unless there is a game-specific reason to move.
What other games tend to show
Tuesday–Thursday accounts for 73/100 high-review games versus 48/100 low-review games. Friday is 15 versus 29. The comparison is not adjusted for audience, release age or publisher support.
A Friday release in the high-review cohort.
Spending on this specific intervention is not disclosed in the corpus.
A Friday release in the low-review cohort. Matching the weekday does not explain either outcome.
Spending on this specific intervention is not disclosed in the corpus.
Budget · Internal first pass
Changing a planning document can be internal work. A delayed launch can create substantial payroll and opportunity costs; calculate those from your actual team and delay length.
Initial internal labour: €150–€3000.5–1 person-days × your chosen rate. Editorial first-pass estimate: Calendar and readiness review. External purchases are separate.
How to check whether it helped
Track readiness, available support and campaign delivery. Compare launch activity without attributing the result solely to the weekday.
No cohort-based expected return is available for moving a date.
Early Access and updates: make the current build and next commitment clear
What could be wrong
Players may be asked to buy a future promise without a clear view of what works now. An update checkbox cannot tell you whether a roadmap is credible or a project is maintained regularly.
How to investigate and improve it
- Describe the playable scope, known limitations and next deliverable in concrete terms.
- Review a timeline of announcements against actual builds and feedback. Separate current Early Access status from the game's release history.
- Estimate one achievable milestone and support workload. Announce a cadence the team can maintain.
What other games tend to show
In the survival snapshot, 31 games are currently in Early Access and 37 have an observed Early Access date. Updates appear on 29/31 current-EA pages and 69/69 released pages; presence does not measure cadence.
Its official update 25 notes describe new onboarding options and accessibility-related changes, an example of naming concrete player-facing work.
The cost of that update is not published in this source.
Current Early Access in the August row; read it as a historical status, not today's live badge.
Spending on this specific intervention is not disclosed in the corpus.
Budget · Internal first pass
Cost the milestone as people × working days × your loaded day rate, then add testing and support. The page/roadmap review is a small first pass, not the cost of finishing a survival game.
Initial internal labour: €300–€6001–2 person-days × your chosen rate. Editorial first-pass estimate: Build-state audit and one milestone scope; development cost remains to estimate. External purchases are separate.
Valve: Early Access expectationsHow to check whether it helped
Measure whether promised work shipped, recurring player issues and response after each dated update. Do not equate more announcements with more traffic.
Released games may have used Early Access. Current status medians do not estimate the effect of choosing Early Access.
DLC: validate a reason to return before funding more scope
What could be wrong
Later content can be mistaken for a launch requirement. Building paid content before understanding the base game's audience can expand costs without resolving the original issue.
How to investigate and improve it
- Identify a player demand or retained audience that a specific addition serves.
- Scope content, technical dependencies, QA, pricing and a communication plan.
- Compare that investment with improving the base game or running a smaller update. Make the decision from your own audience and cost evidence.
What other games tend to show
DLC appears on 75/100 high-review pages versus 16/100 low-review pages, and on 15/31 current-EA survival pages versus 45/69 released pages. These are later product states, not launch-day interventions.
DLC is visible in the snapshot.
Spending on this specific intervention is not disclosed in the corpus.
No DLC recorded despite a high review count.
Spending on this specific intervention is not disclosed in the corpus.
Budget · Internal first pass
Use the team's actual workload estimate. A generic DLC euro range would hide differences in art, levels, engineering and support; the corpus has no comparable production costs.
Initial internal labour: €300–€6001–2 person-days × your chosen rate. Editorial first-pass estimate: Demand check and a costed content brief. External purchases are separate.
How to check whether it helped
Use retained-player demand, uptake of the addition and support load. Track base-game effects separately from DLC sales.
No observed DLC difference proves that commissioning DLC will create an audience.
Demo and festivals: verify the player experience before committing a campaign
What could be wrong
A currently listed demo is not a history of pre-launch testing. Festival participation without a ready, compelling build can leave the team with activity that does not answer the audience question.
How to investigate and improve it
- Distinguish a historical demo, a currently available demo and participation in a particular event.
- Define a representative first session, test it with target players and resolve critical failures.
- Plan support, follow-up and measurement around the event; decide from the build and audience evidence rather than a cohort checkbox.
What other games tend to show
A demo remains listed on 19/100 low-review pages and 12/100 high-review pages. Next Fest is unobserved on 89 and 91 rows. Neither difference measures the value of a demo or festival.
The case describes a historical demo; current availability is a different variable.
Spending on this specific intervention is not disclosed in the corpus.
An observed festival and demo coexist with low review activity.
Spending on this specific intervention is not disclosed in the corpus.
Budget · Internal first pass
Cost actual build work, QA, support and any event-specific expenses. The research does not contain comparable full demo budgets.
Initial internal labour: €300–€6001–2 person-days × your chosen rate. Editorial first-pass estimate: Demo readiness audit and a test scope, excluding demo production. External purchases are separate.
How to check whether it helped
Check session completion, blockers and willingness to continue. Record festival dates and wishlist movement without claiming the event caused all of it.
An unobserved festival is not a no, and a missing current demo is not proof that the game never offered one.
Low visibility: test qualified exposure before buying a large campaign
What could be wrong
A polished, well-liked game can still have little recorded activity. Public reviews cannot separate too few visits, the wrong visitors or poor conversion.
How to investigate and improve it
- Collect the Steam traffic and campaign evidence you can access; write the audience hypothesis first.
- Build a small creator shortlist around actual gameplay and a playable build. Use dated UTM links for the experiment.
- Compare qualified visits and wishlists before scaling spend. Negotiate sponsorships by deliverables, audience fit and usage rights.
What other games tend to show
Favorable reception occurs in both outcome groups, and creator history is missing on 95/100 rows on each side. The corpus cannot give a creator-campaign success rate or a standard sponsorship price.
A highly positive low-volume row shows why reception does not establish reach.
Spending on this specific intervention is not disclosed in the corpus.
A complete page coexists with a low review count; the detailed case supplies context missing from the aggregate.
Spending on this specific intervention is not disclosed in the corpus.
Budget · Internal first pass
Start with research and direct outreach time. Paid placements need channel-specific quotes; spend is not disclosed consistently enough to publish a market average here. Do not treat free outreach as free labour.
Initial internal labour: €300–€6001–2 person-days × your chosen rate. Editorial first-pass estimate: Audience hypothesis, a small shortlist and measurement setup. External purchases are separate.
Valve: UTM analyticsHow to check whether it helped
Compare the same campaign's qualified visits, attributable wishlists and subsequent activity. Steam UTM only attributes eligible tracked visits and can miss cross-device behavior.
No reach diagnosis or expected return is inferred solely from review count. Set any paid test cap from your own budget.
Missing data: instrument the next decision instead of inventing a diagnosis
What could be wrong
Unknown traffic, wishlists or campaign history cannot be treated as zero. Raw event counts also cannot explain a visitor's motive.
How to investigate and improve it
- Name the decision and the observations that would separate competing explanations.
- Collect permitted aggregate metrics and a dated intervention log. Keep source, denominator and missing values visible.
- Compare like-for-like windows and audience segments. Inspect the last observed errors and transitions before proposing a fix.
What other games tend to show
Only three wishlist declarations appear across the 200 launch rows. The snapshot cannot fill Valve's private counter or reconstruct most creator histories.
A documented public wishlist declaration exists; it is a sourced exception, not an estimate for other games.
Spending on this specific intervention is not disclosed in the corpus.
Budget · Internal first pass
Use available Steamworks reporting rather than buying a guessed counter. Tool subscription, engineering and consent implementation costs depend on the chosen setup.
Initial internal labour: €150–€3000.5–1 person-days × your chosen rate. Editorial first-pass estimate: Campaign links and a minimal measurement plan. External purchases are separate.
Valve: UTM analyticsHow to check whether it helped
Check the report's actual denominator, privacy limitations and data freshness. A missing value should remain unknown.
Shypiard's PostHog reports describe usage of Shypiard. They are not access to a studio's private Steam traffic or wishlist totals.
METHOD & SOURCES
How to read and reproduce this study.
This is a purposive comparison of two groups of 100 games, selected by observed review volume. It is not a random sample of Steam and cannot estimate a studio's probability of success. The low-review group contains 99 games with 10–49 reviews and one explicitly retained commercial-failure case, Where the Water Tastes Like Wine, with 1,740 reviews. The high-review group contains games with at least 1,000 reviews.
Store pages, Steam APIs, SteamSpy and SteamCharts were recorded on 17 August 2026. Review counts are all-language counts as recorded in each CSV; the source column identifies the provider. Release years, genres, budgets and time on sale differ. Store features were observed after release, so they may reflect growth as well as precede it. No adjustment for those confounders is applied.
The separate survival corpus is excluded. Empty and unobserved fields stay missing. Review volume measures recorded reception activity, not units sold, profit, page conversion or wishlist stock. Source URLs and observation windows are available in each downloadable row.
This article reanalyses the dated Shypiard corpus. It does not add newly collected games or update the observation date. Editorial revision: 12 October 2026.
CONTINUE THE RESEARCH
FROM RESEARCH TO YOUR GAME
Find the next question worth answering.
Add your game to get a first analysis of its public page and relevant comparisons. Use the research as context for your own launch.