Supporting research

Observed genre performance with visible assumptions.

Emergence Interactive analyzed public Steam data to establish historical reference points for Open World Survival Craft games. The results provide decision-relevant context for Endonaut’s commercial scenarios; they are not a forecast of Endonaut’s sales.

Modeled estimates—not publisher-reported sales. Unless explicitly stated otherwise, every sales figure on this page is derived from public review activity using a configurable review-to-sales model. Public-data coverage varies by title, and the estimates should be interpreted as approximate performance indicators.

A bottom-up benchmark, not a top-down forecast.

Broad total-addressable-market calculations often begin with industry revenue and then assign a hypothetical market share to a future product. For an early-stage game, that approach can conceal the assumptions that matter most. Emergence Interactive instead examined the observed historical performance of Steam titles carrying the Open World Survival Craft tag.

The analysis asks where selected unit-sales scenarios would fall within the analyzed dataset and how unevenly performance is distributed. It helps evaluate the scale of different commercial outcomes and provides context for planning. It does not establish demand for Endonaut or the probability that Endonaut will reach any threshold.

968 tagged titles. 606 usable estimates.

The collection process identified 968 unique Steam titles associated with the Open World Survival Craft tag. Of these, 606 had sufficient public information to produce usable cumulative-sales estimates. Those 606 titles form the evidence-qualified sample used for the primary percentile calculations.

The remaining 362 records lacked usable cumulative-sales estimates. They may include unreleased titles, titles with insufficient review activity, or records with incomplete public data. They are addressed separately in the sensitivity analysis rather than silently treated as equivalent to the evidence-qualified titles.

Steam tags overlap, are applied broadly, and can change. The dataset therefore includes some genre-adjacent games and should not be interpreted as a definitive census of all Open World Survival Craft products.

IDENTIFIED / 968

Steam-tagged titles

The complete tagged set used to define the analysis universe.

QUALIFIED / 606

Titles with usable estimates

The primary evidence-qualified sample used for observed percentiles.

SENSITIVITY / 362

Titles without usable estimates

Retained in a separate conservative test of exclusion effects.

Public reviews become an estimated sales range.

The analysis used public Steam metadata and review timestamps. Because most publishers do not report title-level unit sales, a configurable review-to-sales model based on the Boxleiter approach was used to estimate daily and cumulative sales.

The model begins with a base review-to-sales multiplier, then supports adjustments for genre, current USD price, and time since release. In this run, the Open World Survival Craft label did not match a configured genre override and therefore used the neutral genre factor of 1.0. Confidence-band multipliers produce low, midpoint, and high modeled estimates. The percentile results below use the final cumulative midpoint estimate for each title.

1. Collect public observations

Identify tagged games, retrieve public Steam metadata, and organize review timestamps by date.

2. Apply model factors

Convert review activity into sales estimates using the configured base, genre, price, and release-age factors.

3. Produce estimate bands

Calculate low, midpoint, and high daily estimates, then accumulate them across the observed history of each title.

What the model can—and cannot—say. Review-derived estimates make comparison possible when reported sales are unavailable, but they do not turn public review activity into audited sales records. Differences in player behavior, review rates, pricing history, regional sales, bundles, promotions, free access, and data availability can all affect an individual title’s estimate.

The assumptions for this analytical run.

For each day of observed reviews, the midpoint review-to-sales multiplier was calculated as 40 × genre factor × price factor × release-age factor. Daily modeled sales were then accumulated for each title. These values are analytical assumptions, not empirically verified conversion rates for the individual games.

The exact machine-readable configuration is available as assumptions.json.

Base and genre factors

Base multiplier: 40 estimated sales per review.

Genre factor for this run: 1.0 (neutral). The configuration contains overrides for several other genre labels, but none matched Open World Survival Craft.

Current-price factor

Free: 0.4
$0.01–$9.99: 0.7
$10.00–$19.99: 1.0
$20.00–$39.99: 1.2
$40.00 and above: 1.4

The script uses the current USD price retrieved from Steam, not a title’s historical price on each review date.

Release-age factor

Days 0–30: 0.6
Days 31–180: 0.8
Days 181–730: 1.0
After day 730: 1.2

The factor is evaluated for each review date based on the number of days since release.

Estimate bands

Low: 0.75 × midpoint
Midpoint: 1.0 × midpoint
High: 1.3 × midpoint

Thresholds and percentiles on this page use the cumulative midpoint estimate.

Reference points within the 606-title sample.

A percentile describes placement within this evidence-qualified sample: approximately 68.8% of the analyzed titles with usable estimates were at or below 22,900 estimated units, for example. It does not describe Endonaut’s likelihood of reaching that outcome.

≈68.8TH PERCENTILE

22,900 units

The seed-recovery reference used in the business plan.

≈75TH PERCENTILE

50,000 units

Approximately $1 million in gross base-game revenue at an assumed $19.99 full price.

≈82.3RD PERCENTILE

100,000 units

A higher commercial-performance reference point.

≈87.0TH PERCENTILE

250,000 units

A strong outcome within the evidence-qualified sample.

37 OF 606 TITLES

At least 1 million units

Approximately 6.1% reached the threshold using modeled midpoint estimates.

What changes when all 968 titles are counted?

Excluding 362 titles without usable estimates can affect percentile placement. To test that effect, the analysis assigned each excluded title a nominal value of one unit—below the lowest observed sales range—and recalculated the percentiles across all 968 titles.

This deliberately conservative assumption places every unestimated title below each evaluated threshold. The number of titles at or below a threshold therefore increases, moving the same unit-sales outcome to a higher percentile. The two views answer different questions, so both are reported.

22,900 estimated units

Evidence-qualified sample: approximately 68.8th percentile.
968-title sensitivity case: 80.48th percentile, or approximately 80.5th.

100,000 estimated units

Evidence-qualified sample: approximately 82.3rd percentile.
968-title sensitivity case: approximately 88.9th percentile.

250,000 estimated units

Evidence-qualified sample: approximately 87.0th percentile.
968-title sensitivity case: approximately 91.8th percentile.

A bound, not a claim about the missing titles. Assigning one unit does not assert that every excluded title sold one copy. It tests the most threshold-lowering treatment of the missing group. The primary sample is better supported by public evidence; the sensitivity case shows how strongly the exclusions could affect placement under this conservative assumption.

37 titles in the analyzed dataset.

The following are the titles in the analyzed dataset that reached the million-unit threshold using the model’s cumulative midpoint estimate. The list follows the output order, from the highest modeled midpoint estimate downward.

It is not a definitive ranking of all Open World Survival Craft games. Steam-tag overlap places recognizable core examples such as Rust, ARK: Survival Evolved, Valheim, Subnautica, Satisfactory, and The Forest beside broader or genre-adjacent titles.

Output positions 1–13

  1. Rust (App 252490)
  2. Terraria (App 105600)
  3. ARK: Survival Evolved (App 346110)
  4. No Man’s Sky (App 275850)
  5. Project Zomboid (App 108600)
  6. The Forest (App 242760)
  7. Don’t Starve Together (App 322330)
  8. Dying Light (App 239140)
  9. Palworld (App 1623730)
  10. Valheim (App 892970)
  11. New World: Aeternum (App 1063730)
  12. 7 Days to Die (App 251570)
  13. Raft (App 648800)

Output positions 14–25

  1. Subnautica (App 264710)
  2. Satisfactory (App 526870)
  3. Starbound (App 211820)
  4. Days Gone (App 1259420)
  5. ASTRONEER (App 361420)
  6. Don’t Starve (App 219740)
  7. The Long Dark (App 305620)
  8. Conan Exiles (App 440900)
  9. Grounded (App 962130)
  10. SCUM (App 513710)
  11. Green Hell (App 815370)
  12. Subnautica: Below Zero (App 848450)

Output positions 26–37

  1. Dune: Awakening (App 1172710)
  2. Stormworks: Build and Rescue (App 573090)
  3. Once Human (App 2139460)
  4. The Planet Crafter (App 1284190)
  5. Core Keeper (App 1621690)
  6. Abiotic Factor (App 427410)
  7. ICARUS (App 1149460)
  8. Forager (App 751780)
  9. Stoneshard (App 625960)
  10. Far Cry Primal (App 371660)
  11. Miscreated (App 299740)
  12. Hurtworld (App 393420)

Reference points, with uncertainty retained.

This analysis is useful because it makes the comparison set and assumptions inspectable. Its limitations are part of the result—not qualifications to be hidden after the conclusion.

01

Historical performance is not predictive

Comparable-title results show what occurred in the analyzed dataset. They do not establish the probability of any outcome for Endonaut.

02

Sales are estimated

Review-derived unit sales are modeled estimates, not publisher reports, platform statements, or audited revenue records.

03

Classification is uncertain

Steam tags overlap and can include products that are only partly comparable to Endonaut or to one another.

04

Coverage varies

Titles differ in release history, review activity, pricing, promotions, platform exposure, and completeness of public data.

05

Performance is hit-driven

The observed distribution is highly uneven. A small number of exceptional outcomes coexist with many titles that achieved far smaller estimated sales.

06

The purpose is scenario context

The analysis provides observed reference points for evaluating commercial scenarios while preserving the distinction between a benchmark and a forecast.

A dated analytical snapshot.

The values on this page describe a specific analysis run. Steam tags, review activity, public metadata, and model settings can change; later runs may therefore produce different counts, estimates, and percentile placements.

The model configuration used for this run can be downloaded as JSON. The underlying analysis output and code are not currently available through a public repository. If they are published later, this section is the intended location for the permanent link and any accompanying methodological notes.

ANALYSIS DATE

March 1, 2026

The date this analytical snapshot was run.

DATASET ANALYSIS

rank_gamesV4

Script identifier used for validation, thresholds, percentiles, and sensitivity calculations.

SALES MODEL

salesModel3

Script identifier used for Steam collection and review-derived sales estimates.