Steam Launch Lever Panel — 2026 (10,403 launches, public-source-derived week-1 revenue)

10,403 Steam launches with week-1 gross revenue estimated from public signals, plus the storefront features used to estimate causal lever effects: demo availability, co-op/multiplayer support, and localization breadth. 19 columns, no nulls. Every revenue figure carries a per-row source_quality tier (B/C/D) recording how it was derived. IMPORTANT — survivorship bias: games that sold almost nothing leave little owner-count or review trace, so no revenue can be derived for them and they drop out of the denominator. The median week-1 revenue in this file is NOT 'the median Steam game' — it describes the median MEASURABLE launch. Use this data for relative comparisons (how much a lever moves revenue), not for absolute 'typical game earns X' claims. Canonical analysis: https://steamforecast.app/reports/steam-launch-lever-benchmarks Methodology: https://steamforecast.app/methodology Derived from public Steam storefront data, public review counts, and SteamSpy owner ranges. No licensed, purchased, or partner sales data is included; wishlist counts are absent by construction. Compiled by SteamForecast. License CC BY 4.0.本数据集包含10403款Steam平台游戏的首发数据，其首周总营收通过公开可获取指标估算得到，同时附带用于估算因果杠杆效应的商店页面特征：试玩版可用性、合作/多人游玩支持以及本地化覆盖范围。该数据集共含19列数据，无空值。每一条营收数据均附带一行来源质量层级（source_quality tier）标注（B/C/D），用以记录该营收数据的生成方式。 重要提示——存在幸存者偏差：几乎未产生销量的游戏几乎不会留下拥有量或评论痕迹，因此无法为其估算营收，此类游戏也会被排除在统计分母之外。本文件中的首周营收中位数并非“Steam平台游戏的营收中位数”，而是“可被统计的首发游戏的营收中位数”。本数据集仅可用于相对比较（即某一驱动因素对营收的影响幅度），不可用于“典型游戏营收为X”这类绝对断言。 标准分析报告：https://steamforecast.app/reports/steam-launch-lever-benchmarks 方法论说明：https://steamforecast.app/methodology 本数据集源自公开的Steam商店页面数据、公开评论数以及SteamSpy的游戏拥有量区间数据，未包含授权采购、合作销售等营收数据；因数据集构建规则，未纳入愿望单数量数据。本数据集由SteamForecast汇编，采用知识共享署名4.0国际许可协议（CC BY 4.0）。