初创公司尽调
尽调报告 Developer tools / product experimentation / analytics Series C / acquisition announced 2026-05-28

Statsig

产品平台强,但独立投资逻辑已被 OpenAI 交易基本终结

Statsig 做出了可信的一体化实验平台,但已宣布的 OpenAI 交易和仍未公开的经济性,让新资金应选择回避。

封面要素

最近披露估值 / 交易价值 01
1100 USD M [CV001, CV014]
成立时间 06
2021-02 [CO001]

公司概况

Statsig 是一家私营开发者工具公司,Vijaye Raji 于 2021 年围绕一个判断创办:产品团队应该在同一个实验控制平面里发布、衡量和学习。公开证据支撑一个覆盖功能开关、实验、分析、会话回放和仓库原生工作流的宽栈,也支撑有意义的客户采用和 2025 年 $1.1B 估值锚点。主要保留项是,业务很快就有了战略归属:OpenAI 于 2025 年 9 月宣布 $1.1B 交易,而后续公开材料指向 Amplitude 运营该平台和客户基础,使独立投资逻辑基本关闭,也让长期所有权结构仅凭公开证据更难承保。

官网
www.statsig.com
成立时间
2021-02-01
创始人
Vijaye Raji
创立地点
Kirkland / Bellevue, Washington, USA
总部
Bellevue, Washington, USA
产品
Statsig 销售按用量计费的平台,把功能开关、动态配置、实验、产品分析、会话回放和仓库原生分析,与广泛 SDK 覆盖和企业级控制结合在一起。
客户
以工程驱动的产品团队、数据科学家、增长团队,以及希望用一套集成发布与度量栈的大型软件组织。
商业模式
按用量计费的 SaaS,覆盖自助和企业合同,并为治理要求更高的大型环境提供仓库原生部署。
阶段
Series C company with OpenAI acquisition announced in 2025
融资情况
公开披露融资约 $153.4M,横跨 Series A、B 和 2025 年 5 月以 $1.1B 估值完成的 $100M Series C;随后 OpenAI 于 2025 年 9 月宣布已签署 $1.1B 交易。
[CO001, CO005, CO012, CO026, CO028, CO044, CE001, CV014]

执行摘要

主要优势

  • 产品栈覆盖功能开关、实验、分析、回放和仓库原生工作流,集成度较高。
  • AI、生产力、金融科技、商业和 marketplace 用例都有强命名客户证据。
  • 顶级投资人多轮背书,且已经兑现 $1.1B 战略价格锚点。

主要风险

  • OpenAI 交易之后,独立上行空间基本关闭;公开材料现在也分散在 OpenAI 和 Amplitude 两边。
  • 公开来源仍未清晰披露留存、利润率、集中度或交易完成后的经济性,无法支撑有纪律的承销。
  • 竞争压缩和仓库原生实施复杂度,可能削弱定价权和 TCO 假设。

未决问题

  • OpenAI 与 Amplitude 之间,交易完成后的最终所有权、支持和路线图责任划分。
  • NRR、毛利率、CAC / 回本周期、头部客户集中度和合同期限数据。
  • 当前运营结构下,有代表性的仓库原生成本模型,以及受监管客户的使用证据。

目录

Chapter 01

01公司概况

1.1 身份定位、平台边界与 Bellevue 运营足迹

Statsig 把自己定位为现代产品开发平台,而不是单点 A/B 测试工具。从官网、文档、定价到产品页,公司都在推一个集成栈:实验、功能管理、产品分析、会话回放、Web 分析、仓库原生分析,以及配套发布工具。这个定位重要,因为它解释了定价模型和融资故事:Statsig 试图用一个按量计费的平台,替代一组更窄的开发者和分析产品。官方定价页用与用量而非席位数绑定的 Free、Pro、Enterprise 层级,进一步强化了这个逻辑。 当前地址信号指向 Bellevue,而不只是更宽泛的 Seattle 地区。Statsig 招聘材料、产品页和收购博客都以 “Hello from Bellevue, WA” 收尾;GeekWire 报道,2025 年 7 月公司迁入新的 Bellevue 总部,可容纳 450 至 500 人。不过,地址记录并不完全干净。Statsig 的 2021 年 Series A 新闻稿使用 Kirkland 办公地址,Tracxn 仍保留 Kirkland 标签与 Bellevue 注册地址并存的记录。最干净的公开解读是:Statsig 现在在 Bellevue 运营,而较旧、更新更慢的记录仍反映它早期 Eastside 足迹。[CO001, CO002, CO003, CO004, CO005, CO006]

FO002: Statsig 公司快照逻辑

展示创始人经历、统一产品范围、企业客户、资本、Bellevue 运营和待完成的 OpenAI 交易如何在当前公开记录中相互连接。

[CO002, CO003, CO004, CO011, CO013, CO028]

1.2 创始人背景、领导力集中与治理可见度

Statsig 的公众身份仍异常集中在 Vijaye Raji 身上。留存记录把公司与他在 Microsoft 和 Facebook 的经历紧紧连在一起:Microsoft 资料称他是 Small Basic 的创建者;Meta 工程博客在 Statsig 成立很久之前,就介绍过他在 Seattle Messenger 团队的工作。后续报道称,他曾管理 Facebook 的 Seattle 工程组织,并担任副总裁角色,这些经历与 Statsig 创立时“快速实验、用数据指导产品发布”的投资逻辑直接对上。因此,创始人与市场匹配度格外强,也解释了为什么多家投资者和媒体画像把 Statsig 描述为创始人主导的产品公司。 但治理可见度远薄于产品可见度。最清晰的公开治理信息,是 2025 年 Series C 披露 ICONIQ Growth 合伙人 Murali Joshi 加入董事会。除此之外,本报告审阅的直接公开来源没有给出清晰且最新的董事会名单、委员会结构或详细投票权图谱。这一点重要,因为 OpenAI 交易在本已有限的私营公司治理透明度之上,又叠加了控制权变更。OpenAI 的公告明确表示,Raji 预计将出任 Applications CTO,交易一旦交割,员工和控制动态都会发生实质变化;但这不能替代完整治理材料。[CO016, CO017, CO018, CO019, CO020, CO021]

领导层和创始人表
人物角色背景创始人-市场匹配 / 职能覆盖关键人物依赖
Vijaye Raji创始人 / CEO,直至 2025 年 9 月公告前 Microsoft 工程师,创建 Small Basic;曾任 Meta/Facebook 西雅图工程负责人。直接把 Facebook 实验血统映射到 Statsig 产品逻辑。极高
Murali JoshiICONIQ Growth 合伙人 / 已披露董事会成员成长阶段投资人,随 2025 年 Series C 加入董事会。引入后期 SaaS 投资中的扩张和治理监督。
Joshi 之外的公开董事会名单保留来源未完整披露已审阅直接来源没有列出完整的现任董事会名单或委员会结构。为后续尽调留下部分治理透明度缺口。中高
OpenAI 未来控制点待定收购方和未来母公司OpenAI 表示 Raji 将出任 Applications CTO,员工可能在交割后转入 OpenAI。暗示即将发生的控制权转移,可能重塑领导连续性。

创始人身份清楚,但完整董事会和交割后运营结构并未完全公开。

[CO016, CO017, CO018, CO019, CO020, CO021]
利益相关方或投资人图谱
利益相关方角色控制 / 经济重要性尽调问题
Vijaye Raji创始人 / 运营重心产品逻辑、招聘叙事和待定 OpenAI 领导层交接背后的关键人物。要求提供留任方案、剩余持股,以及交割后授权范围。
Sequoia CapitalSeries A 和 Series B 领投方;Series C 跟投方公司早期和成长期轮次中的核心资本赞助方。要求提供董事会 / 观察员权利、清算堆叠,以及任何老股条款。
Madrona Venture Group所有已披露轮次的连续投资人重要区域支持方,从早期到独角兽轮次保持连续性。要求提供持股比例、保护性条款和任何附函。
ICONIQ Growth / Murali JoshiSeries C 领投方和已披露董事会参与者与 $1.1B 估值和正式董事会监督绑定的后期验证方。要求提供董事会材料、保留事项,以及 2025 年 5 月以来的治理节奏。
员工有望通过交易转入 OpenAI 的人才基础对产品速度,以及评估稀释、留任和整合风险至关重要。要求提供期权池细节、留任方案和当前组织架构图。
OpenAI待定收购方 / 战略交易对手如果交易交割,将成为控制权所有者;已经是具名客户,也是员工未来雇主路径。要求提供最终合并条款、监管路径和客户服务连续性计划。
企业客户收入锚点和产品证据点具名客户验证品类相关性和企业采用。要求提供客户集中度、净留存和参考客户电话名单。

利益相关方图谱强调公开来源中可见且具有经济重要性的群体,而不是完整股权结构表。

[CO019, CO020, CO023, CO025, CO027, CO028]

1.3 融资历史、估值与规模信号

Statsig 的融资路径在金额上很清楚,尽管一些运营指标仍模糊。公司于 2021 年 8 月完成 $10.4 million Series A,2022 年 4 月完成 $43 million Series B,2025 年 5 月以 $1.1 billion 估值完成 $100 million Series C。GeekWire、Tracxn 和 Sacra 的数据合计,累计融资约 $153 million 至 $153.4 million;Sequoia 与 Madrona 连续参与三轮,ICONIQ Growth 作为后期领投方加入。2025 年 Series C 看起来是拐点:Statsig 成为 Bellevue 独角兽,披露新增董事会席位,并开始大幅扩充办公室容量。 规模信号方向强,但精度不均。官方页面称拥有数千家公司、每天超过 1 trillion 事件、每月 2.5 billion 唯一实验对象。第三方和客户证明来源补充了更多细节:Sacra 估计付费客户超过 300 家、ARR 约 $40 million;官方客户页面列出 OpenAI、Microsoft、Notion、Brex、Atlassian、Bloomberg、Eventbrite、Ancestry、Headspace 和 SoundCloud 等品牌。员工数更不清楚。GeekWire 在 Series C 前后报道员工 140 至 145 人,在 2025 年 9 月 OpenAI 公告时报道 155 人;但 Tracxn 当前快照称,截至一个标注含糊的 4 月 26 日,公司有 55 名员工。差距足够大,当前员工数应在尽调中视为未解决事项。[CO022, CO023, CO024, CO025, CO026, CO027]

快照 KPI 表
指标数值 / 状态日期置信度缺口 / 备注
成立Feb 20212021-02创立博客和早期融资报道支持这一创立月份。
当前总部信号Bellevue 运营基地2025-07-15官方页脚和 GeekWire 支持 Bellevue;更早记录仍显示 Kirkland 渊源。
当前阶段已公告、待完成的 OpenAI 收购2025-09-02直接来源描述的是一项已签署、仍受交割条件约束的交易,而不是明显已经完成的交割。
最新估值(USD M)11002025-05-06Series C 估值得到 Business Wire、GeekWire、Reuters 和 OpenAI 交易报道的交叉印证。
累计融资(USD M)153.42025-05-06公开轮次摘要集中在 $153M 到 $153.4M。
ARR 信号(USD M)402025-05-06GeekWire 报道 ARR 为 $40M,Sacra 也估计在同一水平;没有公开审计财务数据。
付费客户信号300+ 付费2025-05Sacra 估计低于官方“数千家公司”的说法,暗示付费客户和更广泛用户混在一起。
员工数信号55 与 140-155 冲突2025-2026当前员工数仍未解决,因为 Tracxn 与多篇 2025 年新闻报道相冲突。

后续章节使用的核心身份和规模标记。若为数值,货币金额均为百万美元。

[CO001, CO005, CO012, CO013, CO026, CO028]
FO003: 快照 KPI

Statsig 的核心运营和融资指标,重点标出公开证据哪里扎实、哪里仍有冲突。

ARR 和当前员工数不是经审计的公司披露;只能作为方向性公开信号使用。

[CO026, CO028, CO029, CO030, CO031, CO039]

1.4 里程碑时间线、交易状态与尽调保留项

里程碑记录显示,这家公司从创立到产品扩张推进很快。Statsig 2021 年 3 月发布创立逻辑,同年 8 月宣布正式可用并完成 Sequoia 领投的 Series A,2022 年 4 月前完成 Series B,2023 年 6 月推出 Warehouse Native。同一 2023 年记录已经显示数百家付费客户、数千名免费层用户,以及用于测试模型成本和延迟的特定 AI 实验功能。到 2025 年 5 月和 7 月,公开报道显示其完成 $1.1 billion Series C、搬入更大的 Bellevue 总部,招聘计划也暗示仍在独立扩张。 最清晰的保留项是交易状态。Statsig 和 OpenAI 均在 2025 年 9 月 2 日表示双方已签署最终协议;OpenAI 与 Reuters 将该交易描述为仍需监管批准和满足惯常交割条件。然而 Tracxn 已将公司标记为已收购。这个不一致意味着,直接来源支持“已宣布的交易”,但表面上不能证明交易已完全完成。第二个保留项是证据质量不对称:产品和融资可见度相对较好,董事会构成、最终交割确认和当前员工数仍较弱。安全可见度居中。Statsig 发布了信任页面,包含 SOC 2、事件响应和治理控制表述;UpGuard 也提供了独立供应商风险视角。但两类来源都没有给出投资者在把可靠性视为已充分承保前会想要的详细运营历史。[CO034, CO035, CO036, CO037, CO038, CO044]

里程碑表
日期事件类型金额 / 估值 / 状态参与方含义
2021-02Statsig 成立创立公司成立Vijaye Raji 和前 Facebook 同事设定 Facebook 到 Statsig 实验叙事的起点。
2021-03-17创立逻辑发布创立创立缘由博文Statsig在永久的一手来源中明确产品理念。
2021-08-05宣布 Series A 和正式可用融资$10.4MSequoia、Madrona、天使投资人联盟验证早期投资人信心和商业发布。
2022-04-20宣布 Series B融资$43MSequoia、Madrona显示 Series A 后约 8 个月内快速跟进融资。
2023-06-14推出 Warehouse Native产品发布Statsig把平台扩展到客户仓库部署和治理敏感用例。
2023-06-14AI 实验功能被公开强调产品发布 / 扩张Statsig、AI 客户显示公司较早围绕模型成本、延迟和 AI 产品评估定位。
2025-05-06Series C 和独角兽估值上调融资$100M,估值 $1.1BICONIQ Growth、Sequoia、Madrona创造最清晰的公开后期估值基准,并加入已披露董事会监督。
2025-07-15宣布 Bellevue 总部扩张规模可容纳 450-500 人的总部Statsig暗示积极招聘目标和强线下运营模式。
2025-09-02宣布加入 OpenAI 的最终协议治理$1.1B 全股票;待交割Statsig、OpenAI引入控制权变更动态,以及 Raji 未来领导岗位迁移。
2026-01发布 AI 工作流和知识图谱文章产品路线图执行Statsig暗示 OpenAI 交易公告后仍在继续产品投入。
2026-05-28公开尽调快照仍显示交割和员工数信号未解决反向待交割表述对比已收购标签;55 对比 140-155 名员工OpenAI、Reuters、GeekWire、Tracxn在承保当前所有权、组织规模或交易完成状态前,需要直接尽调。

该时间线捕捉后续章节需要的主要公开里程碑;并非完整产品更新日志。

[CO001, CO022, CO024, CO026, CO035, CO037]
FO001: Statsig 里程碑时间线

这些精选里程碑显示,Statsig 如何从 2021 年由前 Facebook 团队带出的实验理念,走到 2025 年已签署的 OpenAI 交易;2026 年路线图仍在延续,但交割细节未解。

[CO001, CO002, CO020, CO022, CO024, CO026]

1.5 展示材料

Chapter 02

02市场分析

2.1 市场边界、纳入支出与品类融合

Statsig 处在一个融合的产品开发工具市场,而不是某个清晰的传统品类。核心任务是发布控制加因果度量:功能开关、渐进式交付、A/B 测试、留出组、指标定义和发布后分析。Forrester 明确把功能管理和实验界定为横跨软件交付与产品管理;Azure、AWS 和 PostHog 文档也显示,同一个控制平面用于 beta 访问、暗发布、安全灰度发布、远程配置和实验定向。产品分析属于相邻支出池,因为现代供应商越来越把分析与实验、功能开关打包在一起,但不能未经调整就简单并入核心实验 TAM。因此,合理市场边界应包括实验基础设施、功能管理、仓库原生实验和产品分析邻近项,同时排除通用 BI、后台分析和无关营销技术测试。主要现状替代品是自研功能开关、内部实验工具,以及把发布控制和度量分开的碎片化栈。[CM001, CM002, CM003, CM004, CM005, CM010]

市场定义表
细分 / 品类纳入支出排除支出买方 / 付款方与 Statsig 的相关性
功能管理 / 渐进式交付功能开关、发布控制、熔断开关、定向、远程配置、暗发布没有运行时控制的通用 CI/CD 工具平台工程、应用工程、开发者工具预算核心控制平面切入点,帮助产品安全进入生产环境
实验 / A/B 测试分配、留置组、指标计算、因果读数、实验治理纯网页 CRO 机构和手工电子表格分析产品团队、增长团队、有数据科学支持的产品组织核心因果测量层,在功能开关落地后扩张
仓库原生实验使用治理指标,基于 Snowflake、BigQuery、Databricks 或 Redshift 数据做实验分析没有发布控制或实验工作流的独立仓库数据平台、分析工程、重视安全的企业买家当指标可信度和数据驻留重要时,这是高价值架构切入点
产品分析邻近领域事件分析、漏斗、留存、旅程分析、会话理解、仓库配对通用 BI、财务报告、无关营销技术分析产品、增长、分析和数字团队重要邻近池,因为许多厂商把分析和实验打包销售
AI 产品迭代 / 发布分析AI 驱动体验、发布健康度和上线后影响的测量与产品发布决策无关的模型训练基础设施AI 产品团队、平台团队、创新负责人增长中的邻近领域,因为 AI 提高发布频率,也提高受控测量需求
现状替代方案自研功能开关、内部测试工具、碎片化分析栈、手工发布决策N/A使用更便宜或已有工具的同一批职能买家真实采购替代方案,也往往是切换摩擦的主要来源

边界表把核心 FM&E 控制平面与邻近分析和 AI 迭代预算分开,避免后续测算把每一种产品数据工具都重复计入。

[CM001, CM002, CM003, CM004, CM005, CM011]
FM001: 市场规模测算视角

公开视角堆叠不可相加:狭义测试工具位于更广义的实验软件之下,产品分析则是相邻上限,而不是干净的 TAM 层。

这些层是相邻公开视角,不是嵌套的 TAM/SAM/SOM 桶。图中有意加入边界逻辑,避免读者把每个估计都当成同一个市场。

[CM006, CM007, CM008, CM009, CM010, CM013]

2.2 规模测算视角、市场区间与仓库原生切口

公开来源支持多个可用市场视角,但没有一个权威、适用于 Statsig 的 TAM。在更宽的相邻市场侧,产品分析是 USD 13 billion 至 USD 15 billion 市场,具体取决于来源和年份,Grand View 与 Mordor 都指向双位数增长。在核心实验侧,已发布区间宽得多:Credence 将 2024 年 A/B 测试软件定为 USD 8.18 billion,VWO 引用的 A/B 测试工具估计则窄得多,为 USD 850.2 million。这些并非简单矛盾,更像不同品类外壳。它们框出市场中纯测试工具与更广义实验软件之间的占比。仓库原生实验是该市场内的重要切口,但公开数据没有把它单独作为一个品类测算。Statsig、LaunchDarkly、Optimizely、Harness、Eppo 和 GrowthBook 都在营销仓库原生工作流,把指标、SQL 和实验分析放得更靠近 Snowflake、BigQuery、Databricks 或 Redshift。供应商趋同说明买方在数据治理和指标信任上有拉力;但缺乏独立分拆意味着,仓库原生应被视为一种架构部署模型,而不是独立 TAM。[CM006, CM007, CM008, CM009, CM010, CM013]

TAM/SAM/SOM 或测算视角表
发布方年份地理范围数值CAGR / 增长方法论 / 视角置信度局限
Grand View Research – 产品分析2023全球USD 14.81B19.8% CAGR (2024-2030)覆盖广泛部署用例的邻近产品分析品类测算不是纯实验或功能管理市场
Mordor Intelligence – 产品分析2026全球USD 13.04B14.55% CAGR (2026-2031)当前产品分析品类视角邻近品类;日期基准不同于 Grand View
Credence Research – A/B 测试软件2024全球USD 8.18B15.45% CAGR (2024-2032)更宽的实验软件视角可能包含功能开关和产品实验之外的更多内容
VWO / 引用研究 – A/B 测试工具2024全球USD 0.8502B14.0% CAGR (2024-2031)狭义工具型 A/B 测试视角竞争对手博客上的二级摘要,边界明显更窄
Grand View Research – 产品分析区域组合2023北美>37% 份额N/A产品分析市场的区域份额份额,不是绝对的 Statsig 就绪 SAM
Credence + Mordor 区域读数2024-2026北美领先 / APAC 增长更快定性区域排名N/A来自实验和分析报告的跨来源区域综合仅具方向性,因为来源使用不同品类外壳
分析师综合 – Statsig 相关公开区间2024-2026全球公开视角组合核心: USD 0.85B-8.18B; 邻近: USD 13.04B-14.81B所有已发布视角均为两位数增长受边界约束的公开代理,而非一个干净的 TAM/SAM/SOM非可加总区间,且没有公开的 Statsig 专属 SAM 或 SOM

综合行有意不做加总:它保留狭义核心实验和更宽的分析邻近估计,但不假装这些品类可以直接叠加。

[CM006, CM007, CM008, CM009, CM010, CM013]
FM002: 市场估计区间

公开估计因边界不同而差异很大,但都指向围绕可控产品变更和度量的、有意义且增长中的市场。

各行把不同相邻类别统一到同一个单位(十亿美元)。它们适合作为估值括号,而不是单一权威 TAM。

[CM006, CM007, CM008, CM009, CM010, CM060]

2.3 买方、用户、付款方与采用路径

买方地图横跨产品工程、产品管理、增长和数据。LeadDev 的买方指南把重心锚在持续发布功能的工程团队和产品经理。Firebase 把范围扩到绑定收入和留存的产品及营销实验;PostHog 和 Statsig 都强调,数据团队仍然是度量、指标定义和实验分析的核心。MIT 和 Harvard 增加了审批层:当实验规模化,它会变成业务系统决策,涉及高层领导、数据科学利益相关方和高管赞助人;他们要的是决策质量和更快创新,而不只是一个新 SDK。实践中,采用路径通常从功能开关和受控灰度发布开始,再加入实验定向、指标治理和分析,最后扩展到平台标准化或仓库原生治理。这意味着用户常常是工程师、PM、分析师、增长经理或数据科学家;付款方则会随发布范围,在平台工程、产品运营、数据基础设施或高管数字化转型预算之间切换。[CM023, CM025, CM026, CM027, CM028, CM029]

细分 / 买方图谱
细分场景买方用户付费方工作流采用触发点
平台工程 / 发布控制工程副总裁、平台负责人后端、前端、移动端、发布工程师工程平台或开发者效率预算功能开关、暗发布、熔断开关、分阶段发布不承担全量重新部署风险,也能更快发版
产品管理 / 实验产品总监、增长 PMPM、分析师、产品运营产品或增长运营预算功能实验、引导流程测试、指标读数验证路线图决策,降低功能风险
数据科学 / 分析工程数据负责人、分析工程负责人数据科学家、分析师、实验专家数据平台或分析预算指标定义、数据仓库原生分析、因果护栏让实验结果可信,复用受治理的指标
增长 / 营销增长负责人、生命周期负责人CRM 经理、营销人员、增长分析师增长营销预算活动、消息触达和留存实验提升收入、留存和参与度
安全 / 重治理的企业买方数据平台负责人、安全评审人分析工程、合规、平台团队数据平台、安全或共享基础设施预算把分析留在数据仓库内,减少 PII 流动,保留审计轨迹满足治理、隐私和数据驻留要求
高管发起人 / 标准化买方CPO、CTO、CIO、转型发起人跨职能产品与工程组织中央软件或转型预算标准化发布控制、实验工作流和度量栈工具整合,以及 AI 时代更快迭代

公开来源能清楚看出用户群;付费方常随组织成熟度变化,所以表中区分用户和付费方,而不是强行指定一个固定负责人。

[CM023, CM025, CM026, CM027, CM028, CM029]
FM003: 买方 / 细分市场地图

采用通常从工程发布控制切入,随后扩展到产品和增长度量;要实现标准化,还需要数据治理批准和高管赞助。

[CM001, CM018, CM020, CM025, CM026, CM028]
FM004: 采用漏斗或价值链地图

功能开关入口会放大市场,但一旦团队需要统计严谨性、治理和跨职能标准化,市场又会收窄。

漏斗值是方向性指数,不是观测转化率。它们显示买方从基础发布控制走向受治理平台标准化时,采用摩擦在哪里累积。

[CM016, CM017, CM018, CM020, CM025, CM026]

2.4 增长驱动、采用约束与开放规模缺口

需求驱动真实且多层。产品分析研究者把增长与云采用和个性化联系起来;Credence 将实验需求与数字化转型和移动商务联系起来;Kameleoon 与 VWO 显示,投资和测试频率仍健康,足以支撑品类扩张。最新的需求加速器是 AI 辅助产品开发:DORA 认为 AI 采用现在是与价值流表现绑定的系统问题,Harvard 则把实验描述为组织加快产品学习的战略必需。但约束同样真实。Forrester 和 LeadDev 说明了内部自研为什么昂贵:自研功能开关和实验系统会消耗开发者时间,也难以维护。MIT 又加入了随机化、版本控制和自研还是购买之间的方法复杂性。隐私和治理是另一道刹车。Harness 指出实验会收集敏感行为数据;ICO 和 EDPB 明确,基于 cookie 的数据收集通常需要同意,不能伪装成 “strictly necessary”。结果是,一个增长有吸引力、架构上强烈拉向受治理数据、但持续受切换成本、实施复杂性、隐私和品类模糊拖累的市场。因此,投资者应把 Statsig 放在一组受约束的相邻市场视角中承保,而不是套用一个单一膨胀 TAM。[CM037, CM038, CM039, CM040, CM041, CM042]

增长驱动因素与约束表
驱动因素 / 约束方向时点影响尽调问题
渐进式交付和暗发布驱动因素当前持续交付把功能开关和受控发布变成默认工程需求Statsig 使用中,有多少起点是发布控制,而不是实验?
云与个性化需求驱动因素当前数字产品需要更快个性化闭环时,分析和实验预算会增长多少客户购买实验能力,是为了改善既有个性化或增长工作流?
AI 辅助的产品迭代驱动因素当前至近期AI 原生版本发布更频繁,拉高了度量和回滚纪律的需求Statsig 需求中,有多少来自 AI 产品团队,而非传统 Web / 移动团队?
数据仓库原生治理驱动因素当前指标可信度、SQL 控制权和数据驻留要求,把需求拉向数据仓库原生分析现在有多少赢单需要兼容 Snowflake、BigQuery 或 Databricks?
可见的实验强度驱动因素当前健康的测试频率和明确的功能投入预算,支撑品类韧性有多少客户会从基础功能开关成熟到高频实验?
内部自建维护税约束当前自建功能开关和实验会推迟采购,但之后也会形成迁移压力哪些迁移工具能降低替换内部系统的成本?
隐私、cookie 同意与 PII 风险约束当前部分度量用例需要明确同意或更严格的数据最小化,可用流量因此减少产品实验中,有多少只靠已同意或第一方服务端数据就能运行?
方法复杂度约束当前随机化设计、单元追踪和解读复杂度,限制了实验民主化产品把哪些统计和治理护栏自动化到足够好,能让非专家使用?
集成与资源负担约束当前至近期切换技术栈或跨团队标准化,需要支持、定制和工程时间各细分客群需要多少实施资源,多久能体现价值?
估值中的品类边界模糊约束当前混杂的市场边界会高估 TAM,也会遮住买方到底是在替换单点工具,还是替换更宽的分析栈后续章节哪些应使用狭义核心视角,哪些应使用更宽的邻近视角?

本表把需求驱动因素和约束放在一起,因为此处的市场吸引力不仅取决于品类原始增长,也同样取决于治理和迁移摩擦。

[CM034, CM035, CM037, CM038, CM039, CM040]

2.5 展示材料

Chapter 03

03竞争格局

3.1 直接对手、相邻套件与替代路径

Statsig 已不再只在狭窄功能开关赛道里竞争。最接近的直接对手,是把发布控制与实验结合、并至少带有部分分析或指标层的平台:LaunchDarkly、Optimizely、Harness Split、GrowthBook、Eppo 和 PostHog 都以不同起点重叠同一个买方任务。LaunchDarkly 和 Optimizely 从企业级发布治理和实验严谨性切入该品类。Harness 延续旧 Split 逻辑,把功能开关和实验与发布监控、仓库原生分析绑在一起。GrowthBook 和 Eppo 从仓库原生侧进攻,主张实验应留在买方自己的数据和指标栈内。PostHog 从开源开发者套件角度竞争,VWO 则通过掌控 Web 和 App 优化工作流保持相邻位置,仍能吸收实验预算。替代集合也重要:内部自研、现有分析加点工具,以及 OpenFeature 等标准,都让多供应商并用保持可行。这意味着 Statsig 必须靠集成工作流速度取胜,而不能假设买方在结构上被锁进某一类供应商。[CP001, CP006, CP010, CP013, CP016, CP019]

竞品画像表
竞争对手 / 替代方案类别公开规模 / 采用信号目标买方 / 团队产品范围差异化限制
Statsig集成式实验控制平面1T+ 次事件 / 天、99.99% 可用性、2.5B 月度实验对象、数千家客户信任产品、增长、数据和工程团队分析、实验、功能管理、回放、Web 分析、配置、WHN集成工作流,加上透明的自助式定价未见官方自托管部署;可移植性弱于开源对手
LaunchDarkly治理导向的功能管理老牌厂商独立评论仍把它视为品类领导者;官方范围覆盖运行时控制和智能体控制企业工程与平台团队功能开关、渐进式交付、实验、可观测性、智能体控制审批、可审计性和发布治理可信度公开定价到高阶后转为询价,分析广度窄于集成套件
Optimizely企业实验老牌厂商信任中心材料强调其在云和本地环境中已有 20 多年安全交付数字产品、增长和实验团队功能实验、统计引擎、数据仓库连接、AI 优化统计工具强,能跨技术栈做实验公开定价以询价为主,且并非数据仓库原生 / 开源优先
Harness / Split发布监控与实验老牌厂商Split 品牌仍留在 Harness FME 内;CostBench 突出其成熟的企业归因和广泛 SDK 覆盖平台、DevOps 和产品团队功能开关、定向、动态配置、发布监控、实验、数据仓库原生实验把发布控制与影响数据、发布监控绑定按用户计费可能快速抬升成本,更宽的 DevOps 平台复杂度也会增加采用摩擦
Amplitude分析优先的邻近套件官方定价和信任门户显示,其套件打包很宽,并提供企业级信任展示产品、增长、高管和分析团队分析、功能实验、Web 实验、激活、回放、AI 工具一个套件里同时给 PM 友好的分析和实验高阶价格仍需定制,部署姿态以 SaaS 优先
GrowthBook数据仓库原生开源对手官网显示 3,000+ 家公司、100B+ 次功能开关查询 / 天技术型产品和数据团队实验、功能开关、产品分析、数据仓库原生云或自托管开源、自托管、可预测的席位定价、可审计性需要用户承担更多搭建责任,分发弱于托管型老牌厂商
Eppo数据仓库原生实验专家官网称功能开关支撑每天数十亿次分配;现在被包装为 Datadog Experiments数据科学、产品、工程和营销团队实验、功能开关、上下文 bandit、无代码营销测试零拷贝、数据仓库原生的严谨性,用户级数据不经过平台定价不透明;已审材料未显示自托管部署
PostHog开源邻近产品套件官方定价页显示 60,000+ 客户、10+ 个产品工程驱动型创业公司和技术型产品团队分析、回放、功能开关、实验、问卷、数据仓库、工作流套件范围宽,还有开源和自托管选项云优先的经济性往往优于自托管;对非技术 PM 工作流优化较弱
VWO优化与功能实验邻近厂商安全页显示 3,000+ 客户和广泛合规认证优化、实验和数字体验团队Web 和移动测试、功能实验、发布、审批、护栏Web 优化工作流强,治理界面完整不是完整的产品数据操作系统,企业价格基准仍不完整
内部自建 / 现状替代方案没有共同公开规模信号;成本体现为工程时间,或沿用既有分析栈优先考虑控制权、低前期现金成本或既有技术栈的团队自建功能开关和实验、OpenFeature 抽象层,或分析加点工具可移植性和定制贴合度最高价值兑现慢、持续运营负担重,跨团队治理更难

各行综合了保留的官方产品、定价、文档、信任页面,以及部分独立比较证据;表中只列公开采用信号。

[CP005, CP006, CP010, CP013, CP016, CP019]
FP001: 竞争定位图

0 到 10 的序数分数比较 x 轴上的部署和数据控制,以及 y 轴上的打包能力广度。分数是有证据支撑的综合判断,不是供应商披露指标。

x 轴反映自托管、开源姿态、本地评估,以及数据能否留在买方环境中。y 轴反映每个选项公开打包了多少个实质不同的任务。

[CP001, CP002, CP006, CP010, CP013, CP016]

3.2 能力宽度、包装透明度与部署姿态

在官方页面上,Statsig 把自己呈现为高度集成的实验控制平面:功能管理、实验、分析、回放、Web 分析、配置和仓库原生运营都在同一个产品家族里;定价 FAQ 还公开真实的自助升级路径,包括 2 million 免费计量事件、$150 Pro 层级和明确超额计费。这种宽度真实存在,但方向上已不再独特。Amplitude 在更大套件里销售实验、分析和回放;Optimizely 把功能实验与自己的统计引擎、仓库连接打包;Harness Split 把功能开关连到监控和仓库原生实验;GrowthBook 提供实验、开关和分析,可云端或自托管部署;Eppo 结合仓库原生实验与快速开关;PostHog 销售更宽的开发者套件;VWO 打包测试、灰度发布、审批和护栏。更尖锐的区别在包装和部署哲学。Statsig、GrowthBook、PostHog 以及 Amplitude 的部分入门点在公开信息中可读性更强。LaunchDarkly、Optimizely 和偏企业的大型存量厂商,仅凭公开列表价仍更难对标。仓库原生和自托管权衡形成第二条分界线:Statsig 支持仓库原生模式,但 GrowthBook、Eppo 和内部自研路径更用力把数据所有权和可移植性作为核心购买信息。[CP001, CP002, CP003, CP004, CP006, CP007]

功能 / 能力矩阵
供应商捆绑分析功能开关与发布高级实验统计数据仓库原生 / 零拷贝自托管 / 开源治理 / 信任姿态
Statsig通过 Warehouse Native 提供已审材料未见托管姿态较好;不如 LaunchDarkly 治理优先
LaunchDarkly有限非核心叙事已审材料未见治理、审批、发布控制强
Optimizely通过分析和实验视图部分提供通过数据仓库连接器提供已审材料未见企业信任叙事强
Harness / Split通过影响数据和数据仓库实验部分提供已审材料未见运营治理和安全强
Amplitude非核心叙事已审材料未见企业信任门户和 PM 工作流强
GrowthBook通过 SSO、审计轨迹和自托管体现较强
Eppo部分提供;更像分析层,不是完整分析操作系统已审材料未知数据控制和管理功能强
PostHog通过数据仓库层部分提供对技术型买方中到强
VWO相比套件,分析能力有限有限通过自托管选项提供审批和认证展示强
内部自建 / 现状取决于技术栈取决于自建质量取决于自建质量如果以数据仓库为主,则较强完全取决于内部流程成熟度

矩阵单元格只总结已审公开来源集。「未知」表示保留材料没有显示清晰证据,并不代表能力不存在。

[CP001, CP002, CP006, CP010, CP011, CP013]
定价 / 打包对比
供应商公开入门方案定价模式入门档包含能力透明度影响
StatsigDeveloper 档含 2M 计量事件免费;Pro $150/mo基于用量,对事件和曝光计量门控、配置、实验、分析入门档透明度高对重度实验团队,自助式价格很有进攻性
LaunchDarkly可免费开始;规模化方案定制套餐加定制企业定价运行时控制、功能开关、实验、可观测性、智能体控制入门档以上透明度低很难直接对标透明的自助式竞品
Optimizely各套餐均需询价按产品定制定价按模块提供功能实验、Web 实验、分析、CMS、数据平台透明度低可能适合企业,但公开 TCO 难以核算
Harness / Split免费方案引导,加上大量询价式打包独立评测材料显示按用户和企业式定价功能开关、监控、实验、数据仓库原生实验透明度中到低企业可用,但成本可能比 Statsig 或 GrowthBook 更快上升
AmplitudeStarter 免费;Plus 起价 $49/mo;Growth 和 Enterprise 定制套件加附加模块打包分析优先的平台套件,包含 Feature Experiment 和 Web Experiment透明度中等入门门槛低,但企业经济性仍需拿到报价后验证
GrowthBookStarter 免费,最多 3 个用户;Pro $40/seat/mo基于席位的云定价,可选自托管功能开关和实验不限量;分析 beta 和高级治理位于更高档公开档透明度高技术团队会强力压价封闭式老牌厂商
Eppo未保留公开自助定价页所审材料暗示企业版需报价仓库原生实验与功能旗标高度不透明相较自助替代品,是高价专业工具采购
PostHog多个免费层级,使用费率明确按用量计费,并设账单上限分析、回放、旗标、实验、数据仓库、调查、工作流入门层级不透明度低在所审相邻套件中,透明价格菜单最全
VWO保留页面列出 Growth、Pro 和 Enterprise 层级分层打包,按功能设门槛测试、实验、发布、审批、护栏中等不透明详细功能表便于初筛,但标价底线的经济性仍不完整
内部自建 / 现状方案通常现金入门成本低,但没有开箱即用套件工程时间和既有工具支出形成资本开支团队自己构建或已经拥有的能力无外部标价前期看起来便宜,但往往藏着运营和机会成本

本表只比较保留页面中的公开入门信号和打包逻辑;最低承诺、协商折扣和企业报价底线仍未解决。

[CP003, CP004, CP007, CP008, CP017, CP020]
FP002: 功能广度 / 能力地图

按六项买方标准做战略类别比较。色调概括留存证据,而不是供应商披露分数。

该图有意按战略类别归组竞争者,避免重复详细的逐供应商表。这里的“托管集成套件”主要涵盖 Statsig、Amplitude 和 Optimizely。

[CP035, CP037, CP038, CP039, CP043, CP044]

3.3 治理、信任姿态、切换成本与多供应商并用

信任和切换成本图景是混合的,而非单向。Statsig 可以指向平台级可靠性和仓库原生选项,但治理重的存量厂商在审批链、可审计性和企业发布安全上仍有更强默认心智。LaunchDarkly 的独立定位仍是治理优先,Optimizely 强调长运营历史以及安全的本地和云端经验,Harness 描述同伴评审、回滚纪律、RBAC、SSO 和 2FA,VWO 则在留存页面上直接展示审批、SSO 和大量认证细节。GrowthBook、Eppo 和 PostHog 提供另一种信任叙事:数据留在买方环境内或可自托管,本地评估避免每次检查都发起网络调用,开源或仓库原生架构降低对供应商的硬基础设施依赖。OpenFeature 进一步通过供应商无关 API 降低代码层锁定。结果是,Statsig 周围的切换成本更多基于工作流和治理,而不是代码。指标、灰度发布习惯、实验历史和采购标准化都会制造粘性;但只要买方把数据留在自有仓库、自托管关键组件,或标准化到保留多供应商能力的抽象层,真正硬锁定就更弱。[CP002, CP005, CP009, CP012, CP015, CP018]

3.4 护城河耐久性、商品化风险与反向解读

反向解读很直接:功能管理和实验正在汇聚成更宽的能力层,单靠套件宽度更难持续稀缺。Forrester 明确把功能管理加实验视为一个横跨软件交付和产品管理的演进市场,供应商组合现在也在实践中反映了这个框架。GrowthBook 将自己营销为开源、仓库原生、成本更低且比 Statsig 更可审计,迁移页面还直接攻击 Statsig 的事件计价、自研分析和数据所有权。独立评论呼应了同一警告的较温和版本:LaunchDarkly 仍赢在治理优先的企业动作,GrowthBook 对想要更低成本和更多所有权的团队更有吸引力。这并未抹掉 Statsig 的优势。透明入门定价、集成分析加实验,以及成熟托管工作流,仍让它对想快速落地、减少工具数量的产品主导团队有吸引力。但护城河是分客群的,不是普遍的。Statsig 在团队想要一个托管、以实验为中心的控制平面时最强;在买方优先考虑自托管可移植性、零复制仓库治理,或已经嵌入组织的重审批存量厂商工作流时最弱。[CP033, CP034, CP035, CP036, CP037, CP038]

壁垒耐久度 / 竞争风险登记表
壁垒主张或压力点证据威胁路径严重性当前缓释或抵消因素尽调问题
以实验为中心的一体化套件Statsig 把分析、旗标、实验和回放接进同一套托管工作流Amplitude、PostHog、Optimizely 和 GrowthBook 也在扩展套件边界对产品主导型团队,采购和工作流仍更简单索取按分群划分的多产品附加、扩张和留存数据
透明的自助入门定价Statsig 公布免费和 Pro 入门经济性GrowthBook、PostHog 等透明对手降低了这一信号的稀缺性定价透明仍能提升买方信任和采用速度收集真实企业报价,验证透明度在规模化后是否仍成立
仓库原生选项Statsig 支持 Warehouse Native,且无需 ETLGrowthBook 和 Eppo 把仓库控制放在叙事中心,而不是当作选项Statsig 仍可提供托管或仓库原生的灵活性验证数据团队主导实验栈时的胜率
托管工作流便利性托管配置和集成统计能力降低工程开销开源和自托管选项降低长期锁定,对某些工作负载可能更便宜缺少深度实验基础设施的团队很看重快速上线衡量不同替代方案下首次测试用时和总工程工时
治理与企业审批LaunchDarkly、Optimizely、Harness 和 VWO 强调审批、可审计性、SSO 与发布纪律治理要求重的企业可能认为 Statsig 好用,但不是默认最佳治理需求较轻或产品主导程度更高时,Statsig 更容易赢向 LaunchDarkly 级买方核验企业级部署、RBAC 和审批适配
可迁移性与代码级锁定OpenFeature 和自托管对手降低硬代码锁定买方可以抽象供应商,或把数据留在自有基础设施工作流、实验历史和采购流程仍会制造切换摩擦询问现有客户是否标准化采用 OpenFeature 或类似抽象层
维持现状和内部自建替代原则上仍可内部自建,也吸引重视控制权的团队隐性运营负担可能到后期才被看见专门供应商仍能更快交付价值在低估替代风险前,索取真实内部自建成本对比
品类收敛 / 商品化Forrester 和当前供应商页面显示,实验加发布控制正在变成常见组合单一功能不再稀缺到足以独立支撑溢价Statsig 仍可靠一体化工作流和速度竞争跟踪路线图速度、赢单 / 输单原因,以及相对直接同业的套件附加率
反向迁移叙事GrowthBook 现在用价格、数据所有权和可审计性,直接劝客户从 Statsig 迁移仓库原生和开源对手可以把 Statsig 描述成托管锁定风险Statsig 可用托管便利和一体化工作流反击探查这一叙事在竞争交易和续约中出现的频率

严重性反映其对 2026 年采购流程的可能影响,并非量化财务损失模型。

[CP033, CP034, CP035, CP036, CP038, CP039]
FP003: 护城河 / 准备度 KPI

精选公开指标勾勒出 Statsig 周围入门级和规模级竞争已经有多激烈。

该图有意混用单位,只为概括竞争压力;它不是标准化评分模型。

[CP003, CP004, CP005, CP019, CP020, CP026]

3.5 展示材料

Chapter 04

04财务情况

4.1 收入模式与定价架构

Statsig 的货币化对一家私营基础设施初创公司来说异常清晰,因为公司同时公布免费开发者层和按量计费 Pro 层。官方定价页展示了经典产品主导阶梯:面向个人构建者的免费访问、每月 $150 且包含事件量和付费超额的 Pro 计划,以及定制企业合同。经济信号不是席位货币化;Statsig 反复把定价围绕事件、实验用量和部署规模展开,同时表示席位、功能开关和环境不是限制变量。Warehouse Native 对财务重要,因为它拓宽了企业入口,而不必把每个客户都推入 Statsig 托管算力;Sacra 也明确认为该部署选项可降低 Statsig 自身基础设施负担。收入质量上的保留项同样重要。公开列表价解释了包装和客户如何扩大支出的机制,但没有显示已实现 ASP、折扣、续约率,或自助 Pro 与谈判型企业承诺之间的组合。换句话说,漏斗顶部透明,真正货币化引擎仍是私有信息。[CI001, CI002, CI003, CI004, CI005, CI006]

收入流表
收入流机制计量单位当前数值 / 状态质量尽调问题
Developer 免费计划用免费层级播种产品采用,再转化为付费用量事件 / 回放 / 席位免费;每月 2M 事件和 50K 会话回放索取按分群划分的免费转付费转化、激活和回本周期。
Pro 订阅自助订阅,包含基础用量并收取超量费月度合同每月 $150,包含 5M 事件索取 Pro 客户数、平均超量发生率和流失率。
Pro 超量收入超过包含事件量后的计量收费每 1K 事件每月超过 5M 后,每 1K 事件 $0.05索取按事件计的混合实际价格,以及按账户规模划分的超量收入占比。
Enterprise 事件计价合同面向高事件量大组织的协商合同年度合同定制定价;披露大用量折扣索取 ACV 区间、最低承诺和续约上浮。
Enterprise 实验计价合同按实验用量而非仅按原始事件定价的定制合同年度合同官方定价页称可提供按实验计价合同索取实验量定义和实际单实验定价。
Warehouse Native 企业部署客户把数据 / 计算留在自有仓库,同时购买 Statsig 的统计引擎和工作流企业部署包含在平台内;商业细节未披露索取附加率、相对托管版的利润率差异,以及采用仓库模式后的平均扩张。
支持 / 合规加购优先支持、SSO、RBAC 和 HIPAA 合格打包支撑企业扩张企业套餐捆绑在 Enterprise 层级内;未给出明确独立价格索取支持服务成本,以及合规打包是否改变 ACV 或毛利率。

各行把变现机制和 PLG 漏斗拆开。公开标价有助于理解打包逻辑,但不能代表实际净收入质量。

[CI001, CI002, CI003, CI004, CI005, CI006]
定价 / 变现表
计划 / 计费抓手价格 / 单位 / 合同标价与实际定价折扣 / 未知项来源
Developer免费官方标价转化为付费计划的实际情况未披露Statsig 定价
Developer 事件额度每月 2M 事件官方标价有多少免费账户真正用到上限,仍未知Statsig 定价
Developer 会话回放每月 50,000 次回放官方标价回放存储经济性和转化影响未披露Statsig 定价
Pro 基础计划每月 $150官方标价折扣、年度预付和协商例外未披露Statsig 定价
Pro 包含事件量包含 5M 事件官方标价该阈值下的毛利率未披露Statsig 定价
Pro 超量计费每 1K 事件 $0.05官方标价企业捆绑后的混合实际费率未知Statsig 定价
Enterprise定制仅有官方标价框架事件计价与实验计价合同占比属私有信息Statsig 定价 / TrustRadius
席位限制按事件计价模型不限制席位、旗标或环境官方定位席位数仍可能影响支持和入职成本Statsig 仓库页面

本表有意把公开标价机制与投资测算所需但缺失的实际价格输入拆开。

[CI002, CI003, CI004, CI006, CI010, CI039]
FI001: 收入模型桥接

Statsig 的商业路径从免费开发者采用,走向按量计费的 Pro 使用,再进入谈判式企业扩张;仓库原生部署扩大了企业端边界。

[CI001, CI003, CI004, CI005, CI006, CI039]

4.2 规模、增长与利润率代理指标

Statsig 最强的公开财务支撑是方向性规模,而非审计收入。公司 2025 年回顾称,周活用户 20,000、每天处理 3 trillion 事件、通过实验触达 4 billion 唯一终端用户、实验和功能门控同比增长 2.5 倍,并在 2025 年产生近 7 million 次分析查询。这些产品使用信号让按量计费模型在经济上可信:客户使用更多开关、实验和分析,Statsig 就有更多方式在账户内扩大收入。第三方营收估计仍混杂。Sacra 和 SiliconANGLE 都指向 2025 年 Series C 前后 ARR 约 $40 million,Growjo 则估计较低的 $23.7 million 收入运行率。员工数也是区间而非点估计,官方和第三方来源集中在 145 至 170 人左右,GeekWire 称公司计划到 2026 年初接近 200 人。如果 $40 million ARR 估计方向正确,这一员工区间支持推导出的 ARR/员工约 $235,000 至 $276,000。含义不是成熟 SaaS 效率,而是可信的成长期经营杠杆故事;其利润质量仍取决于未知的托管、支持、销售和留存经济性。[CI015, CI016, CI017, CI018, CI019, CI020]

单位经济性表
指标数值 / 空值置信度为什么重要尽调问题
ARR 估计(USDm)40估值框架中最有支撑的第三方收入锚索取截至 2026 年 Q1 的管理层 ARR、确认收入和净美元留存。
收入估计(USDm)23.7另一组市场数据估计显示公开区间仍有多宽索取经审计或管理层收入桥,解释外部估计为何分歧。
付费客户300有助于粗算 ACV,并分析企业客户集中度索取精确付费账户数,以及头部分群的收入集中度。
公开员工人数区间145-170员工数是公开信息中衡量运营费用负荷的最佳代理索取按职能划分的月末员工数和全负荷薪酬支出。
人均 ARR(USDk)235-276派生生产率代理有助于限定经营杠杆索取人均净收入、销售生产率和按 GTM 角色划分的配额达成率。
隐含 ARR 倍数(x)27.5有助于对标上一轮估值是否需要极端增长预期索取董事会材料,展示 C 轮和收购时的估值假设。
毛利率(%)null软件质量的核心投资测算指标缺失索取托管部署与仓库原生部署的毛利率拆分。
CAC 回收期(月)null判断 PLG 和企业销售是否资本高效的关键索取付费获客支出、销售薪酬,以及按客群划分的 CAC 回收期。
净收入留存(%)null使用量定价产品的扩张耐久性是核心索取按客户分群划分的 NRR、毛留存和收缩驱动因素。

公开证据不足以支撑诚实的投资测算时,空值是有意保留的。数值为公开估计,并非经审计的管理层披露。

[CI021, CI022, CI023, CI024, CI025, CI041]
FI002: 单位经济性桥接

公开使用量增长支撑真实的变现引擎,但从收入桥接到毛利和高效增长时,公开信息在私有指标处断开。

[CI016, CI017, CI020, CI021, CI023, CI025]
FI003: 财务估计区间

ARR、员工数、估值和累计融资都有公开边界,但这些边界仍由质量不一的外部信号和公司自述信号拼成,而不是经审计报告。

低 / 高值混合了公司自述披露和第三方估计。它们是方向性的投资人边界,不是管理层认可的 KPI 报告。

[CI020, CI021, CI022, CI023, CI028, CI032]

4.3 融资历史、估值与收购影响

融资历史比盈利能力更有文档支撑。Tracxn 追踪到三轮已披露融资,合计约 $153 million:2021 年 $10.4 million Series A、2022 年 $43 million Series B,以及 2025 年 5 月 $100 million Series C。官方和媒体来源一致认为,Series C 给 Statsig 的估值为 $1.1 billion,由 ICONIQ Growth 领投,Sequoia 和 Madrona 继续跟投。这轮融资只在几个月后就迎来下一项重大财务事件:OpenAI 于 2025 年 9 月同意以 $1.1 billion 收购 Statsig。GeekWire 最有财务意义的观察是,收购价与上一轮私人估值持平而非高于它,意味着投资人获得流动性,但没有收购溢价。OpenAI 和 CNBC 还表示,在监管批准和整合推进期间,Statsig 将继续独立运营并从 Seattle 服务客户。对财务分析来说,问题从纯风险投资现金跑道转为母公司支持下的运营连续性。如果收购按所述条款交割,短期独立融资依赖应大幅下降。代价是,未来外部投资者可见的财务透明度会更差,因为任何交割后经济性更可能落在 OpenAI 更宽的报告边界内,而不是 Statsig 独立披露节奏里。[CI013, CI028, CI029, CI030, CI031, CI032]

资本充足性表
项目公开数值 / 状态置信度为什么重要尽调问题
累计融资额$153M-$153.4M,三轮融资界定退出前所需外部融资规模索取各轮新股与老股的精确拆分,以及收购前剩余资金。
A 轮$10.4M,2021-08-05显示公司多早进入机构化融资索取原始投后估值和期权池条款。
B 轮$43M,2022-04-20锚定独角兽前的扩张阶段索取 B 轮估值、董事会构成变化和资金用途备忘录。
C 轮$100M,2025-05-06,估值 $1.1B最新独立融资在收购前重置了公司的资本基础索取进入资产负债表的现金、员工老股出售部分和预期烧钱假设。
OpenAI 收购协议2025 年 9 月宣布的 $1.1B 价格投资者和员工的主要流动性事件索取合并协议经济条款、交割状态、托管安排和员工待遇。
收购溢价相较上一轮私人估值,看不到溢价解释为什么这一结果更像流动性加战略契合,而不是突破式价格上台阶索取董事会公平性材料和投资者回报瀑布。
账面现金null没有现金数据,就无法做干净的资金跑道分析索取最新资产负债表和月度现金桥。
月度烧钱 / 资金跑道null公开来源仍无法完成核心融资依赖测试索取过去十二个月经营现金流,以及预算与实际烧钱对比。
债务 / 项目融资义务null即使完成股权融资,债务也可能改变下行情景风险索取债务明细、契约条款和任何云承诺负债。
融资依赖展望若收购完成则较低;若未完成仍不透明短期资本充足性的当前最佳判断索取交割状态和交割后运营资金模型的明确确认。

本表聚焦前瞻资本充足性,而非重述完整融资年表。公开资本获取能力可见;公开流动性和烧钱数据不可见。

[CI013, CI028, CI029, CI030, CI031, CI032]
FI004: 资本强度 / 现金流地图

Statsig 可见的资本故事从风投支持扩张,转向战略收购;短期融资风险下降,但未来披露不透明度上升。

[CI013, CI033, CI034, CI035, CI036, CI051]

4.4 反向视角与承保阻碍

反向财务故事不是可见困境,而是竞争压力、产品摩擦和不透明私营经济性的组合。Sacra 的风险部分清楚指出,实验支出可能被更广泛的可观测性或分析预算吸收;这对任何独立工具供应商都是真实定价风险。同一报告还警告,高级统计能力对高阶用户可能比对主流买方更有价值;如果更广泛客户只需要更简单测试工作流,付费意愿会变弱。TrustRadius 评论从用户侧强化了这一担忧:买方称 Statsig 强大,但也指出界面偏复杂、偏数据科学,工作流存在限制。最大阻碍仍是缺少财务披露。留存来源均未提供毛利率、CAC、回本周期、NRR、客户集中度、手头现金、月度烧钱、债务工具,或清晰的收购后独立报告视图。即便备案线索也不完整:SEC legacy browse 搜索没有在抓取的主要内容中给出可直接读取的 Statsig 公司文件;市场数据服务也可能滞后,Crunchbase 旧快照仍将 Series B 标为最新轮次就是例子。合理财务结论因此是:看好产品市场匹配和资本可得性,但对正常化盈利能力和估值精度保持谨慎。[CI045, CI046, CI047, CI048, CI049, CI050]

公开财务缺口表
缺失的私有指标影响具体尽调路径
确认收入与 ARR 桥接阻断性——投资者无法把产品用量转译成达到 GAAP 质量的收入索取 2024-2026 年每月确认收入、ARR、递延收入和服务收入占比。
按部署模式划分的毛利率阻断性——Warehouse Native 可能显著改善经济性,但没有公开拆分索取托管与仓库原生毛利率、云支出和按客群划分的支持负担。
CAC、回收期和净收入留存重大——只有扩张超过获客支出,使用量定价才有吸引力索取分群留存表、按渠道划分的 CAC,以及 PLG 与企业辅助销售动作各自的回收期。
现金、烧钱、资金跑道和债务阻断性——仅靠公开来源无法测算资本充足性索取最新资产负债表、现金流量表、债务明细和董事会批准的运营计划。
客户集中度和实际净定价重大——标价看不出是否由少数大客户主导收入索取前 10 大客户占比、平均合同价值、折扣区间和用量集中度。
融资历史的直接申报佐证次要但重要——公开来源的申报证据弱于融资数据库和新闻证据向律师或数据室索取每轮融资的实际 Form D 或州申报包。
收购后的独立报告重大——进入 OpenAI 后,Statsig 可能不再留下任何外部可见的独立财务轨迹要求确认并购交割,并提供任何交割后预算、剥离口径 P&L 或管理层报告包。

这些具体缺口使得仅靠公开资料无法做承保判断。多数缺口需要直接公司尽调,而不是更深入的网页研究。

[CI049, CI050, CI052, CI053, CI054]

4.5 展示材料

Chapter 05

05产品与技术

5.1 平台表面与客户工作流

Statsig 的公开表面围绕一个产品开发循环组织,而不是单个独立工具。公司把功能开关、实验、分析和仓库原生分析营销为相互连接的层;文档也强化了这个模型,会根据开发者需要的是布尔发布控制、结构化 JSON payload,还是随机测试,把他们从功能门控路由到动态配置再到实验。这对尽调重要,因为产品定义不只是“功能开关”;它是一套控制与度量栈,意在帮助工程、产品和数据团队发布代码、观察影响,并决定是否扩大或回滚发布。因此,模块地图很宽但仍一致:发布控制和远程配置位于运行时行为前端,实验和留出组建在曝光数据之上,分析和回放位于实验旁用于诊断,仓库原生则把同一工作流扩展给希望在自有数据集上分析的买方。[CE001, CE002, CE003, CE005, CE011, CE014]

产品模块 / 资产矩阵
模块主要用户状态 / 成熟度差异化尽调缺口
功能门控与发布控制工程 / 产品成熟的核心能力与高级定向、发布诊断和回滚工作流绑定的布尔运行时控制公开证据无法确认单个功能门控或规则级授权粒度
动态配置应用工程师成熟但有边界服务端定义的 JSON 让团队无需重新部署代码,就能按用户上下文调整取值100 KB 载荷上限可能把更大的参数集推向相邻配置系统
实验、留出组与 bandit产品 / 数据科学有充分文档的高级能力A/B/n、CUPED、序贯检验、留出组和 Thompson 采样自动调优均有公开文档上下文匹配、个性化上限和发布政策仍需要按买方场景做统计审查
产品分析与看板增长 / 分析成熟的支撑能力漏斗、留存、旅程、生命周期和看板连接到发布与实验产物公开文档对底层云端分析数据模型的披露不如 Warehouse Native 细
会话回放产品 / 支持 / UX有文档,但属于更高风险的附加模块基于 rrweb 的回放提供可视化调试,并带遮罩和资格控制浏览器支持范围和保存期限细节比功能门控能力更薄
Warehouse Native数据平台 / 实验团队企业级,但细节复杂在客户数据仓库中运行实验分析,同时保留 Statsig 的统计引擎和可选 SDK 分配路径公开资料显示的是混合部署模型,而不是通用的纯本地架构

本表映射截至运行日期可见的公开产品界面。“成熟度”反映文档深度和已上线证据,不代表内部收入结构。

[CE001, CE002, CE003, CE004, CE005, CE011]
工作流 / 用例表
用户任务当前工作流Statsig 能力可衡量收益限制
安全发布高风险版本创建控制、定向用户,只有指标守住时才扩大发布功能门控加发布诊断渐进发布,带回滚钩子和直接影响测量取决于 SDK 采用和值班纪律
无需重新部署即可改变应用行为按用户属性提供服务端定义参数动态配置快速调整阈值、文案或排序权重大型或深层嵌套载荷会撞上文档中的 100 KB 上限
运行因果产品测试分配随机变体并测量提升实验、CUPED、序贯检验、留出组用统计依据决定是否发布,而不是依赖历史相关性方法选择和停止规则仍需要实务判断
理解上线后的行为探索旅程、漏斗、留存和看板产品分析共享性能视图,无需从原始 SQL 起步公开文档未完整披露托管分析的仓库级数据血缘
诊断难以复现的 UX 问题回看与产品行为绑定的交互轨迹会话回放用可视化调试支持转化和故障分析回放隐私选项和浏览器兼容性需要单独尽调

只有公开文档或产品页面明确写出的收益,才列入本表。限制来自同一批来源的承认,或公开证据仍然薄弱的地方。

[CE002, CE003, CE004, CE005, CE007, CE010]
FE001: Statsig 产品架构栈

分层展示 Statsig 如何把运行时控制、度量和企业治理组合成一个运营模型。

这是基于公开文档和产品页面综合出的架构图,不是内部部署图。

[CE001, CE014, CE016, CE017, CE023, CE025]

5.2 运行时架构、API 与部署模型

公开材料中可见的技术架构以 SDK 介导的运行时评估,加上控制台和 API 控制平面为中心。功能门控、动态配置和实验都被呈现为通过 Statsig SDK 或 API 发生的运行时检查,事件和曝光日志再进入下游分析。Warehouse Native 改变的是分析在哪里运行,而不是 Statsig 是否参与工作流:公开文档仍依赖 Statsig 控制台做设置、官方 SDK 做分配,以及一条把暂存表和汇总表写入客户仓库的文档化流程。结果是混合架构,而不是纯无控制平面的设备。对想要托管发布管理加仓库分析的买方,这种混合模型可以是优势;但它也是尽调点,因为 “warehouse native” 并不消除围绕分配、摄取、日志回调和中央 API 治理的实施选择。GitHub 代码引用等开发者工具进一步显示,Statsig 试图嵌入日常工程循环,而不只是指标控制台。[CE014, CE015, CE016, CE017, CE018, CE019]

技术 / 运营架构表
层 / 流程 / 组件作用依赖风险
客户端与服务端 SDK 求值在应用或服务运行时执行功能门控、配置和实验官方 SDK、客户端密钥或服务端密钥旧 SDK 12 个月后不再受支持,各语言能力对齐情况未完全公开
曝光与自定义事件日志把下游指标归因到功能和实验SDK 事件 API 或 HTTP API 调用日志配置错误会削弱实验读数和分析归因
Console 和 Console API 控制平面创建、编辑、审查并自动化产品对象statsigapi.net、API 密钥和带版本的标头统一 API 速率限制和治理设置仍属于运营模型的一部分
Warehouse Native 计算管线在客户数据仓库中运行分析,并存储暂存表和结果表客户数据集、数据仓库凭证和 Statsig 管线定义产品演进时,内部表名可能变化
开发者工作流叠加层把 Console 对象连回代码仓库和运行时部署辅助工具用于代码引用的 GitHub 令牌和仓库访问权限集成便利性会增加令牌管理和访问范围负担

本架构表聚焦公开运营模型,而不是内部源代码拆解。它对尽调足够具体,但不能替代供应商提供的架构审查。

[CE015, CE016, CE017, CE018, CE019, CE027]
FE002: 客户工作流 / 运营流程

Statsig 客户如何从发布定义走到运行时评估、度量和更大范围推广。

流程抽象掉分析托管在 Statsig 还是仓库原生环境的差异,因为两种模式仍使用同一套公开产品对象。

[CE002, CE005, CE011, CE017, CE019]
FE003: 关键依赖图

关键外部系统和运营依赖,决定 Statsig 面向企业客户的生产环境表现。

依赖边基于公开记录的集成和部署模式,而不是保密供应商清单。

[CE016, CE022, CE027, CE036, CE037, CE038]

5.3 实验统计、分析与回放

Statsig 的统计表面显著宽于基础固定期 A/B 工具。公开文档覆盖 CUPED 方差降低、序贯测试、留出组和 Thompson-sampling bandits;仓库原生文档也明确把这些方法连回同一实验引擎。这种宽度支撑了公司对数据科学重度买方的推介,但细节也揭示权衡:存在序贯测试,是因为提前窥探是真实风险;bandit 文档明确表示,默认方法无法个性化,且可能对特定用户队列收敛到错误的全局答案。在可观测性侧,分析表面包括漏斗、留存、分布、用户旅程、生命周期和仪表盘;Session Replay 则把诊断扩展到基于 rrweb 的回放。Replay 不只是一个录制按钮;公开文档把它定位为同一反馈循环的一部分,用来指导发布和实验。保留项是,回放有自己的浏览器和隐私边界情况,因此应被承保为比核心功能开关评估风险更高的附加模块。[CE006, CE007, CE008, CE009, CE010, CE011]

FE004: 产品成熟度 / 能力图

公开材料中可见的 Statsig 主要产品模块的相对成熟度和尽调姿态。

评级是定性、以证据为先,不按收入加权。评级反映公开文档深度,不代表内部采用数据。

[CE005, CE009, CE011, CE013, CE014, CE023]

5.4 企业控制、隐私与安全姿态

Statsig 拥有可信的公开企业控制表面,但强项在组织和项目层,而不是最细粒度对象层。访问管理文档清楚区分 SSO 和 SCIM,记录企业 OIDC SSO,并展示基于 Okta 的 SCIM 生命周期自动化。仓库原生角色页面补充了明确的 RBAC 故事,说明谁可以查看、编辑和批准实验、指标及原始数据。安全和隐私材料也不止通用营销:合规文档讨论客户数据隔离、AI 数据边界、静态 AES-256、TLS 1.2+ 和 SOC 2 Type II;隐私政策则区分网站 / 账户处理与客户控制的产品数据。净结论是,Statsig 在核心企业控制上公开可读;但细粒度授权、数据驻留细节,以及部分逐产品控制范围仍需要尽调,不能被公开材料完全解决。[CE020, CE021, CE022, CE023, CE025, CE026]

信任 / 质量 / 合规表
控制 / 姿态状态范围作用缺口
OIDC SSO公开文档可见,企业版组织认证企业可沿用自有身份提供商,并继承 MFA 策略公开资料包未描述对象级授权粒度
SCIM 用户预配Okta 有公开文档用户和用户组生命周期自动把预配、导入、推送和撤销预配流程接入 Statsig非 Okta 身份提供商没有公开 SCIM 文档
可配置 RBAC 与审批Console 或 Warehouse Native 层级有公开文档实验、指标、审批、原始数据可见性限制谁可以查看、编辑和审批敏感对象单个功能门控或规则级策略文档不清晰
会话回放隐私控制公开文档可见回放捕获和遮罩提供基础遮罩级别、CSS 选择器规则和用户资格门控这里未展示公开的保存期限和浏览器支持矩阵
安全与隐私体系公开文档可见平台全局声称提供客户隔离、AI 数据同意边界、AES-256、TLS 1.2+ 和 SOC 2 Type II区域部署细节和更深入审计材料仍需尽调

本表只记录公开资料包中明确可见的控制项。不假设没有直接文档支持的认证、区域部署或审批机制。

[CE020, CE021, CE022, CE023, CE024, CE025]

5.5 技术限制与路线图可见风险

主要公开风险不是 Statsig 缺少核心宽度,而是一些重要运营边界仍只部分记录。公开软件包注册表和代码仓库清楚显示公司维护广泛 SDK 覆盖并积极发货,但这些产物不能证明各语言之间功能完全一致。Session Replay 问题队列显示,即便更大平台故事很强,浏览器特定故障仍会出现。治理方面,公开材料确认 Statsig 具备 RBAC、审批和企业身份功能;但一条关于按门控或按规则做访问控制的公开请求,凸显严格职责分离买方可能遇到的缺口。仓库原生故事也足够强,值得认真对待,但仍留下实施边界问题:公开材料展示的是仓库内分析和混合部署选择,而不是通用 no-ETL 承诺。对承保而言,产品显然有能力,但治理、浏览器支持和部署架构的最后一公里仍需要直接尽调。[CE024, CE028, CE029, CE030, CE031, CE032]

路线图 / 发布 / 开发阶段表
日期 / 阶段功能或信号状态含义来源
2023-06 上线Warehouse Native已上线把 Statsig 从托管分析扩展到 BigQuery、Snowflake、Databricks 和 Redshift 环境Warehouse Native 上线文章
当前文档序贯检验和 CUPED已上线显示其统计引擎不止简单固定周期 A/B 测试实验方法文档
当前文档12 个月内的 SDK 支持现行政策过期 SDK 可能继续工作,但不再获得官方支持SDK 支持政策
当前公开 issueSafari 15 会话回放崩溃抓取页面上的开放风险回放可能独立于核心功能门控或分析栈而失败GitHub 议题 #36
当前公开 issue单个功能门控或规则级 RBAC 粒度已有请求 / 公开信息不清晰企业治理可能落后于买方对委托所有权的预期GitHub 反馈议题 #40
当前包与注册表界面跨语言 SDK 打包活跃但可见度不均Statsig 明显维护广泛的包分发面,但能力对齐细节仍需尽调npm、Go、Dart、.NET、Java 注册表

本表使用公开发布、issue 和包信号,而不是内部路线图。凡是没有明确上线或文档支持的内容,均刻意保守处理。

[CE024, CE028, CE029, CE030, CE031, CE032]

5.6 展示材料

Chapter 06

06客户情况

6.1 客户基础与买方-用户-付款方分层

Statsig 抓取到的客户包显示出宽度,但并非随机或均匀分布。具名案例集中在产品主导的软件和互联网业务,这些场景每天都需要发布控制、实验节奏和产品分析:Notion、Webflow 等生产力与协作平台;Brex 等金融科技和金融软件;HelloFresh、Whatnot、OfferUp 等消费者商业和市场平台场景;Bluesky、SoundCloud、Linktree 等社交、创作者和媒体产品;Code.org 代表的非营利教育;Character.ai 代表的消费者 AI;Runna 代表的健身;Ancestry 代表的家谱订阅;Graphite 代表的开发者工具。在这些例子中,运营买方通常是产品、工程、数据或增长负责人,而不是单独行动的集中采购团队。用户更宽:可能是 B2B 软件内部工程师和分析师,也可能是受已发布体验影响的数百万终端消费者、创作者、教师、学生、跑者、购物者和听众。因此,公开证据对有可度量发布循环的软件原生团队最强,对传统线下或高度监管客户环境则薄得多。[CU001, CU002, CU003, CU004, CU008, CU010]

客户分群表
分群买方 / 用户 / 付款方主要用例代表性公开证据收入 / 战略价值主要缺口
生产力与协作 SaaS买方 = 产品 / 工程 / 数据负责人;用户 = 构建者和终端用户;付款方 = 核心产品预算面向协作产品的实验、发布控制和分析Notion 和 Webflow 案例,加官方公司页面战略价值高,因为发版速度和 UX 质量直接影响增长未公开合同规模、续约率或按账户划分的模块组合
金融科技与金融软件买方 = 数据 / 平台或产品负责人;用户 = 内部财务和运营团队;付款方 = 中央平台或分析预算实验分析、数据整合和发布控制Brex 案例,加 Brex 主页工具整合可让账户价值超出简单功能开关未公开商业条款或支出占比
消费电商与交易平台买方 = 增长 / 产品 / 工程;用户 = 购物者、卖家以及本地买卖双方;付款方 = 消费者产品组织交易平台规模下的转化优化、发布控制和实验HelloFresh、Whatnot 和 OfferUp 案例直接关联结账、激活和变现指标未披露队列留存或卖家 / 买家拆分
社交、创作者与媒体平台买方 = 增长 / 平台 / 数据团队;用户 = 消费者、创作者、听众和社区;付款方 = 平台团队预算参与度测量、发布安全和产品分析Bluesky、SoundCloud 和 Linktree 案例,以及 Bluesky 和 Linktree 官方页面大体量受众面让实验规模具备战略价值公开证据没有揭示收入集中度或模块渗透率
使命驱动型教育买方 = PM / 数据 / 教育平台团队;用户 = 教师和学生;付款方 = 非营利运营预算预算约束下的产品可见性和实验Code.org 案例,加官方影响力页面说明 Statsig 即使在工具预算受限的场景也能赢单未公开定价条款或续约数据
消费者 AI 应用买方 = 产品 / ML / 平台团队;用户 = 与模型交互的终端消费者;付款方 = AI 产品预算安全且有吸引力的模型发布和评估循环Character.ai 案例把平台延展到主观且安全敏感的 AI 产品决策客户侧佐证弱于最强的软件客户参考
开发者工具买方 = 工程平台负责人;用户 = 软件工程师和评审者;付款方 = 工程效率预算功能门控、动态配置和受控发版工作流Graphite 案例,加官方主页发布精度和事故响应是核心时,契合度强未公开 ARR 或账户规模
健身与健康应用买方 = 增长 / 产品团队;用户 = 订阅者和跑者;付款方 = 产品驱动订阅预算测试新用户引导、高级版转化和留存杠杆Runna 案例,加官方主页说明 Statsig 能支撑高速迭代的消费者订阅应用未披露实际留存百分比或续约经济性

分群来自抓取到的具名案例研究,以及用于佐证的客户官方页面。该分群显示出客户类型广度,但公开证据仍偏向互联网原生软件和消费者应用。

[CU001, CU002, CU003, CU004, CU008, CU010]
FU001: 客户旅程图

Statsig 最强公开客户案例中,典型买方、用户和付款方的旅程。

[CU002, CU003, CU021, CU022, CU035, CU038]

6.2 具名客户证明、引用质量与新鲜度

Statsig 客户故事最强的一点,是它远不止 logo 墙。抓取到的案例包含多家具名客户的量化运营结果:Ancestry 称实验从每年 70 次升至 600 次;HelloFresh 称每年达到 1,000 次实验,同时提高决策质量;Brex 报告数据团队效率和工具整合节省;Bluesky 将平台与快速增长和发布活动相连;Whatnot 描述从零爬升到每年数百次实验;Graphite 披露活跃使用功能门控和配置;OfferUp 报告转化率提升 11%;Runna、SoundCloud、Notion、Webflow、Linktree、Code.org 和 Character.ai 都描述了有明确规模背景的生产用例。但引用质量仍混合。几乎所有部署细节都来自供应商撰写、当前仍在线的 Statsig 页面,而不是客户撰写的工程文章、备案文件或采购记录。客户侧官方页面确实佐证许多具名 logo 真实、当前有效且是大型平台,但这些页面很少直接提到 Statsig。因此,本章支持真实采用和部分具体结果,但对多数 logo 仍未达到完全独立证明。[CU005, CU006, CU007, CU009, CU011, CU012]

客户增长 / 采用轨迹表
指标数值日期 / 期间来源置信度含义缺失分母
Ancestry 实验速度70 次手工实验/年增至 600 次实验/年;3.5M 客户受益无日期案例,抓取于 2026-05-28Statsig Ancestry 案例显示成熟的重复使用,以及组织层面对实验的投入未披露商业条款或续约数据
HelloFresh 实验规模1,000 次实验/年;决策时间缩短 25%;形成决策的实验增加 55%无日期案例,抓取于 2026-05-28Statsig HelloFresh 案例规模化实验运营最强公开证据未披露支出、合同期限或按地区划分的模块采用
Brex 效率提升与整合数据科学家时间减少 50%;成本节省 20%+;100+ 次实验无日期案例,抓取于 2026-05-28Statsig Brex 案例同时支撑金融科技账户内的 ROI 和模块整合未披露合同价值或留存队列
Bluesky 增长运营节奏7 个月 30+ 次实验;借助 Statsig 发布 25+ 次;标题 / 上下文提到 25M 用户无日期案例,抓取于 2026-05-28Statsig Bluesky 案例显示早期社交网络规模与频繁发布并存未披露 ARR 贡献或扩张时点
Whatnot 实验节奏年度实验从 0 增至 400 次;第二年三个季度做了 250+ 次实验无日期案例,抓取于 2026-05-28Statsig Whatnot 案例采用从功能开关走向系统化实验的清晰证据未披露客户支出或续约信息
Graphite 工作流深度321 个活跃功能门控;168 个动态配置;事故解决速度提升 50%+无日期案例,抓取于 2026-05-28Statsig Graphite 案例显示深度嵌入运营,而不是轻量试点未披露商业范围
OfferUp 商业结果目标 CTA 点击量增加 11%无日期案例,抓取于 2026-05-28Statsig OfferUp 案例证明可衡量的交易平台转化影响未披露提升是否持续或合同是否扩张
Runna 测试节奏一年内 100+ 次测试;聚焦高级版转化、新用户引导和留存无日期案例,抓取于 2026-05-28Statsig Runna 案例暗示订阅型健身应用频繁迭代未披露实际留存或变现百分比
Linktree 运营面每月 500M 独立访客和 2B 次展示无日期案例,抓取于 2026-05-28Statsig Linktree 案例,加 Linktree 主页大规模使用面提高了实验和 Warehouse Native 分析的战略价值未披露直接结果增量或续约信号
SoundCloud 受众与平台背景375M 首曲目;40M 位艺术家;193 个国家;盈利 / 动态流案例无日期案例,抓取于 2026-05-28Statsig SoundCloud 案例显示 Statsig 用在一个超大消费者媒体场景中抓取资料包中没有独立的客户侧佐证

由于 Statsig 不发布标准化客户 KPI 包,这张表把直接结果指标、实验量指标和运营规模背景放在一起。缺失的分母主要是商业披露缺口,而不是产品采用缺口。

[CU005, CU006, CU007, CU008, CU011, CU012]
具名客户证据表
客户细分领域部署 / 用例生产部署 / 试点量化结果限制
Ancestry消费者订阅 / 家谱贯穿产品生态的实验与留出组生产部署年度实验从 70 个增至 600 个;3.5M 客户受益供应商撰写的案例研究;没有商业合同细节
HelloFresh餐包 / 数字 CPG端到端实验平台和数据仓库原生工作流生产部署1,000 个实验 / 年;决策快 25%;带结论的实验多 55%未披露留存、ARR 或区域级模块拆分
Brex金融科技 / 金融软件整合产品数据、实验和分析生产部署数据科学家耗时减少 50%;成本节省 20%+;100+ 个实验客户侧佐证确认的是 Brex 规模,不是 Statsig 部署
Bluesky消费者社交网络用产品分析、功能门控和实验支撑快速增长生产部署7 个月 30+ 个实验;25+ 次发布;25M 用户背景案例未标日期,也没有公开商业细节
Whatnot直播购物市场在生产环境测试并建立实验文化生产部署年度实验从 0 到 400 个;第二年三个季度内完成 250+ 个所查看的 Whatnot 首页没有提供 Statsig 部署的实质信息
SoundCloud创作者 / 音乐平台从只用门控发布,转向由实验驱动产品开发生产部署盈利 / 信息流叙事,提供 375M 首曲目、40M 艺术家、193 个国家背景已获取材料中的客户侧官方佐证较弱
Linktree创作者工具数据仓库原生实验和决策支持生产部署500M 独立访客;2B 次展示 / 月背景规模背景强,但公开要点未引用直接因果业务变化
Code.org非营利教育科技面向学生和教师的分析与实验可见性生产部署案例称有数百万用户;官网页面显示 107M 学生账户和 3M 教师无公开定价、续约或合同条款
Graphite开发者工具开发工作流中的功能门控和动态配置生产部署321 个活跃门控;168 个配置;事故解决速度快 50%+无公开合同金额或买方席位数
Runna健身 / 订阅应用测试聚焦付费转化、引导和留存生产部署一年内 100+ 次测试虽是留存导向用例,但未发布留存百分比
Notion生产力 SaaS用实验平台提高迭代速度生产部署每季度实验数从个位数到数百个结果细节少于 HelloFresh 或 Brex
WebflowWeb 体验平台统一功能开关、实验和配置,将发布与部署解耦生产部署工作流变化清楚,但已获取要点中没有单一核心 KPI运营收益是定性描述,不是已披露商业指标
OfferUp本地商务市场移动优先市场上的转化与产品优化生产部署目标 CTA 点击增长 11%没有持续使用或续约证据
Character.ai消费者 AI 娱乐用实验判断新模型何时可大范围发布生产部署数千万用户;需要大量并行实验参考质量较低,因为客户侧佐证来源未成功获取

这份枚举优先收录有可用细节的具名生产部署参考,刻意保持部分覆盖而非穷尽;公开记录没有披露完整客户名单、续约情况或所有客户侧佐证。

[CU005, CU006, CU007, CU008, CU009, CU010]
参考质量与新鲜度表
客户 / 队列主要证据来源客户侧佐证结果具体度新鲜度可见性主要保留意见
AncestryStatsig 客户案例Ancestry 官网高:年度实验 70→600,3.5M 客户受益中:运行日期页面可访问,但未标日期供应商撰写案例
HelloFreshStatsig 客户案例HelloFresh 官网高:1,000 个实验 / 年和决策质量指标中:运行日期页面可访问,但未标日期客户侧未提及 Statsig
BrexStatsig 客户案例Brex 官网高:效率和成本指标,加上实验数量中:运行日期页面可访问,但未标日期佐证指向 Brex 本身,而非部署条款
BlueskyStatsig 客户案例Bluesky 关于页面中高:发布与实验节奏,并有用户规模背景中:运行日期页面可访问,但未标日期仍由供应商撰写
GraphiteStatsig 客户案例Graphite 官网高:活跃门控 / 配置和事故解决改善中:运行日期页面可访问,但未标日期商业可见性仍低
Code.orgStatsig 客户案例Code.org 影响力页面中:问题 / 可见性叙事,加上很强的客户规模背景中:页面可访问,但案例研究未标日期没有明确商业 KPI 或定价细节
Notion 和 Webflow 队列Statsig 客户案例Notion 和 Webflow 官方公司页面中:工作流主张强,但硬 KPI 披露少于 HelloFresh / Brex中:页面可访问但未标日期运营改善比商业结果更清楚
SoundCloud / Whatnot / Character.ai 队列Statsig 客户案例已获取材料中的客户侧佐证较弱或有限中:案例中有可用运营细节中:页面可访问但未标日期独立性弱于佐证最佳的行

新鲜度按当前可获取性和页面上的时间信息判断,因为多数 Statsig 案例研究没有明确发布日期。这张表评估证据质量,不评估客户价值。

[CU023, CU024, CU037]
FU003: 客户证据矩阵

按独立性、结果具体性、新鲜度可见性和商业可见性排序的公开客户证据。

[CU023, CU024, CU036, CU037, CU039]

6.3 耐久性、满意度与扩张代理

公开证据通过代理指标支持可信的先落地、再扩张动作。反复出现的客户模式不是小型单用途部署。Webflow 替换了一套自研门控、实验和配置拼出来的系统。Brex 描述整合分析和实验工作流。HelloFresh 从笨重的内部设置迁移出来。Code.org 曾被早期工具的价格和限制约束。SoundCloud 和其他客户在选择统一平台前,明确权衡过自研还是购买。换句话说,胜利条件通常从功能开关或发布控制开始,然后扩展到实验、分析、仓库原生使用或动态配置。独立评论证据部分支持这个故事:TrustRadius 给 Statsig 8.9/10 分,并包含替换 Mixpanel 和 LaunchDarkly 的例子,这与产品表面扩张一致。同时,同一评论来源也反复指出 UI 复杂、分群限制、元数据处理和配置易用性问题。结论是,当技术团队跨模块扩大使用,客户粘性看起来可信;但公开证据仍未披露实际续约率、收缩或队列留存。[CU021, CU022, CU027, CU028, CU029, CU031]

留存 / 重复使用 / 满意度表
指标数值 / 状态细分领域置信度尽调问题
净留存率(NRR)未公开披露整体客户组合索取按产品层级和头部客户队列拆分的 NRR
总留存率(GRR)/ 客户流失未公开披露整体客户组合索取 GRR、客户留存率和收缩率
续约率 / 合同年限未公开披露企业级和中端市场客户索取合同期限中位数、续约时间点和续约率
重复实验节奏代理指标强:Ancestry、HelloFresh、Whatnot、Runna 和 Bluesky 均提到持续测试计划软件原生产品团队拆分每月或每季度跑实验的客户数
多模块扩张代理指标强:Brex、Webflow、Linktree、Graphite 和 HelloFresh 都描述多个 Statsig 产品面采用功能开关以外模块的账户索取模块渗透率和附加率
独立满意度代理指标TrustRadius 评分 8.9/10,来自 5 条评价评价网站用户索取更大样本的 NPS / CSAT 或客户参考名单
独立摩擦代理指标TrustRadius 用户提到 UI 复杂、分群限制、元数据不直观和配置混乱技术买家和操作人员索取实施时间、支持负担和流失原因
供应商筛选证言代理指标Whatnot、SoundCloud 和 Ancestry 负责人提供具名正面引述公开参考集合用带日期的续约记录或客户自写文章验证这些参考

这些是耐久性代理指标,不是真正的商业留存指标。公开记录更能说明客户多频繁使用产品,却很少说明他们是否续约、扩张或流失。

[CU020, CU021, CU027, CU028, CU029, CU031]
FU002: 采用 / 部署漏斗

企业和产品团队从存量痛点走向更广泛 Statsig 采用的常见路径。

[CU007, CU012, CU018, CU021, CU022, CU027]

6.4 集中度风险、证明集中与尽调缺口

缺口在于客户集中度和商业耐久性。已抓取来源没有披露 NRR、GRR、流失、Logo 留存、续约率、标准合同期限、席位数或头部客户收入占比。这不等于客户基础薄弱;只是公开记录无法把那份亮眼的客户名单转化成可承保的收入质量。公开证据也集中在特定客户类型。可见案例主要来自互联网原生软件、消费应用、数据驱动产品团队和实验成熟的组织。Whatnot 的美国和欧洲覆盖、SoundCloud 覆盖 193 个国家、Bluesky 拥有数百万用户、Linktree 的全球创作者基础,以及 Code.org 的全球教育使命,都能证明地理范围够广;但地理分布本身不能化解按客户类型或支出队列划分的集中风险。因此,投资人应把 Statsig 的公开客户叙事看作:具名证据强,独立性和新鲜度中等,留存和集中度披露偏弱。下一步尽调不是再找更多 Logo,而是拿到带日期的续约案例、前 10 大客户敞口、模块渗透率,以及队列层面的留存或扩张数据。[CU004, CU023, CU024, CU030, CU031, CU032]

扩张与集中度风险表
扩张驱动 / 风险类型影响尽调路径
统一平台替代多种工具扩张驱动从功能开关扩展到实验、分析、配置或数据仓库原生工作流后,账户粘性可能提高索取模块附加率、每位客户购买模块数和扩张时间线
数据仓库原生和数据集成用例扩张驱动指标、数据和实验标准化后,数据团队锁定更深索取数据仓库原生客户贡献的 ARR 占比,以及主导的数据仓库
消费者应用和市场的高频实验节奏扩张驱动高频发布和度量闭环可能提高切换成本索取月活实验人员数,以及按队列拆分的头部客户事件量
公开证据集中在互联网原生软件和消费者应用集中度风险如果 ARR 也同样集中在产品驱动的软件客户,公开参考可能高估多元化程度索取按行业 / 客户原型拆分的 ARR
未披露头部客户敞口集中度风险少数大客户可能贡献过高收入或用量,但公开层面看不出来索取前 10 大客户 ARR 占比和最大单一客户占比
最详细案例由供应商撰写,且通常未标日期参考质量风险新鲜度和独立性弱于客户自写工程文章或监管文件索取带日期的客户参考、续约记录或联合公告
评价网站投诉指向入门和 UX 摩擦留存风险产品复杂度可能拖慢高度技术团队以外的采用索取实施时间线、支持工单和丢单原因
不同客户名下的客户侧佐证强弱不一参考质量风险部分客户有强官方身份 / 规模证据,另一些主要依赖 Statsig 页面优先对 SoundCloud、Whatnot、Character.ai 等佐证较弱客户做直接客户访谈

这张表把增长向量和投资判断风险拆开看。商业披露缺失本身就是风险,因为公开 logo 证据难以转化为可持续收入质量证据。

[CU021, CU022, CU029, CU032, CU033, CU038]

6.5 图表

Chapter 07

07风险

7.1 交易、竞争和定价风险

最大残余风险不是产品 Bug,而是平台未来究竟由谁掌控出现漂移。官方收购公告显示,Statsig 同意在 2025 年 9 月加入 OpenAI,Vijaye Raji 将转任 Applications CTO,交易完成后员工预计成为 OpenAI 员工。但当前公开法律材料已经归在 Amplitude 名下,MarTech 报道称 Amplitude 将接管 Statsig 品牌、客户基础、路线图和支持,原团队则留在 OpenAI。这种组合正好制造企业买家最不喜欢的连续性风险:代码可能延续,但关键运营者、统计专家和产品负责人坐在别处。竞争风险还会放大这一过渡。Statsig 低端定价有吸引力,但扩张逻辑依赖把整合平台卖进客户账户;这些客户同样可以选择 PostHog 慷慨的免费层、Amplitude 更低的入门价格,或 Optimizely 的既有企业套件。如果 Amplitude 整理重叠产品或重新定价过猛,流失风险会上升,定价权也会同时被压缩。[CR003, CR004, CR005, CR007, CR008, CR009]

合作伙伴 / 依赖风险台账
依赖项交易对手 / 系统价值链角色失效场景严重性剩余敞口缓释 / 监控
交易后产品归属OpenAI / Amplitude / Statsig品牌、代码、路线图、支持再次过渡后,路线图或支持断裂严重锁定合同负责人、路线图承诺和具名支持升级路径
套件级竞争定价PostHog / Amplitude / Optimizely替代性实验与分析套件扩张停滞,或折扣明显加大每季度将 TCO 与竞争套件对标
客户数仓平台Snowflake / BigQuery / Databricks 等数仓原生模式的执行底座治理或成本架构让部署慢于计划,或成本高于计划全企业推广前先用真实工作负载试点
基础设施与 AI 子处理方AWS / Azure / Google / OpenAI / Snowflake 等托管、监控、模型功能、数仓供应商宕机、政策变化或新子处理方疑虑阻塞部署跟踪子处理方通知,并要求事件与供应商管理报告
参考客户相似度AI 原生和产品驱动型客户支撑企业销售的证明集公开案例无法对应买方复杂度或合规需求与相似的受监管或采购流程重的客户做参考客户访谈
未披露的客户集中度未公开披露的大客户收入韧性一两家大客户贡献的 ARR 高于预期Unknown要求披露前十大客户 ARR 结构、续约时间表和按队列的扩张

最难处理的依赖是组织和商业层面的,不是纯技术问题。

[CR005, CR007, CR008, CR009, CR011, CR020]
FR001: 风险热力图

当产品转型不确定性与仓库原生实施负担、定价重叠交叉时,剩余风险最高。

[CR011, CR018, CR037, CR041, CR043, CR044]
FR003: 依赖图

Statsig 位于客户数据仓库、IAM 控制、云厂商和仍在演进的收购后商业结构交汇处。

[CR001, CR012, CR020, CR021, CR025, CR026]

7.2 仓库原生治理、隐私和运营风险

Statsig 的仓库原生设计是真正的差异化,但也带来不小的实施和治理负担。产品需要服务用户、作业执行权限、读取客户指标和曝光数据的权限,以及写入专用暂存环境的权限。Statsig 表示,一些聚合结果和诊断仍会存储在其服务器上;自家文档也警告,随着实验量上升,实验暂存数据可能快速膨胀。最佳实践指南明确要求客户使用分区、聚类、增量重载和范围化资源,才能控制支出和资源争用。安全姿态显著好于不透明初创公司的基线:Statsig 宣传 SOC 2 Type II、加密、备份、灾难恢复和事件响应。但隐私和法律栈也说得很清楚:客户仍是控制者,标准条款下不应发送敏感数据,HIPAA 支持需要 BAA,供应商链条现在横跨 Amplitude 品牌下的法律所有权以及多个 AI 和基础设施子处理方。纪律严明的团队能管住这些问题,但并非无摩擦。[CR012, CR013, CR014, CR015, CR016, CR017]

监管 / 法律风险登记表
风险法域 / 范围可能性严重性缓释成熟度剩余敞口具体尽调问题
控制者 / 处理者划分将责任上推全球隐私法客户仍需承担合法依据和数据最小化错误获取最新 DPA 修订版,并确认实施指南中的禁止数据政策
标准 DPA 排除敏感或特殊类别数据GDPR / CCPA / 标准合同默认合同可能不适合受监管或医疗健康用例确认是否有部署需要修订处理条款或数据隔离
HIPAA 部署需要 BAA 和客户保障措施美国医疗健康医疗健康扩张受合同和运营控制卡住审查当前 BAA 模板、审计权和 PHI 处理边界
跨境传输和子处理方变更管理EEA / 英国 / 瑞士 / 美国客户必须监控 SCCs、DPF 依赖和新增供应商索取子处理方通知工作流和最新传输影响材料
OpenAI / Amplitude 过渡后的交易对手和所有权不明确商业合同极高买方可能在路线图负责人不清晰时签下长期条款索取当前 MSA、签约实体、支持 SLA 负责人和路线图函
AI 功能治理,以及 AI 应用客户的 AI Act 敞口欧盟和全球 AI 买家AI 评测和提示词功能可能带来额外治理义务梳理已启用哪些 AI 功能、触达哪些数据,以及在哪些地区受限

各行只反映公开法律姿态;内部法律备忘录、执法历史和谈判形成的客户例外均不可得。

[CR006, CR015, CR027, CR028, CR030, CR031]
运营 / 质量 / 安全风险登记表
失效模式可能性严重性缓释成熟度剩余敞口证据 / 含义
数据仓库权限配置错误服务用户的读 / 写 / 查询权限会带来真实的配置影响半径
数据仓库计算和存储成本失控Statsig 文档明确提醒扫描成本、物化成本和实验表增长风险
数据出站和诊断数据留存误解部分汇总结果和短期诊断数据仍会落到 Statsig 服务器
多云或子处理方政策扰动生产与 AI 功能依赖多个基础设施和模型供应商
基线控制虽强,仍发生安全事件严重SOC 2、加密、DR 和 IR 有帮助,但隐私政策仍声明无法保证绝对安全
方法误用带来虚假信心顺序检验、多臂老虎机和 AI 评测文档都说明了误用或解读风险

运营暴露既取决于供应商平台本身,也同样取决于客户实施纪律。

[CR013, CR014, CR015, CR016, CR017, CR018]

7.3 方法论、参考样本和人员风险

Statsig 最强的产品信息是统计能力成熟,但这把刀两面都锋利。平台把贝叶斯和频率学派方法、CUPED、序贯检验、多臂老虎机、AI 评估、防护指标和发布控制直接交给操作人员。文档本身承认,偷看会抬高假阳性,早期统计显著的读数仍可能功效不足,普通多臂老虎机可能收敛到对子群错误的答案,而 LLM-as-a-judge 天然不完美。这些不是避开产品的理由,而是把操作人员质量纳入风险包的理由。人员侧同样重要。公开客户证据集中在数字原生、AI、市场平台和电商 Logo,这些客户本来就有强产品、工程和数据职能。对成熟买家来说这是正向信号,但参考样本也偏向那些能承受仓库设置、策略配置和实验严谨性的团队。与此同时,公开材料没有披露客户集中度或续约敞口。OpenAI 和 Amplitude 过渡之后,买家同时面对执行风险和证据选择偏差。[CR041, CR044, CR046, CR047, CR048, CR049]

人员 / 执行风险台账
角色 / 能力依赖或缺口可能性严重性当前缓释措施剩余敞口尽调要求
创始人 / CEO 领导力Vijaye Raji 转任 OpenAI 领导岗位公开承诺 Statsig 继续独立运营确认现在谁日常负责产品和客户升级
原有统计与产品专家核心建设者可能在 OpenAI,而不在产品 P&L 负责人一侧文档和现有代码库仍在要求提供实验科学、产品和支持团队组织图
企业管理成熟度SSO、SCIM、策略、模板和审核需要配置已有企业访问管理工具验证买方是否有具名平台负责人推动治理落地
数据工程成熟度数仓原生效率取决于分区、聚类和范围化计算最佳实践文档很详细与买方实际数仓团队一起建模部署成本
支持与变更管理客户体验取决于过渡后的支持深度和响应时间企业版售卖优先支持获取 SLA、具名 TAM 覆盖和支持指标

如果买方指望平台在缺少内部数据和平台负责人的情况下自我治理,执行风险会上升。

[CR005, CR018, CR019, CR026, CR039, CR041]
FR002: 风险传导图

几类主要风险会传导到相同下游结果:采用放慢、扩张转弱,以及投资测算信心下降。

[CR010, CR011, CR013, CR017, CR019, CR022]

7.4 缓释措施、否决标准和具体尽调问题

公开缓释措施存在,但大多停留在框架层面,而非结果层面。Statsig 记录了访问控制、评审、审批、策略设置、诊断和安全实践,说明公司理解相关失效模式。未解决的问题是,在公司过渡期里,这些控制是否被持续执行;客户能否在不付出过高内部成本的情况下落地这些控制。对承保来说,最干净的框架是把未决问题改写成明确否决标准。如果交易对手、支持负责人或路线图承诺再次变化,部署就需要重新承保。如果仓库原生计算成本或资源争用结构性高于预期,所谓治理和 TCO 优势会很快变弱。如果 HIPAA、DPA 或子处理方治理无法达到受监管买家的文档标准,可触达市场会收窄。如果相对 PostHog、Amplitude 或 Optimizely 的套件经济性恶化,扩张假设就应下调。因此,剩余尽调包应聚焦过渡后的商业连续性、实际客户集中度、受监管企业证据,以及现实的仓库成本模型。[CR025, CR033, CR037, CR045, CR056, CR057]

缓释与否决标准表
风险领域可监控触发项阈值 / 事件行动含义
交易对手与路线图连续性商务文档漂移尽调后 MSA / DPA / 支持负责人再次变化重新尽调定价,或暂停部署 / 投资
定价与产品重叠相对核心替代方案的 TCO在买方用例上,套件经济性不再优于 PostHog / Amplitude / Optimizely重置扩张假设,并要求价格保护
数仓原生经济性计算、存储与资源争用指标试点工作负载显著超过建模的数仓成本,或阻塞核心 ETL 窗口收缩推广范围,或要求托管替代方案 / 专用计算
隐私与受监管行业准备度必需合规材料无法提供 BAA、DPA、子处理方通知或受监管客户访谈视为不支持垂直行业扩张假设
方法纪律实验与评测治理指标多次提前停止错误、策略覆盖不足,或未评审的 AI 评测上线更广泛采用前要求更严格的策略控制
客户耐久性集中度与续约可见性头部客户结构或续约时间显示集中度高于承销假设提高要求回报,或降低持仓规模

这些否决标准把公开风险证据转成具体的投资组合或采购监控检查点。

[CR056, CR057, CR058, CR059, CR060]

7.5 图表

Chapter 08

08估值

8.1 投资逻辑和反向逻辑

Statsig 估值最强的理由一直是战略价值,而不是纯财务逻辑。公司拼出了一套广泛的产品开发栈,把功能发布、实验、分析、性能监控和会话回放放在一起,并用按量计费模式销售;部署既可跑在 Statsig 云端,也可用仓库原生方式跑在客户基础设施上。第三方估计显示,2025 年 5 月 Series C 前后,公司 ARR 大约 $40 million,付费客户超过 300 家,日事件量超过 1 trillion;客户案例称,买家在选择 Statsig 的整合工作流前,也评估过 Optimizely、LaunchDarkly、Split 和 Eppo 等既有厂商。这些都是真实证据点。反向逻辑在于,保留下来的公开来源没有披露净留存、毛利率、流失或 Series C 的确切条款,因此公开记录无法证明,对于财务投资人而言,27.5x ARR 价格是合理的;它更像战略买家为稀缺实验基础设施支付的价格。[CV001, CV006, CV007, CV008, CV009, CV010]

投资论点 / 反论点表
论点当前证据改变判断的证据
一体化产品开发栈可以获得溢价Statsig 把功能发布、实验、分析、性能监控和会话回放打包进一个平台。证明平台不只是覆盖面广,还能产出软件级留存和利润率指标。
用量计费和数仓原生部署可随客户放大定价按用量计量,文档同时描述数仓原生和托管部署选项。披露 ARR 有多少来自事件量扩张,而不是一次性或低质量用量。
客户证据显示真实竞争胜出客户案例称,买方在选择 Statsig 前评估过 Optimizely、LaunchDarkly、Split 和 Eppo;其中一个部署扩展到数亿用户。提供更独立的客户留存和扩张数据,而不是筛选后的推荐语。
反方:公开可比公司测算低得多按保留的 ARR 估计,Statsig 隐含约 27.5x;Amplitude 交易约 ~2.4x,Datadog 约 ~19.6x。更高的经审计 ARR 基数,或更清晰的战略竞标证据,才会缩小差距。
反方:退出价格与该轮持平OpenAI 的 $1.1B 交易验证了 Series C,但未披露高于该轮的升值。后续披露加价,或更清晰的交割后经济性,会改善财务投资者案例。
反方:治理现在分散在 OpenAI 和 AmplitudeOpenAI 保留公司,Amplitude 接手品牌、客户和平台运营。明确的交割后权利、经济性和支持义务,可以判断连续性风险是否被高估。

正面逻辑是战略稀缺和产品宽度;负面逻辑是实际退出没有上行,且经济性缺失。

[CV009, CV010, CV011, CV012, CV016, CV017]
FV001: 建议逻辑

战略买家看 Statsig 估值合理,但财务投资吸引力不足,因为产品广度和稀缺性抵消了弱公开可比公司的支撑。

图中呈现的是投资测算逻辑,不是产品内部流程。

[CV009, CV006, CV007, CV014, CV016, CV037]

8.2 融资背景和资本历史

融资时间线在标题层面异常清晰,但对后期承保最关键的细节仍不透明。Statsig 在 2025 年 5 月宣布以 $1.1 billion 估值完成 $100 million Series C;Sacra 保留摘要称,本轮包含 $80 million 新股和 $20 million 员工老股流动性。这很重要,因为如果 $1.1 billion 是投后估值,隐含新钱稀释只有约 7% 到 8%,对这个规模的融资来说偏温和;交易其余部分是持有人流动性,而非公司融资。Sacra 还把 Series A、Series B 和 Series C 的累计融资额估计为约 $153.4 million。随后 OpenAI 在 2025 年晚些时候同意以同样 $1.1 billion 收购 Statsig,这验证了战略买家眼中的本轮价格,但没有披露高于该价格的加价。公开来源仍看不到优先股堆叠、清算瀑布或参与权;这些信息才能告诉外部投资人,标题估值在经济上比表面更干净还是更苛刻。[CV001, CV004, CV005, CV014, CV020, CV021]

资本历史 / 稀释桥接
事件资本金额公开口径估值 / 状态大致所有权或稀释信号含义
Series A 轮(2021)~$10.4M留存公开来源未披露估值早期支持方资本;所有权影响未公开确立 Sequoia 和 Madrona 为持续支持方。
Series B 轮(2022)~$43M留存公开来源未披露估值2025 年价格重设前的成长资本把公司推到后来可能完成独角兽轮的阶段。
Series C 轮新股融资(2025)~$80M$1.1B 估值、$100M 轮次的一部分如果 $1.1B 是投后估值,约占投后股权的 7.3%对一个打出独角兽估值的轮次来说,名义稀释看起来不高。
Series C 轮老股流动性(2025)~$20M同一笔 $100M 交易的一部分转移所有权,但不稀释公司显示该轮也服务于员工流动性,而不只是给资产负债表补钱。
出售前累计融资~$153.4MOpenAI 后来同意以 $1.1B 收购 Statsig累计融资约等于出售价值的 14%说明资本效率尚可,但瀑布经济性仍未披露。

稀释计算只是近似值,因为留存公开来源未披露完整股权结构表或精确投前条款;7.3% 假设报道的 $1.1B 估值是投后估值,且只有 $80M 新股部分稀释公司。

[CV004, CV005, CV021, CV022, CV049]

8.3 可比公司和倍数缺口

公开可比缺口很大。已抓取样本中最接近的上市产品分析参照 Amplitude 报告 ARR 为 $374 million,2026 年 Q1 收入为 $93.5 million,但 CompaniesMarketCap 显示其 2026 年 5 月市值仅约 $0.91 billion,按 ARR 约 2.4x。Datadog 是更宽的可观测性和开发者平台,现在通过 Eppo 纳入实验资产;其报告 2026 年 Q1 收入 $1.006 billion、非 GAAP 运营利润率 22%、市值约 $78.95 billion,相当于年化 Q1 销售额约 19.6x。因此,Statsig 隐含 27.5x ARR 倍数高于两个公开参照,尽管规模小得多、披露也更薄。Optimizely 只适合作为战略历史背景,因为 Episerver 在 2020 年收购它时价格未披露。实验领域最接近的 M&A 数据点是 Datadog 收购 Eppo,据报道对价约 $220 million,但分母未披露。换句话说,Statsig 价格唯一干净的理由是战略稀缺,而不是普通的公开可比数学。[CV023, CV024, CV025, CV026, CV027, CV028]

可比估值表
可比对象规模指标倍数 / 估值 / 状态相关性局限
Statsig Series C 轮~$40M ARR;300+ 付费客户2025 年 5 月:$1.1B;~27.5x ARR公司自身的主要融资锚点。ARR 来自第三方估计,未经审计,该轮条款也未公开。
OpenAI 收购 Statsig同一保留 ~$40M ARR 框架;战略买方结果2025 年 9 月:$1.1B 全股票交易最新披露的定价事件,也直接验证了该轮价格。战略买方逻辑未必适用于财务买方定价。
Amplitude2026 年 Q1 ARR $374M;Q1 收入 $93.5M~$0.91B 市值;~2.4x最接近的在市公开产品分析基准。增长较慢的上市公司,利润率和治理画像不同。
Datadog2026 年 Q1 收入 $1.006B;non-GAAP 经营利润率 22%~$78.95B 市值;年化 Q1 销售额 ~19.6x高端可观测 / 实验套件基准。平台更宽、规模更大、盈利能力强得多。
Datadog 收购 Eppo实验资产;收入未披露据报 ~$220M;条款未披露显示 2025 年战略买方对实验资产有需求。没有收入分母,因此不是干净的倍数可比对象。
Optimizely / Episerver 历史签约时 >1,000 家客户;2020 年品类历史价格未披露;战略整合参考可用于理解实验整合的品类历史背景。保留来源中没有披露收购价,也没有公开 2026 年估值。

这是本章使用的完整决策相关可比篮子:两个 Statsig 价格锚、两个公开软件参照、两个战略历史参照。

[CV019, CV023, CV025, CV026, CV027, CV030]
FV002: 估值敏感性

Statsig 实现的 27.5x ARR 锚点,明显高于抓取到的最接近公开参照。

倍数在公开来源只提供这些指标时混用了 ARR 和年化销售额。目的在于方向性比较,而不是精确同口径估值。

[CV019, CV026, CV031, CV037, CV041, CV042]

8.4 情景分析和已实现结果

给 Statsig 估值,最清楚的方法是把已经实现的战略结果和反事实的财务买方案例分开。在保留的约 $40 million ARR 估计上,10x 倍数意味着约 $400 million 股权价值,15x 意味着约 $600 million,20x 意味着约 $800 million。实际结果——先是 Series C,随后是 OpenAI 交易——以 27.5x ARR、即 $1.1 billion 成交。由此,已实现基准情景很容易表述:战略买家愿意支付稀缺性溢价,但没有公开来源显示高于这一价格的加价。悲观情景是,如果没有战略收购方,且公开软件倍数仍设定外沿边界,公司大概率会面对的结果。乐观情景则要求 ARR 基数远高于保留的 $40 million 估计,或出现第二个愿意进一步加价的战略竞标方。截至 runDate,已实现的 $1.1 billion 结果看起来是最相关的锚点,而不是通往下一轮私人加价的跳板。[CV006, CV019, CV020, CV038, CV039, CV041]

牛 / 基准 / 熊情景表
情景核心假设估值逻辑价值区间(十亿美元)概率信号
没有战略买方出现,公开软件可比公司主导定价,业务按 ~$40M ARR 评估,不给稀缺性溢价。高增长但非战略性软件结果隐含 10x-17.5x ARR 区间。0.4-0.7如果 Datadog 从未接触、OpenAI 也没有支付战略价格,这一情景本会更可能。
基准 / 已实现战略买方看重实验基础设施,并以 Series C 同价成交。按保留的 ~$40M ARR 估计,已实现 27.5x ARR。1.0-1.2这是已观察到的结果:OpenAI 同意支付 $1.1B,之后没有更高公开标记。
ARR 远超保留的 ~$40M 估计,或多个战略竞标方争夺该资产。需要明显更高 ARR,或另一场由稀缺性驱动的拍卖,才能支撑高于已实现交易的价值。1.4-1.8需要披露更快放量,或另一个愿意出价高于 OpenAI 的战略竞标方。

基准情景锚定已实现的战略出售;熊情景和牛情景是反事实承销区间。

[CV019, CV038, CV041, CV042, CV043, CV044]
FV003: 估值 / 回报区间

已实现的 $1.1B 战略结果接近基准情景中心,远高于非战略软件区间。

上述区间属于买方驱动的估值框架,不是贴现现金流输出。基准情景采用已实现交易作为最佳可观察锚点。

[CV038, CV041, CV042, CV043, CV044, CV045]

8.5 建议和估值立场

截至 2026-05-28,新财务投资人的正确决策是回避,不是因为 Statsig 产品质量不足,而是因为可投资的独立上行空间实际上已经关闭。OpenAI 同意以 $1.1 billion 收购公司,Amplitude 随后宣布将接管 Statsig 品牌、客户和平台运营,同时与 OpenAI 团队协作。这意味着最重要的经济结果已经确定,运营未来至少分裂在两个交易对手之间。如果问题是旧的 $1.1 billion 标记是否荒唐,答案是否定的:战略买家确实支付了这个价格。如果问题是,对于希望获得额外上行空间的新后期投资人,这个标记是否有吸引力,答案也是否定的:没有披露高于 Series C 的升值,公开可比仍低得多,足以支撑更高价格的缺失指标也从未公开。因此,估值立场是:对战略买家合理,对财务买家偏高;现在除了作为战略估值的已关闭案例研究,已不可操作。[CV014, CV016, CV017, CV020, CV037, CV038]

建议摘要表
维度评估重要性决策含义
建议不投新资本 / 战略结果已落地业务已经进入 OpenAI,运营平台由 Amplitude 共同承接。把 Statsig 视为已落地的战略案例,而不是开放的后期投资机会。
信心水平名义估值锚是真实的,但关键私有指标和条款仍未披露。用区间和买方特定逻辑,而不是精确内在价值结论。
风险评级对新财务投资者为高未披露高于 Series C 的加价,独立上行空间也不清晰。不要按旧 $1.1B 标记承销新进入。
估值立场对战略买方合理,对财务买方偏高OpenAI 确实支付了 $1.1B,但公开可比公司不支持普通财务买家按这个价格买入。将该标记表述为战略逻辑已验证,而不是明显便宜。
价格的验证因素实验基础设施的战略稀缺性Datadog 曾探索交易,OpenAI 支付 $1.1B,实验资产也在整合。即便公开倍数较低,也要承认稀缺价值存在。
什么会改变判断披露的留存、利润率、条款和交割后经济性这些指标能说明名义估值背后是隐藏下行,还是隐藏质量。没有这些证据,就继续观望。

这是买方特定判断表。同样 $1.1B 价格,对战略收购方可能合理,对新财务投资者可能缺乏吸引力。

[CV014, CV020, CV037, CV038, CV039, CV050]
FV004: 投资 KPI

产品质量和战略稀缺性得分较好;公开可比支撑和可投资性得分较差。

评分是内部投委会式判断,基于抓取来源集综合形成,并非外部报告的评级。

[CV022, CV037, CV038, CV047, CV048, CV050]

8.6 退出准备度、逻辑破坏点和最终尽调问题

未解决工作恰好集中在估值信心最该落地的地方:收入质量、轮次条款,以及交割后运营结构的经济性。保留来源没有披露 Statsig 的净留存、毛利率、流失、队列扩张或优先股堆叠,也没有解释 OpenAI 和 Amplitude 之间如何分配经济利益、客户权利、支持义务和 IP 管理权。这些遗漏很关键,因为它们决定 $1.1 billion 标题价代表的是纪律严明的战略承保,还是买家愿意为人才和位置支付溢价。实际尽调路径因此很窄也很具体。买家需要 Series C 和出售文件、队列层 ARR 和留存桥,以及 OpenAI-Amplitude 过渡协议的确切内容,才能把旧估值标记当作可复用先例。没有这些数据,本案应放进“已观察到的战略结果”桶,而不是“可重复估值基准”桶。[CV016, CV017, CV040, CV048, CV049, CV050]

论点破裂与否决触发项表
触发项阈值 / 事件为什么会破坏论点行动含义
OpenAI 与 Amplitude 之间没有清晰连续性有证据显示客户支持、路线图归属或 IP 管理明显碎片化。如果运营资产实质上被拆分,战略价格先例就更难复用。将 $1.1B 标记视为一次性人才或稀缺性采购,而不是可重复的平台价值。
客户质量不及预期过渡后披露任何流失、疲弱 NRR 或负面队列数据。只有收入质量极高,27.5x ARR 价格才说得通。站在财务买方视角,将公司估值标在已实现 $1.1B 战略结果之下。
隐藏轮次条款对投资者不友好参与权、优先清偿权或严苛分配瀑布显著扭曲普通股持有人经济性。名义估值可能夸大实际支付价格的经济性。对先例打折,并逐条承销条款。
公开可比公司压力恶化Amplitude 持续低迷、Datadog 估值下修,同时实验被视为功能而非平台。与 Statsig 的可比差距更难辩护。使用 10x-20x ARR 区间,而不是已实现的 27.5x 战略结果。
战略稀缺性消退没有第二竞标方、没有增量加价,也没有新披露指标支撑高于 OpenAI 价格的出价。已实现的持平结果会变成上限,而不是底线。不要假设历史交易价之上还有上行。

这些否决触发项面向投资者,帮助判断历史 $1.1B 结果能否作为新估值判断的先例。

[CV020, CV037, CV039, CV040, CV045, CV046]
最终尽调要求表
主题缺失证据重要性负责人 / 尽调路径
Series C 与出售条款Series C 与 OpenAI 交易中的优先权结构、参与权、期权处理和清算瀑布。该信息决定名义 $1.1B 数字是对应普通股持有人经济性,还是主要对应优先股保护。从律师和董事会数据室获取融资文件和出售文件。
收入质量出售时的净留存、客户数流失、毛利率、队列扩张和客户集中度。上述指标能说明 27.5x ARR 是有纪律的定价,还是激进定价。向财务团队索取运营模型和队列分析。
OpenAI-Amplitude 过渡经济性Amplitude 接手平台和客户群之后的商业条款、客户归属、支持义务和知识产权。拆分运营模式是复用这一先例时当前最大的疑点。审阅 OpenAI 与 Amplitude 之间的过渡服务和商业协议。
客户延续性过渡后关键账户的续约和扩张行为,尤其是具名企业客户。客户延续证据会降低外界对该战略结果主要是人才收购的担忧。访谈客户推荐人,并索取交易完成后的账户留存数据。
董事会和投资人回报$20M 老股交易和后续出售所得中,有多少流向员工、多少流向早期投资人。由此可厘清该结果按名义口径只是持平,还是按单个持有人口径更好或更差。向公司法律顾问索取收益分配瀑布和股权结构表桥接。

这些索取项按对估值信心影响最直接的程度排序。每一项都在缩小买方特定不确定性,而不是增加泛泛的尽调材料。

[CV016, CV017, CV040, CV048, CV049, CV050]

8.7 图表

免责声明

本报告仅用于尽调和信息参考,完全基于截至 2026-05-28 可获得的公开信息。Statsig 是私营公司,若干财务、所有权和交易后运营细节仍属估计、存在利益冲突或未披露;在作出任何投资或商业决定前,应独立核验。

证据索引

结论
编号陈述可信度来源
CO001 Statsig was founded in February 2021 by Vijaye Raji and other former Facebook colleagues. SO011, SO016
CO002 Statsig's founding thesis was to bring Facebook-style experimentation and feedback-loop tooling to broader product teams. SO011, SO016
CO003 Statsig's official platform scope currently spans experimentation, feature management, product analytics, session replay, web analytics, and adjacent developer tooling. SO001, SO006
CO004 Statsig documentation describes both cloud-hosted and warehouse-native deployment options. SO006, SO009
CO005 Public pricing shows a usage-based model with free Developer, paid Pro, and custom Enterprise tiers. SO005, SO009
CO006 Statsig's Developer tier includes 2 million metered events per month at no charge. SO005
CO007 Statsig's Pro tier starts at $150 per month and includes 5 million metered events before overage pricing begins. SO005
CO008 Statsig's homepage claims the platform processes more than 1 trillion events per day. SO001
CO009 Statsig's homepage claims the platform handles 2.5 billion unique monthly experiment subjects. SO001
CO010 Statsig's homepage claims 99.99% uptime for its API and console. SO001
CO011 Official Statsig pages repeatedly present the company as trusted by thousands of companies. SO001, SO004, SO007
CO012 Current Statsig web properties and careers materials identify Bellevue, Washington as the company's operating home. SO003, SO010
CO013 GeekWire reported in July 2025 that Statsig was moving into a new Bellevue headquarters at 3106 160th Ave. SE with room for 450 to 500 employees. SO022
CO014 Statsig's August 2021 Series A press release listed a Kirkland office address at 10510 Northup Way #205. SO015
CO015 Tracxn's current profile lists a Bellevue registered address while still labeling the company as Kirkland-based. SO028
CO016 Vijaye Raji created Microsoft Small Basic during his time at Microsoft. SO031, SO030
CO017 Meta's engineering blog profiled Raji on the Facebook Messenger team in Seattle in 2012. SO032, SO030
CO018 GeekWire said Raji later led Facebook's Seattle office and held vice-president roles in gaming and entertainment. SO019, SO030
CO019 Funding and acquisition announcements kept Raji in the dual role of founder and CEO through the September 2025 OpenAI deal announcement. SO020, SO021, SO010
CO020 Series C added ICONIQ Growth partner Murali Joshi to Statsig's board. SO021
CO021 Direct sources reviewed did not publish a full current Statsig board roster beyond Murali Joshi's disclosed addition. SO010, SO021, SO023
CO022 Statsig raised $10.4 million in Series A on August 5, 2021. SO015, SO016
CO023 Sequoia led Statsig's Series A, with Madrona and multiple angels also participating. SO015, SO016, SO028
CO024 Statsig raised $43 million in Series B on April 20, 2022. SO017, SO028
CO025 Sequoia led Statsig's Series B and Madrona participated. SO017, SO028
CO026 Statsig raised $100 million in Series C on May 6, 2025 at a $1.1 billion valuation. SO020, SO021, SO028
CO027 ICONIQ Growth led Statsig's Series C, with Sequoia and Madrona also participating. SO020, SO021, SO028
CO028 Public funding profiles put Statsig's cumulative funding at about $153 million to $153.4 million across three rounds. SO021, SO028, SO029
CO029 GeekWire reported $40 million in ARR at the time of Statsig's 2025 Series C round. SO021
CO030 Sacra separately estimated that Statsig reached $40 million in ARR in May 2025. SO029
CO031 Sacra says Statsig served more than 300 paying customers by 2025. SO029
CO032 Official and customer-proof pages name users including OpenAI, Microsoft, Notion, Brex, Atlassian, Bloomberg, Eventbrite, Ancestry, Headspace, and SoundCloud. SO004, SO010, SO018, SO020
CO033 GeekWire said in June 2023 that Statsig had hundreds of paying customers and thousands of active free-tier users. SO019
CO034 Statsig's founding story blog was published on March 17, 2021. SO011
CO035 Statsig launched Warehouse Native on June 14, 2023. SO018, SO019
CO036 The Warehouse Native launch added support for BigQuery, Snowflake, Redshift, and Databricks and framed the product around privacy and governance needs. SO018, SO009
CO037 GeekWire reported in June 2023 that Statsig had released AI experimentation features to test model cost, latency, and other performance metrics. SO019
CO038 January 2026 Statsig blog posts described new AI-assisted experimentation workflows and a knowledge graph linking code, experiments, and metrics. SO012, SO013
CO039 GeekWire said in July 2025 that Statsig had about 145 employees and expected to reach 200 by year-end. SO022
CO040 GeekWire said in May 2025 that Statsig had 140 employees and planned to reach 175 to 190 by early 2026. SO021
CO041 GeekWire said in September 2025 that Statsig employed 155 people when OpenAI announced the acquisition. SO026
CO042 Tracxn's current profile says Statsig has 55 employees as of an "Apr 26" snapshot. SO028
CO043 Public headcount signals therefore conflict, so a current independent employee count is not cleanly supportable from public evidence. SO021, SO022, SO026, SO028
CO044 On September 2, 2025, Statsig said it had signed a definitive agreement to join OpenAI. SO010, SO023
CO045 OpenAI said it was acquiring Statsig and planned to install Vijaye Raji as CTO of Applications. SO023, SO024, SO025
CO046 Reuters said the OpenAI deal valued Statsig at about $1.1 billion in all-stock consideration. SO024, SO027
CO047 Direct acquisition announcements described the transaction as subject to regulatory approval and customary closing conditions. SO010, SO023, SO024
CO048 TechCrunch said all Statsig employees were expected to become OpenAI employees once the transaction completed. SO023, SO025
CO049 OpenAI said Statsig would continue operating independently and serving customers from its Seattle office after closing. SO023, SO024
CO050 Tracxn already labels Statsig as acquired, creating a closing-status mismatch with the direct pending-close announcements. SO023, SO024, SO028
CO051 Statsig's trust page says the company is SOC 2 Type II audited. SO014
CO052 Statsig's trust page says risk assessments and treatment plans are reviewed by a Security Governance Board. SO014
CO053 Statsig's trust page says the company uses SIEM and IDS monitoring, incident response plans, and annual third-party application security assessments. SO014
CO054 UpGuard publishes an independent vendor-risk report on Statsig's security posture, creating an external adverse signal even without alleging a disclosed breach. SO033
CO055 Public diligence remains constrained by partial board disclosure, conflicting headcount snapshots, and unclear public confirmation of final transaction closing. SO014, SO021, SO023, SO028
CM001 Statsig positions itself as a single platform that combines experimentation, feature management, and product analytics. SM001
CM002 Forrester says feature management and experimentation spans both software delivery and product management. SM016
CM003 Feature management supports progressive delivery by keeping flags off, testing in production, and gradually exposing features to larger audiences. SM016, SM012
CM004 Commercial FM&E systems extend basic toggles into managed control over features, target populations, and progressive delivery. SM015
CM005 LeadDev defines feature flags as controls that let developers turn features on or off without redeploying or editing source code. SM018
CM006 The global product analytics market was estimated at USD 14.81 billion in 2023 and Grand View expects 19.8% CAGR from 2024 to 2030. SM022
CM007 Mordor expects the product analytics market to grow from USD 13.04 billion in 2026 to USD 25.73 billion by 2031 at 14.55% CAGR. SM023
CM008 Credence values the A/B testing software market at USD 8.18 billion in 2024 and USD 25.8169 billion by 2032 at 15.45% CAGR. SM024
CM009 VWO cites a much narrower A/B testing tools market of USD 850.2 million in 2024 with 14.0% CAGR through 2031. SM025
CM010 Public size estimates diverge because they measure different boundaries ranging from narrow testing tools to broader experimentation software and adjacent product analytics. SM022, SM023, SM024, SM025
CM011 The narrowest defensible core market for Statsig is experimentation and feature-management infrastructure rather than the entire product analytics category. SM001, SM015, SM016, SM018
CM012 Product analytics remains an adjacent budget pool because GrowthBook, PostHog, and Mixpanel all position analytics alongside experimentation or warehouse alignment rather than inside pure flagging. SM008, SM011, SM028
CM013 Warehouse-native experimentation is best treated as an architectural deployment model inside the experimentation market because public sizing is vendor-led rather than analyst-standardized. SM002, SM003, SM004, SM005, SM007, SM009
CM014 Home-grown feature flags and internal experimentation tooling remain real substitutes because Forrester says many developers still use home-grown solutions. SM015
CM015 Maintaining home-grown FM&E capabilities imposes a tax on developer time that could otherwise go to shipping customer features. SM015
CM016 Statsig Warehouse Native documentation says users connect warehouse data, configure a metric, and get experiment results in record time. SM002
CM017 LaunchDarkly supports BigQuery-native experimentation and warehouse-native metrics. SM003
CM018 Optimizely says warehouse-native experimentation keeps the data warehouse as the single source of truth so data stays secure and centralized. SM004, SM007
CM019 Harness says warehouse-native experimentation runs on governed data without ETL or clunky configuration and with inspectable SQL. SM005
CM020 Eppo says warehouse-native experimentation avoids data duplication and shadow warehouses while leaving an auditable paper trail. SM007
CM021 GrowthBook describes itself as a warehouse-native platform for experimentation, feature flags, and product analytics. SM008
CM022 GrowthBook markets lower cost and higher testing throughput, which indicates price sensitivity is a visible buying theme in the category. SM008
CM023 PostHog says its stack is built for data teams and loved by product teams, reinforcing that analytics and experimentation buying centers overlap. SM028
CM024 Mixpanel frames product analytics and the data warehouse as a pairing problem, showing warehouse alignment is becoming part of analytics tool evaluation. SM011
CM025 LeadDev says software developers and product managers are continuously shipping features and need feature-flag solutions chosen for engineering teams. SM018
CM026 Firebase A/B Testing is used to test app UI, product features, and engagement campaigns against metrics such as revenue and retention before broad rollout. SM014
CM027 Firebase’s use cases show growth and marketing teams participate alongside product teams in experimentation programs. SM014
CM028 Statsig says its platform is built for data teams and can reduce time spent by data scientists. SM001
CM029 MIT says experimentation culture pushes product-specific decisions down to product teams while senior leaders retain the high-level metrics and strategy. SM020
CM030 MIT frames experimentation adoption as a problem for business leaders, data scientists, and researchers, showing that executive and data stakeholders matter alongside engineers. SM020
CM031 Harvard Business School says experimentation has become a strategic imperative for C-suite executives and business leaders. SM021
CM032 PostHog documentation highlights signup and onboarding flows as common experimentation use cases. SM029
CM033 PostHog feature flags support targeting specific users, groups, or percentages of traffic without redeploying code. SM030
CM034 Azure feature-management libraries are designed for beta access, rollout, and dark deployments. SM012
CM035 AWS AppConfig feature flags allow teams to toggle features and configure feature characteristics through flag attributes. SM013
CM036 The buyer map therefore starts in product engineering budgets and expands into product, growth, and data-science workflows as experimentation matures. SM014, SM018, SM020, SM028
CM037 Grand View says growing adoption of cloud-based solutions is a major driver of product analytics market growth. SM022
CM038 Grand View says product analytics helps marketing teams optimize campaigns and target the right audience with personalized messages. SM022
CM039 Credence says experimentation market growth is fueled by digital transformation and personalized customer engagement. SM024
CM040 Credence says mobile commerce and cross-device experience demands are amplifying adoption of experimentation software. SM024
CM041 Kameleoon says 50% of companies expecting growth in 2025 reported high investment in feature experimentation. SM010
CM042 Kameleoon cites Forrester that about 50% of respondents called feature experimentation very important in software-development initiatives. SM010
CM043 Kameleoon says businesses using feature flags can add about 30 new features per application each year, citing Atlassian. SM010
CM044 DORA 2025 says successful AI adoption is a systems problem rather than just a tooling problem. SM019
CM045 DORA 2025 says value stream management is needed to translate local AI productivity gains into measurable product performance instead of downstream chaos. SM019
CM046 Harvard Business School highlights Netflix, Amazon, and Microsoft as examples of companies using experimentation frameworks to improve decision-making and product optimization. SM021
CM047 VWO says 71% of companies run two or more tests each month. SM025
CM048 VWO says 60% of companies use A/B testing for landing pages and 58% use it on paid ads. SM025
CM049 VWO says 63% of companies find A/B testing easy to implement while 7% find it very difficult. SM025
CM050 Harness says experimentation gathers behavioral and performance data that can create significant privacy risk. SM006
CM051 Harness recommends collecting only the data that is strictly necessary for an experiment. SM006
CM052 The ICO says cookies require user consent unless they are strictly necessary to provide the requested online service. SM026, SM027
CM053 The EDPB cookie-banner taskforce flags inaccurately classified essential cookies and deceptive banner practices as enforcement concerns. SM027
CM054 LeadDev says many teams write feature flags directly into source code and later discover that maintaining them becomes a full-time job. SM018
CM055 MIT says build-versus-buy is an early decision in experimentation adoption. SM020
CM056 MIT says advanced experimentation platforms must support units of randomization beyond browser cookies, including devices and network clusters. SM020
CM057 MIT says simple experimentation approaches are rarely preferred when fairness, version maintenance, and unit tracking become complex. SM020
CM058 LeadDev says buyers value vendor support for customization, integration, feature delivery, and resource usage because fitting feature-management tools into a workflow is not always easy. SM018
CM059 Warehouse-native adoption is partly a governance response because Optimizely, Harness, and Eppo all pitch secure data residency, auditability, or governed data. SM004, SM005, SM007
CM060 North America appears to lead current spend in both product analytics and experimentation, while Asia-Pacific is repeatedly described as the faster-growth region. SM022, SM023, SM024
CP001 Statsig’s homepage markets a bundled product family spanning experimentation, feature management, product analytics, session replay, web analytics, developer configs, and warehouse-native operation. SP002, SP003
CP002 Statsig docs say the same core features can run on a customer’s own data warehouse with no ETL or can be hosted in Statsig infrastructure. SP003
CP003 Statsig pricing gives the Developer tier 2 million metered events per month at no charge and includes gates, configs, experimentation, and analytics in that entry tier. SP001
CP004 Statsig’s Pro tier costs $150 per month, includes 5 million metered events, and charges $0.05 per additional 1,000 events beyond that threshold. SP001
CP005 Statsig’s homepage highlights 1+ trillion events processed per day, 99.99% API and console uptime, and under-1ms post-init evaluation latency. SP002
CP006 LaunchDarkly’s reviewed official pages describe a platform scope built around runtime control, feature flags, progressive delivery, experimentation, observability, and agent control. SP004, SP005
CP007 The reviewed LaunchDarkly pricing surface exposes a free-to-start path but does not publish comparable scaled self-serve dollar rates, pushing larger buyers toward tailored pricing. SP004
CP008 CostBench calls LaunchDarkly the category leader but says its pricing is opaque and enterprise-only, which makes it a weak fit for many early-stage SaaS teams. SP029
CP009 Startupik says LaunchDarkly is usually the safest enterprise choice because of governance, approvals, auditability, and low-risk operations. SP030
CP010 Optimizely Feature Experimentation supports experiments across frontend, backend, mobile, and edge environments with SDKs or agent-based REST decisions. SP006
CP011 Optimizely markets a built-in stats engine with false discovery control and outlier smoothing, plus multi-armed bandits and data-warehouse connectivity for experiment metrics. SP006
CP012 Optimizely’s trust center says the company has provided secure solutions for over two decades across both cloud and on-premise systems in complex regulatory environments. SP007
CP013 Harness FME markets feature flags, flexible targeting, dynamic configs, release monitoring, and experimentation inside a single offer that carries forward Split’s feature-management thesis. SP009, SP035
CP014 Harness explicitly markets warehouse-native experimentation in the buyer’s existing data warehouse using governed metrics and SQL. SP009
CP015 Harness says evaluations can run locally inside the application without sending private user data to the cloud, and its security page describes peer review, approvals, RBAC, SSO, and 2FA. SP009, SP010
CP016 Amplitude Experiment markets guided setup, enterprise feature flags, local or remote evaluation, sequential testing, T-tests, bandits, CUPED, mutual exclusion groups, and holdouts. SP012
CP017 Amplitude pricing exposes a free Starter tier and a Plus tier starting at $49 per month, while Growth and Enterprise remain custom-priced inside a broader product suite. SP013
CP018 Amplitude’s trust portal emphasizes compliance certifications, security policies, subprocessors, and controlled access to documentation as part of its enterprise trust posture. SP014
CP019 GrowthBook’s homepage and docs position it as a warehouse-native, open-source platform for experimentation, feature flags, and product analytics that is trusted by 3,000+ companies. SP016, SP017
CP020 GrowthBook pricing offers a free Starter plan for up to 3 users with unlimited feature flags and experiments, while Pro starts at $40 per seat per month with cloud or self-host deployment options. SP015, SP016
CP021 GrowthBook docs say the exact same codebase powers cloud and self-hosted deployments and that evaluations run locally with no network requests. SP017
CP022 GrowthBook security says no PII leaves the customer’s warehouse, governance includes SSO, fine-grained permissions, role-based access, and audit trails, and air-gapped self-host deployment is supported. SP018
CP023 Eppo’s official site and docs frame the product as warehouse-native experimentation plus feature flagging with zero-copy architecture and no user-level data passing through Eppo’s system. SP019, SP020
CP024 Eppo docs say flags and experiments are configured centrally but evaluated locally from CDN-downloaded configs, so normal variant checks do not require per-decision network requests. SP020
CP025 Eppo security describes encryption at rest and in transit, MFA, least-privilege access, and centralized monitoring and logging of sensitive environments. SP021
CP026 PostHog pricing markets 10+ products, says more than 90% of companies use the platform for free, and publishes free allowances for analytics events, feature-flag requests, recordings, and warehouse rows. SP022
CP027 PostHog’s feature-flags page says local evaluation reduced latency from 500ms to under 50ms, removes page flicker, and links flags directly to analytics. SP023
CP028 PostHog’s experiments docs explicitly support experiments with or without feature flags, reinforcing that experimentation buyers do not have to standardize on one flagging vendor. SP024
CP029 PostHog’s self-host docs say the platform is MIT-licensed and self-hostable, but also argue that cloud is usually cheaper and easier while some paid-plan features remain cloud-only. SP025
CP030 VWO pricing markets feature experimentation, web rollouts, approvals, SSO, self-hosting, sequential testing, guardrails, and multi-arm bandits across Growth, Pro, and Enterprise tiers. SP026
CP031 VWO Testing supports website, mobile-app, and server-side experimentation, keeping VWO relevant as an adjacent optimization rival rather than only a classic web-A/B tool. SP027
CP032 VWO’s security page says more than 3,000 customers trust the platform and lists ISO 27001, ISO 27701, ISO 27017, ISO 27018, PCI DSS, and SOC 2 Type II. SP028
CP033 CostBench ranks Statsig best overall for SaaS teams because of transparent pricing, experimentation depth, and SDK quality, but notes that self-hosting is not officially supported and some advanced statistics are Enterprise-only. SP029
CP034 CostBench says Split or Harness remains an enterprise-grade option because of Data Hub attribution and broad SDK coverage, but its per-user pricing can become expensive and enterprise contracts can start above $6,000 per month. SP029
CP035 CostBench says GrowthBook is the strongest open-source experimentation alternative because it connects directly to existing warehouses or analytics backends, while LaunchDarkly still wins on governance maturity. SP029
CP036 Startupik says Statsig works best when teams want to test, launch, measure, and iterate in one system, but it is less natural than LaunchDarkly for governance-first organizations and less appealing if the buyer only wants basic flags. SP030
CP037 Startupik says GrowthBook appeals to teams wanting lower cost, more ownership, and less vendor lock-in pressure, while LaunchDarkly wins when approvals, scale readiness, and non-technical rollout governance matter most. SP030
CP038 Forrester says feature management and experimentation now span both software delivery and product management, signaling a converged market rather than isolated point-solution categories. SP034
CP039 OpenFeature says a vendor-agnostic API can avoid code-level lock-in, let teams use one SDK with any backend, and switch or combine solutions without code refactors. SP031
CP040 GrowthBook’s migration-from-Statsig page attacks Statsig on event-based pricing, proprietary statistics, and data ownership while pitching per-seat pricing, auditable SQL, and warehouse-native control. SP033
CP041 Martin Fowler’s feature-toggle essay shows that internal build remains conceptually feasible because teams can decouple deployment from release and support canary or ops toggles without a commercial platform. SP032
CP042 GrowthBook docs say the top 1% of companies spend thousands of hours building feature-flagging and experimentation platforms in-house, implying that internal build has real opportunity cost. SP017
CP043 The reviewed source set indicates that hard vendor lock-in is lower for open-source, self-hosted, or OpenFeature-aligned options such as GrowthBook, PostHog, and internal build than for managed-first closed platforms such as Statsig, LaunchDarkly, Optimizely, or Eppo. SP017, SP025, SP031, SP002, SP004, SP006, SP019
CP044 Multi-homing remains realistic because buyers can pair analytics or optimization tools with separate flag and experimentation layers, and several reviewed vendors explicitly market interoperability or warehouse connections rather than a single closed stack. SP015, SP020, SP024, SP027, SP031
CP045 Statsig’s moat is strongest where buyers value one managed experimentation-centric control plane with transparent self-serve entry pricing, and weakest where buyers prioritize open-source portability, zero-copy warehouse governance, or incumbent approval-heavy workflows. SP001, SP002, SP030, SP033
CP046 Public scale disclosure is uneven across the reviewed rivals: Statsig publishes throughput metrics, GrowthBook and VWO disclose 3,000+ company or customer figures, and PostHog discloses 60,000+ customers, but like-for-like ARR or customer counts are missing for LaunchDarkly, Optimizely, Harness FME, and Eppo on the retained public pages. SP002, SP016, SP022, SP028, SP004, SP006, SP009, SP019
CP047 Public enterprise price visibility remains incomplete because LaunchDarkly, Optimizely, Harness, and the upper Amplitude tiers move buyers into tailored or request-pricing flows, while VWO exposes tier names and feature gating without a simple comparable floor price on the retained page. SP004, SP008, SP011, SP013, SP026
CI001 Statsig publicly sells three plan tiers: Developer, Pro, and Enterprise. SI001
CI002 The Developer plan is free and includes 2 million events per month, 50,000 session replays per month, one year of analytics retention, and unlimited seats. SI001
CI003 The Pro plan costs $150 per month and includes 5 million events before overage charges apply. SI001
CI004 Statsig charges $0.05 per 1,000 events above the Pro plan’s 5 million included monthly events. SI001
CI005 Enterprise pricing is custom and can be structured as event-based or experiment-based contracts with large-volume discounts. SI001
CI006 Statsig says Warehouse Native is included with the platform and is positioned around security, reliability, and data governance. SI001, SI002
CI007 Statsig says its warehouse page uses an event-based pricing model with no limitations on seats, flags, or environments. SI002
CI008 Statsig says customers can run the experimentation product in their own warehouse or let Statsig host it. SI004
CI009 Statsig’s documentation describes the company as a unified platform for feature flags, A/B testing, and product analytics. SI012
CI010 Statsig’s feature flags page says the company has 30-plus open-source SDKs and infrastructure built for trillions of daily flag checks and billions of users. SI003
CI011 BusinessWire says Statsig’s platform spans experimentation, analytics, feature management, application performance, and qualitative feedback, supporting a bundled-platform revenue story. SI013
CI012 Statsig’s careers page says the company operates a 100% in-person office culture. SI007
CI013 Statsig announced and multiple outlets confirmed a $100 million Series C at a $1.1 billion valuation in May 2025, led by ICONIQ Growth with Sequoia and Madrona participating. SI010, SI013, SI014
CI014 BusinessWire says thousands of companies including Notion, Microsoft, Brex, Electronic Arts, and Atlassian rely on Statsig. SI013
CI015 Statsig’s 2025 recap says the platform earned the trust of 20,000 weekly active users. SI011
CI016 Statsig’s 2025 recap says the platform processed 3 trillion events per day in 2025. SI011
CI017 Statsig’s 2025 recap says experiments and feature gates grew 2.5 times year over year. SI011
CI018 Statsig’s 2025 recap says the platform reached 4 billion unique end users through experiments. SI011
CI019 Statsig’s 2025 recap says nearly 7 million analytics queries were executed on the platform in 2025. SI011
CI020 Statsig’s 2025 recap says the company grew to 170 full-time employees in 2025. SI011
CI021 GeekWire reported that Statsig employed 155 people and planned to expand to nearly 200 by early 2026. SI017
CI022 Growjo estimates Statsig has 145 employees and that employee count grew 54% over the prior year. SI024
CI023 Sacra estimates that Statsig reached $40 million of annual recurring revenue in May 2025. SI016
CI024 SiliconANGLE reported that Statsig had reached $40 million of ARR ahead of its 2025 funding round. SI015
CI025 Sacra estimates that Statsig serves more than 300 paying customers. SI016
CI026 Statsig’s experimentation page says more than 50% of customers increase experiments tenfold within a year. SI004
CI027 Statsig’s 2025 recap says the company works with thousands of companies across industries, business models, and stages of growth. SI011
CI028 Tracxn says Statsig has raised a total of $153 million over three funding rounds. SI020
CI029 Tracxn lists a $10.4 million Series A for Statsig on 2021-08-05. SI020
CI030 Tracxn lists a $43 million Series B for Statsig on 2022-04-20. SI020
CI031 Tracxn lists a $100 million Series C for Statsig on 2025-05-06. SI020
CI032 Tracxn and Sacra both point to a last private valuation of about $1.1 billion for Statsig in 2025. SI016, SI020
CI033 CNBC and OpenAI both say OpenAI agreed to acquire Statsig for $1.1 billion in September 2025. SI018, SI019
CI034 OpenAI and CNBC say Statsig will continue to operate independently and serve customers from Seattle while the acquisition proceeds. SI018, SI019
CI035 CNBC says the acquisition remained subject to customary closing conditions including regulatory approval. SI018
CI036 GeekWire says the $1.1 billion acquisition price matched Statsig’s latest funding-round valuation and therefore carried no visible acquisition premium. SI017
CI037 GeekWire says all Statsig employees were expected to have the option to transition to OpenAI. SI017
CI038 Statsig’s 2025 recap says the company ended the year by joining forces with OpenAI. SI011
CI039 Sacra says Statsig’s usage-based pricing is driven by feature-flag exposures, experiment participants, and analytics events processed. SI016
CI040 Sacra says Statsig’s warehouse-native model lowers Statsig’s own infrastructure costs while helping enterprises with data-governance constraints adopt the product. SI016
CI041 A $1.1 billion valuation against a $40 million ARR estimate implies an approximate 27.5-times revenue multiple. SI016
CI042 Using a $40 million ARR estimate and a 145 to 170 employee band implies about $235,000 to $276,000 of ARR per employee. SI011, SI016, SI017, SI024
CI043 Growjo estimates Statsig generates $23.7 million of annual revenue and about $163,400 of revenue per employee. SI024
CI044 TrustRadius pricing says Statsig has one custom annual paid plan and also offers a free version. SI022
CI045 A TrustRadius reviewer says Statsig replaced Mixpanel and LaunchDarkly in their workflow, which supports willingness to pay for a bundled platform rather than isolated tools. SI023
CI046 TrustRadius reviews cite a complex data-science-focused UI, the inability to test two user segments in one experiment, and non-intuitive metadata handling as product frictions. SI023
CI047 Sacra warns that experimentation budgets can be absorbed into broader observability spending, weakening the case for a standalone platform budget. SI016
CI048 Sacra warns that advanced statistical sophistication may be unnecessary for many mainstream customers, which can pressure premium pricing. SI016
CI049 The retained public sources do not disclose Statsig’s gross margin, CAC, payback, net revenue retention, or customer concentration. SI001, SI010, SI011, SI016, SI020
CI050 The retained public sources do not disclose Statsig’s cash on hand, monthly burn, runway, or debt facilities. SI010, SI011, SI016, SI020, SI021
CI051 The combination of a fresh $100 million Series C and a later $1.1 billion acquisition agreement likely reduced Statsig’s near-term standalone financing dependency versus an independent venture-backed path. SI010, SI018, SI019
CI052 If Statsig becomes a continuing independent unit inside OpenAI, standalone underwriting will become harder because any future economics are more likely to be reported inside OpenAI rather than as a separate company. SI018, SI019
CI053 Crunchbase’s October 2024 snapshot still labeled Series B as Statsig’s last funding type, showing that market-data services can lag real fundraising events. SI025
CI054 The fetched SEC legacy browse search for Statsig did not yield a directly readable company-specific filing result in main-content extraction, so public filing-based corroboration of fundraising remains incomplete. SI026
CE001 Statsig positions its platform as a unified product-development stack spanning feature flags, experimentation, analytics, session replay, and warehouse-native workflows. SE001, SE002, SE003, SE004
CE002 Feature Gates are runtime boolean controls evaluated by Statsig SDKs and support gradual rollouts, kill switches, targeting by user attributes, and environment-based rules. SE007, SE001, SE025
CE003 Dynamic Configs return server-defined JSON and can target different parameter values by user attributes in near real time. SE008, SE025
CE004 Public docs state that a Dynamic Config payload is limited to 100 KB, constraining how much configuration data can be shipped through one object. SE008
CE005 Statsig publicly documents randomized A/B and A/B/n tests, multivariate experimentation, and mutually exclusive experiments that measure changes against product and business metrics. SE009, SE002, SE025
CE006 Statsig’s cloud experimentation product applies CUPED using the seven days before each user’s exposure, while Warehouse Native recommends the same window but allows customization. SE010, SE017
CE007 Statsig documents frequentist sequential testing as a response to the peeking problem in fixed-horizon experiments. SE011
CE008 Statsig Autotune uses Thompson Sampling to allocate traffic in proportion to each variant’s probability of being the best. SE012
CE009 The same bandit methodology page says vanilla multi-armed bandits cannot personalize and can converge to a global winner that is suboptimal for specific user segments. SE012
CE010 Statsig holdouts measure the aggregate impact of multiple features, can be global or selected, and are recommended at small percentages such as 1% to 10% of users. SE013
CE011 Statsig product analytics exposes metric drilldown, funnels, retention, distribution, user journeys, lifecycle views, and dashboards that can also hold feature gate and experiment snapshots. SE014, SE003
CE012 Statsig Session Replay stores a serialized rrweb representation of the page and interaction stream rather than a literal video file. SE015
CE013 Session Replay privacy controls include three baseline masking levels, CSS-selector overrides for mask, unmask, and block behavior, and a global targeting gate that decides which users are eligible for capture. SE016
CE014 Warehouse Native is described as an experimentation platform that runs analysis compute jobs directly inside the customer’s data warehouse while using existing datasets and experiment assignment data. SE017, SE004
CE015 The Warehouse Native pipeline identifies first exposures, annotates metric sources with exposure data, builds user-day staging tables, runs intermediate rollups, and writes per-experiment result tables in the warehouse. SE018
CE016 Even in warehouse-native mode, public docs still rely on Statsig’s console, SDK-driven assignment, optional logging callbacks, and Statsig-managed analysis semantics, so the public architecture is not a pure local-only control plane. SE017, SE019, SE020
CE017 Statsig exposes both a console CRUD API at statsigapi.net and an HTTP API for runtime evaluation, while the HTTP API explicitly recommends official SDKs for most production integrations. SE019, SE020
CE018 Console API mutation requests are rate limited to about 100 requests per 10 seconds and 900 requests per 15 minutes per project, and the documented API version is 20240601. SE019
CE019 Public SDK docs show client and server interfaces for gates, dynamic configs, experiments or layers, custom event logging, and explicit exposure logging support. SE025, SE020
CE020 Statsig’s access-management docs position SSO and SCIM as complementary, with SSO handling authentication and SCIM handling account lifecycle automation. SE021
CE021 Statsig documents enterprise-tier OIDC SSO with providers including Okta, Microsoft Entra ID, Google, Ping Identity, and OneLogin, plus automatic provisioning for new authenticated users and inherited MFA requirements from the IdP. SE023
CE022 Statsig’s published SCIM documentation currently only offers an Okta SCIM integration and supports push users, unassign-based deactivation, import users, import groups, and push groups. SE022, SE040
CE023 Statsig says it has configurable RBAC across the console, and its Warehouse Native roles page emphasizes control over who can view, edit, and approve experiments, metrics, and raw-data visibility. SE028, SE021
CE024 A public feature-request issue asks whether Statsig can restrict edit or delete access on individual feature gates or rules, indicating that fine-grained per-gate governance is not clearly documented on the public surface. SE030
CE025 Statsig’s compliance docs say customer data remains isolated from other customers, AI features do not use customer data to build models without explicit written consent, encryption uses AES-256 at rest and TLS 1.2+ in transit, and the company maintains SOC 2 Type II certification. SE026
CE026 Statsig’s privacy policy says the company acts as a processor or service provider for product data handled on behalf of customers and directs end users to the customer’s own privacy notice in those cases. SE005
CE027 The GitHub Code References integration lets teams locate feature-gate and dynamic-config references from the Statsig console without storing the underlying code. SE027
CE028 Statsig officially supports SDK versions released within the last 12 months, treats older versions as unsupported, and publishes release notes through each SDK repository’s GitHub Releases page. SE024
CE029 The public npm page for statsig-node describes a Node.js SDK for multi-user server-side environments and shows an actively published package surface near the run date. SE034
CE030 The npm page for @statsig/js-client describes a public client package for feature gates, dynamic configs, and A/B/n experimentation. SE033
CE031 The Go package index lists APIs for gate checks, experiment or config retrieval, client initialize responses, event logging, and manual exposure logging. SE035
CE032 pub.dev lists a Statsig SDK for Dart, indicating that package support extends beyond JavaScript and server-side runtimes. SE036
CE033 NuGet lists a Statsig package for both .NET server and client-side applications across modern .NET versions. SE037
CE034 Maven Central and Sonatype both list a com.statsig server SDK artifact for Java, supporting the claim that Statsig maintains JVM server packaging. SE038, SE039
CE035 Statsig’s public Node and Go server SDK repositories say each server SDK is tested across unit, integration, and end-to-end levels, with rule-evaluation consistency checks against backend results. SE031, SE032
CE036 Okta’s integration listing says the Statsig application supports OIDC SSO plus provisioning actions such as create, update, deactivate, sync password, and group push. SE040
CE037 Twilio Segment hosts a Statsig destination listing, indicating partner-surface support for routing data into Statsig through Segment’s catalog. SE041
CE038 A public GitHub issue reports that @statsig/session-replay crashed Safari 15 pages with an invalid regular expression error and describes the problem as reproducible on the page’s referenced version set. SE029
CE039 The replay issue and the lack of a public browser-compatibility matrix imply that session replay adds browser-specific stability risk beyond the core flagging and analytics surfaces. SE015, SE029
CE040 Public docs make the shipped surface legible, but public evidence on item-level roadmap timing and granular authorization scope remains thinner than the live feature surface, leaving diligence gaps around per-feature governance and warehouse-native boundary details. SE030, SE017, SE024
CE041 Statsig’s Warehouse Native launch post says the initial launch covered BigQuery, Snowflake, Databricks, and Redshift and included experiment analysis, holdouts or layers, dashboards, RBAC, approval workflows, and auditing capabilities. SE006
CE042 Statsig’s feature-flag product page says the platform offers automated rollback workflows, real-time release diagnostics, more than 30 open-source SDKs, and infrastructure that serves billions of end users. SE001
CE043 The warehouse product page says customers can keep their own assignment or flagging solution and use existing metric data and exposures, or use Statsig’s SDK and flagging stack, indicating a hybrid rather than all-or-nothing deployment model. SE004
CU001 The fetched public customer proof base spans productivity software, fintech, meal-kit commerce, marketplaces, social and creator platforms, nonprofit education, consumer AI, developer tools, fitness, and genealogy subscriptions. SU001, SU002, SU003, SU004, SU005, SU006, SU007, SU008, SU009, SU010, SU011, SU012, SU013, SU014, SU015
CU002 Across the strongest public references, the operational buyer is usually a product, engineering, data, or growth leader rather than a generic procurement function. SU004, SU008, SU010, SU014, SU015
CU003 The end users affected by Statsig vary widely—internal builders in B2B software and millions of consumers, creators, students, shoppers, listeners, and runners in consumer products—so the user is often broader than the buyer. SU003, SU006, SU009, SU011, SU012, SU013, SU017, SU019, SU022, SU024, SU025
CU004 The visible reference set over-indexes to internet-native software and consumer app companies, making public proof strongest for product-led teams and thinner for traditional offline or heavily regulated customer archetypes. SU001, SU002, SU003, SU004, SU005, SU006, SU007, SU008, SU009, SU010, SU011, SU012, SU013, SU014, SU015
CU005 Statsig's Ancestry story says experimentation grew from 70 manual experiments per year to 600 experiments per year. SU002
CU006 The same Ancestry story says 3.5 million customers were benefitting from scalable experimentation. SU002
CU007 Statsig's Brex story says Brex reduced time spent by data scientists by 50% and generated 20%+ cost savings after consolidating tooling. SU004
CU008 Brex describes itself as serving 30,000+ customers and the Statsig story says Brex created 100+ experiments with the platform. SU004, SU018
CU009 Statsig's Character.ai story says the platform serves tens of millions of users and treats experimentation as critical to deciding when models are ready for broad release. SU005
CU010 Statsig's Code.org story says the nonprofit needed better product visibility for a platform used by millions of students and teachers, and Code.org's own impact page corroborates current scale with 107 million student accounts and 3 million teachers supported. SU006, SU019
CU011 Statsig's Graphite story says Graphite actively uses 321 feature gates and 168 dynamic configs in production and cut incident-resolution time by 50%+. SU007
CU012 Statsig's HelloFresh story says HelloFresh scaled to 1,000 experiments per year, cut time to decision by 25%, and increased the share of experiments with a decision by 55%. SU008
CU013 Statsig's Linktree story says Linktree operates at over 500 million unique visitors and 2 billion impressions per month, and Linktree's official homepage corroborates a very large creator base. SU009, SU022
CU014 Statsig's Notion story says experimentation grew from single digits to hundreds of experiments per quarter. SU010
CU015 Statsig's OfferUp story says a target call-to-action improved by 11%, while OfferUp's own page confirms it is a local marketplace for buying and selling. SU011, SU024
CU016 Statsig's Runna story says the company ran 100+ tests within a year and tied the program to premium conversion, onboarding completion, and retention goals. SU012
CU017 Statsig's SoundCloud story describes a shift from rollout-only gating to experimentation and frames the platform as helping SoundCloud achieve profitability and release its flagship news feed. SU013
CU018 Statsig's Webflow story says the platform replaced a patchwork of homegrown systems so launches could be decoupled from deployments. SU014
CU019 Statsig's Whatnot story says experimentation scaled from 0 to 400 annual experiments, including 250+ experiments over three quarters in year two. SU015
CU020 Statsig's customer landing page includes named positive quotes from leaders at Whatnot, SoundCloud, and Ancestry, which adds direct testimonial color but remains a vendor-controlled surface. SU001
CU021 Across multiple stories, the win pattern starts with rollout or experimentation pain and then expands into analytics, warehouse-native workflows, dynamic configs, or broader product-development usage. SU004, SU007, SU008, SU009, SU014
CU022 Procurement friction is often triggered by fragmented homegrown or point-tool stacks: Brex had validation and data-trust issues, HelloFresh described a clunky in-house setup, Code.org had been constrained by earlier tools, SoundCloud faced build-vs-buy tradeoffs, and Webflow had three internal systems. SU004, SU006, SU008, SU013, SU014
CU023 The most detailed deployment evidence in this chapter is vendor-authored on Statsig's site; customer-side official pages mostly corroborate who the customers are and how large they are, not the specific Statsig deployment. SU002, SU003, SU004, SU006, SU007, SU008, SU009, SU010, SU011, SU012, SU014, SU016, SU017, SU018, SU019, SU020, SU021, SU022, SU023, SU024, SU025, SU026
CU024 Customer-side official corroboration is strongest for Ancestry, Bluesky, Brex, Code.org, Graphite, HelloFresh, Linktree, Notion, OfferUp, Runna, and Webflow because fetched official pages confirm current category, audience, or scale. SU016, SU017, SU018, SU019, SU020, SU021, SU022, SU023, SU024, SU025, SU026
CU025 The fetched pack contains at least 14 named Statsig customer stories, which is materially better than a generic logo wall but still not a complete roster. SU001, SU002, SU003, SU004, SU005, SU006, SU007, SU008, SU009, SU010, SU011, SU012, SU013, SU014, SU015
CU026 FeaturedCustomers says Statsig has 45 reviews and testimonials, 12 case studies, and 57 customer reviews or references, implying broader reference availability than the individual pages fetched for this chapter. SU028
CU027 TrustRadius rates Statsig 8.9 out of 10 from five reviews in 2026, which is a positive but small-sample customer-satisfaction proxy. SU027
CU028 A TrustRadius reviewer says Statsig replaced Mixpanel and LaunchDarkly, which is consistent with a broader consolidation and expansion motion inside some accounts. SU027
CU029 TrustRadius surfaces recurring friction points including a complex data-science-focused UI, inability to test two different user segments in the same experiment, non-intuitive metadata handling, and confusing default-parameter behavior. SU027
CU030 Neither the fetched case studies nor the third-party reference aggregator disclosed top-customer ARR share or concentration by named logo. SU001, SU028
CU031 No fetched public source disclosed NRR, GRR, logo churn, or renewal rates for Statsig customer cohorts. SU001, SU027, SU028
CU032 No fetched public source disclosed standard contract length, seat counts, or cohort-level commercial pricing by customer segment. SU001, SU027, SU028
CU033 Because retention and contract data are absent, public durability evidence is stronger on repeat product usage than on commercial renewal. SU002, SU008, SU012, SU015, SU027
CU034 Geographic and audience breadth is visible in public materials: Whatnot cites the United States and Europe, SoundCloud says 193 countries, Linktree describes a global creator base, Bluesky refers to millions of users, and Code.org frames global education reach. SU003, SU006, SU009, SU013, SU015, SU017, SU019, SU022
CU035 Several of Statsig's best public outcomes are experiment-volume or workflow metrics rather than hard commercial-retention metrics, so adoption proof is better than durability proof. SU002, SU004, SU008, SU012, SU014, SU015
CU036 The clearest public production proofs pair a named customer, a concrete operating use case, and a quantified outcome—for example HelloFresh, Brex, Ancestry, Graphite, OfferUp, and Bluesky. SU002, SU003, SU004, SU007, SU008, SU011
CU037 Reference freshness is mixed: the pages were live and fetchable on 2026-05-28, but most Statsig case studies do not display publication dates, so freshness must be inferred from current availability and on-page context rather than timestamps. SU001, SU002, SU003, SU004, SU005, SU006, SU007, SU008, SU009, SU010, SU011, SU012, SU013, SU014, SU015
CU038 Taken together, the public record supports a credible land-and-expand story inside technical product teams, but not a quantified portfolio-level retention or revenue-quality story. SU004, SU008, SU014, SU027, SU028
CU039 Because public proof is concentrated in software-native and experimentation-mature teams, investors should underwrite concentration by customer archetype and product maturity rather than raw logo count alone. SU001, SU003, SU004, SU005, SU006, SU007, SU008, SU009, SU010, SU011, SU012, SU013, SU014, SU015
CU040 The highest-value next diligence asks are cohort retention, top-10 customer ARR share, average contract term, module penetration by customer, and dated renewal references from named accounts. SU001, SU027, SU028
CR001 Statsig markets itself as a unified platform spanning feature flags, experimentation, analytics, and adjacent product-development workflows. SR001, SR013
CR002 Statsig publicly claims scale of more than 1 trillion events processed per day, 2.5 billion unique monthly experiment subjects, 99.99% infrastructure uptime for API and console, and sub-millisecond evaluation latency. SR001
CR003 Statsig's Developer tier is free and includes 2 million events per month, unlimited flag and config checks, and 50,000 session replays. SR002
CR004 Statsig's Pro tier is priced at $150 per month, includes 5 million events, and then charges $0.05 per 1,000 events. SR002
CR005 Statsig's Enterprise plan is custom-priced and is the tier that adds event- or experiment-based contracts, warehouse-native deployment, SSO, role-based access controls, teams, and priority support. SR002
CR006 Statsig does not advertise blanket HIPAA compliance; it advertises HIPAA-eligibility only when a Business Associate Agreement is required and executed. SR002, SR036
CR007 PostHog markets a broad free tier across analytics, feature flags, experiments, data warehouse, AI, and other modules, with usage-based pricing after the free limits. SR030
CR008 Amplitude's pricing page shows a free Starter plan and a paid Plus plan starting at $49 per month. SR031
CR009 Optimizely uses request-pricing across experimentation and analytics products rather than transparent self-serve pricing on its plans page. SR032
CR010 Because Statsig repeatedly sells consolidation and all-in-one platform value, a large part of the economic upside depends on customers expanding into multiple modules rather than staying on a single-product footprint. SR001, SR002, SR007
CR011 Competitive compression is plausible because PostHog, Amplitude, and Optimizely each sell overlapping analytics and experimentation suites that can be framed against Statsig's bundle. SR030, SR031, SR032, SR006
CR012 Statsig explicitly supports both hosted deployment and warehouse-native deployment with no-ETL positioning on customer data warehouses. SR006, SR008, SR013
CR013 Warehouse Native requires a service user with read access to metric and exposure data, write access to a dedicated Statsig staging environment, and permission to run jobs and queries. SR014, SR015
CR014 The Warehouse Native quickstart requires customers to connect metric sources, map timestamp and user identifiers, and connect assignment sources before results can be analyzed. SR014
CR015 Statsig says warehouse-native analysis keeps intermediate data in customer-managed staging areas, but Pulse and health-check result sets are then stored on Statsig servers. SR016
CR016 Statsig says customer data contained within SDK assignment exposure events can be retained on Statsig for up to 30 days for diagnostics and debugging. SR016, SR024
CR017 Statsig warns that experimentation staging datasets can multiply data by the number of experiments and recommends dropping temporary artifacts within days or when experiments are concluded. SR016
CR018 Statsig's warehouse-native best practices explicitly warn about query scan cost, join and materialization cost, warehouse resource contention, and the need for careful scheduling. SR017
CR019 Statsig recommends partitioning, clustering, incremental reloads, and scoped compute resources, which means customers need active data engineering discipline to operate warehouse-native deployments efficiently. SR017, SR037, SR038, SR039
CR020 Statsig lists Snowflake, Athena, BigQuery, Databricks, and Redshift among supported warehouse-native back ends. SR014, SR015
CR021 BigQuery, Snowflake, and Databricks each layer their own governance models such as lineage, access control, auditability, object ownership, and AI governance on top of the storage layer. SR037, SR038, SR039
CR022 The practical burden of a warehouse-native rollout is therefore shifted into the customer's existing warehouse governance model rather than removed by Statsig. SR015, SR017, SR037, SR038, SR039
CR023 Statsig's trust center says the company is SOC 2 Type II audited and certified. SR009, SR024
CR024 Statsig says customer data is encrypted in transit and that data persisted to disk is encrypted with 256-bit AES. SR009, SR024
CR025 Statsig says it uses AWS, Azure, and Google Cloud in production and maintains disaster recovery, backups, and incident-response processes. SR009
CR026 Pricing and documentation indicate that stronger IAM controls such as SSO, teams, and related access management are tied to select packages or enterprise deployment patterns rather than being guaranteed in every tier. SR002, SR009, SR018
CR027 Statsig's privacy policy is published under Amplitude, Inc. and covers website, marketing, account, support, and service-interaction data collection. SR010
CR028 The privacy policy permits disclosure to service providers, payment processors, analytics providers, legal authorities, and counterparties involved in business transfers. SR010
CR029 The privacy policy also says no internet transmission is fully secure or error free, which limits how far public assurance language can be taken without customer-specific diligence. SR010
CR030 The DPA assigns controller responsibility to the customer and processor responsibility to Statsig, and says the customer is responsible for the legality of the data it sends. SR011, SR034
CR031 The standard DPA says customers should not provide sensitive or special-category personal data to Statsig under the default terms. SR011
CR032 The DPA incorporates SCC-style transfer mechanics and gives customers a limited process to object to newly added subprocessors on reasonable security grounds. SR011, SR012
CR033 HHS guidance says a covered entity must obtain written business-associate assurances before disclosing PHI, which is why Statsig's HIPAA message is explicitly conditioned on a BAA. SR036, SR002
CR034 Statsig's AI governance documentation says customer data is not used to build or develop AI models unless the customer provides explicit written consent. SR024
CR035 The same AI governance documentation says third-party LLM providers used by Statsig are not allowed to retain, access, or use customer data processed through those AI features. SR024
CR036 The published subprocessor list now lives on Amplitude and includes AWS Bedrock, Google Vertex AI, OpenAI, Snowflake, MongoDB, Azure, Datadog, Wiz, and Fivetran for Statsig-branded services. SR012
CR037 The combination of Amplitude-branded legal pages, third-party AI providers, and multiple infrastructure subprocessors means security and compliance diligence has to extend well beyond the core product UI. SR010, SR011, SR012, SR024
CR038 Statsig announced on 2025-09-02 that it had signed a definitive agreement to join OpenAI and that closing remained subject to customary conditions including regulatory approval. SR026, SR025
CR039 OpenAI said Vijaye Raji would become CTO of Applications and that Statsig employees would become OpenAI employees once the acquisition closes, while the product continues to serve customers independently from Seattle. SR025, SR027, SR028
CR040 Reuters and CNBC both reported the transaction at roughly $1.1 billion. SR027, SR028
CR041 MarTech reported in May 2026 that Amplitude would take over the Statsig brand and customer base while the original Statsig team remained at OpenAI. SR029
CR042 MarTech argues that this structure leaves Amplitude managing roadmap and support for a product whose original builders now work elsewhere. SR029
CR043 MarTech also highlights the risk that overlapping experimentation and analytics products inside Amplitude could be consolidated over time. SR029, SR032
CR044 The founder transition and product-handoff structure create a genuine key-person and continuity risk even if the software keeps operating. SR025, SR029
CR045 Public contracting and governance materials already reflect Amplitude-branded ownership language, so buyers need to verify the actual counterparty, roadmap owner, and support obligations before underwriting a multi-year deployment. SR010, SR011, SR012, SR029
CR046 Statsig's public customer stories are dominated by digital-native software, AI, marketplace, and e-commerce brands such as OpenAI, Notion, Brex, SoundCloud, Whatnot, Linktree, and Bluesky. SR004, SR006, SR007, SR008
CR047 That public reference set is strong evidence of fit with high-velocity product teams, but it is not the same as proof that slower-moving regulated enterprises can adopt the platform with equal ease. SR004, SR018, SR029
CR048 No reviewed public source provides a top-customer revenue breakdown, renewal profile, or quantitative concentration disclosure for Statsig's customer base. SR002, SR003, SR004
CR049 Statsig explicitly offers org-level experiment policies, gate policies, SSO, SCIM, and role-based access patterns to reduce governance risk as deployments scale. SR018, SR019, SR020
CR050 Those controls only help if customers have admins, data scientists, or platform owners willing to configure them, so self-serve deployments can still under-govern releases or experiments. SR018, SR019, SR020, SR040
CR051 Statsig's sequential-testing documentation says peeking at fixed-horizon experiments inflates false positives and that early decisions may remain underpowered even when they look statistically significant. SR021
CR052 Statsig's bandit documentation says the base Thompson-sampling multi-armed bandit can converge to a global maximum that is still suboptimal for specific user subgroups. SR022
CR053 Statsig's AI Evals documentation says LLM-as-a-judge is not perfect and is used because it is fast and consistent enough to compare outputs at scale. SR023
CR054 Because Statsig surfaces Bayesian and frequentist methods, CUPED, holdouts, sequential testing, bandits, and AI evals directly in-product, weak design discipline can create false confidence rather than merely noisy results. SR002, SR006, SR021, SR022, SR023
CR055 Statsig partially mitigates methodological misuse with diagnostics, guardrail signals, approvals, and policy controls, but those protections are configuration-dependent rather than automatic for every workspace. SR014, SR019, SR020
CR056 A diligence process should treat any material change in contractual counterparty, support owner, or roadmap commitments after the OpenAI and Amplitude transition as a thesis-damaging signal. SR010, SR011, SR029
CR057 Sustained warehouse-native cost overruns, repeated query contention, or an inability to isolate Statsig compute are concrete indicators that implementation complexity is outrunning ROI. SR017, SR037, SR039
CR058 A second material security incident, failure to execute the required BAA or DPA controls, or an inability to document subprocessor governance would be a clear escalation trigger for regulated customers. SR009, SR011, SR036
CR059 If pricing changes or product consolidation materially weaken the value proposition against PostHog, Amplitude, or Optimizely, any expansion-based revenue thesis should be reset. SR029, SR030, SR031, SR032
CR060 The highest-priority diligence asks are a current MSA and DPA pack, subprocessor-notice workflow, roadmap and support letter post-transition, regulated-industry reference calls, top-customer concentration data, and a representative warehouse cost model. SR011, SR012, SR029, SR004
CV001 Statsig announced a $100 million Series C at a $1.1 billion valuation led by ICONIQ Growth with participation from Sequoia and Madrona. SV001, SV002, SV003
CV002 Fortune reported that Statsig raised the Series C after ending acquisition talks with Datadog. SV004
CV003 GeekWire reported that Datadog had approached Statsig about an acquisition before ultimately agreeing to buy Eppo. SV003, SV004
CV004 Sacra said Statsig’s $100 million Series C consisted of about $80 million of primary capital and $20 million of employee secondary sales. SV011
CV005 Sacra said Statsig had raised about $153.4 million across three rounds, including a $10.4 million Series A and a $43 million Series B. SV011
CV006 ARR Club and Sacra each retained a roughly $40 million ARR estimate for Statsig around May 2025. SV010, SV011
CV007 Sacra said Statsig served more than 300 paying customers. SV011
CV008 Sacra said Statsig processed more than 1 trillion events per day. SV011
CV009 Statsig’s official materials describe a single platform spanning feature releases, experimentation, product analytics, performance monitoring, and session replay. SV001, SV014
CV010 Statsig’s pricing page offers 2 million metered events per month on a free developer tier before paid Pro and Enterprise plans. SV012
CV011 Statsig’s docs say customers can run warehouse-native with no ETL or use Statsig-managed infrastructure. SV014
CV012 Statsig’s customer references say some buyers evaluated Optimizely, LaunchDarkly, Split, and Eppo before choosing Statsig for integrated experimentation workflows and large-scale use. SV013
CV013 OpenAI said Statsig would continue operating independently from Seattle after the acquisition closes. SV005, SV006
CV014 OpenAI acquired Statsig for $1.1 billion. SV006, SV007
CV015 GeekWire reported the OpenAI transaction was all-stock and that all Statsig employees were expected to have the option to transition to OpenAI. SV008
CV016 Amplitude said it would take on Statsig’s brand and customers and would maintain and develop the current Statsig platform across the cloud and data warehouse. SV024
CV017 Amplitude said it would work closely with the Statsig team at OpenAI during the transition. SV024
CV018 Optimizely’s compare page argued that the original Statsig team no longer maintains the product after the Amplitude handoff. SV019
CV019 A $1.1 billion valuation on a retained ~$40 million ARR estimate implies about a 27.5x ARR multiple. SV010, SV011, SV007
CV020 The OpenAI deal appears flat to the Series C headline valuation, with no disclosed markup above the $1.1 billion round price. SV001, SV006, SV007
CV021 If only the $80 million primary component diluted the company, implied new-money dilution was roughly 7% to 8% of post-money equity. SV011
CV022 Total raised of about $153.4 million equaled roughly 14% of the later $1.1 billion sale value. SV011, SV007
CV023 Amplitude reported Q1 2026 ARR of $374 million and Q1 revenue of $93.5 million. SV021, SV022, SV026
CV024 Amplitude’s March 31 2026 10-Q said it had 727 customers that each represented more than $100,000 of ARR. SV026
CV025 CompaniesMarketCap showed Amplitude at roughly $0.91 billion of market capitalization in May 2026. SV023
CV026 On Amplitude’s reported $374 million ARR base, the public market valued Amplitude at only about 2.4x. SV021, SV023
CV027 Datadog reported Q1 2026 revenue of $1.006 billion and a 22% non-GAAP operating margin. SV027, SV031
CV028 Datadog reported about 4,550 customers with $100,000 or more of ARR in Q1 2026. SV027, SV031
CV029 Datadog said fiscal 2025 revenue grew 28%, operating cash flow reached $1.05 billion, and free cash flow reached $915 million. SV028
CV030 CompaniesMarketCap showed Datadog at roughly $78.95 billion of market capitalization in May 2026. SV029
CV031 On annualized Q1 2026 revenue, Datadog traded at about 19.6x sales. SV027, SV029
CV032 TechCrunch reported Datadog acquired Eppo in May 2025 and that outside reporting suggested a price around $220 million, though terms were undisclosed. SV016
CV033 Statsig’s own blog called the Datadog-Eppo deal a huge move in experimentation. SV015
CV034 Episerver completed its acquisition of Optimizely in 2020 at an undisclosed price, making Optimizely a strategic-history reference rather than a clean valuation multiple comp. SV017, SV018
CV035 TechCrunch reported Optimizely had more than 1,000 customers when Episerver agreed to buy it. SV018
CV036 Optimizely’s current web experimentation page shows experimentation remains an enterprise software category even though retained sources do not disclose a current Optimizely valuation. SV020
CV037 Statsig’s implied 27.5x ARR anchor sat far above Amplitude’s ~2.4x public multiple and above Datadog’s ~19.6x sales multiple. SV011, SV021, SV023, SV027, SV029
CV038 The later $1.1 billion OpenAI transaction shows the Series C price was fair for at least one strategic buyer. SV006, SV007
CV039 For a financial investor, the same flat $1.1 billion exit implied limited upside above the Series C entry price. SV001, SV006, SV007
CV040 The Amplitude handoff splits team, platform stewardship, and customer custody across OpenAI and Amplitude, making standalone-SaaS valuation comparisons less clean. SV019, SV024
CV041 A 10x ARR framework on the retained ~$40 million ARR estimate would value Statsig at about $400 million. SV010, SV011
CV042 A 15x ARR framework on the retained ~$40 million ARR estimate would value Statsig at about $600 million. SV010, SV011
CV043 A 20x ARR framework on the retained ~$40 million ARR estimate would value Statsig at about $800 million. SV010, SV011
CV044 The realized 27.5x ARR outcome corresponds to the actual $1.1 billion strategic price. SV010, SV011, SV007
CV045 A bull case above the realized $1.1 billion outcome would have required much higher ARR or another strategic bidder willing to pay a scarcity premium. SV010, SV011, SV003, SV004
CV046 Without strategic bidding, the bear case would likely have converged closer to rich public software multiples than to the realized $1.1 billion strategic outcome. SV004, SV023, SV027, SV029
CV047 Datadog’s earlier interest before buying Eppo suggests Statsig had genuine strategic scarcity even if public-market math looked aggressive. SV003, SV004, SV016
CV048 Retained public sources used in this chapter do not disclose Statsig’s net retention, logo churn, or gross margin. SV001, SV011, SV012
CV049 Retained public sources used in this chapter do not disclose the Series C preference stack, liquidation waterfall, or participation rights. SV001, SV006, SV011
CV050 Because those private metrics and terms are missing, the valuation call must be framed as a range and a buyer-specific stance rather than as a precise intrinsic value. SV006, SV011, SV024
CV051 As of the run date, there is no clean standalone late-stage entry left because Statsig is inside OpenAI and the platform and customer base are being operated with Amplitude. SV006, SV024
CV052 The best-fit bottom-line stance is fair for a strategic buyer but unattractive for a new financial investor, which supports an avoid recommendation. SV007, SV011, SV024
来源
编号出版方标题引文
SO001 Statsig Statsig | The modern product development platform
SO002 Statsig Statsig | The modern product development platform
SO003 Statsig Careers at Statsig
SO004 Statsig Statsig is the best, say our customers
SO005 Statsig Statsig | The modern product development platform
SO006 Statsig Documentation Statsig Overview - Statsig Documentation
SO007 Statsig Statsig | The world's leading experimentation platform
SO008 Statsig Statsig | Smarter feature flags for fast-moving teams
SO009 Statsig Statsig | Warehouse Native
SO010 Statsig Statsig is joining OpenAI
SO011 Statsig Why we started Statsig
SO012 Statsig How we're making Statsig smarter with AI
SO013 Statsig Statsig's Knowledge Graph: Connecting code, experiments, and metrics
SO014 Statsig Security at Statsig | Statsig
SO015 Business Wire Statsig Announces $10.4 Million in Series A Funding Led by Sequoia
SO016 TechCrunch Former Facebook teammates raise $10.4M in Sequoia-led round to launch features development | TechCrunch
SO017 GeekWire Statsig raises $43M Series B round for rapid, live testing and analysis of new product features
SO018 Business Wire Statsig Launches Warehouse Native, Bringing Powerful Experimentation to Product Teams on the Modern Data Stack
SO019 GeekWire Interview with Vijaye Raji, CEO of fast-growing Statsig, which lets developers test new AI features
SO020 Business Wire Statsig Secures $100M Series C Funding to Transform Product Development, Valuing Startup at $1.1 Billion
SO021 GeekWire Statsig lands $100M at $1.1B valuation, aiming to unify product development as AI upends industry
SO022 GeekWire Statsig doubles down on in-person work with new headquarters for up to 500 people
SO023 OpenAI Vijaye Raji to become CTO of Applications with acquisition of Statsig
SO024 Reuters OpenAI to acquire product testing startup Statsig, appoints CTO of applications
SO025 TechCrunch OpenAI acquires product testing startup Statsig and shakes up its leadership team | TechCrunch
SO026 GeekWire OpenAI acquires Statsig for $1.1B, names CEO to key role in surprise exit for Seattle-area unicorn
SO027 CNBC OpenAI acquires Statsig for $1.1 billion, brings on CEO as applications executive
SO028 Tracxn Statsig
SO029 Sacra Statsig revenue, funding & news
SO030 Wikipedia Vijaye Raji
SO031 Microsoft Community Hub Featured Article: Interviews with Vijaye Raji, the creator of Small Basic | Microsoft Community Hub
SO032 Engineering at Meta Meet a Facebook Engineer: Vijaye Raji
SO033 UpGuard Statsig Security Rating, Vendor Risk Report, and Data Breaches | UpGuard
SM001 Statsig Statsig | The modern product development platform Statsig gives your team 5+ products in a single platform.
SM002 Statsig Warehouse Native Quickstart - Statsig Documentation This page walks you through connecting your data, configuring a metric, and getting experiment results.
SM003 LaunchDarkly BigQuery native Experimentation | LaunchDarkly | Documentation Creating experiments using warehouse native metrics
SM004 Optimizely Get started with Warehouse-Native Experimentation Analytics This ensures your data warehouse remains the single source of truth, keeping your data secure and centralized.
SM005 Harness Warehouse Native Experimentation | Feature Management & Experimentation | Harness Use your Data Warehouse to run experiments on governed data without waiting for ETL or clunky configuration.
SM006 Harness Ensure data privacy in feature experimentation for user trust | Harness Glossary | Harness While experimentation is crucial for rapid product iteration, it involves gathering user data—everything from clicks and navigation patterns to demographic and performance metrics.
SM007 Eppo Data Warehouse Native Experimentation Run experiments from a single source of truth; no data duplication.
SM008 GrowthBook GrowthBook | Experimentation, Feature Flags & Product Analytics Platform The warehouse-native platform used by modern product teams for experimentation, feature flags, and product analytics.
SM009 GrowthBook Best 7 Warehouse Native A/B Testing Tools | Growthbook Warehouse-native A/B testing tools flip that model: analysis runs directly inside your Snowflake, BigQuery, Redshift, or Databricks.
SM010 Kameleoon Feature Management & Feature Experimentation | Kameleoon Indeed, among companies expecting growth in 2025, 50% reported a high investment in feature experimentation.
SM011 Mixpanel Mixpanel: Product Analytics for Mobile, Web & More Product Analytics and the data warehouse: A long road to a perfect pairing
SM012 Microsoft Feature Management Overview - Azure App Configuration Use these libraries to secure a consistent experience when developing applications that use patterns such as beta access, rollout, dark deployments, and more.
SM013 Amazon Web Services Creating a feature flag configuration profile in AWS AppConfig You can use feature flags to enable or disable features within your applications or to configure different characteristics of your application features using flag attributes.
SM014 Google Firebase Firebase A/B Testing It gives you the power to test changes to your app's UI, features, or engagement campaigns to see how they impact your key metrics (like revenue and retention) before you roll them out widely.
SM015 Forrester Key Insights From The Feature Management And Experimentation Solutions Landscape, Q1 2024 Maintaining them creates a tax on developer time that could potentially be better spent delivering features to customers.
SM016 Forrester Feature Management And Experimentation — An Evolving Market Feature management and experimentation is a broad set of capabilities that spans both software delivery and product management.
SM017 LeadDev The State of Feature Management and Experimentation
SM018 LeadDev The best feature management and experimentation software 2025 Many teams write feature flags directly into their source code, only to discover that maintaining them is a full-time job.
SM019 Google Cloud 2025 DORA State of AI Assisted Software Development Successful AI adoption is a systems problem, not a tools problem.
SM020 MIT Press Online Experimentation: Benefits, Operational and Methodological Challenges, and Scaling Guide An early and important decision that leaders have to make when adopting experimentation is to determine if they should build or buy the necessary tools.
SM021 Harvard Business School AI Institute Scaling Experimentation for a Competitive Edge For C-suite executives and business leaders, embracing a culture of experimentation is no longer optional—it’s a strategic imperative.
SM022 Grand View Research Product Analytics Market Size, Share & Growth Report, 2030 The global product analytics market size was estimated at USD 14.81 billion in 2023 and is expected to grow at a CAGR of 19.8% from 2024 to 2030.
SM023 Mordor Intelligence Product Analytics Market Size, Competitive Landscape, Trends 2026 – 2031 The product analytics market size is expected to grow from USD 11.39 billion in 2025 to USD 13.04 billion in 2026 and is forecast to reach USD 25.73 billion by 2031 at 14.55% CAGR.
SM024 Credence Research AB Testing Software Market Size, Share and Growth Report 2032 The AB testing software market size was valued at USD 8180 million in 2024 and is anticipated to reach USD 25816.9 million by 2032, at a CAGR of 15.45%.
SM025 VWO 30 Key A/B Testing Statistics: A Comprehensive Guide | VWO The global A/B testing tools market is projected to reach USD 850.2 million in 2024, with a compound annual growth rate (CAGR) of 14.00% from 2024 to 2031.
SM026 Information Commissioner’s Office Cookies and similar technologies You must also get the user’s consent.
SM027 European Data Protection Board Report of the work undertaken by the Cookie Banner Taskforce Type I practice: inaccurately classified “essential” cookies.
SM028 PostHog PostHog – We make dev tools for product engineers PostHog data stack, built for data teams and loved by product teams.
SM029 PostHog Experiments - Docs - PostHog Test changes to signup and onboarding flows.
SM030 PostHog Feature flags - Docs - PostHog They’re the foundation for safe rollouts, A/B testing, and remote configuration.
SP001 Statsig Statsig | The modern product development platform
SP002 Statsig Statsig | The modern product development platform
SP003 Statsig Statsig Overview - Statsig Documentation
SP004 LaunchDarkly Pricing | LaunchDarkly
SP005 LaunchDarkly Platform Overview | LaunchDarkly
SP006 Optimizely Optimizely Feature Experimentation
SP007 Optimizely Privacy, security and compliance
SP008 Optimizely Plans and pricing for Optimizely products
SP009 Harness Feature Management & Experimentation | AI Powered | Harness
SP010 Harness Harness Security Practices: Protecting Data & Ensuring Compliance
SP011 Harness Harness Pricing | Flexible Plans for DevOps & CI/CD
SP012 Amplitude Amplitude Feature Experimentation: A/B testing & Product Experimentation Platform
SP013 Amplitude Amplitude Pricing Options | Fast, Intelligent Customer Behavior Insights with Affordable Pricing Plans
SP014 Amplitude Amplitude Trust Center | Powered by Wolfia
SP015 GrowthBook Predictable Pricing – Free Tiers, Enterprise Plans | GrowthBook
SP016 GrowthBook GrowthBook | Experimentation, Feature Flags & Product Analytics Platform
SP017 GrowthBook GrowthBook Documentation | GrowthBook Docs
SP018 GrowthBook Security & Compliance | GrowthBook
SP019 Eppo Eppo is now Datadog Experiments | Next-Gen Experimentation Platform
SP020 Eppo Welcome to the Eppo Docs | The Eppo Docs
SP021 Eppo Security
SP022 PostHog PostHog pricing – Transparent, usage-based, generous free tier
SP023 PostHog Feature Flags – Ship safely and control rollouts with PostHog
SP024 PostHog Experiments - Docs - PostHog
SP025 PostHog Self-host PostHog - Docs - PostHog
SP026 VWO Pricing & Plans | VWO Platform
SP027 VWO #1 A/B Testing application for websites, mobile apps, server-side, and more | VWO Testing
SP028 VWO Data & Information Security | VWO
SP029 CostBench Best Feature Flag Tools 2026: Statsig, Flagsmith, GrowthBook Ranked
SP030 Startupik Top Feature Flag Tools Compared (LaunchDarkly vs Statsig vs GrowthBook) - Startupik | Startup magazine
SP031 OpenFeature OpenFeature
SP032 Martin Fowler Feature Toggles (aka Feature Flags)
SP033 GrowthBook How to migrate from Statsig to GrowthBook | GrowthBook Blog | Growthbook Blog Migrate from Statsig to an open-source platform where your data never leaves your warehouse. Auditable results, per-seat pricing, and a free AI-powered migration kit.
SP034 Forrester Feature Management And Experimentation — An Evolving Market Feature management and experimentation is a broad set of capabilities that spans both software delivery and product management.
SP035 Split.io Split.io
SI001 Statsig Pricing Pro $150/mo 5M events included, then $0.05 per 1K events.
SI002 Statsig Warehouse Native Statsig uses an event-based pricing model with no limitations on seats, flags, or environments.
SI003 Statsig Feature Flags We serve billions of users and trillions of daily flag checks, plus offer simple flat-rate pricing that scales with your team.
SI004 Statsig Experimentation Run Statsig in your warehouse or let us host it.
SI005 Statsig Product Analytics Statsig is the most affordable analytics platform at any scale - from our free tier to enterprise plans.
SI006 Statsig Company
SI007 Statsig Careers at Statsig We are a 100% in-person office and it may not be for everyone.
SI008 Statsig Customers Parafin started self-serving with Statsig, built on their wins, and expanded their investments in experimentation to directly impact revenue growth.
SI009 Statsig Testimonials
SI010 Statsig Statsig secures series C funding to bring science to the art of product development Statsig has raised a $100M Series C at a $1.1B valuation, led by ICONIQ Growth, with participation from Sequoia and Madrona.
SI011 Statsig Statsig's 2025 year in review We shipped a new feature every 3 days on average, grew the team to 170 full-time employees, earned the trust of 20,000 weekly active users, raised our Series C, and ended the year with a bang - joining forces with the team at OpenAI!
SI012 Statsig Statsig Overview - Statsig Documentation Statsig is a unified platform for feature flags, A/B testing, and product analytics.
SI013 BusinessWire Statsig Secures $100M Series C Funding to Transform Product Development, Valuing Startup at $1.1 Billion Statsig... today announced it has raised $100 million in Series C funding at a $1.1 billion valuation.
SI014 VentureBeat Statsig Secures $100M Series C Funding to Transform Product Development, Valuing Startup at $1.1 Billion
SI015 SiliconANGLE Statsig secures $100M in funding for its product testing platform The company reportedly reached $40 million in annual recurring revenue ahead of its funding round.
SI016 Sacra Statsig revenue, funding & news A key challenge for Statsig is the potential for experimentation budgets to be absorbed into existing observability spending, complicating the justification for standalone platform costs.
SI017 GeekWire OpenAI acquires Statsig for $1.1B, names CEO to key role in surprise exit for Seattle-area unicorn The acquisition price of $1.1 billion matches Statsig’s valuation in its latest funding round in May.
SI018 CNBC OpenAI acquires Statsig for $1.1 billion, brings on CEO as applications executive Statsig will continue to operate independently and serve customers out of its Seattle office, OpenAI said.
SI019 OpenAI Vijaye Raji to become CTO of Applications with acquisition of Statsig Once the acquisition is finalized, Statsig employees will become OpenAI employees. It will continue operating independently and serving its customer base out of its Seattle office.
SI020 Tracxn Statsig - 2026 Funding Rounds & List of Investors Statsig has raised a total of $153M over 3 funding rounds.
SI021 Tracxn Statsig - 2026 Company Profile, Team, Funding & Competitors Statsig is an acquired company based in Kirkland (United States), founded in 2021.
SI022 TrustRadius Statsig Pricing 2026 Statsig has 1 pricing plans(s)... A free version is available for Statsig.
SI023 TrustRadius Statsig Reviews & Ratings 2026 Cons: Complex data science focussed UI.
SI024 Growjo Statsig: Revenue, Competitors, Alternatives Statsig's estimated annual revenue is currently $23.7M per year... Statsig has 145 Employees.
SI025 Crunchbase Statsig - Crunchbase Company Profile & Funding Founded Date Feb 2021... Last Funding Type Series B... Legal Name Statsig, Inc.
SI026 U.S. Securities and Exchange Commission EDGAR Search Results for Statsig Home | Search the Next-Generation EDGAR System | Previous Page
SE001 Statsig Statsig | Smarter feature flags for fast-moving teams
SE002 Statsig Statsig | The world's leading experimentation platform
SE003 Statsig Statsig | Analytics for teams that ship
SE004 Statsig Statsig | Warehouse Native
SE005 Statsig Privacy | Statsig
SE006 Statsig Announcing Statsig Warehouse Native
SE007 Statsig Documentation Feature Flags - Statsig Documentation
SE008 Statsig Documentation Dynamic Config - Statsig Documentation
SE009 Statsig Documentation Experiments Overview - Statsig Documentation
SE010 Statsig Documentation CUPED - Statsig Documentation
SE011 Statsig Documentation Frequentist Sequential Testing - Statsig Documentation
SE012 Statsig Documentation Methodology - Statsig Documentation
SE013 Statsig Documentation Holdouts - Statsig Documentation
SE014 Statsig Documentation Product Analytics Overview - Statsig Documentation
SE015 Statsig Documentation Session Replay Overview - Statsig Documentation
SE016 Statsig Documentation Privacy Options for Session Replay - Statsig Documentation Use a feature gate to define which users are eligible for replays, ensuring that recordings are limited to specific users or cohorts.
SE017 Statsig Documentation About Warehouse Native - Statsig Documentation Statsig Warehouse Native runs experimentation compute jobs directly in your data warehouse, using your existing datasets to calculate metrics and enrich experiment analysis based on your data.
SE018 Statsig Documentation Pipeline Overview - Statsig Documentation
SE019 Statsig Documentation Console API Overview - Statsig Documentation
SE020 Statsig Documentation HTTP API - Statsig Documentation
SE021 Statsig Documentation Workspace Management Overview - Statsig Documentation
SE022 Statsig Documentation SCIM User Provisioning - Statsig Documentation
SE023 Statsig Documentation Single Sign-On With OIDC - Statsig Documentation
SE024 Statsig Documentation SDK Support Policy - Statsig Documentation
SE025 Statsig Documentation JavaScript Client SDK (Web) - Statsig Documentation
SE026 Statsig Documentation AI Governance, Security & Privacy - Statsig Documentation
SE027 Statsig Documentation Github Code References - Statsig Documentation
SE028 Statsig Documentation Roles - Statsig Documentation
SE029 GitHub community @statsig/session-replay v3.25.5 crashes on Safari 15 with "Invalid regular expression: invalid group specifier name" · Issue #36 · statsig-io/js-client-monorepo The @statsig/session-replay package (v3.25.5) causes the website to completely fail to load on Safari version 15.6.
SE030 GitHub community Role Based Access Control on Individual Created Feature Gates Or Rules in Statsig · Issue #40 · statsig-io/statsig-feedback Is it Possible to have Role based Edit/Delete access on individual Feature Gates Or Rules in Statsig?
SE031 GitHub / statsig-io GitHub - statsig-io/node-js-server-sdk: Statsig's SDK for server-side Node.js applications.
SE032 GitHub / statsig-io GitHub - statsig-io/go-sdk: A golang SDK for interfacing with Statsig Feature Gates, Dynamic Configs, and A/B Experiments
SE033 npm @statsig/js-client - npm
SE034 npm statsig-node - npm
SE035 Go Packages statsig package - github.com/statsig-io/go-sdk - Go Packages
SE036 pub.dev statsig | Dart package
SE037 NuGet Statsig 2.4.1
SE038 Maven Central Maven Central: com.statsig:serversdk
SE039 Sonatype Maven Central: com.statsig:serversdk
SE040 Okta Integrate Statsig with Okta
SE041 Twilio Segment Statsig Integration
SU001 Statsig Statsig is the best, say our customers "Excited to bring Statsig to Whatnot! We finally found a product that moves just as fast as we do and have been super impressed with how closely our teams collaborate."
SU002 Statsig Statsig customer story - Ancestry goes from 70 to 600 annual experiments "I know that we are able to impact our key business metrics in a positive way with Statsig."
SU003 Statsig Statsig customer story - How Bluesky reached 25 million users in record time - and keeps growing!
SU004 Statsig Statsig customer story - How Brex’s data teams achieved a +50% time efficiency gain
SU005 Statsig Statsig customer story - How Character.ai scales AI entertainment with AI experimentation
SU006 Statsig Statsig customer story - How Code.org unlocked data visibility for millions of users
SU007 Statsig Statsig customer story - How Graphite built speed and control into their development workflow
SU008 Statsig Statsig customer story - HelloFresh scales to 1000 experiments per year with Statsig Warehouse Native
SU009 Statsig Statsig customer story - How Linktree drives product-led growth with Statsig Warehouse Native
SU010 Statsig Statsig customer story - Notion went from single-digit to hundreds of experiments per quarter
SU011 Statsig Statsig customer story - OfferUp increases target CTA clicks by 11%
SU012 Statsig Statsig customer story - Runna runs 100+ tests within a year to drive user growth
SU013 Statsig Statsig customer story - SoundCloud achieved profitability and released its flagship news feed with experimentation
SU014 Statsig Statsig customer story - How Webflow decoupled launches from deployments
SU015 Statsig Statsig customer story - Whatnot goes from 0 to 400 annual experiments
SU016 Ancestry Ancestry - Family Tree, Genealogy & Family History Records
SU017 Bluesky Bluesky
SU018 Brex Brex: The Modern Finance Software Platform | Spend Smarter
SU019 Code.org About Code.org – Our Mission, Impact, and Approach
SU020 Graphite Graphite - Code review for the age of AI
SU021 HelloFresh HelloFresh® Meal Kits | Free Breakfast for Life + 10 Free Meals
SU022 Linktree Link in bio tool: Everything you are, in one simple link | Linktree
SU023 Notion The AI workspace that works for you. | Notion
SU024 OfferUp OfferUp | Buy, Sell, and Discover Locally
SU025 Runna Runna | #1 rated personalized training plans for running
SU026 Webflow About: our mission, vision, and principles | Webflow
SU027 TrustRadius Statsig Reviews & Ratings 2026 | TrustRadius Complex data science focussed UI
SU028 FeaturedCustomers 57 Statsig Customer Reviews & References
SR001 Statsig Statsig | The modern product development platform Statsig gives your team 5+ products in a single platform.
SR002 Statsig Statsig | The modern product development platform HIPAA-eligibility (BAA required)
SR003 Statsig Statsig | The modern product development platform
SR004 Statsig Statsig is the best, say our customers
SR005 Statsig Statsig | Smarter feature flags for fast-moving teams
SR006 Statsig Statsig | The world's leading experimentation platform
SR007 Statsig Statsig | Analytics for teams that ship
SR008 Statsig Statsig | Warehouse Native
SR009 Statsig Security at Statsig | Statsig SOC2 Type II audited and certified
SR010 Statsig Privacy | Statsig Amplitude, Inc. (“Statsig,” or “we”) is an experimentation platform...
SR011 Statsig Data Processing Addendum (DPA) | Statsig Customer is solely responsible for ... ensuring that no sensitive or special categories of personal data is provided to Statsig.
SR012 Amplitude Amplitude Subprocessor List Amplitude may use Subprocessors to Process Customer Data in its provision of the Amplitude Services.
SR013 Statsig Documentation Statsig Overview - Statsig Documentation
SR014 Statsig Documentation Warehouse Native Quickstart - Statsig Documentation
SR015 Statsig Documentation Connect Your Warehouse - Statsig Documentation
SR016 Statsig Documentation Egress, Privacy, & Storage - Statsig Documentation Experimentation staging datasets can generate a lot of data, since they can potentially blow up your data by the number of experiments you run.
SR017 Statsig Documentation Data Best Practices - Statsig Documentation
SR018 Statsig Documentation Workspace Management Overview - Statsig Documentation
SR019 Statsig Documentation Experiment Policy - Statsig Documentation
SR020 Statsig Documentation Feature Gates Policy - Statsig Documentation
SR021 Statsig Documentation Frequentist Sequential Testing - Statsig Documentation
SR022 Statsig Documentation Methodology - Statsig Documentation
SR023 Statsig Documentation AI Evals Overview - Statsig Documentation
SR024 Statsig Documentation AI Governance, Security & Privacy - Statsig Documentation Your data is not used to build or develop any AI models unless you provide explicit written consent.
SR025 OpenAI Vijaye Raji to become CTO of Applications with acquisition of Statsig
SR026 Statsig Statsig is joining OpenAI
SR027 Reuters OpenAI to acquire product testing startup Statsig, appoints CTO of applications
SR028 CNBC OpenAI acquires Statsig for $1.1 billion, brings on CEO as applications executive
SR029 MarTech Amplitude and Statsig deal raises questions for customers | MarTech Amplitude is getting Statsig’s code without the talent, it is a race car without a driver.
SR030 PostHog PostHog pricing – Transparent, usage-based, generous free tier
SR031 Amplitude Amplitude Pricing Options | Fast, Intelligent Customer Behavior Insights with Affordable Pricing Plans
SR032 Optimizely Plans and pricing for Optimizely products
SR033 National Institute of Standards and Technology Cybersecurity Framework
SR034 European Union Regulation - 2016/679 - EN - gdpr
SR035 European Union Regulation - EU - 2024/1689 - EN
SR036 U.S. Department of Health and Human Services Business Associates
SR037 Google Cloud BigQuery overview  |  Google Cloud Documentation
SR038 Snowflake Overview of Access Control | Snowflake Documentation
SR039 Databricks What is Unity Catalog? | Databricks on AWS
SR040 The CX Lead Statsig Review: Pros, Cons, Features and Pricing
SV001 Statsig Statsig secures series C funding to bring science to the art of product development
SV002 Business Wire Statsig Secures $100M Series C Funding to Transform Product Development, Valuing Startup at $1.1 Billion
SV003 GeekWire Statsig lands $100M at $1.1B valuation, aiming to unify product development as AI upends industry
SV004 Fortune Exclusive: Statsig raises $100 million at $1.1 billion valuation after abandoned Datadog acquisition attempt Datadog bought its rival Eppo for $220 million, Upstarts Media first reported.
SV005 Statsig Statsig is joining OpenAI
SV006 OpenAI Vijaye Raji to become CTO of Applications with acquisition of Statsig Once the acquisition is finalized, Statsig employees will become OpenAI employees. It will continue operating independently and serving its customer base out of its Seattle office.
SV007 CNBC OpenAI acquires Statsig for $1.1 billion, brings on CEO as applications executive OpenAI has acquired Statsig for $1.1 billion.
SV008 GeekWire OpenAI acquires Statsig for $1.1B, names CEO to key role in surprise exit for Seattle-area unicorn
SV009 Sequoia Capital Statsig and OpenAI: A New Chapter for Product Experimentation
SV010 ARR Club Statsig ARR Hits $40M
SV011 Sacra Statsig revenue, funding & news At a $1.1B valuation, Statsig's $40M in ARR values it at approximately a 27.5x revenue multiple.
SV012 Statsig Statsig Pricing
SV013 Statsig Statsig is the best, say our customers
SV014 Statsig Statsig Overview - Statsig Documentation
SV015 Statsig Why Datadog bought Eppo for $220M, and what it means for the future of experimentation
SV016 TechCrunch Datadog acquires Eppo, a feature-flagging and experimentation platform
SV017 PR Newswire Episerver Completes Acquisition of Optimizely, Creating the Industry's Most Advanced Digital Experience Platform
SV018 TechCrunch Optimizely acquired by content management company Episerver
SV019 Optimizely Optimizely vs. Statsig Amplitude took over the Statsig platform. The engineers who built it are staying at OpenAI.
SV020 Optimizely Optimizely Web Experimentation
SV021 Amplitude Amplitude Announces First Quarter 2026 Financial Results
SV022 Amplitude Investor Relations Amplitude Announces First Quarter 2026 Financial Results | Amplitude
SV023 CompaniesMarketCap Amplitude (AMPL) - Market capitalization
SV024 Amplitude Amplitude and Statsig Partnership Amplitude will take on Statsig’s brand and customers. Amplitude will maintain and develop the current Statsig platform across the cloud and data warehouse.
SV025 Securities and Exchange Commission EDGAR Filing Documents for 0001193125-26-209631
SV026 Securities and Exchange Commission Amplitude 10-Q for quarter ended 2026-03-31
SV027 Datadog Datadog Announces First Quarter 2026 Financial Results
SV028 Datadog Datadog Announces Fourth Quarter and Fiscal Year 2025 Financial Results
SV029 CompaniesMarketCap Datadog (DDOG) - Market capitalization
SV030 Securities and Exchange Commission EDGAR Filing Documents for 0001628280-26-032328
SV031 Securities and Exchange Commission Datadog 10-Q for quarter ended 2026-03-31