[{"data":1,"prerenderedAt":409},["ShallowReactive",2],{"tool-marketing-thinkingdata":3,"footer-total":384,"related-marketing-thinkingdata":385},{"id":4,"title":5,"author":6,"bestFor":7,"body":8,"category":343,"cons":344,"cover":349,"date":350,"description":351,"draft":352,"extension":353,"meta":354,"navigation":355,"notFor":356,"path":357,"pricing":358,"pros":362,"region":367,"score":368,"scoreBreakdown":369,"seo":375,"stem":376,"tags":377,"toolName":209,"updated":350,"url":381,"vendor":382,"__hash__":383},"tools\u002Ftools\u002Fmarketing\u002Fthinkingdata.md","数数科技 评测","元析评测","有中重度游戏产品、需要按区服和付费分层做运营的数据与运营团队",{"type":9,"value":10,"toc":315},"minimark",[11,15,23,26,29,45,48,52,55,58,61,64,67,70,73,76,80,83,86,89,93,96,100,103,107,110,114,117,121,124,127,192,195,198,272,278,282,286,289,293,296,300,303,306,309],[12,13,14],"h2",{"id":14},"结论先行",[16,17,18,22],"p",{},[19,20,21],"strong",{},"推荐,总分 7.6,官网未公开报价,按事件量年付,需联系销售。"," 适合有中重度游戏、按区服和付费分层做运营的团队。数数科技(ThinkingData)从游戏数据分析起家,数据模型里区服、角色、等级、充值档位都是原生字段,不用自己硬凑。",[16,24,25],{},"换成非游戏业务,这套模型的优势就变成负担:模板大多围绕留存和付费,电商要看的是转化漏斗和客单结构,配置工作量明显更大。选它之前先看自己的业务形态是不是事件流清晰、可按角色或账户分层。",[12,27,28],{"id":28},"这篇评测基于什么数据",[30,31,32,36,39,42],"ul",{},[33,34,35],"li",{},"官网与产品页:2026-11 查看了 thinkingdata.cn 的产品页与行业方案,核对 ThinkingAnalytics 的分析模块、私有化部署说明与 ThinkingAI 的定位。",[33,37,38],{},"公开资料:参考官方文档中的事件模型规范、渠道归因与 LTV 分析说明,以及公开分享里提到的游戏行业落地流程。",[33,40,41],{},"对比基准:横向对比神策数据与火山引擎 DataFinder 在数据模型、部署方式与计费口径上的差别。",[33,43,44],{},"实测边界:未做统一环境实测与性能压测,文中关于配置量、查询性能的判断来自官方文档与使用者访谈,不涉及具体客户数据。",[12,46,47],{"id":47},"核心功能",[49,50,51],"h3",{"id":51},"游戏场景的数据模型",[16,53,54],{},"游戏的数据结构和电商不一样:一个玩家可能在多个区服有多个角色,充值是角色级的,留存要按角色算也要按账号算。数数的模型把这几层拆开了,区服、角色 ID、等级、VIP 档位、充值金额都是标准属性。",[16,56,57],{},"好处是运营提问时不用翻译。「上周新开的三个服里,首充率最低的是哪个」这种问题,后台选几个条件就出图,不用找数据开发写 SQL。跨端产品的账号打通要自己传用户 ID,默认按设备算。",[49,59,60],{"id":60},"分析与看板",[16,62,63],{},"留存、漏斗、路径、分布、留存矩阵这些常规分析都有,还多了游戏特有的付费分析:首日付费率、付费档位分布、ARPPU、按渠道拆的 LTV 曲线。",[16,65,66],{},"看板可以按项目、按区服、按渠道做多套,给发行和研发看不同的视图。看板权限能细到单张图表,做多项目并行时比较实用。",[49,68,69],{"id":69},"私有化与数据治理",[16,71,72],{},"私有化是它的主力交付方式,国内不少游戏公司要求数据不出机房。部署形态包括单机、集群、以及对接已有的大数据平台。",[16,74,75],{},"数据治理这块有统一的事件管理、属性管理、用户表管理入口,新埋点要走审核。这套流程刚开始觉得繁琐,项目多了之后能避免同名事件满天飞。",[49,77,79],{"id":78},"thinkingai-与智能分析","ThinkingAI 与智能分析",[16,81,82],{},"近两年主推的 ThinkingAI 是把大模型接到数据上,支持用自然语言提问出图,比如「上周付费金额下降的原因」。实际效果取决于数据规范和问题表述,复杂归因仍然要人来看。它还能按自然语言生成一段分析思路,给新人当参考,但结论要人工核对。",[16,84,85],{},"对不写 SQL 的运营,这个入口降低了门槛;对数据团队,它更像一个查询加速器,还不能替代建模。",[12,87,88],{"id":88},"评分明细",[49,90,92],{"id":91},"功能完整度8110","功能完整度:8.1\u002F10",[16,94,95],{},"游戏分析需要的能力基本齐全,从埋点到付费分层到 LTV 都能覆盖。缺的是通用 BI 那种自由拖拽和外部数据源接入,做跨系统联合分析要绕。",[49,97,99],{"id":98},"易用性7310","易用性:7.3\u002F10",[16,101,102],{},"后台功能密度高,菜单层级深,新用户第一周主要在找入口。埋点规范和事件命名要有人统一管,否则三个月后事件列表会失控。",[49,104,106],{"id":105},"数据质量7910","数据质量:7.9\u002F10",[16,108,109],{},"事件采集稳定,离线与实时链路分开,报表口径清楚。数据准确性依赖埋点评审,漏埋和重复上报是常见问题,上线前要做对数。",[49,111,113],{"id":112},"性价比7010","性价比:7.0\u002F10",[16,115,116],{},"按事件量年付,中重度游戏每天上报量大,费用随产品规模一起涨。功能密度对得起报价,小体量产品摊下来偏贵。",[49,118,120],{"id":119},"客户支持7710","客户支持:7.7\u002F10",[16,122,123],{},"有专属对接群和实施团队,游戏行业的经验足,能给出埋点方案建议。非游戏客户的支持响应相对慢一些,行业 know-how 也弱。",[12,125,126],{"id":126},"定价说明",[128,129,130,146],"table",{},[131,132,133],"thead",{},[134,135,136,140,143],"tr",{},[137,138,139],"th",{},"套餐",[137,141,142],{},"价格",[137,144,145],{},"说明",[147,148,149,161,171,182],"tbody",{},[134,150,151,155,158],{},[152,153,154],"td",{},"SaaS 版",[152,156,157],{},"需联系销售",[152,159,160],{},"按年事件量计费,含分析与看板模块",[134,162,163,166,168],{},[152,164,165],{},"私有化版",[152,167,157],{},[152,169,170],{},"一次性授权加年度服务费,含实施与升级",[134,172,173,176,179],{},[152,174,175],{},"定制开发",[152,177,178],{},"按项目",[152,180,181],{},"数据对接、模型定制、专属看板",[134,183,184,187,189],{},[152,185,186],{},"ThinkingAI 模块",[152,188,157],{},[152,190,191],{},"按账号或调用量开通,与主产品分开计价",[16,193,194],{},"官网未公开报价,需联系销售,2026-11 核对。谈价时把三件事写进合同:事件量超限后的单价、私有化的升级是否另收费、以及数据导出的格式与频次。这三项决定了三年期的总成本。产品有海外发行的话,还要确认数据存储地域与合规要求。",[12,196,197],{"id":197},"和竞品对比",[128,199,200,216],{},[131,201,202],{},[134,203,204,207,210,213],{},[137,205,206],{},"维度",[137,208,209],{},"数数科技",[137,211,212],{},"神策数据",[137,214,215],{},"火山引擎 DataFinder",[147,217,218,231,245,258],{},[134,219,220,223,226,228],{},[152,221,222],{},"起步价",[152,224,225],{},"未公开,按事件量年付",[152,227,225],{},[152,229,230],{},"有试用额度,正式版按量",[134,232,233,236,239,242],{},[152,234,235],{},"行业强项",[152,237,238],{},"游戏,区服与付费模型",[152,240,241],{},"多行业,数据治理严谨",[152,243,244],{},"增长与实验,字节实践",[134,246,247,250,253,255],{},[152,248,249],{},"部署",[152,251,252],{},"SaaS 与私有化",[152,254,252],{},[152,256,257],{},"以云服务为主",[134,259,260,263,266,269],{},[152,261,262],{},"短板",[152,264,265],{},"非游戏模板少",[152,267,268],{},"实施重、报价高",[152,270,271],{},"迁移成本与计费复杂",[16,273,274,277],{},[19,275,276],{},"怎么选",":产品是游戏、要按区服和付费分层做运营,选数数;多业务线并行、需要统一数据治理,神策更稳;已经在用火山引擎、要做增长实验,DataFinder 更顺。",[12,279,281],{"id":280},"faq","FAQ",[49,283,285],{"id":284},"q-我们游戏日活-8-万事件量大概什么量级","Q: 我们游戏日活 8 万,事件量大概什么量级?",[16,287,288],{},"A: 中重度游戏人均每天可能上报几十到上百个事件,按 8 万日活估算,月事件量在几亿条这个区间。具体取决于埋点密度,签约前让技术按事件清单算一遍再谈档位。",[49,290,292],{"id":291},"q-私有化部署要准备多少服务器","Q: 私有化部署要准备多少服务器?",[16,294,295],{},"A: 取决于事件量和查询并发,小体量项目通常几台机器起步,集群版要更多。官方会给资源评估表,建议按当前量的一点五倍预留,避免开新服时临时扩容。",[49,297,299],{"id":298},"q-thinkingai-能替代数据分析师吗","Q: ThinkingAI 能替代数据分析师吗?",[16,301,302],{},"A: 目前不能。它擅长把常见问题快速出图,比如看某天的付费波动;涉及口径定义、归因判断和多表关联的问题,还是要有分析师把关。",[12,304,305],{"id":305},"评测方法说明",[16,307,308],{},"元析评测按功能完整度、易用性、数据质量、性价比、客户支持五项打分,每项满分 10 分,五项均值折算总分。文中功能范围、部署形态与计费口径于 2026-11 对照官网核对;未做压测与接入实测的部分已在正文说明。",[16,310,311,314],{},[19,312,313],{},"最后更新",":2026-11-16",{"title":316,"searchDepth":317,"depth":317,"links":318},"",2,[319,320,321,328,335,336,337,342],{"id":14,"depth":317,"text":14},{"id":28,"depth":317,"text":28},{"id":47,"depth":317,"text":47,"children":322},[323,325,326,327],{"id":51,"depth":324,"text":51},3,{"id":60,"depth":324,"text":60},{"id":69,"depth":324,"text":69},{"id":78,"depth":324,"text":79},{"id":88,"depth":317,"text":88,"children":329},[330,331,332,333,334],{"id":91,"depth":324,"text":92},{"id":98,"depth":324,"text":99},{"id":105,"depth":324,"text":106},{"id":112,"depth":324,"text":113},{"id":119,"depth":324,"text":120},{"id":126,"depth":317,"text":126},{"id":197,"depth":317,"text":197},{"id":280,"depth":317,"text":281,"children":338},[339,340,341],{"id":284,"depth":324,"text":285},{"id":291,"depth":324,"text":292},{"id":298,"depth":324,"text":299},{"id":305,"depth":317,"text":305},"marketing",[345,346,347,348],"非游戏行业的分析模板少,电商与内容类客户要自己配看板","按事件量计价,重度游戏每天上报量大,费用增长快","自助分析的自由度不如通用 BI,复杂口径仍要找服务方","官网不标价,采购要走销售与技术评估两轮,周期一个月起",null,"2026-11-16","数数科技评测:ThinkingAnalytics 从游戏数据分析起家,区服与付费模型做得细,支持私有化,按事件量年付且官网未公开报价需联系销售;非游戏行业模板偏少。",false,"md",{},true,"只有官网和小程序、没有事件量积累的中小电商或内容站","\u002Ftools\u002Fmarketing\u002Fthinkingdata",{"free":352,"startingPrice":316,"currency":359,"billingPeriod":360,"notes":361},"CNY","年","按年事件量与部署方式报价,支持 SaaS 与私有化;官网未公开报价,需联系销售,2026-11 核对,私有化另含实施、运维与版本升级费用。",[363,364,365,366],"数据模型原生支持区服、角色、等级、付费档位,游戏团队不用自己造字段","付费分析与 LTV 预测做成模板,能按渠道看首日到三十日的回收曲线","私有化部署成熟,数据留在自有服务器,已在国内多家游戏公司跑生产环境","数据治理做得细,事件、属性、用户表有统一管理入口,长期使用不容易乱","cn",7.6,{"features":370,"ease":371,"data":372,"price":373,"support":374},8.1,7.3,7.9,7,7.7,{"title":5,"description":351},"tools\u002Fmarketing\u002Fthinkingdata",[378,379,209,380],"市场营销","游戏分析","ThinkingAnalytics","https:\u002F\u002Fwww.thinkingdata.cn","数数科技(ThinkingData)","XVE8tPEtEucI_GPGKwZS3FbZd2Z9nP8mUR_WCemuu9s",1016,[386,395,403],{"path":387,"toolName":388,"description":389,"score":390,"pricing":391,"vendor":394,"date":350,"category":343},"\u002Ftools\u002Fmarketing\u002Foceanengine","巨量引擎","巨量引擎评测:不卖套餐,按 CPC\u002FOCPM 竞价消耗,小额预算也能开跑。抖音日活 7 亿量级的流量池,素材产能强的电商与线索商家值得投;计划三五天衰退,审核严。",8.3,{"free":352,"startingPrice":316,"currency":359,"billingPeriod":392,"notes":393},"按量","官网未公开报价,需联系销售,2026-11 核对。平台只公开计费方式:信息流按 CPC\u002FCPM\u002FOCPM 竞价,电商走巨量千川按成交目标出价,品牌广告按 CPM\u002FCPT 排期刊例,均无固定套餐,消耗即费用。","字节跳动",{"path":396,"toolName":397,"description":398,"score":399,"pricing":400,"vendor":402,"date":350,"category":343},"\u002Ftools\u002Fmarketing\u002Falimama","阿里妈妈","阿里妈妈评测:官网不公开报价,直通车按点击计费、淘客按成交付佣金。淘系站内流量最直接的买法,适合已有淘宝天猫店的卖家;站内流量见顶,点击单价涨,智能投放偏黑盒。",8.2,{"free":352,"startingPrice":316,"currency":359,"billingPeriod":392,"notes":401},"官网未公开报价,需联系销售,2026-11 核对。直通车按关键词点击扣费,万相台无界版按成交目标出价,引力魔方按 CPC\u002FCPM 竞价,淘宝联盟按成交抽佣,均无公开套餐价。","阿里巴巴",{"path":404,"toolName":212,"description":405,"score":399,"pricing":406,"vendor":212,"date":408,"category":343},"\u002Ftools\u002Fmarketing\u002Fsensorsdata","神策数据评测:事件模型加多维分析,用户画像与智能运营可分开买,支持私有化部署,官网未公开标准报价需联系销售,提供 1 个月免费试用。实施与运维另算,适合有数据团队的中大型公司。",{"free":352,"startingPrice":316,"currency":359,"billingPeriod":360,"notes":407},"官网未公开标准报价,需联系销售,按数据量、模块组合与部署方式(SaaS 或私有化)报价。官网提供 1 个月免费试用,2026-11 核对产品与试用说明。","2026-11-04",1794788239053]