[{"data":1,"prerenderedAt":335},["ShallowReactive",2],{"blog-facebook-custom-audience-guide-zh-CN":3},{"id":4,"title":5,"excerpt":6,"content":7,"coverImage":275,"meta":283,"site":287,"status":308,"slug":309,"author":310,"category":321,"publishDate":19,"featured":47,"updatedAt":330,"createdAt":331,"contentHtml":332,"previewUrl":333,"localeSlugs":334},559,"Facebook 自定义受众指南 2026：四种来源、名单准备与匹配率","四种自定义受众来源各自真正的用途、怎么准备一份真的匹配得上的客户名单、决定网站受众质量的几条规则，以及四个悄悄让受众失效的错误。",{"root":8},{"children":9,"direction":19,"format":15,"indent":13,"type":274,"version":18},[10,21,27,33,61,67,77,81,85,89,95,101,107,113,119,127,131,137,143,149,155,159,165,171,177,183,187,224,228,232,238,244,250,256,262,266,270],{"children":11,"direction":19,"format":15,"indent":13,"type":20,"version":18,"textFormat":13,"textStyle":15},[12],{"detail":13,"format":13,"mode":14,"style":15,"text":16,"type":17,"version":18},0,"normal","","自定义受众（Custom Audience）在投放后台里看起来很简单：选个来源、起个名、拿去用。但决定这个受众到底有没有用的事，发生在更早的地方——来源怎么准备、匹配出来的规模有多大、以及这个受众衡量的到底是不是你以为的那件事。这篇讲清四种来源类型、决定匹配率的名单准备规则，以及那些悄悄让受众变得没用的错误。","text",1,null,"paragraph",{"children":22,"direction":19,"format":15,"indent":13,"type":25,"version":18,"tag":26},[23],{"detail":13,"format":13,"mode":14,"style":15,"text":24,"type":17,"version":18},"四种来源，各自真正的用途","heading","h2",{"children":28,"direction":19,"format":15,"indent":13,"type":20,"version":18,"textFormat":18,"textStyle":15},[29,31],{"detail":13,"format":18,"mode":14,"style":15,"text":30,"type":17,"version":18},"客户名单。",{"detail":13,"format":13,"mode":14,"style":15,"text":32,"type":17,"version":18}," 你上传自己已经持有的标识符。这是唯一从头到尾由你掌控的来源，也是唯一能覆盖「从没碰过你网站或 App 的人」的来源。它的天花板是你的数据质量。",{"children":34,"direction":19,"format":15,"indent":13,"type":20,"version":18,"textFormat":18,"textStyle":15},[35,37,39,50,52,59],{"detail":13,"format":18,"mode":14,"style":15,"text":36,"type":17,"version":18},"网站受众。",{"detail":13,"format":13,"mode":14,"style":15,"text":38,"type":17,"version":18}," 由像素或转化 API 的事件构建——访客、看过特定页面的人、完成过特定动作的人。最常用的来源，也最受度量缺口影响：追踪漏事件，受众就会悄无声息地比应有的规模小。所以先把度量层做对是前提，见 ",{"children":40,"direction":19,"format":15,"indent":13,"type":43,"version":44,"fields":45,"id":49},[41],{"detail":13,"format":13,"mode":14,"style":15,"text":42,"type":17,"version":18},"Meta 转化 API 接入指南","link",3,{"linkType":46,"newTab":47,"url":48},"custom",false,"https://deepclick.com/zh-CN/resources/blog/meta-conversions-api-setup/","6a9408adc2bdf500c8ac4085",{"detail":13,"format":13,"mode":14,"style":15,"text":51,"type":17,"version":18},"与 ",{"children":53,"direction":19,"format":15,"indent":13,"type":43,"version":44,"fields":56,"id":58},[54],{"detail":13,"format":13,"mode":14,"style":15,"text":55,"type":17,"version":18},"Facebook Pixel 接入指南",{"linkType":46,"newTab":47,"url":57},"https://deepclick.com/zh-CN/resources/blog/facebook-pixel-setup-guide/","6a9408adc2bdf500c8ac4086",{"detail":13,"format":13,"mode":14,"style":15,"text":60,"type":17,"version":18},"。",{"children":62,"direction":19,"format":15,"indent":13,"type":20,"version":18,"textFormat":18,"textStyle":15},[63,65],{"detail":13,"format":18,"mode":14,"style":15,"text":64,"type":17,"version":18},"App 活动受众。",{"detail":13,"format":13,"mode":14,"style":15,"text":66,"type":17,"version":18}," 移动端的对应物，由 SDK 上报的 App 事件构建。需要先注册应用、先定义事件，之后才可能有受众。",{"children":68,"direction":19,"format":15,"indent":13,"type":20,"version":18,"textFormat":18,"textStyle":15},[69,71,73,75],{"detail":13,"format":18,"mode":14,"style":15,"text":70,"type":17,"version":18},"互动受众。",{"detail":13,"format":13,"mode":14,"style":15,"text":72,"type":17,"version":18}," 由用户在平台内与你的资产的互动构建——视频观看、表单打开与提交、主页或资料页互动。它",{"detail":13,"format":18,"mode":14,"style":15,"text":74,"type":17,"version":18},"完全不依赖你自己的追踪",{"detail":13,"format":13,"mode":14,"style":15,"text":76,"type":17,"version":18},"，因此在你的站内度量还不完整时，它是最可靠的来源——也正因为这个原因，它是被浪费得最多的一种。",{"children":78,"direction":19,"format":15,"indent":13,"type":20,"version":18,"textFormat":13,"textStyle":15},[79],{"detail":13,"format":13,"mode":14,"style":15,"text":80,"type":17,"version":18},"一个实用的排序：互动受众立刻可用，网站受众需要追踪是好的，客户名单需要数据是干净的，App 受众需要先做 SDK。",{"children":82,"direction":19,"format":15,"indent":13,"type":25,"version":18,"tag":26},[83],{"detail":13,"format":13,"mode":14,"style":15,"text":84,"type":17,"version":18},"怎么准备一份真的匹配得上的客户名单",{"children":86,"direction":19,"format":15,"indent":13,"type":20,"version":18,"textFormat":13,"textStyle":15},[87],{"detail":13,"format":13,"mode":14,"style":15,"text":88,"type":17,"version":18},"上传的标识符会在离开你的机器之前先哈希。匹配率——也就是你的名单里能对应到真实账号的比例——几乎完全由准备工作决定。",{"children":90,"direction":19,"format":15,"indent":13,"type":20,"version":18,"textFormat":18,"textStyle":15},[91,93],{"detail":13,"format":18,"mode":14,"style":15,"text":92,"type":17,"version":18},"用不止一列标识符。",{"detail":13,"format":13,"mode":14,"style":15,"text":94,"type":17,"version":18}," 同时有邮箱、手机号、姓名的名单，比只有邮箱的匹配得好，因为任何一列都可能在某个人身上失效。",{"children":96,"direction":19,"format":15,"indent":13,"type":20,"version":18,"textFormat":18,"textStyle":15},[97,99],{"detail":13,"format":18,"mode":14,"style":15,"text":98,"type":17,"version":18},"上传前先规范化。",{"detail":13,"format":13,"mode":14,"style":15,"text":100,"type":17,"version":18}," 全部转小写、去首尾空格、手机号去掉标点并加国家区号、国家用标准两位代码。格式不一致不会报错，它只是匹配不上。",{"children":102,"direction":19,"format":15,"indent":13,"type":20,"version":18,"textFormat":18,"textStyle":15},[103,105],{"detail":13,"format":18,"mode":14,"style":15,"text":104,"type":17,"version":18},"把混装字段拆开。",{"detail":13,"format":13,"mode":14,"style":15,"text":106,"type":17,"version":18}," 一种标识符一列。一个「联系方式」列里有时装邮箱有时装手机号，匹配会很差。",{"children":108,"direction":19,"format":15,"indent":13,"type":20,"version":18,"textFormat":18,"textStyle":15},[109,111],{"detail":13,"format":18,"mode":14,"style":15,"text":110,"type":17,"version":18},"有国家就带上国家。",{"detail":13,"format":13,"mode":14,"style":15,"text":112,"type":17,"version":18}," 它能消解手机号和常见姓名的歧义，而且几乎不花成本。",{"children":114,"direction":19,"format":15,"indent":13,"type":20,"version":18,"textFormat":18,"textStyle":15},[115,117],{"detail":13,"format":18,"mode":14,"style":15,"text":116,"type":17,"version":18},"别指望名单能被全量匹配上。",{"detail":13,"format":13,"mode":14,"style":15,"text":118,"type":17,"version":18}," 一份准备良好的名单，现实区间大致在「较大的少数」到「多数」之间。匹配率极低几乎总是格式问题而不是数据问题——在下「我的客户触达不到」这个结论之前，先检查规范化。",{"children":120,"direction":19,"format":15,"indent":13,"type":20,"version":18,"textFormat":13,"textStyle":15},[121,123,125],{"detail":13,"format":13,"mode":14,"style":15,"text":122,"type":17,"version":18},"受众可用之前有最小规模门槛，而算数的是",{"detail":13,"format":18,"mode":14,"style":15,"text":124,"type":17,"version":18},"匹配后的规模",{"detail":13,"format":13,"mode":14,"style":15,"text":126,"type":17,"version":18},"，不是上传的行数。",{"children":128,"direction":19,"format":15,"indent":13,"type":25,"version":18,"tag":26},[129],{"detail":13,"format":13,"mode":14,"style":15,"text":130,"type":17,"version":18},"网站受众：决定质量的几条规则",{"children":132,"direction":19,"format":15,"indent":13,"type":20,"version":18,"textFormat":18,"textStyle":15},[133,135],{"detail":13,"format":18,"mode":14,"style":15,"text":134,"type":17,"version":18},"回溯窗口。",{"detail":13,"format":13,"mode":14,"style":15,"text":136,"type":17,"version":18}," 受众往回看多少天。窗口短意味着意图更新、规模更小；窗口长意味着有量但信号更陈旧。按用途分别选，而不是一律拉满。",{"children":138,"direction":19,"format":15,"indent":13,"type":20,"version":18,"textFormat":18,"textStyle":15},[139,141],{"detail":13,"format":18,"mode":14,"style":15,"text":140,"type":17,"version":18},"规则的具体程度。",{"detail":13,"format":13,"mode":14,"style":15,"text":142,"type":17,"version":18}," 「所有访客」和「到达过某个特定步骤的人」是两种完全不同的资产。后者更小，通常也更值钱，而且正是它让「一串有层次的受众」成为可能，而不是一个不加区分的大池子。",{"children":144,"direction":19,"format":15,"indent":13,"type":20,"version":18,"textFormat":18,"textStyle":15},[145,147],{"detail":13,"format":18,"mode":14,"style":15,"text":146,"type":17,"version":18},"刷新方式。",{"detail":13,"format":13,"mode":14,"style":15,"text":148,"type":17,"version":18}," 网站受众随事件到达持续更新；客户名单是一张快照，不重新上传就会过期。团队常常记得前半句、忘了后半句，于是一份名单在早已不代表任何在册客户之后还被用了好几个月。",{"children":150,"direction":19,"format":15,"indent":13,"type":20,"version":18,"textFormat":18,"textStyle":15},[151,153],{"detail":13,"format":18,"mode":14,"style":15,"text":152,"type":17,"version":18},"排除和包含一样重要。",{"detail":13,"format":13,"mode":14,"style":15,"text":154,"type":17,"version":18}," 受众定向里大部分被浪费的花费，来自没有排除已转化的人，或者来自互相重叠的受众自己跟自己竞价。受众结构要让每个 campaign 都有明确的包含和明确的排除。",{"children":156,"direction":19,"format":15,"indent":13,"type":25,"version":18,"tag":26},[157],{"detail":13,"format":13,"mode":14,"style":15,"text":158,"type":17,"version":18},"四个悄悄让受众失效的错误",{"children":160,"direction":19,"format":15,"indent":13,"type":20,"version":18,"textFormat":18,"textStyle":15},[161,163],{"detail":13,"format":18,"mode":14,"style":15,"text":162,"type":17,"version":18},"建在坏掉的追踪上。",{"detail":13,"format":13,"mode":14,"style":15,"text":164,"type":17,"version":18}," 网站受众的完整度不会超过喂给它的事件。度量有损，受众就会少收，而表现出来的样子是「效果差」而不是「数据有问题」。",{"children":166,"direction":19,"format":15,"indent":13,"type":20,"version":18,"textFormat":18,"textStyle":15},[167,169],{"detail":13,"format":18,"mode":14,"style":15,"text":168,"type":17,"version":18},"互相重叠的受众自己跟自己竞价。",{"detail":13,"format":13,"mode":14,"style":15,"text":170,"type":17,"version":18}," 两个包含大体相同人群的受众并行跑，抬高的是你自己的成本。放量之前先查重叠，不要放量之后再查。",{"children":172,"direction":19,"format":15,"indent":13,"type":20,"version":18,"textFormat":18,"textStyle":15},[173,175],{"detail":13,"format":18,"mode":14,"style":15,"text":174,"type":17,"version":18},"客户名单从不刷新。",{"detail":13,"format":13,"mode":14,"style":15,"text":176,"type":17,"version":18}," 半年前的快照定向的是你当时服务的人，不是你现在服务的人。",{"children":178,"direction":19,"format":15,"indent":13,"type":20,"version":18,"textFormat":18,"textStyle":15},[179,181],{"detail":13,"format":18,"mode":14,"style":15,"text":180,"type":17,"version":18},"把匹配后很小的受众当成很大的用。",{"detail":13,"format":13,"mode":14,"style":15,"text":182,"type":17,"version":18}," 上传行数和匹配规模是两个数字，只有第二个是真的。基于第一个做的决定等于基于空气。",{"children":184,"direction":19,"format":15,"indent":13,"type":25,"version":18,"tag":26},[185],{"detail":13,"format":13,"mode":14,"style":15,"text":186,"type":17,"version":18},"一套能跑起来的起步结构",{"children":188,"direction":19,"format":15,"indent":13,"type":221,"version":18,"listType":222,"start":18,"tag":223},[189,196,203,209,216],{"children":190,"direction":19,"format":15,"indent":13,"type":195,"version":18,"value":18},[191,193],{"detail":13,"format":18,"mode":14,"style":15,"text":192,"type":17,"version":18},"一个互动受众",{"detail":13,"format":13,"mode":14,"style":15,"text":194,"type":17,"version":18},"作为基线——不依赖站内追踪也能用。","listitem",{"children":197,"direction":19,"format":15,"indent":13,"type":195,"version":18,"value":202},[198,200],{"detail":13,"format":18,"mode":14,"style":15,"text":199,"type":17,"version":18},"两到三个不同漏斗深度的网站受众",{"detail":13,"format":13,"mode":14,"style":15,"text":201,"type":17,"version":18},"，各自配不同的回溯窗口。",2,{"children":204,"direction":19,"format":15,"indent":13,"type":195,"version":18,"value":44},[205,207],{"detail":13,"format":18,"mode":14,"style":15,"text":206,"type":17,"version":18},"一份已购买客户名单",{"detail":13,"format":13,"mode":14,"style":15,"text":208,"type":17,"version":18},"，按周期刷新，主要用作排除项和种子。",{"children":210,"direction":19,"format":15,"indent":13,"type":195,"version":18,"value":215},[211,213],{"detail":13,"format":18,"mode":14,"style":15,"text":212,"type":17,"version":18},"每个拉新 campaign 都设显式排除",{"detail":13,"format":13,"mode":14,"style":15,"text":214,"type":17,"version":18},"，别花钱去触达你已经拥有的人。",4,{"children":217,"direction":19,"format":15,"indent":13,"type":195,"version":18,"value":220},[218],{"detail":13,"format":18,"mode":14,"style":15,"text":219,"type":17,"version":18},"加任何新受众之前先做一次重叠检查。",5,"list","number","ol",{"children":225,"direction":19,"format":15,"indent":13,"type":20,"version":18,"textFormat":13,"textStyle":15},[226],{"detail":13,"format":13,"mode":14,"style":15,"text":227,"type":17,"version":18},"这个结构小到能想得清楚，同时覆盖了受众定向真正产生价值的那些场景。",{"children":229,"direction":19,"format":15,"indent":13,"type":25,"version":18,"tag":26},[230],{"detail":13,"format":13,"mode":14,"style":15,"text":231,"type":17,"version":18},"常见问题",{"children":233,"direction":19,"format":15,"indent":13,"type":20,"version":18,"textFormat":18,"textStyle":15},[234,236],{"detail":13,"format":18,"mode":14,"style":15,"text":235,"type":17,"version":18},"受众最小规模是多少？",{"detail":13,"format":13,"mode":14,"style":15,"text":237,"type":17,"version":18}," 存在一个可用门槛，而且是按匹配后的规模算的。规模很小的受众即使技术上可用，投放表现通常也很差。",{"children":239,"direction":19,"format":15,"indent":13,"type":20,"version":18,"textFormat":18,"textStyle":15},[240,242],{"detail":13,"format":18,"mode":14,"style":15,"text":241,"type":17,"version":18},"匹配率为什么低？",{"detail":13,"format":13,"mode":14,"style":15,"text":243,"type":17,"version":18}," 几乎总是格式问题：手机号没规范化、标识符列混装、缺国家。在对客户基数下任何结论之前，先检查准备工作。",{"children":245,"direction":19,"format":15,"indent":13,"type":20,"version":18,"textFormat":18,"textStyle":15},[246,248],{"detail":13,"format":18,"mode":14,"style":15,"text":247,"type":17,"version":18},"没有像素能用自定义受众吗？",{"detail":13,"format":13,"mode":14,"style":15,"text":249,"type":17,"version":18}," 互动受众和客户名单可以，网站受众不行。",{"children":251,"direction":19,"format":15,"indent":13,"type":20,"version":18,"textFormat":18,"textStyle":15},[252,254],{"detail":13,"format":18,"mode":14,"style":15,"text":253,"type":17,"version":18},"客户名单多久刷新一次？",{"detail":13,"format":13,"mode":14,"style":15,"text":255,"type":17,"version":18}," 按你的客户基数发生实质变化的周期。每月是个合理默认值；而常见做法是——从不。",{"children":257,"direction":19,"format":15,"indent":13,"type":20,"version":18,"textFormat":18,"textStyle":15},[258,260],{"detail":13,"format":18,"mode":14,"style":15,"text":259,"type":17,"version":18},"该用一个大受众还是几个小的？",{"detail":13,"format":13,"mode":14,"style":15,"text":261,"type":17,"version":18}," 几个小的，分处不同漏斗深度，并配排除。一个大受众既没有可比对象，也没有可编排的顺序。",{"children":263,"direction":19,"format":15,"indent":13,"type":25,"version":18,"tag":26},[264],{"detail":13,"format":13,"mode":14,"style":15,"text":265,"type":17,"version":18},"一句话版本",{"children":267,"direction":19,"format":15,"indent":13,"type":20,"version":18,"textFormat":13,"textStyle":15},[268],{"detail":13,"format":13,"mode":14,"style":15,"text":269,"type":17,"version":18},"按你手上真正跑得通的东西来选来源：互动受众什么都不需要、网站受众需要可靠追踪、客户名单需要多列且规范化的数据。名单看匹配规模而不是上传行数，按周期刷新，并且刻意设好排除，别让自己的受众互相竞价。",{"children":271,"direction":19,"format":15,"indent":13,"type":20,"version":18,"textFormat":13,"textStyle":15},[272],{"detail":13,"format":13,"mode":14,"style":15,"text":273,"type":17,"version":18},"DeepClick 与广告主一起处理受众定向所依赖的度量与分发层。","root",{"id":276,"alt":277,"updatedAt":278,"createdAt":278,"url":279,"thumbnailURL":19,"filename":280,"mimeType":281,"filesize":282,"width":19,"height":19},1040,"Abstract illustration of four groups of dots feeding into one overlapping cluster with a highlighted segment","2026-08-30T10:40:21.366Z","https://cms-r2.deepclick.com/cover723-aae424e7085c.jpg","cover723-aae424e7085c.jpg","application/octet-stream",145363,{"title":284,"description":285,"image":286},"Facebook 自定义受众指南：来源、名单准备与匹配率","互动/网站/客户名单/App 四种受众对比，规范化怎么决定匹配率，回溯窗口与排除项怎么设，以及放量前必做的重叠检查。",{"id":276,"alt":277,"updatedAt":278,"createdAt":278,"url":279,"thumbnailURL":19,"filename":280,"mimeType":281,"filesize":282,"width":19,"height":19},{"id":18,"key":288,"name":289,"prodHost":290,"testHost":291,"blogPath":292,"docPath":293,"zhPrefix":294,"deployHookTest":295,"deployHookProd":296,"brandAuthor":297,"brandKit":301,"enabled":305,"updatedAt":306,"createdAt":307},"deepclick","DeepClick","https://deepclick.com","https://www-test-deepclick.qiliangjia.one","/resources/blog/{slug}","/docs/{slug}","/zh-CN","https://api.cloudflare.com/client/v4/pages/webhooks/deploy_hooks/60a9adef-153b-4c89-8d07-7118e91e9522","https://api.cloudflare.com/client/v4/pages/webhooks/deploy_hooks/05323321-f694-4ce8-a5af-173c507b8bae",{"id":202,"name":289,"avatar":298,"updatedAt":299,"createdAt":300},25,"2026-04-22T08:09:35.299Z","2026-04-22T06:42:49.116Z",{"logoLight":19,"logoDark":19,"primaryColor":302,"accentColor":303,"tagline":304,"wechatName":19,"xhsHandle":19,"wechatQr":19},"#1a73e8","#ff6600","一次点击，多重价值",true,"2026-08-25T06:40:57.663Z","2026-07-14T06:26:38.962Z","published","facebook-custom-audience-guide",{"id":311,"name":312,"avatar":313,"updatedAt":320,"createdAt":320},6,"Grace Whitmore",{"id":314,"alt":315,"updatedAt":316,"createdAt":316,"url":317,"thumbnailURL":19,"filename":318,"mimeType":281,"filesize":319,"width":19,"height":19},924,"Grace Whitmore — app compliance editor","2026-07-27T07:17:21.278Z","https://cms-r2.deepclick.com/gpt_1785136442811_0-20e667237702.png","gpt_1785136442811_0-20e667237702.png",1976962,"2026-07-27T07:17:23.388Z",{"id":322,"site":323,"titleZh":325,"titleEn":326,"slug":327,"order":220,"updatedAt":328,"createdAt":329},7,{"id":18,"key":288,"name":289,"prodHost":290,"testHost":291,"blogPath":292,"docPath":293,"zhPrefix":294,"deployHookTest":295,"deployHookProd":296,"brandAuthor":202,"brandKit":324,"enabled":305,"updatedAt":306,"createdAt":307},{"logoLight":19,"logoDark":19,"primaryColor":302,"accentColor":303,"tagline":304,"wechatName":19,"xhsHandle":19,"wechatQr":19},"技术导航","Tech Guides","tech-guides","2026-04-27T08:37:10.576Z","2026-04-23T02:59:13.436Z","2026-08-30T10:40:54.739Z","2026-08-30T10:40:37.311Z","\u003Cdiv class=\"payload-richtext\">\u003Cp>自定义受众（Custom Audience）在投放后台里看起来很简单：选个来源、起个名、拿去用。但决定这个受众到底有没有用的事，发生在更早的地方——来源怎么准备、匹配出来的规模有多大、以及这个受众衡量的到底是不是你以为的那件事。这篇讲清四种来源类型、决定匹配率的名单准备规则，以及那些悄悄让受众变得没用的错误。\u003C/p>\u003Ch2>四种来源，各自真正的用途\u003C/h2>\u003Cp>\u003Cstrong>客户名单。\u003C/strong> 你上传自己已经持有的标识符。这是唯一从头到尾由你掌控的来源，也是唯一能覆盖「从没碰过你网站或 App 的人」的来源。它的天花板是你的数据质量。\u003C/p>\u003Cp>\u003Cstrong>网站受众。\u003C/strong> 由像素或转化 API 的事件构建——访客、看过特定页面的人、完成过特定动作的人。最常用的来源，也最受度量缺口影响：追踪漏事件，受众就会悄无声息地比应有的规模小。所以先把度量层做对是前提，见 \u003Ca href=\"https://deepclick.com/zh-CN/resources/blog/meta-conversions-api-setup/\">Meta 转化 API 接入指南\u003C/a>与 \u003Ca href=\"https://deepclick.com/zh-CN/resources/blog/facebook-pixel-setup-guide/\">Facebook Pixel 接入指南\u003C/a>。\u003C/p>\u003Cp>\u003Cstrong>App 活动受众。\u003C/strong> 移动端的对应物，由 SDK 上报的 App 事件构建。需要先注册应用、先定义事件，之后才可能有受众。\u003C/p>\u003Cp>\u003Cstrong>互动受众。\u003C/strong> 由用户在平台内与你的资产的互动构建——视频观看、表单打开与提交、主页或资料页互动。它\u003Cstrong>完全不依赖你自己的追踪\u003C/strong>，因此在你的站内度量还不完整时，它是最可靠的来源——也正因为这个原因，它是被浪费得最多的一种。\u003C/p>\u003Cp>一个实用的排序：互动受众立刻可用，网站受众需要追踪是好的，客户名单需要数据是干净的，App 受众需要先做 SDK。\u003C/p>\u003Ch2>怎么准备一份真的匹配得上的客户名单\u003C/h2>\u003Cp>上传的标识符会在离开你的机器之前先哈希。匹配率——也就是你的名单里能对应到真实账号的比例——几乎完全由准备工作决定。\u003C/p>\u003Cp>\u003Cstrong>用不止一列标识符。\u003C/strong> 同时有邮箱、手机号、姓名的名单，比只有邮箱的匹配得好，因为任何一列都可能在某个人身上失效。\u003C/p>\u003Cp>\u003Cstrong>上传前先规范化。\u003C/strong> 全部转小写、去首尾空格、手机号去掉标点并加国家区号、国家用标准两位代码。格式不一致不会报错，它只是匹配不上。\u003C/p>\u003Cp>\u003Cstrong>把混装字段拆开。\u003C/strong> 一种标识符一列。一个「联系方式」列里有时装邮箱有时装手机号，匹配会很差。\u003C/p>\u003Cp>\u003Cstrong>有国家就带上国家。\u003C/strong> 它能消解手机号和常见姓名的歧义，而且几乎不花成本。\u003C/p>\u003Cp>\u003Cstrong>别指望名单能被全量匹配上。\u003C/strong> 一份准备良好的名单，现实区间大致在「较大的少数」到「多数」之间。匹配率极低几乎总是格式问题而不是数据问题——在下「我的客户触达不到」这个结论之前，先检查规范化。\u003C/p>\u003Cp>受众可用之前有最小规模门槛，而算数的是\u003Cstrong>匹配后的规模\u003C/strong>，不是上传的行数。\u003C/p>\u003Ch2>网站受众：决定质量的几条规则\u003C/h2>\u003Cp>\u003Cstrong>回溯窗口。\u003C/strong> 受众往回看多少天。窗口短意味着意图更新、规模更小；窗口长意味着有量但信号更陈旧。按用途分别选，而不是一律拉满。\u003C/p>\u003Cp>\u003Cstrong>规则的具体程度。\u003C/strong> 「所有访客」和「到达过某个特定步骤的人」是两种完全不同的资产。后者更小，通常也更值钱，而且正是它让「一串有层次的受众」成为可能，而不是一个不加区分的大池子。\u003C/p>\u003Cp>\u003Cstrong>刷新方式。\u003C/strong> 网站受众随事件到达持续更新；客户名单是一张快照，不重新上传就会过期。团队常常记得前半句、忘了后半句，于是一份名单在早已不代表任何在册客户之后还被用了好几个月。\u003C/p>\u003Cp>\u003Cstrong>排除和包含一样重要。\u003C/strong> 受众定向里大部分被浪费的花费，来自没有排除已转化的人，或者来自互相重叠的受众自己跟自己竞价。受众结构要让每个 campaign 都有明确的包含和明确的排除。\u003C/p>\u003Ch2>四个悄悄让受众失效的错误\u003C/h2>\u003Cp>\u003Cstrong>建在坏掉的追踪上。\u003C/strong> 网站受众的完整度不会超过喂给它的事件。度量有损，受众就会少收，而表现出来的样子是「效果差」而不是「数据有问题」。\u003C/p>\u003Cp>\u003Cstrong>互相重叠的受众自己跟自己竞价。\u003C/strong> 两个包含大体相同人群的受众并行跑，抬高的是你自己的成本。放量之前先查重叠，不要放量之后再查。\u003C/p>\u003Cp>\u003Cstrong>客户名单从不刷新。\u003C/strong> 半年前的快照定向的是你当时服务的人，不是你现在服务的人。\u003C/p>\u003Cp>\u003Cstrong>把匹配后很小的受众当成很大的用。\u003C/strong> 上传行数和匹配规模是两个数字，只有第二个是真的。基于第一个做的决定等于基于空气。\u003C/p>\u003Ch2>一套能跑起来的起步结构\u003C/h2>\u003Col class=\"list-number\">\u003Cli\n          class=\"\"\n          style=\"\"\n          value=\"1\"\n        >\u003Cstrong>一个互动受众\u003C/strong>作为基线——不依赖站内追踪也能用。\u003C/li>\u003Cli\n          class=\"\"\n          style=\"\"\n          value=\"2\"\n        >\u003Cstrong>两到三个不同漏斗深度的网站受众\u003C/strong>，各自配不同的回溯窗口。\u003C/li>\u003Cli\n          class=\"\"\n          style=\"\"\n          value=\"3\"\n        >\u003Cstrong>一份已购买客户名单\u003C/strong>，按周期刷新，主要用作排除项和种子。\u003C/li>\u003Cli\n          class=\"\"\n          style=\"\"\n          value=\"4\"\n        >\u003Cstrong>每个拉新 campaign 都设显式排除\u003C/strong>，别花钱去触达你已经拥有的人。\u003C/li>\u003Cli\n          class=\"\"\n          style=\"\"\n          value=\"5\"\n        >\u003Cstrong>加任何新受众之前先做一次重叠检查。\u003C/strong>\u003C/li>\u003C/ol>\u003Cp>这个结构小到能想得清楚，同时覆盖了受众定向真正产生价值的那些场景。\u003C/p>\u003Ch2>常见问题\u003C/h2>\u003Cp>\u003Cstrong>受众最小规模是多少？\u003C/strong> 存在一个可用门槛，而且是按匹配后的规模算的。规模很小的受众即使技术上可用，投放表现通常也很差。\u003C/p>\u003Cp>\u003Cstrong>匹配率为什么低？\u003C/strong> 几乎总是格式问题：手机号没规范化、标识符列混装、缺国家。在对客户基数下任何结论之前，先检查准备工作。\u003C/p>\u003Cp>\u003Cstrong>没有像素能用自定义受众吗？\u003C/strong> 互动受众和客户名单可以，网站受众不行。\u003C/p>\u003Cp>\u003Cstrong>客户名单多久刷新一次？\u003C/strong> 按你的客户基数发生实质变化的周期。每月是个合理默认值；而常见做法是——从不。\u003C/p>\u003Cp>\u003Cstrong>该用一个大受众还是几个小的？\u003C/strong> 几个小的，分处不同漏斗深度，并配排除。一个大受众既没有可比对象，也没有可编排的顺序。\u003C/p>\u003Ch2>一句话版本\u003C/h2>\u003Cp>按你手上真正跑得通的东西来选来源：互动受众什么都不需要、网站受众需要可靠追踪、客户名单需要多列且规范化的数据。名单看匹配规模而不是上传行数，按周期刷新，并且刻意设好排除，别让自己的受众互相竞价。\u003C/p>\u003Cp>DeepClick 与广告主一起处理受众定向所依赖的度量与分发层。\u003C/p>\u003C/div>","https://deepclick.com/zh-CN/resources/blog/facebook-custom-audience-guide",{"en":309,"zh-CN":309},1788086618173]