主题：「专业播客后期制作服务合作」。这句话对每一个播客主都成立，所以对他不成立——他分类时看到的是「一批」，不是「我」。
主题：「你上期说的那 4 小时」。他知道那 4 小时是什么，别人不知道；只有他能认领这句，于是它自动落进「写给我的」那一堆。
1. 放一个具体名词：他的项目名、他说过的数字、他所在的城市。抽象词（升级 / 赋能 / 合作机会）一律删掉。
2. 写成同事发消息的样子：不加书名号、不加感叹号、不首字母全大写，六到十二个字最好。
3. 不勾好奇心、不伪造上下文：「Re:」这类假前缀换来一次打开和一次永久的不信任——这是说服与操纵的界线，跨过去就不是话术问题了。
4. 别拿打开率当成绩：算数的只有回复。打开率高、回复率零，说明主题行和正文不是一封信。
EN Nobody reads their inbox; they sort it, in about half a second, into "written to me" and "sent to a batch." So a subject line isn't competing on interest or intrigue — its whole job is to land in the first pile, and the reliable way to do that is to put one concrete fact inside it that is true of this person and nobody else: something they said last week, the internal name of their project, the street they work on. Write it the way a colleague writes one — six to twelve words, no exclamation marks, no title case, no abstractions like "synergy" or "growth opportunity." Curiosity-gap lines ("quick question") buy you an open and no reply, and they pre-classify your next message as well. Fake context — a "Re:" on a thread that never existed — buys one open and permanent distrust; that one is the persuasion-manipulation line, not a wording choice. And never score yourself on opens: the only number that means anything is replies.
「与我有关」不是偏好，是注意与记忆的一条优先通道：跟自我挂上钩的信息编码更深、更容易从背景噪声里被捞出来。主题行放一个只对他成立的事实，等于接上这条通道；放一句对谁都成立的话，它就只能在噪声里排队。
边界：这些实验测的是记忆与注意捕获，不是回复率——给方向，不给效应量。反向风险也在这：勾好奇心和伪造上下文同样能拿到注意，代价是信任；注意力是手段，被信才是目的。另外实验里也只有约三分之一的人捞到自己的名字，所以你必然会被大量写对了的信忽略。
{{名字}}、{{公司}} 的伪造成本是零秒、所有群发工具都自带，所以它现在是负信号；复述一条公开新闻的成本是五秒；只有认真看过才说得出的观察成本足够高。而且它必须连着一句因果——不是夸他，是「因为这件事，所以我来找你」。1. 花十分钟找一个具体物证：他说过的一句原话、一个数字、他产品里的一处细节。要带位置或时间（哪一期、哪一页、上周三），可核对才贵。
2. 接一句因果，别接一句夸奖：「你上期说剪了 4 小时——我来找你正是因为这个」，而不是「你的节目很棒」。夸奖不解释你为什么出现，因果句解释了。
3. 用具体的一个，别用统计：「你那位剪辑同事每周要花掉一个半天」比「行业平均节省 40% 后期时间」有力——一个能被想象，另一个只能被同意。
4. 成本不够就换人：找不到值十分钟的物证，说明这个人不该在这轮名单里（名单怎么减，见「找客户 prospecting」那期）。
EN Personalization doesn't work by flattering anyone; it works by proving the message was written for one person. And a signal is credible in proportion to what faking it would cost. Merge fields — the name, the company — cost zero seconds and ship inside every mass-mail tool, so today they are a negative signal: they prove you used the tool. Restating a public news item costs five seconds and is worth about as much. Only an observation you could not have made without actually looking costs enough to count, and it has to be attached to a causal clause rather than a compliment: not "your show is great" but "you said at minute 32 that the edit took four hours — that's exactly why I'm writing." A compliment doesn't explain why you appeared; a because-clause does. Cite one imaginable person or detail rather than an industry average, and give your fact a location — which episode, which page, which Wednesday — since checkability is what makes it expensive. If you can't find a fact worth ten minutes, that name doesn't belong on this list.
陌生人核实不了你的诚意，只能核实你付出的成本——这是信号理论的核心：只有伪造足够贵，信号才携带信息。另一半在「因为」：给出理由会提高顺从率，但理由的内容要不要紧，取决于请求有多大。
边界：Langer 那个实验测的是当面小请求，跨到邮件要打折，也别读反——不是「加个因为就有效」，是「请求越大，理由越必须真」。至于「个性化能提升多少回复率」，流传的数字几乎都来自外联工具厂商对自家客户的统计，样本自选、无对照组，所以这里只给机制和方向，幅度你自己测。
「附件是我们的服务介绍和报价表，方便约个 30 分钟电话详聊？这是我的日程链接。」——附件、链接、日历、三十分钟，四道门槛叠在一起，每一道都能单独让他放下这封信。
「我把你上期那段口型没对上的地方标出来了，就三处，时间码在下面，你直接拿去用，不用回我。
只顺便问一句：这些现在是你自己剪还是外包？」——先给一件他能马上用的东西并明说不用回，再问一个两个字能答的问题。
1. 第一行 · 为什么是你：那个只对他成立的具体观察 ＋ 一句因果。
2. 第二行 · 先给：一件你已经替他做完的小事（标出来的三处问题、一个查到的数、一句同行的原话）。明说「不用回我」——一给就要，那不是给，是账单。
3. 第三行 · 十秒问题：一个不用思考、只需确认的问题。「现在是自己做还是外包？」比「你们有这方面需求吗？」好，因为后者要他先替你判断。
4. 第四行 · 退路：「不合适就直说，我不再打扰。」他知道回一个「不」就能干净地结束，回复的心理成本就降了。
5. 别放：附件、日历链接、三个时间选项、图片签名。全文控制在手机一屏内。
EN "Do you have thirty minutes?" is an expensive request from a stranger: it means opening a calendar, spending future time, and accepting the risk that the call turns out to be a pitch. So the first message shouldn't ask for a meeting — it should ask for something answerable in ten seconds: a yes or no, a name, one line. Someone who has replied once is far likelier to grant twenty minutes next, because they now file themselves as a person who talks to you. "Leading with value" likewise doesn't mean attaching a brochure; it means putting a small piece of finished work directly in the body — three timestamps you marked, a number you looked up — and explicitly saying no reply is needed, because a gift with an invoice stapled to it isn't a gift. Structure it in four lines: why you, the small gift, the ten-second question, and an exit ("if it's not a fit, say so and I'll stop"). Ask what they do today rather than whether they have a need; the second one makes them do your qualifying for you. No attachments, no scheduling links, no three time options, one phone screen long.
答应过一个小请求的人，会把自己更新成「会答应这类事的人」，后面的大请求就顺理成章——这叫登门槛效应（foot-in-the-door），机制是自我知觉：人从自己的行为反推自己是谁。第二条机制是启动成本：含糊的请求要对方自己去想「什么时候、怎么做」，缺这一步正是大量本来愿意的人最终没动的原因。
边界：请求跨度太大时登门槛会断——回一个词不等于半年后签合同，中间还得有真实价值。反向风险也在这：把小请求当套人的钩子（先骗一个「是」再顺杆上），一旦被察觉你失去的比这一单多得多。区别不在动作，在于你给的那一小块是不是真有用、且真的不要求交换。
能测（差异大，样本够得着）：换人群、换触发条件、换渠道、序列长度（3 封 vs 5 封）、送达本身（带链接的信是不是进了垃圾箱）。
基本测不出来（效应太小）：主题行改几个字、称呼、结尾敬语、发送时间是周二还是周四。在你那点样本量下，这些差异全被噪声盖住（为什么，见「找客户 prospecting」那期的方差账）。
操作规矩：一批不少于 30 个人、跑完再改；每批只改一个变量，且只改上面第一行里的；只记绝对数——这批产生了几次真实对话。
先测送达再测一切：把这封信原样发到两三个朋友的邮箱，看进没进垃圾箱。十分钟排除一整类原因，也是「零回复」时唯一值得先查的东西。
EN Most replies arrive after the first message, but what makes follow-ups work isn't that persistence is touching — it's sampling. How urgent this is for someone fluctuates over time: a budget lands, an owner changes, the system breaks again. Each additional touch is another draw at the week when they happen to be thinking about it. Which means every message must enter from a new angle — a different pain, a different kind of evidence, a different person, a different channel — because "just following up, did you see this?" carries zero new information and burns a slot to remind them to delete you. Space the intervals out rather than bunching them, write each message so it could be read as a first message, and fix the ending before you start: a final note that says you'll stop, then archive. On testing: audience, trigger, channel, sequence length and deliverability are detectable at your sample size; subject-line wording, salutations and send-day are not. And test deliverability first — mail yourself through two friends' inboxes and see whether you land in spam. Ten minutes rules out a whole class of cause.
序列有效有两条独立机制。一是时机：紧急程度随时间波动，多次触达等于多次抽样，命中概率随次数上升，跟你写得多好无关。二是间隔：分散的重复比密集的留下更牢的痕迹。还有必须点破的第三条——重复本身会让一句话「感觉更真」，这恰恰是不该靠它去推一个站不住的主张的原因。
边界：节奏受渠道和行业约束，邮件比私信容忍度高。还有一条硬线：对方说了不需要，序列立刻结束、名单里划掉，不换个借口重启。这不只是礼貌——这是你在这个圈子里还能不能被推荐的问题，而那个池子比这一单值钱得多。
45 分钟，把「一封写得很用力的信」换成「一封能通过分类题的信 ＋ 一条写好终点的序列」。