「负责社群日常运营与活动策划，参与用户增长项目，协助完成月度报表。」读完不知道发生了什么改变，也无从查证。它描述了你每天坐在哪，没描述你走后留下了什么。
「把 12 个快死的群做成 3 个活群：退群率从每月 9% 降到 3%。做法是把活动从『每周一场』改成按用户进群天数触发（方案见附件）。同一套打法在新品线复用了一次，结果类似。」从什么到什么、怎么做到、去哪核对——一次给全。
EN Treat the CV as a landing page, not a biography. A landing page has exactly one conversion goal, and yours is not "get hired" — it is "get a conversation". Screening is an elimination task, not a selection task: someone facing two hundred applications can only survive by shrinking the pile, so they read looking for a reason to discard you. Two consequences follow. First, if the top three lines give no reason to keep you, nothing further down was ever read. Second, effort of reading is itself a reason to discard — fluently processed material feels better, and that faint good feeling gets misattributed to the thing being judged. So: one falsifiable positioning line, three results written as from-what-to-what with the method attached, one thing that can be opened and checked, one clear next step. Verbs like "responsible for", "involved in", "assisted with" describe where you sat, not what changed after you left.
机制：筛简历是淘汰任务，不是挑选任务。一个人面对两百份材料，唯一可行的策略是尽快把堆变小，所以他在找「扔掉它的理由」，而不是找亮点。由此推出两件事：前三行没给出留下你的理由，后面写了什么都没被读到；而读起来费劲本身就是一个扔掉的理由——大脑会把「加工顺畅」错当成「这东西不错」，反过来，术语堆砌和排版拥挤会被记成「这人自己也没想清楚」。
边界：落地页逻辑适用于有活人真的会读的场合。大厂 ATS 关键词初筛是另一台机器（先过词、再过人），所以关键词照抄岗位描述里的原词，别自创同义词；而小公司、内推、直接寄给用人经理，落地页写法回报最高。另外别把「结果化」做成吹牛：每个数字都要经得起一句追问——这怎么算的。
"Here's the before and after — the method's in the link." —— 用对照代替形容词。
"Happy to walk you through how that number was calculated." —— 主动邀请核对。
"I'd rather show you than describe it." —— 把说服推到下一步。
「在我讲我自己之前，想先确认一下：这个岗位设出来，是因为现在哪件事没人管、或者管得不够好？」
「如果一年后你觉得这个人招对了，那时候什么变了？」
拿到答案之后，你后面每一个回答都从他刚说的那句话里取词——同一段经历，讲法完全不同。
「我 2019 年毕业于……然后在 A 公司做了两年，主要负责……后来去了 B 公司……」他手里正拿着这份简历。你花五分钟重复他已经知道的信息，还把最宝贵的开场用在了让他走神上。
「我挑跟你刚说的那件事最相关的一段讲。你说现在最头疼的是新用户第二周就流失——我上一家正好啃过这个：三个月里把第二周留存从 18% 提到 27%，关键不是做活动，是发现流失集中在没完成首次配置的人身上。这个判断怎么来的我可以现在拆给你看，你想听哪一步？」
EN An hour of interview spent as question-and-answer is an hour wasted on both sides. Open with two diagnostic questions instead — "what isn't getting handled, that this role exists to fix?" and "a year from now, what has to be different for you to say this hire worked?" — then spend the middle of the hour taking apart the problem they just named, out loud, using their own words. Two mechanisms make this work. Talking about oneself is intrinsically rewarding: self-disclosure engages the mesolimbic dopamine system, and people will give up real money for the chance to do it, so letting them talk makes the conversation feel good to them in a way they'll later file under "we clicked". And interviewers are far more confident about reading people than unstructured interviews deserve — which means you cannot rely on being discovered through chat. You have to put something evaluable into the room yourself, and a genuine piece of work beats ten polished answers. One caution: turning the questions round is diagnosis, not a coup. Two at the start, one at the end. And demo only the step you can actually do.
机制两层。社交层：人在谈论自己时，中脑边缘奖赏系统真的在放电——所以让对方多说，他对这场对话的体感更好，而这份好感会被他记成「聊得很投缘」；你抢着表现，只有你自己爽。预测层：面试官对自己「看人准」的信心，远高于闲聊式面试的真实预测力——这意味着你不能指望他从聊天里发现你，你得主动把可评估的东西放进这一小时。
边界：反过来问是诊断，不是夺权——开场一两问、结尾一问足够，全程反问会让人觉得自己在被面试。有些环节（大规模校招、HR 初筛）根本不给诊断空间，那一轮就老实过关键词，把诊断留给用人经理那一轮。还有：现场拆解要挑你真的会的那一步；硬上一个不熟的环节，比不 demo 更糟。
"Before I talk about myself — what isn't getting handled right now?" —— 诊断先于开方。
"What would have to be true in a year for you to say this hire worked?" —— 把成功标准问出来。
"I'd rather work through your actual problem than talk in general." —— 把答题换成干活。
「这个岗位我是想做的，把最后这块谈定我就签。」——先给确定性再谈条件：你不是在犹豫，你是在成交。
「基于我在这块的经验和同类岗位的区间，我的期待在 A 到 B。」——给区间加依据，不要一个孤零零的数字。
「如果基数这次动不了，我们能不能看签字费、职级、或者半年后的评估点？」——价格谈不动就谈结构，把单一维度换成多维度。
EN Through every interview round you are the one being assessed; the leverage sits with them. The instant they extend an offer, that reverses — they have now sunk several rounds of time, an internal approval, and quite possibly a rejection sent to someone else. Re-running the search is expensive for them, which is precisely where your leverage comes from, and you no longer have to prove yourself to use it. The window is usually days. Three mechanisms govern what you do inside it. Your BATNA — the alternative you actually have — sets your floor; you never state it, but it determines where you can stop. The first number spoken becomes an anchor, and subsequent adjustment from it is typically insufficient. And this is a repeated game: you'll be working with this person on Monday, so winning too visibly gets collected later, which is why the framing is "how do we close this" rather than "what I want". If the base salary genuinely can't move, move to structure — level, reporting line, a review date, a starting budget. Two hard rules: don't negotiate before the offer exists, and never invent a competing offer.
机制三条。一、BATNA（最佳替代方案，即谈崩了你手上还剩什么）决定你的下限：它不需要说出口，但决定你敢停在哪；真没有替代选项时，最诚实的做法是承认自己在弱势，把筹码押在非现金条件上。二、先出的数字会成为锚，对方随后的调整往往不充分。三、这是重复博弈——你下周就要和他共事，赢得太难看会在入职后收账，这正是把「我要多少」换成「我们怎么成」的原因。
边界：没拿到 offer 前不要谈价，那只会白送对方一个淘汰你的理由。别编造不存在的竞争 offer——圈子小、重复博弈，被戳穿是长期损失。还有一条更实际的：能谈的前提是你真的准备好走；如果你其实无论如何都会接受，就把力气花在非现金条件上（职级、汇报线、启动预算、评估节点），这些常常比几千块更决定你两年后的位置。
"I want this job — let's close out this last piece and I'll sign." —— 先给确定性。
"Based on what I'm seeing for comparable roles, I'm looking at A to B." —— 区间加依据。
"If base can't move, can we look at level or a review point at six months?" —— 换维度。
EN The same person moving through two different channels gets two different outcomes, and the difference isn't effort — it's who is carrying the risk of being wrong. In the applicant pile, the question in the reader's head is "why shouldn't I reject this one"; when someone vouches for you, the question becomes "when can we talk". A referral works because the referrer is staking their own standing inside the company: it costs a careless recommender more than it costs an honest one, which is exactly what makes it a credible signal, and it also spares the employer the most expensive step — figuring out who is worth an hour. So build the channel deliberately: do a small piece of real work for three teams you actually want, unpaid and unasked; route it through someone you know slightly rather than your closest friends, since close contacts mostly know what you already know; make the ask small enough to refuse cheaply. And honestly: referrals amplify existing network inequality, which is why manufacturing weak ties by doing work first matters most for people who start without a network.
机制：推荐是一个昂贵信号——推荐人押上的是自己在公司里的信誉，押错要还；正因为对乱推荐的人成本更高，雇主才把它当真。同时它替雇主省掉了最贵的那一段：从两百份材料里判断谁值得聊。你「先干一件小事再开口」也是同一个道理——愿意先付出成本的人，和只会说自己很有热情的人，从此不再被归为一类。
边界：先干活不等于免费打工。一次性的小东西是信号，持续无偿产出是被占便宜——对方收下东西却只想让你继续做，就停。另外别把「被需要」演成饥饿营销：编造不存在的竞争、假装很忙，一旦被识破，损失的是你唯一真正值钱的东西。至于怎么打磨那句「别人转述你时会说的话」，「卖自己是元技能」那期已经拆过。
"I put something together for your team — no strings, just thought you'd find it useful." —— 先付出。
"Would you be willing to pass this to whoever owns growth? If it's not a fit, just say so." —— 给退出口。
"Here's a line you can copy-paste if you do mention me." —— 替推荐人省事。
六十分钟，四件事。做完你手上会多一份能用的材料、少一次靠临场发挥。