于是继续追爆款，涨到十万粉，课卖了二十份——转化率还降了。被「99% 的人不知道」钩来的人心里没有待解决的问题，只是被反差挠了一下。这类人再多也不构成买家。
钩子内容照发，但每条末尾多一句：「如果你每个月都要手动合并十几张部门表，我这里有个免费模板。」——这句不涨粉，它把泛注意力筛成「此刻正被这件事折磨的人」。两万人里也许只筛出三百个，但那三百个才是第二段的入口。
EN Attention, trust and revenue are three separate funnels running on three different fuels, and most people treat them as one. What spreads is surprise and emotional arousal; what earns trust is checkable specifics accumulated over time; what closes is someone happening to have that problem right now. Those requirements actively conflict, which is why "it went viral and nobody bought" is the default outcome rather than bad luck — the people a shock headline pulls in were never carrying a problem, they were just tickled. Getting more of them changes nothing. So stop reading one number. Count the three stages separately, find the narrowest one, and fix that. And notice what trust actually requires: not just looking capable but looking like you're on their side — people read competence and warmth as two distinct dimensions, and warmth is judged first and weighted heavier. A stream of impressive tips builds only half of it. The half you're missing is usually the reason a big audience converts like a small one.
三段各有各的门槛。传播看唤醒水平：敬畏、愤怒、焦虑推动分享，平静的悲伤反而不会。印象沿两条独立的轴走：他行不行（能力）、他站不站在我这边（善意），善意判断得更早、权重更大——只输出干货只长了一条腿。成交还要加时机：再高的信任也换不来一个此刻没这个问题的人的钱。
边界：分段不等于每段都要做。客单价高、每月只需成交两三个的人，第一段几乎可以不做；只有低客单价标准品才真需要量。另外善意信任装不出来，它的可核对形式是你替读者省过钱、劝过他别买、承认过自己不会——短期都在减少收入。
"Big audience, small business — those are two different problems." —— 先分清你缺哪一段。
"This one's not for everyone; here's who should skip it." —— 用劝退换善意信任。
"If that's not your situation, don't buy it." —— 说得出这句，你的推荐才有分量。
打到 5,000，成交变成三单，还把付过全价的老客户得罪了。降价没碰到真正的障碍——对方怕的不是八千贵，是「花八千给一个只在网上看过的人」。你降的是价格，他担心的是风险。
加一个 199 元的「品牌一致性体检」：他把物料发来，你三天内出两页问题清单，指出最伤转化的三处，附一句「这三处你自己也能改，改完还想做系统方案再来找我」。——两百块买到的不是折扣，是一次真实交付体验：你的判断力、响应速度、跟你合作是什么感觉。做完这步的人再听到八千，评估对象已从「陌生人」变成「合作过的人」。
一、他要冒什么风险？钱、时间、还是「买了显得蠢」的面子。答不上来说明你在按价格分级，不是按风险分级。
二、风险跳跃有多大？0 到 8,000 是悬崖；0 → 199 → 8,000 是两级台阶。
三、这一级能独立交付完吗？取样品必须自己就有用。做成「只为让你买下一级」的半成品，读者闻得出来。
四、能不能把风险接过来一部分？「不满意退款，清单你留着」——退款率通常远低于它带来的成交增量。它比降价有效，因为它直接对着损失厌恶下手而不是绕开（即「做一个无法拒绝的报价」那期的风险逆转）。
EN An account with one expensive offer converts badly not because the price is high but because there's no step between "scrolled past you" and "wire five thousand". What a ladder actually sells isn't a cheaper tier, it's a cheaper way to check you out — so the bottom rungs shouldn't be measured in revenue at all. Their KPI is how many people received one real delivery from you. The mechanism is loss aversion: the dread of "what if it's useless" weighs roughly twice as much as the equivalent hope that it works, so the barrier is risk, not price. That's also why discounting is the wrong fix — it lowers the number they're worried about losing without touching the reason they're worried. Two rules keep a ladder honest. Every rung has to be genuinely useful on its own; a sample built purely to sell the next rung reads as bait, and that smell lingers. And take some risk onto yourself — a refund with the deliverable kept usually costs far less than the sales it unlocks, because it attacks the loss term directly instead of dodging it.
障碍是风险，不是价格。人对损失的敏感度显著高于同等收益，「花了钱结果没用」压得过「可能挺有用」。既然障碍在损失那一侧，能做的只有两件：把单次可能损失的金额切小（台阶），或把损失搬一部分到自己身上（退款、保证、先做后收）。降价两件都没做，它只是把赌注调小，赌局结构没变。另有一层：小额买过一次之后「我是你的客户」已经成立，后面的大额决定不再从零起步。
三个失效场景。一、取样品太薄：拿到手发现只是广告，损失的信任比多卖几单值钱得多。二、台阶太多：每级都消耗一次决策成本，两三级够了。三、错配：高价 B2B 里对方要的是同行推荐和案例，不是 199 元试用——那里的台阶该是一次免费的深度诊断会，形式不同，机制一样。
"Try the small one first — it's useful even if you never buy the big one." —— 让取样品自足。
"If it doesn't help, I'll refund it and you keep the file." —— 把风险接过来。
"It's not that it's expensive. It's that you don't know me yet." —— 说破真实障碍。
EN As long as you're the only source of trust, your ceiling is your own time: stop posting and the value goes to zero. The turning point is the day members start answering each other — that's the first asset you own that keeps compounding without you. Two mechanisms carry it. Belonging is a basic human motive, not a nice-to-have, so people stay for "others like me" long after the content stops being novel; and a peer's account of a result is weighed differently from yours, because you have an obvious reason to say it and they don't. Be honest about the cost, though. A quiet group is worse than no group: every day it silently tells each member that nobody's here. Small and specific beats big and vague — thirty people doing the same thing outperform three thousand who came for a free file. And resist answering everything. If you reply fast and well every time, the star shape locks in permanently; the first peer-to-peer answer is worth more than a hundred brilliant ones from you.
两条。一、归属本身就是动机：人留在群里很大一部分不为信息，而为「有人和我处境一样」本身——这是基本需要不是附加值，所以社群的留存曲线和内容账号完全不同。二、谁说的，决定这句话值多少：同一句「这东西真有用」，从你嘴里说带着明显的利益动机，从一个和你没关系的成员嘴里说则没有。社群就是把后一种声音变得常见、可追问。
边界：社群是所有变现方式里运营成本最高的，它要每天在场，不是每周更新；三个月热闹、半年冷掉的群，伤害大于从没做过。另外同伴证言一旦被你安排（找托、给好处换好评），就从社会证明变成造假——一次被发现整个群的可信度归零，先毁掉的正是最忠诚的那批人。
"I'll let someone who's actually done it answer that one." —— 主动把回答权交出去。
"Small on purpose — everyone here is working on the same thing." —— 把「小」讲成筛选。
"Who here has solved this already?" —— 一句话把星型转成网状。
EN Good content is a deposit and every pitch is a withdrawal, but the exchange rate is badly asymmetric: one moment where people feel farmed outweighs dozens of good posts. That isn't readers being stingy, it's how attention to bad news works — negative events carry systematically more psychological weight than equivalent positive ones, so the crash in the curve is steeper than any climb. There's a second, quieter mechanism: once an obvious external incentive is visible, people discount the internal explanation for your behaviour, so everything you said before gets silently reassessed. Four rules follow. Let people know from day one that you sell things, so selling is never a reversal. Only recommend what you've actually used or delivered. Label ads as ads at the top — disclosure does cost you conversions on that post, and being caught costs you a category. And always ship the version that's useful to someone who buys nothing. The most dangerous overdraft isn't the audience's trust, though. It's that revenue metrics quietly retrain you to write what sells rather than what's true, and that one is invisible from the inside.
三层。一、负面偏差：同等强度的坏事分量系统性大于好事——注意得更快、记得更牢、影响更久，所以信任曲线天然「爬得慢、掉得快」。二、动机折扣：存在明显外部诱因（你在卖东西）时，人会把「他是真心这么想」打折，而且是回溯性的——你过去说的话会被重新评估一遍。三、最容易被忽略的是它反过来驯化你：收入一旦成为反馈信号，你会不知不觉多写好卖的、少写真实但难卖的，而每一步都显得合理，所以从内部看不见。
最重要的一条边界：本期所有东西提高的都是兑现已有信任的效率，它们不创造信任。底下没真本事时，漏斗越顺翻车越快——你只是让更多人更快发现这一点。说服与操纵的界线也在这里：让对方在充分知情下更容易做出对他有利的决定是说服；靠他不知道的信息差促成交易是操纵（「影响力六原理」那期讲的是同一条线）。自检：如果买家事后知道你现在知道的全部，还会觉得划算吗？答不出「会」就别卖。
"Heads up: this one's a paid recommendation." —— 放在开头，不放在角落。
"You can absolutely do this yourself — the paid version just saves you a weekend." —— 给不买的人一条路。
"I turned that sponsorship down; I haven't used the product." —— 说出来的拒绝，本身就是存款。
九十分钟，四件事。做完你会知道卡在哪一段，手上还有一个能立刻上线的取样品。