「这个我们正好有——AI 筛简历，周期能压掉一半，我发份方案您看看？」——从这秒起你不再是和他一起看问题的人，而是一张报价单。他接下来只会做一件事：再问两家。
「从发 JD 到发 offer，哪步耗时最长？」——「面试排期。」——「排不动是面试官不接，还是协调来回？」——「面试官不接。」——「那筛简历再快一倍，这卡点会松吗？」
——最后一句是全场关键：你亲手否掉了自己本来想卖的东西。对方这时会替你把话说完：「所以我该解决的是面试官时间。」这句是他说的。
EN The instinct to name a solution the moment you hear a symptom is not a sales flaw, it's a general professional reflex, and medicine has measured it three separate times. In 1984 physicians were recorded interrupting patients a mean of eighteen seconds into their opening statement, with only 23% getting to finish; a replication fifteen years later found 28% finishing and a mean of 23.1 seconds; a 2019 study found a median of eleven seconds. Selling has no recording study of comparable quality, but the same impulse shows up wearing a different costume: you start describing the product. The cost is specific. The moment a solution is on the table, the buyer's evaluation frame switches from "what is actually wrong here" to "is this price right for these features", and you have handed them a quote to shop around. The alternative isn't slower for its own sake — you keep segmenting the complaint until you reach the actual bottleneck, and the strongest move available is to rule out your own product out loud when it doesn't touch that bottleneck. What you're buying with that is the one thing a vendor can't manufacture: a conclusion the buyer said himself.
报方案是一次身份声明。方案出口那一刻，你从「和他一起看问题的人」变成「要卖他东西的人」，他的评估框架跟着切换：不再问「我有什么问题」，改问「这家报价对不对得起功能」。而诊断真正的价值不在收集信息，在于结论从他嘴里出来——自己生成的结论和听来的信息不走同一条路。
边界：诊断不是拖延，更不是查户口。对方已清楚要买什么、只差比价时（复购、标品），继续问只会让人烦。判据：他答完这个问题，如果他自己没得到新东西，这问题就不该问。
"Before I suggest anything — walk me through how it works today." —— 把「先诊断」说在明处。
"What happens if you do nothing for another quarter?" —— 问代价，比问需求有用。
"Honestly, faster screening wouldn't fix that." —— 否掉自己的方案，是顾问身份最便宜的证明。
「没问题，全套最划算，长期看反而省钱。」——你说的可能是真的，但他没法验证，只能按「卖方当然这么说」打折。你把一次建立可信度的机会换成了当期收入。
「按你刚说的用量，全套里的多地协同和审批流今年根本用不上，这部分现在买是浪费。建议先上中间那档，把面试官排期这段跑通。等开到第三个城市再升，升级不重收实施费。」
——你少签了钱，但他对你后面每句话的折扣率都降了，而且他记住的是：这人会拦我。
EN Every sentence a vendor says arrives pre-discounted, because the listener knows you have a stake in it. "This package is perfect for you" carries almost no information — you would say that regardless — so the listener attributes it to your position rather than to the facts. The one thing that lowers that discount rate is saying something true that costs you: it removes the "he says that because it pays him" explanation and leaves only "he actually believes it". This is the classic finding on communicators whose position disconfirms what their interests predict: they are judged as less biased and shift attitudes more. The practical form is a three-tier prescription where the first tier is the part you're telling them not to buy, the second is the smallest thing that fixes today's actual bottleneck, and the third is the condition under which upgrading would make sense. Two caveats worth holding. It only works when it's true — strategic self-deprecation about something irrelevant gets read eventually, and backfires harder than a straight pitch. And if the thing you cut is something they genuinely need, that isn't honesty, it's malpractice.
每个卖方的话都自带折扣率，来源是对方对你动机的推断：他知道你有立场，于是把你说的一切按「你当然会这么说」打折。唯一能把折扣率降下来的，是一句违背自己利益的真话——它排除了「他这么说因为对他有好处」。三档处方真正在卖的不是那三档，是你有资格说「别买」。
边界：这招只在你说真话时成立。把「不推荐」做成套路（先砍一个反正没人买的模块显得诚实），对方第二次就看穿——一旦开始表演诚实，你就得记住自己演过什么。还有个反向失效：砍掉的东西他其实真需要，那不是诚实，是失职。
"You don't need that module — not at your size." —— 具体、可验证。
"Here's where we're genuinely worse than the alternative." —— 说清劣势比夸自己有用。
"I'd wait a quarter, honestly." —— 敢说这句，说「该买」才有人信。
半年后系统没人用，他不会怪自己没配人，他会记住「这东西没用」——而这句话带着你的名字在他圈子里传。你以为赚了一单，实际是拿一单的钱买了条负面口碑。
「你这规模买是合适的，但有件事我得先说：这类项目失败最常见的原因不是产品，是没有一个人在内部盯落地，而你们现在没这个人。两个办法：要么先定一个人，把他的时间排进实施计划；要么只上 A 模块，跑通再扩——我更建议后者，单子小一半，成活率完全不同。」
EN Selling to someone who shouldn't buy pays today and bills you in instalments: they can't get it working, they don't renew, and they mention it to three peers in the same industry. The mechanism worth internalising is that satisfaction is a difference, not a level. The expectancy-disconfirmation model — the foundation of consumer satisfaction research since 1980 — holds that satisfaction is driven by performance relative to prior expectation rather than absolute performance, which means every promise made at signing is also a price set for the future. The neural version is the same shape: midbrain dopamine neurons encode reward prediction error, the gap between what happened and what was expected, firing above baseline when things beat expectation and dipping below when they don't. So overpromising doesn't excite the buyer, it raises the baseline you then have to clear. Worse, the difference is asymmetric: negative events outweigh equivalent positive ones across impression formation, relationships and word of mouth, so one bad sale is not cancelled by one good one. The four signals to walk away are: your product only fixes a fraction of their problem, the conditions for it to work aren't there and won't be soon, their expectations have already been inflated beyond what you can correct, and — sufficient on its own — you'd have to conceal something to close it.
满意度是个差值，不是绝对值。同一份交付，预期抬高一档变成失望，预期说准变成信任——成交那天的每句承诺都在给未来定价。更麻烦的是这个差值不对称：坏的那侧权重系统性更大，一次失望要好几次满意才补得回来。
边界：别把「敢不卖」变成姿态。拿「我劝你别买」当高级话术、指望对方反过来求你，那是操纵，识破一次就永久失效（检验跟「影响力六原理」那期一样：说破了还成立的才是说服）。真正的不卖有代价：这月数字会难看。这条对拿全额提成的新人最难，也最该由公司在考核上兜住，而不是靠个人道德硬扛。
"I don't think you should buy this yet — here's what I'd fix first." —— 劝退要给替代动作，否则只是拒绝。
"Let me tell you where this usually fails." —— 先讲失败模式，比讲案例可信。
"Smaller scope, much better odds of it actually working." —— 把单子做小是保住关系最实的一招。
「太好了，那我回去整理一份方案发您？」——他一件事没做。方案发出去之后你会等两周，然后开始猜他为什么不回。（沉默多数时候不是拒绝，是根本没有下一步——「处理没回复与死单」那期讲过。）
「我把刚梳理的东西整理成一页发你。在那之前想先落两件事：一，面试官排期这块，能不能拉上负责的 W 聊 30 分钟？二，你们上季度的到岗周期给我一个大概区间——我按你的数字算，不按行业平均。
这两件要是有一件推不动，说明还不到做方案的时候。」
谁：具体人名或角色，不是「你们团队」。
什么时候：一个日期，不是「下周找个时间」。
做什么：一个动词——拉一个人进来 / 给一份数据 / 开一个只读账号 / 内部过一遍某个页面。
他能得到什么：这动作就算最后没买对他也有价值。做不到这条，它就只是逼单。
EN The characteristic way consultative selling fails is not shallow diagnosis, it's a deep and accurate one that ends with you volunteering to do all the remaining work: you write the beautiful proposal, they take it in-house or use it to brief a cheaper vendor. What separates a meeting that moved from one that merely felt good is whether the other side left holding an action. Rackham's distinction is the useful one — an advance is a specific step the buyer commits to taking, a continuation is a warm meeting with no agreed action, and a lot of people log the second as progress. Two reasons to insist on it. Vague intentions don't execute: a meta-analysis of 94 studies found that specifying when, where and how — an if-then implementation intention — produces a medium-to-large effect on follow-through, which is the difference between "let's find time next week" and "Thursday, 10am, you, me and W, thirty minutes". And the small action is the most honest qualifying signal you have, because saying "very interested" costs nothing while pulling a colleague into a meeting spends their internal credit. The test that keeps this from becoming pressure: the action has to be worth something to them even if they never buy.
诊断创造的价值不落在对方的一个动作上，就变成他免费拿走的资产；没有动作的会面不产生承诺压力，你在他日程上的优先级第二天归零。要一个小动作还有第二重作用：它是最诚实的资格判定信号——嘴上说「很有兴趣」成本为零，拉个同事进来花的是他的内部信用。
边界：要推进不等于逼单，区别在那个动作对他有没有独立价值——「拉 W 聊 30 分钟」就算最后不买，他也把内部一个卡点谈明白了；「先签个意向书」只对你有价值。另一种失效：那动作在他们组织里根本不归他管，你却把推不动读成没兴趣。这时该问的不是「你还有兴趣吗」，是「这事要动起来得谁点头」。
"Before I write anything up — could we get W in the room for thirty minutes?" —— 把方案换成他的一个动作。
"If that's hard to arrange, it probably isn't the right time." —— 给推不动一个体面解释，也给你一个判据。
"Who has to nod for this to move?" —— 推不动时问这句，别问「还有兴趣吗」。
四十分钟，三步。挑一个你手上真在跑、却推不动的单子——找客户、找工作、找合作都算。