起源故事 · 从《目标》到一套通用流程 Origin Story · From The Goal to a Reusable Process
《目标》出版后,Eliyahu Goldratt 把书里的洞见提炼成一套可复制的流程 —— 五个聚焦步骤(POOGI,持续改善流程)。 它的革命性在于:放弃"面面俱到"的改善幻想,承认系统里永远只有极少数约束,因此每一刻都该把全部精力对准那一个。 先免费榨干现有产能,逼到极限后再花钱扩能;扩能后约束会跳到别处,于是回到第一步重新识别 —— 改善因此永不停歇。 After The Goal landed, Eliyahu Goldratt distilled its insights into a repeatable procedure — the Five Focusing Steps, also known as POOGI: the Process of Ongoing Improvement. What made it radical was its honesty: drop the fantasy of "improving everywhere at once" and accept that a system always has only a tiny number of true constraints. So at any given moment, all your effort belongs on that one constraint. First exploit existing capacity for free; only after you've pushed it to its limit do you spend money to elevate it. Once you do elevate it, the constraint jumps somewhere else — so you loop back to step one. Improvement, by design, never ends.

1 五步循环作用在一条产线上 The five-step loop, acting on one production line

瓶颈:工序 ?

上半部是五步循环(高亮当前步),下半部是一条五工序产线。点"下一步"看每一步如何作用在那道最窄的工序上。 Top half: the five-step loop (current step highlighted). Bottom half: a five-station line. Click "Next step" to see how each step acts on the narrowest station.

2 每一步具体怎么做 How to do each step in practice

① 识别约束 Identify:找出限制系统产出的那一环 —— 它前面会堆在制品、后面会饿肚子。先看堆积,别看忙碌。 ① Identify the constraint: find the one link that caps system throughput — WIP piles up in front of it, the stations behind it sit starved. Look for piles, not for who looks busy.
② 挖尽约束 Exploit:不花一分钱,先把现有产能榨到极限 —— 约束不许停工、不做废品、不等料、午休轮班别歇。 ② Exploit the constraint: without spending a cent, squeeze every drop out of existing capacity — no downtime, no scrap, no waiting for material, no idling through lunch breaks or shift changes.
③ 一切迁就约束 Subordinate:让所有非约束环节配合约束的节奏 —— 不要超产堆库存,只按约束能消化的速度供料。 ③ Subordinate everything else: pace every non-constraint to the constraint's beat — stop over-producing, stop piling inventory, feed material only at the rate the constraint can absorb.
④ 提升约束 Elevate:榨干仍不够,才投资扩能 —— 加设备、加班、外包,真正抬高约束的天花板。 ④ Elevate the constraint: only when exploitation isn't enough do you spend the money — add machines, add shifts, outsource — and actually raise the constraint's ceiling.
⑤ 回到第一步 Repeat:约束被打破后会转移到新工序。回头重新识别,别让惯性(旧规则、旧政策)成为新约束。 ⑤ Repeat — and beware inertia: once you break a constraint, it shifts somewhere new. Go back and re-identify. Above all, don't let old rules and old policies harden into the next constraint.

3 现实里的五步聚焦 The five steps in the real world

工厂改善:先把瓶颈机台的换型时间压到最短(挖尽),不够再买第二台(提升),约束随即跳到下游。 Factory improvement: first cut the bottleneck machine's changeover to the minimum (exploit). Only if that's not enough do you buy a second machine (elevate) — and the constraint promptly jumps downstream.
医院流程:手术室是约束 → 优化排程不让它空转(挖尽),再增加麻醉师与术后床位(提升)。 Hospital flow: the OR is the constraint. Tighten scheduling so it never sits idle (exploit), then add anesthesiologists and recovery beds (elevate).
研发交付:评审是瓶颈 → 先精简评审流程(挖尽),再增配评审人(提升),约束转到部署环节。 Engineering delivery: code review is the bottleneck. Streamline the review process first (exploit), then add reviewers (elevate) — and the constraint shifts to the deployment pipeline.
政策约束:当物理瓶颈都解决后,真正的约束常是"一条过时的规定" —— 第五步要警惕的惯性约束。 Policy constraints: once the physical bottlenecks are gone, the real constraint is often a stale rule on paper — exactly the inertia step five warns you to watch for.
一句话In one line
五个聚焦步骤把"改善"从一场到处救火的疲劳战,变成一次只攻一个点的杠杆战。 关键顺序是:先免费挖尽(②),再花钱提升(④) —— 太多团队跳过②直接买设备,钱花了产出却没涨。 而最深的智慧在第五步:约束永远会转移,改善没有终点;最难打破的,往往不是机器,而是一条"我们一直这么做"的旧规则。 The Five Focusing Steps turn improvement from an exhausting firefight on every front into a leverage play that hits one point at a time. The order matters: exploit for free first (②), then spend to elevate (④). Far too many teams skip step ② and buy equipment straight away — the cash leaves, the throughput doesn't move. The deepest wisdom sits in step five: the constraint always migrates; improvement has no finish line. And the hardest constraint to break is rarely a machine — it's the policy of "this is how we've always done it."
常见误用Common mistakes
跳过"挖尽"直接花钱"提升"。先把现有产能免费榨干,再考虑投资 —— 顺序不能反。 Skipping Exploit and jumping straight to Elevate. Squeeze existing capacity dry — for free — before you authorize a single dollar of investment. The order is not negotiable.
多个改善项目齐头并进。一次只聚焦当前那一个约束,分散兵力等于没改善。 Running many improvement projects in parallel. Aim everything at the current constraint. Splitting your effort across the system is indistinguishable from no improvement at all.
解决了瓶颈就以为大功告成。回到第一步:约束已转移,警惕惯性成为新约束。 Treating a broken bottleneck as the end of the job. Loop back to step one — the constraint has already moved, and the next one might be a policy hiding in plain sight.

五个聚焦步骤