---
author: TechButMakeItReal
date: '2026-02-11'
guest: ''
layout: post.njk
source: https://www.youtube.com/watch?v=zUrwjSWzOLM
speaker: TechButMakeItReal
tags:
  - consulting-industry
  - ai-disruption
  - business-models
  - future-of-work
  - labor-arbitrage
title: 企业咨询的兴衰：AI如何重塑行业格局
summary: 本文深入剖析了传统管理咨询的盈利模式——以瑞士航空的案例揭示其潜在的战略风险，并阐述了基于‘劳动套利’和‘金字塔结构’的盈利机制。随后，重点探讨了AI技术，特别是麦肯锡的‘Lily’平台，如何通过自动化核心研究环节颠覆咨询业的经济基础。最后，预测行业将分化为精品咨询与软件服务模式，并分析这对初级咨询师职业生涯的深远影响。
insight: ''
draft: true
series: ''
category: tech-trends
area: tech-engineering
project: []
people: []
companies_orgs:
  - McKenzie
  - Accenture
  - Deloitte
  - Palantir
products_models:
  - Lily
media_books: []
status: evergreen
---
### 咨询业的困境：从瑞士航空的陨落看战略失误

瑞士航空（Swiss Air）曾不仅仅是一家航空公司，它更是瑞士国民的骄傲，一个象征着国家特质的标志。它完美诠释了瑞士引以为傲的品质：精准、准时、优质服务，以及最重要的——财务稳健。数十年间，这家航空公司因其坚实的财务基础而赢得了“飞行银行”的美誉。几代瑞士儿童甚至将瑞士航空的股票作为礼物。作为瑞士的全球大使，它一直运营到2001年10月。然而，其衰落的种子早在十年前就已埋下，当时瑞士公民投票否决了加入欧洲联盟（EU）。

瑞士选择留在欧盟之外，这一决定导致瑞士航空被排斥在与其他航空公司的主要联盟之外。作为一个“局外人”，它只能眼睁睁地看着英国航空（British Airways）、汉莎航空（Lufthansa）、法国航空（Air France）和KLM等公司不断增加航班。瑞士航空试图独立发展的努力，却因欧盟国家的限制而举步维艰，最终陷入了进退两难的境地：在国内这个狭小的市场里它过于弱小，无法竞争；同时又被阻挡在更广阔的欧盟市场之外。为了寻求突破，瑞士航空聘请了麦肯锡（McKenzie）。麦肯锡提出了一种名为“猎手战略”（Hunter strategy）的方案。该战略的核心并非加入或组建联盟（这本身就是一场艰苦的战斗），而是收购多家欧洲航空公司的少数股权，并以此为基础从外部“编织”出一个网络。瑞士航空采纳了这一策略，花费数亿美元收购了比利时萨贝娜航空（Sabina）49%的股份，以及德国LTU航空和几家陷入困境的法国航空公司的重要股权。

这个计划听起来相当周密，但却忽略了一个关键细节：欧盟法律不允许非欧盟航空公司拥有控股权。因此，瑞士航空虽然收购了股份，却无法真正影响这些航空公司，却要承担它们的所有亏损。萨贝娜航空的工会阻挠了成本削减措施，员工数量激增，瑞士航空不得不在一_年内冲销了全部股权投资。然而，管理层并未因此退缩，反而变本加厉，因为他们在1998年创下了利润新高，这似乎是对麦肯锡战略的“验证”。随后，当燃油价格上涨、需求下降时，这座“纸牌屋”轰然倒塌。到2000年，瑞士航空公布了29亿瑞士法郎的巨额亏损。2001年，KPMG的审计显示，其债务高达170亿瑞士法郎，而股本仅为5.55亿瑞士法郎，负债权益比高达30:1。9·11事件后，在现金几乎耗尽且银行不愿提供更多信贷的情况下，瑞士航空于2001年10月2日破产，其存在一夜之间戛然而止。麦肯锡的“猎手战略”建议瑞士航空采取一种扩张模式，这种模式在PPT上看起来光鲜亮丽，但在欧盟法律框架下却难以实现。收购战略被包装成保持竞争力的唯一途径，却从未被问及一个核心问题：如果法律禁止你拥有这些航空公司，你是否真的能够控制并扭转它们的困境？瑞士航空严格按照建议执行了这一战略，最终导致了其毁灭，损失了约150亿瑞士法郎和一个国家象征。而麦肯锡，因未参与执行细节，则带着咨询费和声誉安然退场。瑞士人民不禁要问，世界上最负盛名的咨询公司，是否在事实上助长了其国家旗舰航空公司的覆灭？

<details>
<summary>Original English</summary>
Swiss
 Air
 was
 more
 than
 an
 airline
 to


the
 Swiss
 people.
 It
 was
 a
 national


symbol,
 an
 icon
 that
 embodied
 the
 very


quality
 Switzerland
 prided
 itself
 on


precision,
 punctuality,
 service,
 and


above
 all,
 financial
 property.
 For


decades,
 the
 airline
 was
 so
 financially


solid
 that
 it
 earned
 the
 nickname
 the


flying
 bank.
 Generations
 of
 Swiss


children
 received
 shares
 of
 Swiss
 Air
 as


gifts.
 The
 airline
 operated
 as


Switzerland's
 global
 ambassador
 until


October
 2001.
 The
 collapse
 began
 a


decade
 earlier
 when
 Swiss
 voters


rejected
 to
 join
 the
 European
 Union.


Switzerland's
 vote
 to
 stay
 out
 of
 the
 EU


left
 Swiss
 Air
 locked
 out
 of
 the
 big


alliances
 with
 other
 airlines.
 As
 an


outsider,
 the
 airline
 could
 only
 watch


as
 British
 Airways,
 Luansa,
 Air
 France,


and
 KM
 kept
 adding
 flights
 to
 their


schedule.
 Swiss
 Air's
 own
 attempts
 to


grow
 independently
 were
 hampered
 by
 the


EU
 countries
 and
 the
 airline
 was


trapped.
 Too
 small
 to
 compete
 in
 its


tiny
 home
 market
 and
 yet
 blocked
 from


the
 broader
 EU
 market.
 Desperate
 for
 a


way
 in,
 Swiss
 Air
 hired
 McKenzie.


McKenzie
 proposed
 something
 called
 a


Hunter
 strategy.
 Instead
 of
 joining
 or


forming
 an
 alliance,
 which
 is
 an
 uphill


battle
 in
 and
 of
 itself,
 buy
 minority


stakes
 in
 multiple
 European
 airlines
 and


stitch
 together
 a
 network
 from
 the


outside.
 Swiss
 Air
 followed
 the
 playbook


and
 spent
 hundreds
 of
 millions
 to


acquire
 49%
 of
 Belgium's
 Sabina
 as
 well


as
 big
 stakes
 in
 Germany's
 LTU
 and


several
 struggling
 French
 carriers.
 And


that
 plan
 sounds
 pretty
 solid
 except
 for


one
 tiny
 detail.
 EU
 law
 did
 not
 allow
 a


non-EU
 airline
 to
 take
 majority
 control.


So
 Swiss
 Air
 could
 not
 truly
 influence


these
 airlines
 that
 had
 bought,
 but
 it


was
 still
 on
 the
 hook
 for
 their
 losses.


Sabina's
 unions
 blocked
 cost
 cuts.


Headcount
 ballooned
 and
 Swiss
 Air
 had
 to


ride
 off
 its
 entire
 equity
 stake
 within


a
 year.
 Instead
 of
 backing
 off,


management
 doubled
 down
 because
 they
 hit


record
 profit
 in
 1998.
 And
 that
 was
 the


apparent
 validation
 of
 McKenzie


strategy.
 When
 fuel
 prices
 rose
 and


demand
 went
 down,
 the
 House
 of
 Cards


collapsed.
 By
 2000,
 Swiss
 Air
 posted
 a


2.9
 billion
 frank
 loss.
 And
 a
 KPMG


reodit
 in
 2001
 showed
 17
 billion
 franks


of
 debt
 supported
 by
 555
 million
 in


equity,
 which
 comes
 to
 30:1
 debt
 to


equity
 ratio.
 After
 911,
 with
 cash


almost
 gone
 and
 banks
 unwilling
 to


extend
 more
 credit,
 Swiss
 Air
 collapsed


on
 October
 2nd,
 2001
 and
 ended
 the


airlines
 existence
 overnight.
 McKenzie's


Hunter
 strategy
 advised
 Swiss
 Air
 to


pursue
 an
 expansion
 model
 that
 looked


brilliant
 in
 a
 deck,
 but
 was
 impossible


in
 reality
 under
 EU
 laws.
 The


acquisition
 strategy
 was
 framed
 as
 the


only
 way
 to
 stay
 competitive,
 but
 never


asked
 the
 hard
 question.
 Can
 you


actually
 control
 and
 fix
 these


struggling
 airlines
 if
 you're
 legally


barred
 from
 owning
 them?
 Swiss
 Air


executed
 the
 strategy
 exactly
 as
 advised


and
 it
 destroyed
 them,
 wiping
 out
 about


15
 billion
 Franks
 and
 a
 national
 icon.


McKenzie,
 excluded
 from
 execution,


walked
 away
 with
 its
 fees
 and
 reputation


largely
 intact.
 While
 Switzerland
 was


left
 to
 ask
 whether
 the
 most
 prestigious


consultants
 in
 the
 world
 had
 effectively


helped
 kill
 their
 flag
 carrier,
 today's


episode
 is
 about
 the
 rise
 and
 fall
 of


corporate
 consulting
 and
 where
 it's


headed
 now
 that
 AI
 is
 part
 of
 the


equation.
 The
 good,
 the
 bad,
 and
 the


ugly.
 Let's
 dive
 in.
</details>

### 管理咨询的黄金时代：劳动套利与金字塔模型

管理咨询的黄金时代，本质上是一场精心策划的**劳动套利**（Labor Arbitrage: 利用不同地区或国家劳动力成本差异以降低成本的行为。在咨询业中，指利用初级、低成本员工提供服务，并以高级、高费率收费）。咨询公司通过雇佣大量初级员工，让他们大规模地投入工作，从而获得可观的利润率。尽管咨询公司将自身服务包装为战略洞察和合伙人级别的专业知识，但实际上，其中很大一部分是营销术语。正如镀金首饰中的“金”字一样，一位名叫**大卫·梅斯特**（David Maester）的教授系统化地阐述了一个理论：一个咨询公司要运作良好，需要维持约70%的高利润率。这种70%的利润率将咨询从一个专业服务转变为一个劳动套利机器。

在传统的咨询公司中，层级结构通常是这样的：分析师（Analyst）、顾问（Consultant）、经理（Manager）、高级经理（Principal）和合伙人（Partner）。合伙人年收入在100万至500万美元之间，但这并非来自他们自己的计费工时。他们的收入来源于**梅斯特教授**所称的“非合伙人员工产生的盈余”。换句话说，合伙人通过出售他们“合伙人级别”的知识和专业经验，但实际上是由非合伙人团队来执行。如果听众觉得这听起来很熟悉，却又说不上来是什么，那是因为这本质上就是一个经典的**金字塔骗局**（Pyramid Scheme），一个以金字塔为商业模式本身的企业。其根本的经济模型基于“人均收入”的指标，而麦肯锡、BCG和Bain等公司在这方面表现得惊人地一致。

过去两年，人均咨询师收入平均约为每年43万美元。然而，麦肯锡的入门级商业分析师年薪仅为12万美元左右（基本工资），总收入约15万美元，这意味着存在3倍的加价。现在，让我们放大视角，看看一个典型的咨询项目通常的**人员配置结构**（Staffing Structure）是怎样的。你可能会问，什么是“典型”？我们以一个最常见的、中等规模的项目为例。

谁是咨询服务最大的消费者？通常是**中型市场企业**（Mid-market enterprises）。它们之所以高度依赖外部资源，是因为它们存在一个独特的“差距”：它们面临着大型企业才会遇到的问题，如并购（M&A）、市场扩张、新市场开拓等，但却缺乏大型企业所拥有的资源。它们有预算聘请咨询公司，但没有预算或时间来建立自己的内部专业团队。一个典型的咨询项目人员配置模式是：一位高级合伙人投入20%的时间（这意味着他同时管理其他四个客户），一位项目经理（Engagement Manager），两名顾问（Consultants），以及两名分析师（Analysts），项目周期为三个月。人员配置还会根据项目类型而变化。

**梅斯特教授**称之为“**杠杆决策**”（Leverage Decision）。高杠杆项目会投入大量初级员工，成本倍数可达4到6倍，为公司带来约70%的毛利润。低杠杆项目则由高级合伙人负责，虽然对客户而言成本更高，但会压缩咨询公司的利润率至30%至40%。这意味着，咨询公司的**单位经济效益**（Unit Economics）更倾向于高杠杆项目。这也是为什么合伙人的薪酬并非来自他们自己的计费工时，而是来自他们所能调动的下属团队的“杠杆”。一个拥有100名初级顾问、30名经理和10名合伙人的公司，实现了10:1的杠杆比例，每位合伙人从其下属的10名员工那里提取利润。

这种结构催生了与咨询相关的诸多文化现象：频繁的客户现场拜访、耗费心力的PPT制作、巨大的工作压力。这都是为了证明那些“高得离谱”的成本是合理的。初级顾问会锁定数千小时的计费工时，以证明他们每小时75美元的费率是合理的，并产生每小时300美元的利润，从而支撑合伙人的薪酬。为了更好地理解这一点，我们来做一个类比。假设我开了一家名为“Daria's Lemonade Stand”的公司，通过榨取柠檬、加糖和水来制作柠檬水。制作一杯的成本是0.5美元，售价为5美元，每杯利润4.5美元，利润率高达90%。我的商业模式是线性增长的。我制作的柠檬水越多，我就会越富有。

但有一天，我醒悟过来：“我怎么能只做个业余的？”我决定开设更多的柠檬水摊位，并将柠檬水卖给餐厅。我雇佣了玛丽亚（Maria）作为我的吧台经理，她负责日常运营、培训员工和质量控制。然后我雇佣了劳伦（Lauren）来制作柠檬水，她负责服务顾客和补充杯子。劳伦代表了我77%的利润。接着我又雇佣了杰克（Jack）来负责库存和准备柠檬。现在，我不再销售单个杯子，而是销售柠檬水套餐，为附近的办公室提供持续的柠檬水供应。我最终获得了一份价值每月20万美元的持续柠檬水供应的大型企业合同。这份合同需要我的监督、玛丽亚负责与办公室沟通、劳伦制作柠檬水，以及杰克准备原料。我唯一的重大限制是人力。劳伦每月只能制作有限数量的柠檬水。如果我想扩大规模，就需要雇佣更多的调酒师并购买更多的柠檬。

但随后，我购买了一台价值5万美元的自动化柠檬水机。它能生产出与劳伦或杰克组合相当的质量，每月运营成本仅为2美元（假设的电费），并且可以24/7不间断地满足订单。现在，我的利润率不再是77%，而是99%。你可能会认为这是个利润丰厚的生意。是的，但它有局限性。这种**杠杆模型**的盈利能力依赖于最大化计费工时。我投入到项目中的工时越多越好，使用的劳动力越便宜越好。但这同样意味着线性扩展。一位合伙人只能有效监督10到20名直接下属。如果我有一个由五到七人组成的更大团队，我就不得不雇佣额外的管理人员。这种模式在三十年间创造了非凡的回报。

<details>
<summary>Original English</summary>
The
 golden
 age
 of
 management
 consulting


was
 basically
 a
 well
 orchestrated


arbitrage.
 You
 get
 a
 lot
 of
 junior


labor.
 You
 put
 them
 to
 work
 at
 scale
 and


benefit
 from
 a
 pretty
 sizable
 margin.


Yes,
 consultancies
 sold
 their
 services


as
 strategic
 insights
 and
 partner
 level


expertise.
 But
 in
 reality,
 a
 lot
 of
 it


was
 marketing
 lingo.
 Just
 like
 the
 word


gold
 in
 a
 goldplated
 jewelry,
 there
 is
 a


professor
 by
 the
 name
 of
 David
 Maester


who
 formalized
 a
 theory
 that
 for
 a


consulting
 firm
 to
 function,
 it
 needs
 to


have
 a
 very
 high
 margin
 around
 70%.
 This


70%
 margin
 transformed
 consulting
 from
 a


professional
 service
 into
 a
 labor


arbitrage
 machine.
 In
 a
 traditional


consulting
 firm,
 the
 hierarchy
 goes
 like


this.
 Analyst,
 consultant,
 manager,


principal,
 and
 partner.
 partners
 make


between$1
 and
 $5
 million
 annually,
 but


not
 through
 their
 own
 billing
 hours.


They
 make
 it
 through
 what
 Professor


Maester
 calls
 the
 surplus
 generated
 from


nonpartner
 staff.
 Once
 again,
 partners


make
 millions
 selling
 knowledge
 and


expertise
 through
 the
 surplus
 generated


from
 non-partner
 staff.
 In
 plain


English,
 this
 means
 that
 I
 am
 selling
 my


partner
 level
 knowledge,
 except
 I'm
 not


doing
 it
 myself.
 If
 you're
 listening
 to


this
 and
 thinking,
 "This
 sounds
 awfully


familiar,
 but
 I
 can't
 put
 a
 finger
 on


what
 it
 is."
 Let
 me
 help
 you.
 It's


called
 the
 good
 old
 pyramid
 scheme.
 A


business
 where
 the
 pyramid
 is
 the


business
 model
 itself.
 The
 underlying


economic
 model
 is
 based
 on
 the
 revenue


per
 employee
 metrics.
 and
 McKenzie,
 BCG,


and
 Bane
 are
 strikingly
 consistent
 with


it.
 As
 of
 the
 last
 two
 years,
 revenue


per
 consultant
 averages
 around
 $430,000


per
 year.
 But
 entry-level
 business


analysts
 at
 McKenzie
 earn
 around
 $120.


That
 is
 their
 base
 and
 around
 $150,000


total,
 which
 means
 that
 there
 is
 a
 3x


markup.
 Now,
 let's
 zoom
 out.
 What
 is
 the


typical
 staffing
 structure
 on
 a
 typical


consultant
 project?
 You
 may
 ask,
 what
 do


you
 mean
 by
 typical?
 Let's
 say
 a
 most


common
 mid-range
 project.
 Now,
 who
 are


the
 biggest
 consumers
 of
 consulting


services?
 Mid-market
 enterprises.
 And


the
 reason
 for
 it
 is
 because
 they
 have
 a


unique
 gap
 that
 makes
 them


hyperdependent
 on
 outside
 resources.


Mid-market
 enterprises
 deal
 with
 big
 boy


problems,
 M&A,
 market
 expansion,
 new


territories.
 But
 what
 they
 don't
 have
 is


the
 big
 boy
 enterprise
 resources.
 They


have
 the
 budget
 for
 consulting
 firms
 but


not
 the
 budget
 or
 time
 to
 build
 their


own
 in-house
 expertise.
 A
 typical


staffing
 model
 for
 a
 consulting
 project


puts
 one
 senior
 partner
 who
 dedicates


20%
 of
 their
 time,
 meaning
 that
 they


manage
 four
 other
 clients.
 One


engagement
 manager,
 two
 consultants,
 two


analysts
 over
 a
 three-month
 period.
 The


staffing
 depends
 on
 the
 project
 type
 and


Professor
 Maester
 calls
 it
 the
 leverage


decision.
 High
 lever
 projects
 put
 armies


of
 junior
 staff
 at
 4
 to6x
 cost
 multiples


and
 bring
 the
 firm
 around
 70%
 gross


margins.
 Low
 leverage
 projects
 are


staffed
 with
 senior
 partners
 and
 yes


they
 cost
 more
 to
 the
 client
 but
 they


also
 compress
 margins
 to
 about
 30
 to
 40%


for
 a
 consultancy
 which
 means
 that
 the


unit
 economics
 of
 a
 consulting
 firm


favors
 high
 leverage
 projects.
 And
 this


is
 why
 partner
 compensation
 derives
 not


from
 partners'
 own
 billable
 hours
 but


from
 the
 leverage
 they
 have
 underneath.


A
 firm
 with
 100
 junior
 consultants,
 30


managers,
 10
 partners
 achieves
 10
 to
 one


leverage
 and
 each
 partner
 extracts


margin
 from
 10
 subordinates
 underneath.


This
 structure
 gave
 life
 to
 a
 myriad
 of


cultural
 artifacts
 associated
 with


consulting.
 client
 site
 visits,
 weakened


slide
 decks
 preparations,
 enormous


stress,
 and
 it
 makes
 sense
 because
 you


have
 to
 somehow
 justify
 costs
 that
 are


blown
 out
 of
 proportion.
 Junior


consultants
 would
 lock
 thousands
 of


billable
 hours
 to
 justify
 their
 $75
 an


hour
 rate
 and
 generate
 the
 $300
 an
 hour


margin,
 which
 would
 fund
 the
 partner


compensation.
 To
 put
 this
 in


perspective,
 you
 guys
 know
 I
 love


analogies,
 so
 let's
 use
 one.
 Let's
 say
 I


start
 a
 company
 called
 Daria's
 Lemonade


Stand.
 I
 make
 lemonade
 by
 squeezing


lemons,
 adding
 sugar
 and
 water.
 The
 cost


to
 make
 one
 cup
 is
 50.
 I
 sell
 each
 cup


for
 $5.
 Profit
 per
 cup
 $4.50.


That
 is
 90%
 margin.
 My
 business
 model


scales
 linearly.
 The
 more
 lemonade
 I


make,
 the
 richer
 I
 get.
 Then
 one
 day
 I


wake
 up
 and
 go,
 what
 am
 I,
 an
 amateur?


I'm
 going
 to
 make
 more
 lemonade
 stands


and
 I'm
 going
 to
 sell
 my
 lemonade
 to


restaurants.
 I
 hire
 Maria,
 my
 bar


manager.
 She
 runs
 the
 stand
 daily.
 She


trains
 staff.
 She
 handles
 the
 quality.


And
 then
 I
 hire
 Lauren
 who
 makes
 the


lemonade.
 She
 would
 serve
 customer
 and


she
 would
 restock
 the
 cups.
 Lauren
 is
 my


77%
 margin.
 And
 then
 I
 hire
 Jack
 who


stocks
 shelves
 and
 prepares
 the
 lemons.


Now
 instead
 of
 individual
 cups,
 I
 sell


lemonade
 packages,
 a
 continuous
 supply


of
 lemonade
 to
 offices
 nearby.
 And
 I


finally
 find
 my
 big
 corporate
 contract


for
 $200,000
 a
 month
 of
 continuous


lemonade
 supply.
 This
 contract
 requires


my
 oversight,
 Maria
 managing


communications
 with
 offices,
 Lauren


making
 lemonade,
 and
 Jack
 preparing
 the


ingredients.
 My
 only
 major
 constraint
 is


people.
 Lauren
 could
 only
 make
 so
 many


cups
 a
 month.
 If
 I
 want
 to
 scale,
 I
 need


to
 hire
 more
 baristas
 and
 buy
 more


lemons.
 But
 then
 I
 buy
 an
 automated


lemonade
 machine
 for
 50
 grand.
 It


produces
 the
 same
 quality
 as
 Lauren
 or


Jack
 combined.
 And
 it
 cost
 me
 $2
 a
 month


to
 operate
 an
 electricity
 bill,


hypothetical
 one,
 but
 still.
 Fulfills


orders
 24/7
 without
 any
 breaks.
 And


guess
 what?
 Now
 my
 margin
 is
 no
 longer


77%.
 It's
 99.
 You
 may
 think
 what
 a


profitable
 business.
 Yes,
 but
 it
 has


limitations.
 The
 profitability
 of
 the


leverage
 model
 depends
 on
 maximizing


billable
 hours.
 The
 more
 hours
 I
 put


into
 a
 project,
 the
 better.
 The
 cheaper


the
 labor,
 the
 better.
 But
 what
 this


also
 means
 is
 linear
 scaling.
 A
 partner


could
 only
 supervise
 only
 10
 to
 20


direct
 reports
 effectively.
 And
 if
 I


have
 a
 larger
 team
 of
 five
 to
 seven


people
 or
 more,
 I
 would
 have
 to
 hire


additional
 management.
 This
 model


generated
 extraordinary
 returns
 for


three
 decades.
</details>

### AI的颠覆：麦肯锡“Lily”重塑咨询业态

然而，AI对咨询业最根本的影响在于，它颠覆了咨询业务模式的**单位经济效益**。AI甚至不需要完全自动化所有工作，它只需要能够赋能中层顾问，就能极大地改变局面。但如果它完全自动化了工作，它就会摧毁咨询公司的利润基础，而利润正是咨询公司得以运营的首要原因。咨询公司销售洞察，但利润来源于**杠杆**（Leverage: 在咨询业中，杠杆指项目中初级员工与高级员工的比例。高杠杆意味着初级员工相对于合伙人数量更多，从而能带来更高的利润率）。一旦移除了杠杆，一个价值160亿美元的麦肯锡公司其经济基础将可能蒸发。

管理咨询的经济学建立在一个前提之上：客户支付高昂的费用，是因为咨询公司提供的服务需要大规模部署的人工专业知识。但这一观念已被证明是错误的，而证明它错误的，恰恰是麦肯锡自己内部的AI平台——“**Lily**”。2023年7月，麦肯锡推出了其内部AI平台Lily，该平台基于公司100年的知识产权进行训练，涵盖了数十万份文档、案例研究、访谈记录和框架。在一年半的时间里，麦肯锡45,000名顾问中的75%每月都在使用它。再次强调，公司中有三分之二的员工将AI作为核心研究工具，而非仅仅用于撰写邮件或总结笔记。

Lily成为了研究工作的关键工具，而研究正是咨询业务的核心。设身处地为一名初级顾问着想：你需要回答一个问题，比如“如何清除纽约的老鼠？”你首先会分解问题，使其易于理解。然后呢？你想知道过去是如何解决类似问题的。现在你为一家大型咨询公司工作，这正是咨询师们的工作方式。你拥有数据——过去100年的数据、框架、方法论、案例研究、演示文稿。但你需要花费大量时间去筛选这些海量案例和数据。过去，当客户向麦肯锡提交一份提案请求（RFP）时，一名初级顾问会被指派来撰写。他们会花费数小时甚至数天时间翻阅内部知识库，试图弄清楚这个问题是否以前被处理过，能从过往案例中学到什么，如何处理问题，哪些方法有效，哪些无效。

但有了Lily，他们可以立即获得一份摘要。他们可以快速浏览先例、查找联系人、找到框架、综合各种来源的数据，并附带链接。这项研究工作原本需要数天，现在可以在3小时内完成。之后，他们会为启动会议做准备，而项目经理（Engagement Manager）会负责规划整个项目：工作流、范围、人员配置、时间表。这一切通常需要两到三天的时间。有了Lily，你可以获得关于哪些项目结构有效、权衡了哪些因素、需要哪些资源等所有关于过往案例的信息。你拿来稍作修改，一天之内就能完成。

然后，我们来到了定义咨询行业传奇部分——演示文稿（decks）。Lily可以根据提示自动生成完整的PowerPoint演示文稿。任何准备过演示文稿的人都知道，制作PPT更多的是关于移动颜色和形状，而不是内容本身。当然，我们都有公司批准的模板，但说实话，很多时候它们并不适用。Lily消除了初级分析师约20%的工作量，包括研究、综合、演示文稿制作和客户互动。对于一个拥有45,000名员工的公司来说，这意味着约8,000名全职员工被“消除”。麦肯锡裁员了5,000人，其余人员则被吸收到更高级别的工作中。毫不奇怪，在这样做之后，他们并没有损失任何收入，这反过来证明了客户支付的是“**人为稀缺**”（Artificial Scarcity）的费用。

令人瞩目的是咨询公司内部的文化转变。当客户因为好奇能否更便宜地获得答案而主动使用AI时，这是一种情况。而当AI成为公司员工的一部分时，则是另一种情况。当会议开始时有人说：“你问过Lily了吗？”你就已经认识到，价值不再是研究本身，而是你如何利用Lily的输出。因此，旧的顾问技能集——详尽的研究能力——变得过时了。具有讽刺意味的是，这个以销售优化为生的行业，最终优化了它自己。当他们引入AI并削减了30%自身的工作量时，他们证明了自己的商业模式是人为膨胀的。

<details>
<summary>Original English</summary>
But
 the
 most
 fundamental


impact
 of
 AI
 on
 consulting
 is
 in


realizing
 that
 it
 disrupts
 the
 unit


economics
 of
 the
 consulting
 business


model.
 It
 doesn't
 even
 have
 to
 fully


automate
 the
 work.
 All
 it
 has
 to
 do
 is


to
 boost
 a
 midlevel
 consultant.
 But
 if


it
 automates
 fully,
 it
 destroys
 the


margin.
 And
 the
 margin
 is
 the
 reason
 why


consultancies
 operate
 in
 the
 first


place.
 Consulting
 firms
 sell
 insights


but
 profit
 from
 leverage.
 Remove
 the


leverage
 and
 the
 economics
 of
 a
 16


billion
 dollar
 McKenzie
 fully
 evaporate.


The
 economics
 of
 management
 rest
 on
 a


premise
 that
 clients
 pay
 premium
 rates


because
 the
 work
 that
 consultants


provide
 requires
 human
 expertise


deployed
 at
 scale.
 But
 this
 very
 notion


was
 proven
 wrong.
 Guess
 by
 whom?
 by


McKenzie's
 very
 own
 internal
 AI
 platform


called
 Lily.


July
 2023,
 McKenzie
 launches
 Lily,
 their


internal
 AI
 platform
 trained
 on
 100


years
 of
 firm's
 intellectual
 property.


Hundreds
 of
 thousands
 of
 documents,


cases,
 interview
 transcripts,


frameworks,
 and
 within
 a
 year
 and
 a


half,
 75%
 of
 Mackenzie's
 consultants,


which
 let
 me
 remind
 you,
 it's
 45,000


people
 we're
 talking,
 are
 using
 it


monthly.
 Once
 again,
 34
 of
 the
 firm
 uses


AI
 as
 a
 core
 research
 tool,
 not
 for


writing
 emails
 or
 summarizing
 notes.
 It


serves
 as
 an
 essential
 tool
 for


research.
 And
 research
 is
 what


consulting
 business
 sells.
 Put
 yourself


in
 the
 shoes
 of
 a
 junior
 consultant.
 You


got
 to
 answer
 a
 question
 like
 how
 to
 get


rid
 of
 rats
 in
 New
 York.
 You
 break
 down


the
 problem.
 You
 make
 it
 digestible.
 And


then
 what?
 Then
 you
 want
 to
 know
 how


similar
 problems
 were
 solved
 in
 the


past.
 Now
 you
 work
 for
 a
 major


consultancy
 and
 that's
 exactly
 what


consultants
 do.
 You
 have
 the
 data.
 You


have
 the
 data
 from
 the
 past
 100
 years.


You
 have
 frameworks.
 You
 have
 methods.


You
 have
 case
 studies.
 You
 have
 networks


decks.
 But
 you
 need
 time
 to
 sort
 through


piles
 and
 piles
 of
 cases
 and
 piles
 and


piles
 of
 data.
 Here's
 what
 used
 to


happen.
 When
 a
 client
 sent
 McKenzie
 a


request
 for
 proposal,
 a
 junior


consultant
 would
 get
 assigned
 to
 write


it.
 They
 would
 then
 spend
 hours
 or
 even


days
 digging
 through
 internal
 knowledge


base
 trying
 to
 figure
 out
 whether
 it
 has


been
 done
 before
 and
 what
 can
 we
 learn


from
 previous
 cases,
 how
 to
 approach
 a


problem,
 what
 worked
 and
 what
 didn't.


But
 with
 Lily,
 they
 get
 a
 summary
 of
 it


right
 away.
 They
 go
 through
 precedents,


they
 find
 contacts,
 they
 find


frameworks,
 synthesize
 data,
 all
 kinds


of
 sources
 with
 links.
 And
 this
 research


can
 be
 done
 in
 3
 hours.
 They
 would
 then


prepare
 for
 a
 kickoff
 call
 and
 someone


called
 an
 engagement
 manager
 would


structure
 the
 entire
 engagement,
 work


streams,
 scope,
 staffing,
 timelines.
 All


of
 this
 would
 require
 two
 to
 three
 days


of
 planning.
 With
 Lily,
 you
 get
 answers


what
 project
 structure
 worked,
 what


trade-offs
 you
 had,
 resources,
 and


everything
 you
 need
 to
 know
 about


previous
 cases.
 You
 take
 it,
 you
 refine


it,
 and
 it's
 done
 in
 less
 than
 a
 day.


And
 now
 we
 get
 to
 the
 legendary
 part


that
 defines
 consulting
 as
 an
 industry,


the
 decks.
 And
 Lily
 can
 autogenerate
 an


entire
 PowerPoint
 from
 prompts.
 Anyone


who
 has
 prepped
 a
 deck
 before
 knows
 that


a
 deck
 is
 a
 lot
 less
 about
 content
 and
 a


lot
 more
 about
 moving
 the
 freaking


colors
 and
 circles
 around.
 And
 sure,
 we


all
 have
 company
 approved
 templates,
 but


let's
 be
 honest,
 half
 the
 time


doesn't
 work.
 Lily
 wiped
 around
 20%
 of
 a


junior
 analyst
 work.
 research,


synthesis,
 decreation,
 client


interaction,
 and
 for
 a
 45,000
 people


firm,
 that's
 about
 8,000
 full-time


employees
 eliminated.
 McKenzie
 cut
 5,000


and
 the
 rest
 got
 absorbed
 into
 higher


level
 work.
 And
 no
 surprise,
 when
 they


did
 that,
 they
 did
 not
 lose
 any
 revenue,


which
 in
 turn
 proved
 that
 the
 clients


were
 paying
 for
 artificial
 scarcity.


What's
 remarkable
 is
 the
 culture
 shift


within
 the
 consultancy
 itself.
 It's
 one


thing
 when
 your
 customer
 prompts
 AI


because
 they're
 curious
 if
 they
 can
 do


it
 cheaper
 and
 get
 the
 answers
 quickly.


It's
 another
 thing
 when
 AI
 becomes
 part


of
 your
 staff.
 When
 you
 start
 a
 meeting


with,
 "Have
 you
 asked
 Lily?"
 you're


recognizing
 that
 the
 value
 is
 no
 longer


research.
 It's
 what
 you
 do
 with
 Lily's


output.
 And
 therefore,
 the
 old


consultant
 skill
 set,
 which
 is


exhaustive
 research,
 becomes
 obsolete.


The
 irony
 is
 the
 fact
 that
 the
 business


that
 sells
 optimization
 optimized


itself.
 When
 they
 introduced
 AI
 and
 cut


30%
 of
 their
 own
 work,
 they
 proved
 that


their
 own
 business
 model
 was


artificially
 inflated.
 So
 what
 does
 this


mean
 for
 consulting
 as
 an
 industry?
 Is


it
 dead?
 No,
 it's
 not
 and
 it
 won't
 be.
</details>

### 咨询业的未来：分化趋势与职业生涯重塑

那么，这对咨询业意味着什么？它是否已经死亡？不，它并没有，也不会。但它确实正在分化为两个截然不同的领域，而这两个领域在五年内可能几乎认不出彼此。

第一条路径是**精品咨询公司**（Boutique Consultancies）。这本质上是回归咨询业最初的定位：提供昂贵的精英专业知识。这类公司通常由五到十人组成，专注于高度专业化的领域，例如医疗AI伦理、ESG合规、供应链韧性等。它们将明显偏向于高级人员配置，专注于开发原型而非演示文稿，侧重于概念验证（Proof of Concept）而非通用框架。

第二条路径则是成为**软件公司**，并披上咨询的外衣。它们将削减金字塔的底部——那些初级顾问和刚毕业的大学生，他们缺乏专业知识。这条路径本质上是将精品咨询模式规模化，但拥有更高比例的领域和技术专家。

你可能还记得，在2022年之前，像埃森哲（Accenture）或德勤（Deloitte）这样的公司，如果获得一份联邦合同（例如，负责一家国防机构的物流系统全面改造），它们会作为**主承包商**（Prime Contractor）中标。这意味着它们拥有客户关系，签署了价值3亿美元的主要合同，并部署数百名顾问进行实施。它们将SAP、Salesforce、Oracle等实际软件视为需要外包的子项目。

但在AI配置的模式下，合同首先会流向**Palantir**这样的技术提供商，然后埃森哲或德勤则成为**首选实施伙伴**（Preferred Implementation Partners），本质上是子承包商。这就是金字塔结构如何被颠倒过来。Palantir每个季度营收近12亿美元，利润率高达50%，这几乎是纯利润。在软件即服务（SaaS）领域，投资者使用“**40法则**”（Rule of 40）来衡量公司表现：增长率加上利润率，任何超过40的都算精英。Palantir的这一数字高达114，在同等规模的软件企业中几乎是无敌的。

再次强调，这种利润率差距的存在，是因为咨询业是一个服务型业务，其扩展是线性的。如果你想增加100万美元的收入，你就需要再雇佣三名顾问。每位顾问的全部成本（fully loaded）约为13万美元，你的成本会按比例增长。人越多，收入越多，支出也越多。但软件的扩展是指数级的。要增加100万美元的收入，你需要销售更多的软件。如果你的软件每年收费10万美元，卖出10套就能实现目标。新客户的边际成本约为1万至2万美元，这主要是云基础设施的费用。

那么，这对咨询行业的职业生涯意味着什么？我的一位好朋友曾给我看了一张描绘咨询行业现状的梗图，我认为它非常贴切。市场正在发生变化，这无疑对初入职场的年轻人来说是件好事，因为它将不再制造“轻松赚钱”的幻觉。我曾有与那些整个职业生涯都在咨询行业工作的人共事和学习的经历。我形成了一个印象，认为这是一种非常特定且狭窄的技能集，技能的可转移性很低，因为你将自己局限于与特定类型和规模的公司打交道，而你的整个职业生涯都围绕着处理该层级的问题。但在一个充满极端不确定性的市场中，一个需要广泛技能的市场里，这并没有多大帮助。

咨询曾是少数几家能在毕业后就提供10万美元起薪的精英职业之一。但客观地说，一种通用的技能组合不应如此昂贵。我们现在看到的是2026年的**入门级招聘崩溃**。咨询职位发布量下降，初级顾问招聘人数骤降54%。普华永道（PWC）计划到2028年削减30%的入门级职位。那些在社会标准下被认为是“令人称羡”的艰苦工作，正是最容易被自动化的。职业道路已经相当受损，一个20岁的年轻人可能在进入这个领域前需要三思。但另一方面，咨询业终于要回归它本应有的样子：精英的领域专业知识。

<details>
<summary>Original English</summary>
So
 what
 does
 this
 mean
 for
 a


career
 in
 consulting?
 My
 good
 friend


once
 showed
 me
 a
 meme
 that
 illustrates


the
 consulting
 industry
 pretty
 well
 in


my
 opinion.
 I
 think
 the
 fact
 that
 the


market
 is
 changing
 is
 definitely
 for
 the


better,
 especially
 for
 early
 career


folks
 because
 it
 will
 no
 longer
 create


an
 illusion
 of
 easy
 money.
 I've
 had


experience
 working
 and
 studying
 with


people
 whose
 entire
 career
 was
 built
 in


consulting.
 And
 I
 formed
 an
 impression


that
 it's
 a
 very
 specific
 and
 a
 very


narrow
 skill
 set
 with
 low


transferability
 of
 skills
 because
 you're


locking
 yourself
 into
 working
 with


companies
 of
 a
 very
 certain
 profile
 and


a
 certain
 size.
 And
 your
 entire
 career


is
 about
 dealing
 with
 the
 problems
 that


happen
 at
 that
 level.
 But
 if
 you're


working
 in
 a
 market
 with
 extreme


uncertainty,
 a
 market
 where
 you
 need
 a


wide
 variety
 of
 skills,
 that
 doesn't
 do


you
 any
 good.
 Consulting
 was
 once
 one
 of


a
 few
 elite
 careers
 in
 business
 that


could
 land
 you
 $100,000
 in
 starting


salary
 right
 out
 of
 school.
 But


objectively,
 a
 generic
 skill
 set
 should


not
 cost
 that
 much.
 What
 we're
 seeing
 in


26
 is
 the
 entry-level
 hiring
 collapse.


Consulting
 job
 postings
 are
 down.


Entry-level
 consultant
 hiring
 plunged


54%.
 PWC
 cutting
 30%
 of
 entry-
 level


roles
 by
 2028.
 The
 infamous
 and


prestigious
 by
 social
 standards
 grind
 is


exactly
 what
 gets
 automated.
 The
 career


path
 is
 fairly
 eroded
 and
 a
 20-year-old


should
 probably
 think
 twice
 before


entering
 the
 field.
 But
 on
 the
 flip


side,
 consulting
 is
 finally
 coming
 back


to
 what
 it
 was
 always
 supposed
 to
 be,
 an


elite
 domain
 expertise.
 We
 hope
 this
 was


helpful.
 Let
 us
 know
 what
 you
 think
 in


the
 comments.
 Till
 next
 time.
 Bye.
</details>