All Session Merged Yadav

BUSINESS MODEL INNOVATION
& HERNIAS, HEARTS AND DIAPERS

COURSE INTRODUCTION
PRASHANT YADAV
PRASHANT.YADAV@INSEAD.EDU
TRADITIONAL PATHS TO INNOVATION
New Technology
New Design
INTRO-2
COURSE THEME:
Process Innovation:
existing products with existing technologies to existing markets
Analyze and create superior business models that disrupt

industries, change lives and create wealth through better
processes!
(classical operations concepts needed for a tight grasp)
INTRO-3
WILL IT BE ENTERTAINING ?
INTRO-4 4
WILL THERE BE MATH?
MATH CREATES PRECISENESS OF THOUGHT AND BUILDS CHARACTER
INTRO-5
6
ABOUT ME- PRASHANT YADAV Twitter : Yadav_supplychn
Current roles
Academia- B-School
Academia- Medical School
Policy Think Tank
Board Roles
Advisory Board
Previous Roles
INTRO-6
CLASSROOM NORMS
• Rules of engagement
– Come prepared
– while always behaving tolerantly and with respect towards your peers
– and respecting the role of the professor to ultimately direct the discussion
• Contact
- Professor : Prashant Yadav (Prashant.Yadav@insead.edu)
- Assistant: Magalie NKIANI (Magalie.NKIANI@insead.edu)
- Tutor: Thomas BREUGEM (thomas.breugem@insead.edu)
INTRO-7
COVID-19 Code of Conduct Reminders
Strictly no food in the classroom
Drinks must be in closed containers
Wear mask at all times & properly – cover mouth & nose
Put mask back on after eating/drinking
Questions? Go to inse.ad/covid-19 / email prevent.coronavirus2019@insead.edu
Wear your Respect social Avoid physical Wash your Clean as Download contact
mask distancing contact hands you go tracing app
COURSE INFORMATION/POLICY
► Read the Outline. Full Information. Clearly Defined Rules. No

Surprises.
► Course Components
► Qualtrics Questions
► Not every class
► Counts towards participation
► Home-works – graded for effort
► Class participation – Good citizenship
► Exam
► Exams
► Class – Big Ideas and Intuition– Innovation Exercise
► Tutorials & Review Sheets – The analysis backing the Ideas, Application –
Analysis Exam
► Four review sessions available. Thomas BREUGEM
(thomas.breugem@insead.edu) is a very knowledgeable and helpful tutor
► Class + Tutorials are necessary to ace the exams
INTRO-9
Tools
User Manual Classic Exam Prep Geek Alert!
FOUR LEVERS TO MOVE RISK
WHO: DECISION MAKER WHEN: DECISION TIMING

Transfer to best Informed Delay Decisions
Transfer to who cares most Resequence
Transfer to who can bear risk best Split Decisions
MANAGING RISK
WHY: INCENTIVES WHAT: SUBSTANCE
Change Cash Streams Focused vs. Diversified
Align Time Horizons Real Options (Hedge/Complement)
Integrate Reduce/Increase # of Offerings
Flexible/Reactive
INTRO-11
THIS CLASS

MANAGING RISK
Flexible/Reactive
INTRO-12
SHOULDICE
WORD OF THE DAY: FOCUS
BUSINESS MODEL INNOVATION: FOCUS
Advantages Disadvantages
Leads to more efficient utilization of labor, capital Targets only fraction of market
Leads to specialization and perfection Competencies may not translate to other market
segments
Clear mission and goals for management &
employees Places natural limits and challenges on growth
Clear offering for customers Dependent on one market, risky, no diversification

in source of revenues
Low cost, high quality, enjoyable customer
experiences Limits frontline innovation
Challenges in finding, retaining and motivating

workforce
Focused strategies place constraints on growth, innovation and risk-diversification
INTRO-15
SHOULDICE HOSPITAL, NARAYANA HRUDAYALAYA
INTRO-16
PRODUCT PROCESS MATRIX
Choice of Product/Market
High Product Variety Standardized Product
Low Volume/Pdt High Volume/Pdt
Flexible Process
(Functional/ Job Shop)
Choice of Process
High Unit Cost

General Hospital
Rigid Well Defined

Process
(Assembly Line) Shouldice
Low Unit Costs
Operational Efficiencies
can be derived from rigid,
well Defined Processes
Matching the Product and Process is Vital
INTRO-17
PRODUCT PROCESS MATRIX: FOOD INDUSTRY
Highly Customized
Product Standardized Product
Low Volume High Volume
Job shop type kitchen.

Each dish is made to
order
Choice of Process
Each dish is slightly

customized
Assembly Line
Production
INTRO-18
PRODUCT PROCESS MATRIX: AIRLINES
Multiple service Single service class

classes
Multiple in-flight
procedures, and
aircraft types
Choice of Process
Single in-flight
procedure and
aircraft type
INTRO-19
THE COMMON THREAD?
INTRO-20
KEY LESSONS
► Choosing What your Business Model does is a key lever for Innovation
► Focused vs. Diversified
► Focused strategies deliver superior operational performance (lower variety risks)
► Focused strategies expose firms to market risk
► Risk Driven Business Models have the potential to change the world
MANAGING RISK
Focused vs. Diversified
Reduce/Increase # of Offerings
INTRO-21
REALIZING BUSINESS MODEL INNOVATIONS: WHAT-FOCUS
QUESTIONS FOR YOU TO REFLECT ON
Pick your favorite Industry-Market
Are there focused & diversified players in this industry?
If yes, Who are they and what is the relative profitability?

What processes do they employ?
What are the key perceived differences between the players?
If no, What are the challenges to pursuing a focused strategy in this industry?
What are the operational consequences of focus?
Is there an opportunity for a focused player to enter the market?
INTRO-22
DIAGNOSING BUSINESS PROCESSES
BOTTLENECKS & CAPACITY MANAGEMENT
WHAT WE LEARN FROM THE GOAL
PRASHANT YADAV
INNOVATION EXERCISE
INNOVATION EXERCISE
(Must be submitted at the course website before the day of the final exam. This is a group project. Please, work in your
MBA pre-assigned groupings).
Provide one example of a setting where one of the innovation concepts discussed
in class can be used to build a business model that may have, or has already had, a
game changing, disruptive impact. Opportunities, which are novel, not mentioned
in the class or in the book and not yet implemented are encouraged. State clearly
the context, provide the traditional way of doing business, describe the innovation
you have in mind and analyze how this could be game changing. You can use any of
the cases or examples used in class to provide a comparison or reference. Your
total answer should be < 2500 words. A precise, focused analysis is preferred. You
don’t need to do any financial or market analysis, but feel free to include it if it
helps your case, or if you are planning on pursuing this idea further (perhaps for
the IVC).
THE GOAL: 2-2

A JOURNEY THROUGH TIME: AUTOMOBILES AND TRANSPORTATION
HORSELESS CARRIAGE
THE GOAL: 2-4

WAVE #1: FORD "Any customer
can have a car
painted any color
that he wants so
long as it is black."
Disruptive Cost
Reduction
Assembly Line
Welfare Capitalism Vertical Integration

THE GOAL: 2-5
MODEL-T AND CHOICE OF COLOR
THE GOAL: 2-6

HENRY FORD AND FULLY INTEGRATED SUPPLY CHAINS
Retailer
Parts
Manufacturing WH Consumer
Manufacturing
WH
Retailer
Logistics
Sales
Marketing
Vertical
Integration
Firm boundary was drawn to cover as many activities as possible
Ford Rouge plant
THE GOAL: 2-7

RIVER ROUGE COMPLEX DEARBORN, MI CIRCA 1950
Source: Detroit News/ Sent to me by Joseph Butler ZLOG Class of 2006
THE GOAL: 2-8 8

COMMUNICATIONS WAS EXPENSIVE
(AND TRANSPORT WAS SLOW)
Source: Presentation by G. Long, UPS
THE GOAL: 2-9 9

HENRY FORD’S LEGACY
Henry Ford (July 30, 1863 – April 7, 1947) was an American industrialist, the founder of the Ford Motor Company, and
sponsor of the development of the assembly line technique of mass production. Although Ford did not invent the
automobile or the assembly line, he developed and manufactured the first automobile that many middle class
Americans could afford. In doing so, Ford converted the automobile from an expensive curiosity into a practical
conveyance that would profoundly impact the landscape of the twentieth century. His introduction of the Model T
automobile revolutionized transportation and American industry. As owner of the Ford Motor Company, he became one
of the richest and best-known people in the world. He is credited with "Fordism": mass production of inexpensive
goods coupled with high wages for workers. Ford had a global vision, with consumerism as the key to peace. His
intense commitment to systematically lowering costs resulted in many technical and business innovations, including
a franchise system that put dealerships throughout most of North America and in major cities on six continents. Ford left
most of his vast wealth to the Ford Foundation and arranged for his family to control the company permanently .
THE GOAL: 2-10
THE TOYOTA WAY…
THE GOAL: 2-11

PRODUCT VARIANTS--TOO FEW, TOO MANY?
THE GOAL: 2-12 12

WAVE #2: TOYOTA*
Managing for the Suppliers: From Competition to

Long Run Collaboration
Disruptive Quality
Improvements
Continuous
Just In Time
Improvement
Session 9: WHY Relationships? * Extra: Read “Standing on the Shoulders of

THE GOAL: 2-13 Giants” from The Goal, p. 339-362
PRODUCT VARIANTS--TOO FEW, TOO MANY?
THE GOAL: 2-14

Supply chain cost as a fraction of overall cost
70% 30%
Suppliers Car Manufacturer
Adapted from: John Kay - NSF Workshop

COST IN THE AVERAGE NEW VEHICLE
on?SCM, University of Florida
THE GOAL: 2-15

WHAT’S NEXT FOR TRANSPORTATION?
THE NEWEST TRANSPORTATION REVOLUTION
Platform to Match Supply with Demand

THE GOAL: 2-17
THE COMMON THREAD?
THE GOAL: 2-18

WHAT IS THE GOAL?
► ROIC = (throughput – operating expense) / inventory
THE GOAL: 2-20

THE (TRADITIONAL) GOAL OF AN ENTERPRISE
Maximize Increase
Shareholder Earnings Sales – Operating Expenses – Capital Expenses
Value
THROUGHPUT: The Rate at which the System Generates Revenue/Sales

►Production is not Revenue; Revenue = Min {Production, Demand}
►Capacity Utilization is not the Goal, only a possible means to achieve the goal
OPERATING EXPENSE: The Rate at which the System Generates Costs
INVENTORY: The level of capital invested in system

►It costs money to make money.
►Money costs money

► Debt: Interest, shareholder risk increases
► Equity: Dilution of shareholding
$$ Going in the System $$ Coming from the System

= Throughput $$ Sitting in System = Inventory = Operating Expense
THE GOAL: 2-21

FROM OPERATIONAL MEASURES TO FINANCIAL MEASURES
Net Operating Profit after Taxes (NOPAT) = Throughput – Operating Expenses †
Economic Value Added (EVA) = NOPAT - Weighted Average Cost of Capital * Inventory
Margin Capital or
Inventory Turns
† Operating Expenses include taxes; Alternately Throughput- Operating expense = Operating Profit
Assuming no change in inventories
THE GOAL: 2-22

BUSINESS MODELS AND INEFFICIENCIES
TWO KINDS OF INEFFICIENCIES

• INFORMATION BLINDSPOTS: DECISIONS MADE WITH POOR INFORMATION
• INCENTIVE MISALIGNMENT: DECISIONS MADE WITH “BAD” INTENTIONS
THE GOAL: 2-23

WHAT IS THIS SYSTEM’S THROUGHPUT?
Step I Step II Step III
Service Time 2 Minutes/Unit 3 Minutes/Unit 1 Minute/Unit
Capacity
Bottleneck
Bottleneck Capacity
System Capacity
Utilization at Max
Capacity
Where should additional resources be allocated?

How should the work be re-allocated?
Assume Sufficient Demand
THE GOAL: 2-24

IDENTIFY THE BOTTLENECK (ANSWER)
Step I Step II Step III
Capacity 30 Units/hr 20 Units/hr 60 Units/hr
Bottleneck Step II Bottleneck = Step with Lowest Capacity
Bottleneck Capacity 20 Units/hr
System Capacity 20 Units/hr System Capacity = Bottleneck Capacity
Utilization at Max
20/30= 66.67% 20/20=100% 20/60=33.33%
Capacity
Benefit of additional
0 + 0
Resources
Additional Resources should be allocated at the Bottleneck Step

Work should be re-allocated from the bottleneck step to other steps
THE GOAL: 2-25

IDENTIFY THE BOTTLENECK: QUALITY LOSSES
Step I Quality Step II Step III

Control
60%
40%
Capacity
Bottleneck
System Capacity
Utilization at Max
Capacity
What is the System Capacity? Assume Sufficient Demand

Very Quick Quality Inspection
Step II and III have no quality issues
THE GOAL: 2-26

IDENTIFY THE BOTTLENECK: QUALITY LOSSES
Step I Quality Step II Step III

Control
60%
40%
Capacity 30 Units/hr 20 Units/hr 60 Units/hr
18 Good Units/hr 20 Good Units/hr
Bottleneck Step I
System Capacity 18 Units/hr
Utilization at Max
30/30= 100% 18/20=90% 18/60=30%
Capacity
Step I become the bottleneck, System Capacity = 18 Good Units/hr

Very Quick Quality Inspection
Step II and III have no quality issues
THE GOAL: 2-27

MANAGING BOTTLENECKS: WHAT WOULD JONAH DO?
Bottlenecks Drive System Capacity
► Finding The Bottleneck

► Symptoms
► The resource with minimum capacity
► The resource which is always busy
► The resource with longest wait
► The resource with most inventory
► Improving Bottleneck Performance

► Investments in improving bottleneck capacity improve system capacity
► Lost capacity at bottleneck is lost cost capacity for whole system
► Always utilize the bottleneck fully
► Some strategies
► Offload work to non-bottleneck steps (contract out steps, etc.)
► Create inventory buffer before bottleneck to avoid starving
► Quality check before bottleneck
► Increase capacity by decreasing number of batches; Reduce setups (more on batches
later)
THE GOAL: 2-28

COVID CAPSULE
BOTTLENECKS AND COVID-19

RESPONSE
COVID-19 TESTING
PPE
“It’s a supply chain which has

dynamically shifting bottlenecks,
and the administration is trying to
overcome them one at a time, as
they pop up. What we need is a far
more comprehensive approach,“
said Prashant Yadav
THE GOAL: 2-29

WHERE DO YOU THINK IS THE
BOTTLENECK?
IS IT STATIC?
THE GOAL: 2-30

INVENTORY MANAGEMENT
► Beyond Financial Costs, inventory in system is costly in terms of

► Quality problems
► Design changes
► Responsiveness to market conditions
► While bottlenecks need some inventory to avoid starving, minimize inventory at non-
bottlenecks
► Drum and rope scheduling to let the bottleneck set the pace of the line
► Reduce batch sizes at non-bottlenecks (but not so much that they become a bottleneck)
(more on batches later)
► Where to place the bottleneck? Cost vs. Management Complexity

► Early in the process simplifies flow management
► Where capacity is costliest
THE GOAL: 2-31

BATCH PROCESSES WITH SETUP TIME: BAKING COOKIES
► Baking Cookies requires preparation of each individual cookie and then baking them in
the oven.
► Preparing each cookie takes some time, say p minutes per cookie
► But, the cookies must all also be baked. Irrespective of the number of cookies it takes the same
amount of time to bake the cookies♪, say B minutes every cycle.
► How many cookies should be baked at once ?
► Many Industrial Processes share these characteristics

► The heat treatment step from the Goal
► Most pharma manufacturing is batch manufacturing (including all COVID-19 vaccines)
► Also applies when cost structure is similar to time: Ordering units with fixed order costs
How many cookies should be baked at one go?
♪ This is an assumption which is accurate for reasonable numbers of cookies (that can fit in a commercial oven).
THE GOAL: 2-32

WHAT WOULD JONAH DO?
INCREASE THROUGHPUT (CAPACITY) REDUCE CAPITAL COSTS (INVENTORY)
Q cookies are produced every B+Q.p minutes At any point in time, there are Q cookies that are
work in progress inventory.
Capacity =
Inventory Costs will increase linearly in Q
250
1600
Unit Capital Costs ($)

200
1200
Capacity
150
800
100
50 400
0 0
0 500 1000 1500 0 500 1000 1500 2000
Batch Size, Q
Batch Size, Q
Throughput Increases with Batch Size, but so do Inventory Costs !!
THE GOAL: 2-33

BATCH SIZE: PRESCRIPTIONS
► At bottleneck steps; goal is to increase Utilization/Throughput/Capacity. An increase in

bottleneck capacity is an increase in system capacity, so capacity increase takes
priority.
► At non-bottleneck steps an increase in capacity is not helpful at all! So focus is on

reducing costs.
At Bottleneck Steps, use Large Batch Sizes
At Non-Bottleneck Steps, use Smaller Batch Sizes
THE FINE PRINT

► This analysis assumes that costs (operating expenses) do not depend on batch size
► These prescriptions hold within reasonable ranges of values that are feasible for the process technology in use
THE GOAL: 2-34

KEY LESSONS
► The Goal of an enterprise is to make money and generate social value. Operationally
that corresponds to increasing Throughput (rate of sales), decreasing Operating
Expenses and Inventory (capital costs)
► The Bottleneck Step determines the performance of the whole process

► Step with lowest capacity is the bottleneck
► Increase Capacity of the bottleneck by reducing its work, adding more resources, etc.
► Minimize inventory at non-bottleneck steps
► In processes with Setup Times and Batches

► Choose larger batch sizes for bottleneck steps
► Choose smaller batch sizes for non-bottleneck step
► A general approach to Manage Change

► Identify the goal
► Identify the obstacles to achieve it (bottlenecks)
► Manage the bottlenecks
THE GOAL: 2-35

ROLE OF FLEXIBILITY
&
REAL OPTIONS
OUR BUSINESS MODEL INNOVATION TOOLKIT- SO FAR
MANAGING RISK
Real Options (Hedge/Complement)
Flexible/Reactive
1. Information Risks vs. Product/Market Risks
2. Which process step to focus the most on?

SESSION 3- COVID-19 VACCINES AND FLEXIBILITY
DECISION PROBLEM FACED BY GOVERNMENTS C SEP 2020
mRNA mRNA Curevac DNA Innovio

Pfizer/BioNTech
AdV Oxford/AZ
WHICH ONES SHOULD
I CONTRACT FOR
ADVANCE PURCHASE?
AdV J&J HOW MUCH?
SHOULD I GIVE
MONEY TO ENHANCE
AdV Sinopharm
PRODUCTION
CAPACITY?
Recomb Protein TO WHICH ONES?

GSK+Sanofi
AdV Gamaleya
Inactivated Inactivated
virus Sinovac virus BBIL
WHAT INFORMATION DO YOU NEED TO ANSWER THAT QUESTION?

PROBABILITY OF SUCCESS & TIME TO APPROVAL
Probability of
success (At least
Platform 1 approved)
Live-attenuated 0.0%
Protein subunit 86.6%
Inactivated 54.7%
RNA 74.0%
Non-replicating viral vector 53.8%
Replicating viral vector 18.8%
DNA 7.1%
Other 0.03%
All platforms 99.6%
Source: McDonnell, A., Van Exan, R., Lloyd, S., Subramanian, L., Chalkidou, K., La Porta, A., . . . Yadav, P. (2020, October).
COVID-19 Vaccine Predictions: Using Mathematical Modelling and Expert Opinions to Estimate Timelines and Probabilities of
Success of COVID-19 Vaccines

WHAT MORE INFORMATION DO YOU NEED?

PROBABILITY OF SUCCESS CORRELATION MATRIX
POS Corr LAV Inactivated mRNA ..
LAV
Inactivated
mRNA
..
Platform Correlation
Live attenuated virus Medium
Protein subunit+ adjuvant Medium
Inactivated virus Medium
mRNA Low
Non-replicating adenoviral vector High
High
Replicating adenovirala
DNA Medium
Other vaccines None
Source: Anthony McDonnell, Robert Van Exan, Steve Lloyd, Laura Subramanian, Kalipso
Chalkidou, Adrian La Porta, Jiabin Li, Eddine Maiza, David Reader, Julie Rosenberg, Jack
Scannell, Vaughan Thomas, Rebecca Weintraub and VPrashant
SESSION 3- COVID-19 Yadav. 2020
ACCINES AND FLEXIBILITY
FOR WHICH ONES SHOULD I BUILD MORE MANUFACTURING CAPACITY?
REMEMBER WE ARE IN SEP 2020

COVID-19 vaccine supply chain
Uncertainty about product type and

Vials
quantity allocated over time
Also makes planning for ancillary

Antigen
Fill &
supplies more challenging e.g. syringes
manufacturing
Finish
Excipients, Lipids,
Adjuvants
manufacturing
Syringes
Other
materials for
cold chain?
(e.g. dry ice)

Adapted from a graphic in Essential Medicines Supply Chain Disruptions by Anthony McDonnell, Subramanian, Kalipso Chalkidou, and Prashant Yadav. 2020
DEDICATED PRODUCTION / NO FLEXIBILITY
Vaccine
What can wrong with this
manufacturing network?
When would it be OK to
manufacture with this network?

NO FLEXIBILITY VS TOTAL FLEXIBILITY
Vaccine Vaccine
Is it feasible?
How expensive to build such flexibility?
NO FLEXIBILITY VS TOTAL FLEXIBILITY
Vaccine Vaccine
IS DEMAND FOR VACCINE A INDEPENDENT OF THE DEMAND FOR

VACCINE B?
MANUFACTURING CAPACITY FLEXIBILITY IN C-19 VX MANUFACTURING
Antigen/Drug Substance Fill & Finish

Manufacturing Manufacturing
• Highly platform dependent manufacturing

• “Reasonably” fungible/interchangeable across Vx candidates
steps
• Capacity depends on packaging type (glass vial, BFS, etc) and #
of doses/vial etc.
• Limited fungibility across Vx platforms

13
FLEXIBILITY AND COVID-19 VACCINE MANUFACTURING
Production of drug substance and manufacture of drug product for an

RNA vaccine
Production of drug substance and manufacture of drug

product for replicating and non-replicating viral vaccines

CYTIVA
https://www.youtube.com/watch?v=jk_9UibqyDQ
UNIVERCELLS
https://www.youtube.com/watch?v=uadVBWOZxH4

CHAINING AND FLEXIBILITY
► A chain is a group of plants and products connected via links.
► Under some conditions, flexibility is most effective if it is added to create long chains.
► A configuration with 20 links can produce nearly the results of total flexibility as long
as it constructs one large chain:
► Hence, a little bit of flexibility is very useful as long as it is designed correctly
Vaccine Vaccine
Vaccine Vaccine
< << ≈

THE VALUE OF FLEXIBILITY
► Adding flexibility increases capacity utilization, expected sales (and social welfare):
1000
Total flexibility
20 links
950
Expected sales, units
These data are

12 links collected via simulation
900
11 links
850
No flexibility
800
80 85 90 95 100
Expected capacity utilization, %

Does horizontal
collaboration in
manufacturing
solve the capacity
flexibility problem?

HOW MUCH WOULD YOU BE WILLING TO PAY FOR FLEXIBILITY?
► Flexibility comes at higher initial investment/capital expenditure

► Flexibility may also have higher operating costs
► Depends on future demand uncertainty. What factors?

REAL OPTIONS AS A TOOL FOR UNDERSTANDING AND VALUING STRATEGIC FLEXIBILITY?
► Traditional discounted cashflow approaches (such as the NPV rule) cannot

properly capture management’s flexibility to adapt and revise later decisions in
response to unexpected market developments.
► Traditional DCF assumes
► an expected/average scenario of cashflows
► Static operating strategy (no changes as informational evolves)
► In reality, as new information is revelead and uncertainty about market

conditions is resolved, management can alter its operating strategy.
► Strategic Options, or Real Options.
► Source of value in an option: The asymmetry from having the right but not
the obligation to exercise the option.
20
To wait before taking decision on capacity until more is known
Defer or timing is expected to be more favorable
To increase or decrease capacity in response to demand

Expand
Abandon To discontinue production capacity and liquidate the assets
Change product To alter the mix of inputs or outputs of a production process in

mix response to market prices

PANDEMIC INFLUENZA VACCINE MANUFACTURING
AND REAL OPTIONS
Chen, JR., Liu, YM., Tseng, YC. et al. Better influenza vaccines: an industry perspective. J Biomed Sci 27, 33 (2020).
https://doi.org/10.1186/s12929-020-0626-6
EXPECTED DCF UNDERVALUES INVESTING IN PANDEMIC INFLUENZA VACCINE MANUFACTURING

WORD OF THE DAY:
FLEXIBILITY
KEY LESSONS
► With information risks, flexibility (and the way you make

decisions about it) can be a powerful source of Business Model
Innovation

INNOVATION TOOLKIT
MANAGING RISK
Flexible/Reactive

NEXT TIME
► We will look at more kinds of Risk and how it impacts Business Models
► Read Note on Queueing Models.

► Group HW1 due
► Improvement & Analysis to Innovation….

COVID-19 vaccines have reminded us of the need for “equitable” global
distribution.
XENON OPERATING STRATEGY
VARIABILITY & QUEUING

EMERGENCY ROOM CROWDING AND AMBULANCE DIVERSION
Pictures from various Newspapers; put together by L. Green
XENON-2
EMERGENCY ROOMS AROUND THE WORLD
Brazil – “National Medical Council Canada – "Canada has longest Emergency Room
‘declares’ wait time of at most 120min” wait times among top 11 developed nations"
France – “It’s not right to have to wait six hours

UK – “waiting times in England are worst
in the emergency room, but it has become
for a decade”
common”
XENON-3
WAITING TIMES IN THE EMERGENCY ROOM
Waiting Time Service Time
Total Time in the System
XENON-4
A SOMEWHAT ODD SERVICE PROCESS
Arrival Inter-Arrival Service Gantt Chart illustrating Service
Patient
Time Time Time
1 7:00 4
2 7:05 5 4
3 7:10 5 4
4 7:15 5 4
5 7:20 5 4
6 7:25 5 4
7 7:30 5 4
8 7:35 5 4
9 7:40 5 4
10 7:45 5 4
11 7:50 5 4
12 7:55 5 4
7:00 7:10 7:20 7:30 7:40 7:50 8:00
Average Inter-Arrival Time is 5 minutes, Average Service Time is 4 minutes. No Waits!
What is odd about this Service Process?
XENON-5
A MORE REALISTIC SERVICE PROCESS
Arrival Inter-Arrival Service
Patient Patient 1 Patient 3 Patient 5 Patient 7 Patient 9 Patient 11
Time Time Time
1 7:00 5 Patient 2 Patient 4 Patient 6 Patient 8 Patient 10 Patient 12
2 7:07 7 6
Time
3 7:09 2 7
4 7:12 3 6 7:00 7:10 7:20 7:30 7:40 7:50 8:00

4
5 7:18 6 5
6 7:22 4 2 3
7 7:25 3 4
2 2 2
2
8 7:30 5 3
Number of cases
9 7:36 6 4
1 1
1
10 7:45 9 2
11 7:51 6 2
0
2 min. 3 min. 4 min. 5 min. 6 min. 7 min.
12 7:55 4 2
Service times
Average Inter-Arrival Time is 5 minutes, Average Service Time is 4 minutes.
Will the service performance be the same as before?
XENON-6
VARIABILITY LEADS TO WAITING TIME
Arrival Inter-Arrival Service
Patient
Time Time Time
1 7:00 5
2 7:07 7 6
Service time
3 7:09 2 7
4 7:12 3 6
5 7:18 6 5
6 7:22 4 2
7 7:25 3 4
8 7:30 5 3
Wait time
9 7:36 6 4
10 7:45 9 2
11 7:51 6 2
12 7:55 4 2
7:00 7:10 7:20 7:30 7:40 7:50 8:00
3
Variability is Evil!
2
1
Inventory
(Patients at lab)
0
7:00 7:10 7:20 7:30 7:40 7:50 8:00
XENON-7
https://www.youtube.com/watch?v=mygmoUzjrB4
XENON-8
PERFORMANCE
► Throughput
► Cycle time
► Utilization
► Customer service
► Best possible Performance

► Maximum throughput
► Minimum cycle time
► 100% utilization of all tools
► 100% customer service
XENON-9 9
REMEMBER!
► Variability is EVIL
► Variability is the norm not the exception
XENON-10
VARIABILITY OR RANDOMNESS
Sometimes we fail to differentiate between Variability and

Randomness.
Variability can be due to Randomness or due to Bad Control
XENON-11 11
XENON-12
Average Service Time

𝜏 (hours/patient)
Average
Arrival Rate
XENON-13
Average Waiting Time Average Service Time

Average
Arrival Rate
Avg. # of Patients in Queue
XENON-14
Average Waiting Time Average Service Time

Average
Arrival Rate
Avg. # of Patients in Queue

Utilization
Little’s Law
XENON-15
WHAT DOES CV (COEFFICIENT OF VARIATION) MEAN?
A PROCESS WITH CV=0

3 mins 3 mins 3 mins 3 mins 3 mins
Perfectly Scheduled Arrivals, Output from
an assembly line.
9:55 9:58 10:01 10:04 10:07 10:10
A PROCESS WITH CV=1

No relationship between arrivals,
Independent Arrivals. 3 mins 2 mins 4 mins 2.5 mins 3.5mins
Inter-arrival times are exponential, Arrivals
according to a Poisson Process, Memory- 9:55 9:58 10:00 10:04 10:07 10:10
less Arrivals.
A PROCESS WITH CV>>1
Bunched Arrivals
“Fat-Tail” inter-arrival times
10:01 10:10
XENON-16
THE WAITING TIME FORMULA (APPROXIMATION FOR SINGLE SERVER)
    CVa2 + CVs2 
Wq =    
1−    2 
Variability effect
Utilization effect
Scale effect
Average flow
time Wq+ 
Increasing
Variability
Theoretical Flow Time
0% Utilization 100%
XENON-17
MECHANICS: A SINGLE-SERVER EXAMPLE
General Single Server approach Emergency Operating Room Example
1. What are the parameters? 1. ER Data
1. Find 2. Utilization
2. Find from formula 3. Waiting time in queue
XENON-18
HOW TO ADD EXTRA CAPACITY?
AVG. WAIT TIME
2 SINGLE SERVER SYSTEMS

AVG. LENGTH OF EACH QUEUE
TOTAL WAITING CUSTOMERS
UTILIZATION
A 2 SERVER SYSTEM
TOTAL WAITING CUSTOMERS
AVG. WAIT TIME
XENON-19
LOOKING UP THE TABLE FOR MULTIPLE SERVERS (APPENDIX 1)
LOOKUP S, # OF SERVERS
ON HORIZONTAL AXIS
LOOKUP ρ, COMPUTED
UTILIZATION, TO GET LQ
XENON-20
THE POWER OF POOLING Waiting
Time Tq
70.00
2 SINGLE SERVER SYSTEMS 2X (S=1) s=1
60.00
50.00
40.00
s=2
30.00
20.00
Waiting Time for Patients is 27 Hours* s=5
10.00 s=10
0.00
60% 65% 70% 75% 80% 85% 90% 95%
A 2 SERVER SYSTEM S=2
Utilization, ρ
Implications:
+ Balanced utilization
+ Shorter waiting time (pooled safety capacity)
Waiting Time for Patients is 12.78 Hours* - Changeovers and set-ups
Pooled Systems perform better!

* For consistent comparison demand (arrival rate) is double
of single server cases
XENON-21
THE CUSTOMER CONTACT CENTER INDUSTRY
MASSIVE OUTSOURCED CALL CENTERS
Service Economics Human Resources
Wide variability in demand during a shift Labor intensive: low cost countries
Shifts are Staffed to manage peak demand Relatively low talent labor
Typical utilization ~ 30%
XENON-22
THE LIVEOPS BUSINESS MODEL
The LiveOps Business Model
Interested Mompreneurs register and do basic on-line

training. Mompreneurs when available, wait for calls (at
home)
Received Calls are directed to highest Ranking Agent that is What is the utilization of an agent at LiveOps?
available
Mompreneur is paid for the time on the call Who would work at LiveOps?
“On Demand Contact Centers: LiveOps Inc.”, INSEAD Case , 2011.
XENON-23
THE LIVEOPS BUSINESS MODEL: RESEQUENCING
TRADITIONAL CALL CENTER
Employee Selection Client sign-up Staffing decisions Incoming calls
Client sign-up Incoming calls Employee Selection Staffing decisions
LIVEOPS BUSINESS MODEL
XENON-24
LIVEOPS: THE SUCCESS STORIES
HTTPS://WWW.YOUTUBE.COM/WATCH?V=VZE8JL-IGAU
Read more about customer success stories here
XENON-25
LiveOps reduces the risk of demand/supply (calls/agents) mismatch

“On Demand Contact Centers: LiveOps Inc.”, INSEAD Case , 2011.
XENON-26
INNOVATING INVENTION: R&D UPSIDE DOWN
Traditional model
Selection amongst Investment in Revelation of

candidates development performance
Investment in Revelation of Selection amongst

development performance candidates
Tournaments model
XENON-27
HOW MICHAEL DELL REVOLUTIONIZED THE PC INDUSTRY
HP: Build To Forecast Dell: Build To Order
Manufacture Demand Demand Manufacture
XENON-28
THE COMMON THREAD?
XENON-29
THE COMMON THREAD: RESEQUENCING WHEN DECISIONS ARE MADE
DECIDE AFTER INFORMATION

TRADITIONAL TIME-LINE
Assortment, Pricing Production Sales Markdowns Receivables
NEW TIME-LINE
Component Sales Production Markdowns Receivables

Assortment, Pricing
THE INNOVATION WORKS BEST WHEN

► Change sequence of events– Bets (Production, ► High Mismatch Costs (Demand Uncertainty)
Hiring) and Information (Demand Information) ► Modular Product, Desire for Variety
► Changes Exposure to Sales Uncertainty Risks, ► Limited benefits of commitment
Increase Costs
XENON-30
WORDS OF THE DAY
► Variability is EVIL
► Pooling helps to deal with variability (with some

caveats)
► Resequencing can work even better
XENON-31
REALIZING BUSINESS MODEL INNOVATIONS: TEMPLATE WHEN-RESEQUENCING
METHOD I: TRANSFER TO OTHER INDUSTRY-MARKET

Pick your favorite Industry-Market
What are the consequential decisions made in this industry?

Are they made with perfect, ok or poor information (information risk)?
When are these decisions made?

Map out important events before and after these decisions.
Does the critical decision have to be made at the time it is made– Can it be resequenced?
If yes, you have a game changing idea…
METHOD II: RESOLVE PAIN POINTS OF AN INDUSTRY
Think of industries characterized by high Information Risk.

• Frequent dramatic departures from budgeted performance metrics (sales, resource utilization, etc.);
• Wide variations in year-to-year performance;
• High exposure to prices and actions out of the control of the firm (for instance, energy prices, partners’ behavior);
• Vulnerability of business performance to a few high-impact decisions subject to significant uncertainty;
• Expensive, frequently underutilized assets;
What are the critical decisions in these industries?
Can they be re-sequenced?
XENON-32
INNOVATION TOOLKIT

Delay Decisions
Resequence
MANAGING RISK
Flexible/Reactive
XENON-33
NEXT TIME
► How do you make resequencing work?
► Read TIMBUKT2 case
ALSO, in the meanwhile

► Do additional reading/review of queueing models
► Reflect on Tutorial session #1 and see if you have internalized the concepts
XENON-34
BUSINESS MODEL INNOVATION IN ACTION
HOW TO RESEQUENCE?
FOUR LEVERS TO MOVE RISK

MANAGING RISK
Flexible/Reactive
MASS CUSTOMIZATION-2 INSEAD

TODAY’S SESSION
BUSINESS MODEL INNOVATION IN ACTION

HOW TO RESEQUENCE?
BUILD TO ORDER & BUILD TO FORECAST

SPACKLING
DELAYED DIFFERENTIATION
MASS CUSTOMIZATION
WHEN DOES EACH MAKE SENSE, WHAT ARE THE TRADEOFFS?

THE COMMON THREAD: RESEQUENCING WHEN DECISIONS ARE MADE
DECIDE AFTER INFORMATION

NEW TIME-LINE
Component Sales Production Markdowns Receivables

Assortment, Pricing
• Changes the uncertainty your are facing (components vs mix)

• Aggregate instead of granular forecasts

THE POWER OF CUSTOMIZED PRODUCTS
Customized Products enjoy a Significant Premium

THE BENEFITS OF CUSTOMIZATION
► Customization allows a customer to more closely match her wants to the available products. This
can be used to extract a premium
► Customers don’t care about all dimensions equally. A technique to identify which dimensions are
most important to customers is called conjoint analysis
► Conjoint Analysis surveys people using appropriately designed choices to find out what they really
care about
The “Most Wanted” painting in America, according to Conjoint Analysis conducted by Komar and Melamid.
See Komar and Melamid’s website on paintings based on attribute surveys: http://www.diacenter.org/km

TIMBUK2: AN EXAMPLE OF PRODUCTION TO ORDER
www.Timbuk2.com

RECALL THE DELL EXAMPLE
HP: Build To Forecast Dell: Build To Order
Manufacture Demand Demand Manufacture

No Finished Goods Inventory!
Source: Dell
MASS CUSTOMIZATION-9 9
INSEAD
LOWER INVENTORY: OTHER ADVANTAGES
“If I’ve got 11 days of inventory and my competitor has 80, and Intel comes out with a new
… chip, that means I’m going to get to market 69 days sooner” - Michael Dell
Inventory Velocity ?
INSEAD
NO FG INVENTORY and CUSTOMER EXPERIENCE
Merge in Transit
Anticipated Customer Order from Valencia

Monitor merge date Committed a Delivery Date
Supplier Production
Release date
DELL
Limerick Plant
A A A
A B C
Merge in Transit Facility

Girona Shipment to Valencia
Keyboard
Supplier
Source: Prashant Yadav and Richard Pibernik, 2006
INSEAD
SHOULD WE HAVE FINISHED GOODS INVENTORY OR NOT ?
Build To Forecast Build To Order
Final
Demand Demand Final Product
Product
Ex. Airplanes
Ex. Ready to wear clothes
Tailored Suits
Fast Food
Gourmet Restaurant
Buffet Table
Dell Computers
Circored
A La Carte
Goal
Services

TO HAVE FINISHED GOODS INVENTORY OR NOT?
Build To Forecast (BTF)

► Systems that have Finished Goods Inventory (FGI) are called Build To Forecast (or
Make to Stock) systems.
► In such systems we produce well in advance of customer arrival based on our forecast
of demand (Think: Buffet Table). When a customer shows up, we can just supply him
from the Finished goods inventory, and the customer doesn’t have to wait. The
downside is that we have wasted inventory, and we have limited ability to customize
the product to individual taste.
Build To Order (BTO)

► Systems where no finished goods inventory is allowed are called are called Build To
Order (or Make to Order) Systems.
► In such systems, production is only initiated after a customer arrives, so she has to
wait for as long as it takes to produce the product (including waiting time). (Think: A la
Carte). The main upsides are no wasted inventory, possibility of customization.

BUILD TO ORDER V/S BUILD TO FORECAST
Product
Wait Waste
Variety
BTF (FGI) Shorter (Good) Small (Bad) High (Bad)
BTO (no FGI) Longer (Bad) More (Good) Low (Good)
What does the market dictate?
Customers prefer more and more customized products.

Increased Variety is a very important competitive weapon.
One needs accurate forecasts for build to stock.
Lost Sales in BTF are also a kind of waste (Mismatch between Supply & Demand)

BTO V/S BTF. WHEN DOES BTO MAKE SENSE?
► Build To Order is ideal when …
► Customers are patient (willing to wait to receive their order)
► Assembly must be fast, easy and cheap:

► Most likely with a modular product design.
► Customers have a strong preference for variety.
► Inventory is expensive to hold

► High carrying costs, short product life cycles, component price deflation.
► Cannot be too expensive to deliver individual units
► Market forecasts are very inaccurate

SPACKLING AT TIMBUK2 USING FLEXIBLE CAPACITY
Average Demand for Custom + Std

200
Production Volume
150
Standard products (BTF)
100
Demand for Custom
50
Custom products (BTO)
0
0 5 10 15 20 25 30
Day
A combination of Customized and Standard products smooths production levels
Cattani, K., E. Dahan, and G. Schmidt. 2007. Spackling: Smoothing Built-to-Order Production of Mass-Customized Products with Build-To-Forecast production of Standard Items..

THE PRODUCT PROCESS MATRIX REVISITED…
High Product Variety Standardized Product
Low Volume/Pdt High Volume/Pdt
Flexible Process
(Functional/
Job Shop)
High Unit Cost
Choice of Process
Rigid Well Defined

Process
(Assembly Line)
Low Unit Costs

LUCENT’S TRES SANTOS EXPERIENCE W BTS AND BTO
Push Pull Boundary
Order Penetration
Point
Source: Lee, Hoydt and Lopez-Tello
INSEAD
ASSORTMENT CHOICES: THE COSTS OF CUSTOMIZATION
► Variety of products has increased dramatically due to COSTS OF AN ADDITIONAL DIMENSION
customer preferences
► Mismatch costs: Demand Pooling
► This implies at the SKU level demand uncertainty has
effects are diminished by having
become higher
too many products#
► Elicitation Costs: Is it feasible for

the customer to customize this
dimension? Are there ergonomics/
technical feasibility constraints?
► Process Costs: How does allowing

customization of this dimension
change the economics of the
production and delivery process?
Recall there are economies of scale
► Product Assortment is the set of options that the in production.
customer can choose from. Assortment choice is a key
part of a firm’s product strategy.
# FOR DETAILS ON DEMAND POOLING LOOK UP SLIDES FROM SESSION ON ZARA

► How you go about deciding the optimal assortment
depends on the cost of adding an additional
customizable dimension and the benefit of adding that
dimension to the assortment.

HOW TO DELIVER ON THE OPTIONS YOU ARE PROMISING?
► Offer Variety on Modular dimensions, such that

modules with limited number of interfaces can be
mixed and matched.
► With modular products you will Assemble-to-Order

rather than building to order. Building to Forecast is
generally only possible for components rather than
final goods.

TIMBUK2 AND YOU: ATTACKING UNDERSERVED LONG-TAIL SEGMENTS
THE LONG
TAIL EFFECT
THE INNOVATION
► Identify underserved market segments that you can
address by mass-customization, Sell less of more.
► Provide Recommendations, customization tools
► Increase Revenue and Cost
WORKS BEST WHEN

► High desire for variety
► Low cost of customization (postponement, modularity,
production on demand)

ILLUSTRATIVE EXAMPLE: VALUE OF PLATFORM DESIGNS (DELAYED DIFFERENTIATION)
Early Differentiation Process†
Dyed Yarns Finished Sweaters
Dyeing Knitting
Delayed Differentiation Process (Postponement)

White Garments Finished Sweaters
Knitting Dyeing
► In an early differentiation process, dyeing and knitting quantities have to be decided on the basis of individual
demand for colors.
► In a delayed differentiation process, knitting quantities can be decided on basis of total demand.
► Total Demand has lower variance than the sum of variances of individual demand. This will reduce total mismatch
costs and increase profits
Delayed Differentiation also leads to pooled demand, and reduces mismatch costs!
Anand K. S. and K. Girotra, “The Strategic Perils of Delayed Differentiation,” Management Science, 53:5, May 2007, pp. 697-712.

MORE EXAMPLES OF DELAYED DIFFERENTIATION
► Campbell Soup:
► Manufacturers brand name and private label soup (same soup)
► Problem: many different private labels (Giant, Kroger, A&P, etc)
► Solution: Hold inventory in cans without labels, add label only when demand is realized.
► Black and Decker:

► Sell the same drill to different retailers that want different packaging.
► Store drills and package only when demand is realized.
► Nokia:
► Customers want different color phones.
► Design the product so that color plates can be added quickly and locally.
► Hewlett Packard
► Sells printers in different countries. Different markets have different power supply
conventions and require different product labeling.
► Designs and produces generic printers with special power supply modules that can be
plugged in after demand is realized.

POSTPONEMENT /DELAYED DIFFERENTIATION
Delay the point of differentiation as much as possible

► Keep products as “standard vanilla boxes” as possible
► Ideally until an order is received
► If not, at least as late as possible

COVID CAPSULE: EACH CONCEPT APPLIED TO FRONT LINES OF THE
PANDEMIC
Delay the point of differentiation as much as possible
► Keep products as “standard vanilla boxes” as possible
► If not, at least as late as possible
► EXAMPLES from CURRENT CONTEXT?

BENEFITS OF DELAYED DIFFERENTIATION
► Lower inventory carrying costs ?

► Lower risk of unsold items
► More standardized planning
► Fewer part numbers and Fewer operations
► Faster response to customers ?
► Product only has to be completed, not manufactured from the beginning
► Lower customs and tariffs if postponement implies close to market customization
► Components and subassemblies may incur lower tariffs than completed goods
► Some of the localization material can be locally sourced (cheaper)
► Compliance with local content laws
INSEAD
SUMMARY
► Resequencing customer order and assembly allows you to do mass customization

► Resequencing allows you to do statistical pooling and aggregate forecasts (Dell)
► Build-to-Order provides product variety to customers, but requires flexible production
methods
► Platforming and modularization done right decrease the operational costs of
customization (Timbuk2). You achieve high customization with low unit costs.
► It must be easy/cheap to elicit customer preferences
► Platforming and modularization are examples of adding cheap flexibility to your
business model (Auto Industry)

NEXT CLASS
► Managing Risks: Newsvendor to Nintendo
► Optional reading: Chapter “What” from the “Risk Driven Business Model”
► Advanced optional reading (maybe useful for the exam): Chapter 11 of

“Matching Supply and Demand”

DEALING WITH UNCERTAINTY:
THE NEWSVENDOR MODEL
RISK!
INFORMATION RISK: DECISIONS MADE WITH POOR INFORMATION
INCENTIVE ALIGNMENT ISSUES: DECISIONS MADE WITH SELF INTEREST (AS OPPOSED TO TEAM INTEREST)
MANAGING INFORMATION & INCENTIVE ALIGNMENT RISKS IS KEY TO SUCCESS
Newsvendor-2
MANAGING RISK: THE NEWSVENDOR MODEL
Demand Forecast
0.12
0.10
0.08
Frequency
Probability
0.06
0.04
0.02
0.00
40000
44000
48000
52000
56000
60000
64000
68000
72000
76000
80000
84000
88000
Newspaper Demand
Daily output (thousands of barrels)
0 200 400
SITUATION
The local newsstand purchases copies of the weekly paper Progessivo at the start of every week.
The observed sales (demands) during each of the last 52 weeks are :
15 19 9 12 9 22 4 7 8 11
14 11 6 11 9 18 10 0 14 12
8 9 5 4 4 17 18 14 15 8
6 7 12 15 15 19 9 10 9 16
8 11 11 18 15 17 19 14 14 17
13 12
How many units should they buy next week ?
What information do you need to make this decision?
Newsvendor-4 4
DISTRIBUTION OF DEMAND
0
0 2 4 6 8 10 12 14 16 18 20 22
Newsvendor-5 5
“TOO MUCH” AND “TOO LITTLE” COSTS
► Co = overage cost
► The consequence of ordering one more unit than what you would have ordered had you
known demand.
► Suppose you had left over inventory (you over ordered). Co is the increase in profit you would have
enjoyed had you ordered one fewer unit.
► Cu = underage cost
► The consequence of ordering one fewer unit than what you would have ordered had you
known demand.
► Suppose you had lost sales (you under ordered). Cu is the increase in profit you would have enjoyed
had you ordered one more unit.
Newsvendor-6
CONTINUOUS APPROXIMATION
0.12
0.10
0.08
0.06
0.04
0.02
0
-4 -2 0 2 4 6 8 10 12 14 16 18 20 22 24 26
Newsvendor-7 7
KEY INPUT PARAMETERS
► p = sale price
► s = outlet or salvage price
► c = purchase price
► g = loss of goodwill cost
► D = demand ----mean, variance, distribution
► Q = order quantity
Newsvendor-8 8
HOW MUCH DOES THE NEWSVENDOR SELL ON AVERAGE ?
 D if D  Q 
Sales =  
Q if D  Q 
► The expected units sold is

E[min{Q, D}] =  min{Q, d} f (d )
−
Newsvendor-9 9
WHAT IS THE OBJECTIVE?
► Choose Q to maximize the total expected profit?
E[Profit {Q, D}] = pE[ Sold ] + sE[ Left ] − gE[ Short ] − cQ
Newsvendor-10
BALANCING THE RISK AND BENEFIT OF ORDERING A UNIT
► Ordering one more unit increases the chance of overage …
► Expected loss on the Qth unit = Co x Pr(DQ).
► … but the benefit/gain of ordering one more unit is the reduction in the chance of
underage:
► Expected gain on the Qth unit = Cu x Pr(D ≥ Q)= Cu x (1-Pr(D  Q)).
► To maximize expected profit, order Q units so that the expected gain on the Qth unit is the same as the
expected loss on the Qth unit:
Co  Pr( D  Q) = Cu  (1 − Pr (D  Q ))
Expected loss Expected gain
► Rearrange terms in the above equation:

Cu
Pr( D  Q) =
Co + Cu
► The ratio Cu / (Co + Cu) is called the critical ratio.
► Hence, to maximize profit, choose Q such that the probability we satisfy all demand (i.e., demand is Q
or lower) equals the critical ratio.
Newsvendor-11
A MORE GENERAL DESCRIPTION OF THE SETTING
► Product Demand is Random

► Product has a single selling season
► Newspaper, books, doughnuts
► Fashion, Style goods,
► Increasingly all major non consumable items are falling into the style
goods category
► Books, Toys, Fashion garments, CDs, DVDs, PDAs., Laptops, Printers
► Limited salvage at the end of the season
► Long lead time - a single opportunity to procure at the start of the
season
► For many products production occurs in a far away location so no
opportunity to purchase again during the “season”
Newsvendor-12 12
THINKING BROADLY ABOUT ALTERNATE APPLICATIONS OF THE NEWSVENDOR MODEL
The main features of the Newsvendor model are:
• An uncertain quantity (demand) and

a bet on that quantity (made before the exact value of the uncertain quantity is revealed).
• There are consequences of getting this bet wrong.
• This structure extends to many situations beyond the traditional Newsvendor Setup
An Uncertain Number Ex-Ante Bet

Product Demand FGI/Stock Level
# of Guests at a Party Amount of Food to order
Talk Time Demand Cell-Phone Plan
Market Size for new Product Capacity of Production Facility
Project Complexity Number of employees on project
Withdrawals at a hedge fund Cash reserve/Short Term Investments
The Expected Profit Maximizing Bet balances the too-much and too-little costs
Newsvendor-13
WORDS OF THE DAY:
WHEN MAKING A DECISION

UNDER UNCERTAINTY, TILT YOUR
BET IN THE DIRECTION OF THE
LEAST CONSEQUENCE.
THE NINTENDO WII
Newsvendor-15
NINTENDO: A WII BIT CAUTIOUS!
Newsvendor-16
A YEAR LATER: TWO SIDES OF THE STORY
We don’t feel like we’ve made any mistakes,” said

“It’s staggering,” said James Lin, senior analyst at the George Harrison, SVP for marketing at Nintendo
MDB Capital Group in Santa Monica, Calif., who of America.
estimates that Nintendo is leaving $1.3 billion on the He said there was a shortage because the
table. “They could easily sell double what they’re company must plan its production schedule five
selling.” months ahead, and projecting future demand is
difficult.
Newsvendor-17
APPLYING THE NEWSVENDOR MODEL TO THE NINTENDO WII
► Reasons to consider a different Cur-c

► Sales of complementary products, Network effects, goodwill, repeat business
► Reasons to consider a different Coc-s

► Pilferage, brand dilution…
► Reasons to consider a different objective

► Cash constraints, bankruptcy prevention, etc.
► Who makes these decisions?

► Individual Incentives, Operations - Marketing conflict,
► Who estimates the costs?

► Operations vs. Marketing
► Organizational Culture
► Risk Aversion, meeting street forecasts.
Newsvendor-18
PERFORMANCE METRICS FOR THE NEWSVENDOR PROBLEM
For Self Study
► For any order quantity we would like to evaluate the following performance
metrics:
► Expected lost sales
► The expected number of units by which demand will exceed the order quantity
► Expected sales
► The expected number of units sold.
► Expected left over inventory
► The expected number of units left over after demand (but before salvaging)
► Expected profit
► Only lost sales need to be calculated, the rest follow trivially.
► For the Normal distribution, we have tables that help us get the Lost Sales
Newsvendor-19
COVID-19 CAPSULE
Q1
► You are the head of vaccine purchasing for a large regional country bloc. Demand is
uncertain. How many COVID-19 vaccines would your purchase?
-------------------------------------------------------------------
Q2
► Pfizer/BioNTech’s COVID-19 Vaccine is packaged in a specialized container with ~1000
doses
► After opening the container, vials inside can be used for 48hours
► After opening a vial, you have use all of the doses within 6 hours
► You run a small rural clinic/pharmacy in Southern France which is offering this vaccine to
its patients/customers
► Last time you offered flu vaccines to your customer population, you had approx. 5 walk-
ins each day and 2 scheduled
► How many units will you pre-order?
Newsvendor-20
KEY LESSONS
► Modern business is a risk-laden team sport
► The Newsvendor model provides the intuition for making bets in the face of uncertainty
► Essentially, in making these decisions we must take into account the relative
consequences of betting too high or too low. This tells us if we bet more or less than the
mean
► Further, how much more or less than the mean we go, depends on the uncertainty
involved
► Incentive alignment is key and different people in your organization will perceive
underage and overage differently.
Newsvendor-21
NEXT TIME
► How to obtain the demand forecast!
► HW #2 on New Product Demand Forecasting is due! No research is necessary. Simply

brainstorm with your team. Be creative. See what you can come up with!
► HW #3 can be completed based on today’s lecture! Start on it now!
► For the remainder of the course, we will draw on the learnings from the Newsvendor
model, so make sure you understand the intuition well
Newsvendor-22
FORECASTING DEMAND
DEMAND FORECASTING AND THE PUSH-PULL BOUNDARY
The importance of demand forecasting in a system depends on the location

of the push-pull boundary or Order Penetration Point.
Forecasting-2 2
EXAMPLE I – TACO BELL BURRITO
Push/Pull Interface
Refrigerator Cooking Assembly Packaging

& Sales
Customer
Forecast Driven Order Driven
Forecasting-3 3
EXAMPLE II – BURRITOVILLE BURRITO
Push/Pull Interface
Refrigerator Cooking Assembly Packaging

& Sales
Customer
Forecast Driven
Order Driven
Forecasting-4 4
DEMAND FORECASTING & DEMAND MANAGEMENT
Historical Data
Forecasting Promotions calendar
Forecast
Demand Historical Data

Planning Promotions history
Forecast Inventory
Promotion Calendar
Information
Business Processes Demand Historical Data

Demand Model
that depend on the Shaping Promotions history
forecast
Inventory Price Elasticity

Information
Forecast Demand Management
Forecasting-5 5
FORECASTING--MATCH TYPE OF METHOD WITH TYPE OF PRODUCT
Forecasting Methods
New Products Existing Products
Mostly Qualitative Methods Mostly Quantitative Methods
Forecasting-6
METHODS
Qualitative
► Grass Roots - Talk to Sales Force, Talk to Customers
► Market Research - Surveys of Customers, Experimental Test Markets
► Panel Consensus - Bring in Experts
► Executive Judgment - Surveys or Formal Input from Executives
► Historical Analogy - For New Products/ Technology, Find a Similar Product
► Delphi Method - Formal, Sequential Method of Polling and Pooling Expert Opinions
Quantitative
► Time Series Analysis - THE PAST WILL REPEAT
► Causal Relationships (Regression)
► AI methods
Forecasting-7 7
WHAT DO COMPANIES DO?*
► Customer/market research– 57%
► Jury of executive opinion – 44%
► Sales force composite – 39%
► Looks-like analysis – 30%
► Trend line analysis – 19%
► Moving average – 15%
► Scenario analysis – 14%
► Exponential smoothing – 10%
► Delphi method – 8%
► Linear Regression – 7%
► Decision trees – 5%
► Simulation – 4%
► Others: Quant Methods – 9%
*Based on a survey of 168 companies

Forecasting-8
EXISTING PRODUCT FORECASTING
► Demand or sales history is available.

► Train a statistical model on that history.
► Use the model to predict future demand.
Forecasting-9
AI, ML AND ADVANCED ALGORITHMS FOR SALES FORECASTING
Forecasting-10
WHAT LEVEL TO FORECAST AT ?
► Time
► Daily, Weekly, Monthly, Quarterly ?
► Guideline: Tailor the level of time granularity to the decisions
► Precondition: Sufficient data must exist to enable forecasts at that level
► Geography
► Store, Sales District, Region, State, Country
► Channel
► Retailer, Wholesaler, Key Accounts, Channel type (online, brick and mortar)
► Product
► SKU, Product Category, Product Line/Product Group, Business Unit, Division
Forecasting-11 11
AGGREGATION
► Aggregation reduces variability in the forecasts

► Smoother series are easier to forecast
► Higher R2 is observed in models on aggregated series than in lower level series
TOP DOWN VS. BOTTOM UP
► Top down achieves a better forecast at the aggregate level

► Top level forecasts are also typically target based
► It maybe easier to generate econometric or more advanced models for aggregate level forecasts
► If you do top level forecasts how do you proportion sales to each of the lower levels?
► Bottom-up is grassroots level forecast and hence may reflect the true nature of business
Forecasting-12 12
EXAMPLE: MEN’S WEDDING BANDS--AT THE SKU LEVEL
BandW07
14
12
10
Sales in Units
0
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61 63 65 67 69 71 73 75
Month Number
Forecasting-13 13
EXAMPLE: MEN’S WEDDING BANDS--AT THE AGGREGATE STYLE LEVEL
BandW
100
90
80
70
60
Sales in Units
50
40
30
20
10
0
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61 63 65 67 69 71 73 75
Month Number
Forecasting-14 14
EXAMPLE: A PROMINENT ANTI INFECTIVE DRUG--AGGREGATE LEVEL
ALL ALL
0.3
0.25
0.2
Scaled Units
FORECAST
0.15
ACTUAL
0.1
0.05
0
0
2
00
00
01
01
02
02
0
2
-00
-01
-02
v-0
v-0
v-0
l-0
l-0
l-0
n-0
n-0
n-0
ar-
p-
ar-
p-
ar-
p-
ay
ay
ay
Ju
Ju
Ju
No
No
No
Se
Se
Se
Ja
Ja
Ja
M
M
M
Month M
Forecasting-15 15
Scaled Units
Scaled Units Ja
0
0.05
0.1
0.15
0.2
0.25
0.3
n-0
0
Ja
M
0
0.05
0.1
0.15
0.2
0.25
0.3
n-0
0 ar-
00
Forecasting-16
M M
ar- ay
00 -00
M
ay Ju
-00 l-0
0
Ju Se
l-0 p-
0 00
Se No
p-
00 v-0
0
No Ja
v-0 n-0
0 1
Ja
n-0 M
ar-
1 01
M M
ar- ay
01 -01
M
ay Ju
-01 l-0
1
Month
Ju Se
OTH ALL
l-0 p-
1 01
Month
Se No
PC ALL
p- v-0
ALL
01 1
No Ja
ALL
v-0 n-0
1 2
Ja M
n-0 ar-
2 02
M M
ar- ay
02 -02
M Ju
ay l-0
-02 2
Se
Ju p-
l-0 02
2
Se No
p- v-0
02 2
No
v-0
2
ACTUAL
FORECAST
ACTUAL
FORECAST
Scaled Units
Scaled Units
Ja
0
0.05
0.1
0.15
0.2
0.25
Ja n-0
0
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
n-0
0
M
ar-
M 00
ar-
00 M
ay
M -00
ay
-00
Ju
l-0
0
Ju
l-0
0 Se
p-
00
Se
p- No
00
v-0
0
No
v-0 Ja
0 n-0
1
Ja M
n-0 ar-
1 01
M M
ar- ay
01 -01
M Ju
ay l-0
-01 1
Month
Se
ID ALL
Ju p-
l-0 01
1
No
Month
ALL
Se v-0
UR ALL
p-
01 1
Ja
No n-0
2
ALL
v-0
1 M
ar-
Ja 02
n-0
2 M
ay
-02
M
ar-
02
EXAMPLE: A PROMINENT ANTI INFECTIVE DRUG- SEGMENT LEVEL
Ju
l-0
M 2
ay
-02 Se
p-
02
Ju
l-0 No
2 v-0
2
Se
p-
02
No
v-0
2
ACTUAL
FORECAST
16
ACTUAL
FORECAST
FORECASTING BASED ON HISTORICAL SALES DOES NOT ACCOUNT
FOR VARIABLES WHICH ARE CHANGING
e.g. PRODUCT PROMOTIONS
FOR ESTABLISHED PRODUCTS WITH SUFFICIENT HISTORY, NOT
CAPTURING THE IMPACT OF PROMOTIONS IS A KEY REASON FOR
POOR FORECASTS
Why does a manufacturer do a Trade promotions

trade promotion ? Sales contests
Off-invoice pricing
 Increase Sales Volume Free product with order
 Maintain Volume/Share Co-op dollars
 Increase Market Share Guaranteed consumer advertising & promotion
support
What does a retailer expect Consumer promotions
from a trade promotion ?
Sweepstakes
Gift with Purchase
 Increase Store Traffic Purchase with Purchase
 Improve Category Profitability Rebates
 Increase Customer Loyalty Coupons/IRC
Sampling
Cents-Off
Free Trial
Cross Promotion
Forecasting-18 18
FORECASTING WITH PROMOTIONS
The “1, 2 STEP” APPROACH

► Baseline sales
► measure what retail sales would be in the absence of a promotion
► "Last like" promotion
► Quantify Sales Lift
► Extend it to Fixed Effect vs. Fluid Effect
► Baseline sales
► Sales Lift
► Hysteresis
► Pantry Loading
► Holdover
► Cannibalization
► Halo Effect
Forecasting-19 19
“HYSTERESIS”
If you push on something, it will yield: when you release, does it spring back completely ?
If not, it is exhibiting hysteresis
sales increase quickly as a result of promotion, then

remain the same or decline slowly
Time
Forecasting-20
DELAYED RESPONSE OR HOLDOVER
Customers wait for the last few days of the promotion
Time
Forecasting-21
WEAR OUT
Promotion causes some customers to try but after

that the product stays unaffected
Time
Forecasting-22
PANTRY LOADING
consumers buy in advance in response to a

promotion, leading to post-decline in sales
Time
Forecasting-23
CANNIBALIZATION
➢ A promotion on Coors Light 12 pack cannibalizes the

sales of Coors Light 18 pack cans
Halo Effect
➢ A promotion on Jose Cuervo 650ml increases sales for

Margaritaville cocktail mix and all SKUs of nachos and
tortilla chips
Forecasting-24
RETAILER FORECASTING AND REPLENISHMENT PROCESS IN CPG RETAIL
Retailer POS Data

Market/item Create item-level
knowledge, store Collect POS Data and forecast and special Create purchase orders
planning, item other supporting event calendar (e.g..., for items
planning by information promotions, store
individual stores openings, item
distribution)
No Retailer and
manufacturer
?
Decision - Is
manufacturer able
discuss other
options
to meet retailer’s
purchase order?
MANUFACTURER
Yes
Marketing Programs and

Promotional developed with Input
Production planners Forecast drives production
from Sales/Marketing (e.g...,
validate item-level Product shipped to meet
pricing, item additions/deletions)
forecast purchase order
and market/customer knowledge
specifications
Forecasting-25
WHY WOULD FORECASTS BE DIFFERENT?
► One forecast is based on information that the other forecast did not consider
► One forecast makes a different assumption about the nature of relationship between
the variables
Forecasting-26 26
WITH COLLABORATIVE FORECASTING
Historical Data
Retailer Manufacturer
Adjust Item Forecast
Market Knowledge Baseline Pricing

Store Planning Forecast(s) Pre-launch tests
POS Data National advertising
In-store surveys Sales from other retailers
Local advertising Forecast
Competitive intelligence
reconciliation
Promotions Calendar Capacity constraints
adjustments
Similar product sales
Returns
Collaborative
Forecast
Forecasting-27
FORECAST SIGNAL DIVERSITY
The difference between the

two modifiers is a (crude)
indicator of forecast signal
diversity
National
Promotion Advertising
Capacity
Local Events Constraints
Weather Retailer’s Supplier’s Weather

Baseline Forecast Baseline Forecast
Modifier Modifier Competitive
Surveys Intelligence
.. ..
.. ..
The diversity is due to :

◆Different causal variables (information sets)
◆Differences in predicting the extent of influence of each factor (coeff. estimates in the model)
ZURICH
Assumption: Both generate the same baseline forecast ?
Forecasting-28 28
PRODUCT LIFE CYCLE AND FORECAST SIGNAL DIVERSITY
Less forecast
Sales
Moderate to high diversity
growth
High forecast diversity
forecast
diversity
Maturity
Decline
Growth
Launch
Time since launch
Forecasting-29
NEW PRODUCT FORECASTING
NEW PRODUCT FORECASTING
► A recent survey shows that almost 30% of company revenues from new products
launched in the previous three years (study by Cooper & Edgett 2012).
► Starts many months or years before the launch date. Very high information risk!
► A study found the average forecast accuracy across all types of new products to be 58%.
(Kahn 2002)
► Little or no demand history.
► Has become more relevant due to the increasing rate of new product introductions
but also because of the drive of technology advances, product life cycles are
shortening.
Forecasting-31
NEW PRODUCTS LAUNCHED MORE AND MORE
Forecasting-32
METHOD #1: USE SIMILAR PRODUCTS (LOOKS-LIKE ANALYSIS)
PROCEDURE
► Find a similar product with demand history.
► Use this to forecast the new product’s demand.
ASSUMPTIONS
► The product is not completely new (as opposed to, say,
Nintendo Wii)
► Market conditions are similar.
PROS/CONS
► + Data driven, less bias
► - Similar product not always available. May look similar on
the surface but driven by completely different underlying
variables (trial rates, repeat purchase rates, etc.)
BUILDING FORECAST DISTRIBUTION

► Use the error or std. dev. of your regression method (or
of any other estimation method that you are using)
EXAMPLE ” … companies like Dell could use historical

► Dell uses past products’ life cycle curves to predict the data from previous products to improve
demand of a new product. forecasting accuracy on new products by as
much as 9%. This translates into millions of
dollars of savings.”
-KelloggInsight
Forecasting-33
METHOD #1 EXAMPLE: FORECASTING MTS PRODUCTS AT DELL AND ITS SUPPLIERS
► Dell shifted part of its production to Make-To-Stock (mostly

B2B sales)
► It uses past products’ life cycle curves to predict the demand
of a new product.
► Since these are B2B sales, promotions and discounts are
limited
Kejia Hu, Jason Acimovic, Francisco Erize, Douglas J. Thomas, Jan A. Van Mieghem (2018) Forecasting New Product Life
Cycle Curves: Practical Approach and Empirical Analysis. Manufacturing & Service Operations Management
Forecasting-34
METHOD #2: EXPERT OPINION
PROCEDURE
► Identify experts on the uncertainty you want to forecast.
► Ask experts for forecasts, use average forecast as mean and
standard deviation of forecasts by different experts as standard
deviation.
► Example: Retail buyers (consumer goods, fashion),
Pharmaceuticals probability of success of new drugs

► Weighted average of experts’ past accuracy
► Ask experts for level of certainty and use aggregate uncertainty
► Make sure information sets are similar (see Leitax example later in class)
► Aggregation can get complicated fast (My favourite topic: Happy to chat more about this over coffee!)
PROS/CONS
► + Works for new products, incorporates human judgement
► - People tend to be biased
► - Does not take into account level of confidence or information of experts
PEOPLE ARE BIASED BUT CAN INCORPORATE UNSEEN, CHANGING CONDITIONS.

Forecasting-35
HOW TO MAKE THIS WORK?
1. Group has relevant information

• Diversity is important
• Commitment to quality of forecast
2. Proper incentives
• Hold individuals accountable for their accuracy
• Reward each individual on their own performance.
3. Ask independently
• In group settings, personality can play a major part.
Forecasting-36
METHOD #3: CROWDSOURCING/PILOT
PROCEDURE
► Poll a large number of people, the crowd!
► Run a small pilot/experiment
► Example: BestBuy, Google, City of Paris, INSEAD
► Platforms: Kickstarter, mTurk,

► Similar logic to the expert case
► A well designed pilot can give you information if a product will
be a hit or not (which might be the most significant source of
uncertainty)
PROS/CONS
► + Works well for new feature development
► - Infeasible in many corporate contexts
► - Works only in cases where there is relevant distributed
information
Forecasting-37
METHOD #3: CROWDSOURCING/PILOT EXAMPLE
Invest Sales
NEW TIMELINE
Pre-Sales Invest Sales
Forecasting-38
CLASSIC EXAMPLES:
• Country fair in Plymouth, 1906

• 800 people, weight of the ox
• Average: 1197 lbs.
• Truth: 1198 lbs.
• Number of jellybeans in a jar

• Average: 871
• Truth: 850
• Temperature in a class room

• Average: 72.4 F
• Truth: 72.0 F
METHOD #4: PREDICTION MARKETS (ALSO A FORM OF CROWDSOURCING)
PROCEDURE
► Establish a trading platform where participants can buy and sell securities (using
real or play money)
► The value of tradable securities is directly linked to events in the future.
EXAMPLE
► Holder of Security X is paid D$ if demand for product is D units, after demand is revealed.
► This security can be bought and sold now.
► The current price of the security captures the market’s estimates of demand.
► If you believe this is an underestimate, you would like to long (buy) this security, because its fundamental value is
greater than what you have to pay to acquire it.
► If you believe this is an overestimate, you would like to short (sell) this security, because its fundamental value is
lower than what one has to pay to acquire it.
► In equilibrium (no arbitrage), the price of a security should reveal the best estimate of demand.
Forecasting-40
PREDICTION MARKETS: MORE EXAMPLES
PUBLIC MARKETS
HOLLYWOOD STOCK EXCHANGE, HSX.COM PREDICTIT IOWA ELECTRONIC MARKETS
SOFTWARE PROVIDER
CORPORATE MARKETS
Forecasting-41
PREDICTION MARKETS: ASSUMPTIONS, PROS, CONS
ASSUMPTIONS
► Efficient Markets create the right incentives for information revelation and aggregation
PROS/CONS
► + Markets also create incentives for confidence weighting for information
► The amount of money you will bet will depend on your confidence.
► + Aggregate information from multiple sources taking into account confidence levels.
► + Can provide precise estimates.
► - Liquidity Limitations: Markets require a lot of investors.

► - Potential for manipulation
► Programmer places a $100k bet that a project run behind schedule and then
introduces a virus into the intranet
► - Prediction markets may be illegal, or against religious norms.
► - Hard to obtain a distribution if it is not a binary (yes/no) decision.

► For a binary decision (yes/no) the price is the forecast
Forecasting-42
WORDS OF THE DAY:
FORECASTING IS A PROCESS!
FORECAST PRINCIPLES
1. Forecast is a distribution - not a number
2. Forecasts should always include measure of error
3. The longer the time horizon, the larger the error
4. A distribution captures all these features.
5. Accuracy of a method cannot be judged by one outcome.

6. Method should be chosen based on need and context
0.025
0.02
0.015 Value
0.01
0.005
0
Now Time
0 200 400 600
Forecasting-44
SUMMARY:
PRELUDE:
► Forecast is a distribution – not a number.
► Match your forecasting method with the context.
EXISTING PRODUCTS:
► Statistical prediction is like driving a car by only looking in the rear mirrors.
► Adjust methods for promotions and other factors
► Beware of generalizability of ML models – take courses at INSEAD related to this
NEW PRODUCTS:
► Method 1: Look-like analysis for for similar products when market conditions don’t change too fast
► Method 2: When using expert opinion: 1. crowd must be ``wise’’, 2. proper incentives, 3. ask
independently.
► Method 3: Crowdsourcing and pilots can help identify if a product is a hit or not
► Method 4: Prediction markets have had a mixed track-record but can work for binary (yes/no)
decisions
► Forecast diversity and forecast combination improves forecast accuracy
EPILOGUE:
► Always hold people accountable for their forecasting mistakes
► To avoid incentive misalignment, share information and consider setting-up a dedicated forecasting
team
► Remember: Forecasting is a process!
Forecasting-45
PREP FOR NEXT CLASS
► Read Sport Obermeyer case

► HW3 is due.
Forecasting-46
THE WHEN STRATEGY: SPLIT DECISIONS
REACTIVE CAPACITY
SPLITTING DEMAND DECISIONS
DISCOVERY DRIVEN PLANNING
BACK TO THE NEWSVENDOR MODEL: ORIGINS OF THE MODEL
Since
Demand = Sales + Lost Sales,
then,
Sales = Demand – Lost Sales.
If revenue is $r per unit, cost is $c per unit and salvage value is $s per unit, then
Profit = (r-c) × Sales – (c-s) × Leftover Inventory.
Since Sales = Demand – Lost Sales, we obtain

Profit = (r-c) × Demand – (r-c) × Lost Sales – (c-s) × Leftover Inventory
In expectation (since we choose the ordering quantity before we see demand),

Demand is indep. of order quantity Newsvendor Cost
E[Profit] = (r-c) × E[Demand] – (r-c) × E[Lost Sales] – (c-s) × E[Leftover Inventory]
“Underage Cost” “Overage Cost”
When-Split-2 INSEAD
BACK TO THE NEWSVENDOR MODEL
E[Profit] = (r - c) × E[Demand]
When-Split-3 INSEAD
BACK TO THE NEWSVENDOR MODEL
Information
Risk
E[Profit] = (r - c) × E[Demand]
– cu × E[Lost Sales]
– co × E[Leftover Inventory]
► Last time: lower information risk by forecasting better
► Today: lower information risk by injecting flexibility (recourse) into
business models and, more generally, splitting decisions
Innovative business models are not just about minimizing
costs/maximizing gross margins, but about minimizing risk
When-Split-4 INSEAD
WHAT SHOULD WALLY DO?
When-Split-5 INSEAD
SPECULATIVE AND REACTIVE CAPACITY
Speculative
Production Capacity Reactive Capacity
Material LT Prod. LT
New Info
TOTAL ORDER QTY
OVERSEAS ORDER LOCAL ORDER

(SPECULATIVE CAPACITY) (REACTIVE CAPACITY)
When-Split-6 INSEAD
EFFECT OF REACTIVE CAPACITY
Reactive Capacity impacts Supply-Demand Mismatch Costs
When-Split-7 INSEAD
When-Split-8 INSEAD
COSTUME GALLERY: BUSINESS AND SOURCING FROM CHINA
► Dance costume wholesaler

► Based in New Jersey, U.S.
► Annual sales US$ 30-40 million
Ellen Ferreira,
Owner and CEO,
Costume Gallery
► Demand is concentrated in April (highly seasonal)

► Production cost in China: $3; Lead time: 3 Months
► Average Sales Price: $35
► If Ellen sources from China, she must order before the Chinese
new year holiday in January-February.
► Overage Cost ~ $3 (assuming no salvage)

► Underage Cost ~ $32
When-Split-9 INSEAD
BUSINESS MODEL INNOVATION: REACTIVE CAPACITY
► In addition to sourcing from China, Ellen decides to also use local sources
Average Cost: $15

OVERSEAS LOCAL ORDER
Local Lead time: 1-2 days ORDER
Production Min. Order Size: 10 Dec Jan Feb Mar
Early Bird Overseas Most (All) Local

Overseas Average Cost: $3 Avg. Price: $35 Orders orders must orders come in Production
Contractors Lead time: 2-3 Months be placed Deliveries
Min. Order Size: 100s
PROFITS
THE OVERSEAS ORDER ► What happens to profits by having both

the local and overseas production
► Overage Cost ~ $3 (assuming no salvage) possibilities?
► Underage Cost = Consequence of producing too few in china
► Underage cost will no longer be $32 but $12. ► Profits are increased substantially by
having the local option in addition to the
overseas option
THE LOCAL ORDER
► Order is placed after observing demand
► The overseas capacity is known as
► Order more if china production falls short speculative capacity, the local capacity is
► Do not order, if China production is known as reactive capacity
excessive
► Having Reactive capacity is a type of a Real
option
When-Split-10 INSEAD
WORD OF THE DAY: SPLITTING
(DECISIONS & BETS)
BET $$$$$ UNCERTAINTY
BET $$ BET $$ BET $$ UNCERTAINTY
BET $$ BET $$ UNCERTAINTY NO-RISK BET $$

INNOVATING THE TIMELINE: ADVANCE INFORMATION (SPLIT CONSUMER DECISIONS)
THE INNOVATION
TRADITIONAL TIME-LINE Time
NEW TIMELINE Time
Advance Sales Assortment, Pricing Production Sales Receivables

Markdowns
THE INNOVATION WORKS BEST WHEN

► Obtain Early demand information ► Multiple Styles, Large Assortment
► To obtain credible information, offer discount ► High Mismatch Costs, Long Production Lead Times
► Reduces Risks, Reduces Revenue ► Early Information is available
► Early ordering is possible
SPLITTING DECISIONS: DELAYED DIFFERENTIATION AT BENETTON
Early Differentiation Process
Dyeing Knitting

Knitting Dyeing
► In an early differentiation process,dyeing and knitting

quantities have to be decided on the basis of individual demand
for colors.
► In a delayed differentiation process, knitting quantities can be
decided on basis of total demand. YARN QUANTITY COLORS
► Total Demand has lower variance than the sum of variances of UNCERTAINTY
individual demand. This will reduce total mismatch costs and
increase profits
OTHER INDUSTRIES AND PRODUCTS WHERE THIS APPLIES
THE RIGHT WAY TO GO TO CHINA!

BUSINESS MODEL INNOVATION II: ADVANCE PURCHASE DISCOUNTS
ORDERS COMMITTED BY DEC
Average Cost: $15 15TH GET A 15% DISCOUNT
Local Lead time: 1-2 days
Production Min. Order Size: 10
OVERSEAS LOCAL
ORDER ORDER
Dec Jan Feb Mar
Overseas Average Cost: $3 Avg. Price: $35
Contractors Lead time: 2-3 Months
Early Bird Overseas Most (All) Local
Min. Order Size: 100s
Orders orders must orders come in. Production
be placed Deliveries
DANCE SCHOOLS COSTUME GALLERY

► In the original model, they have no ► It is losing money by giving a discount.
incentive to place early orders… Why take
the risk. (Kids change sizes, drop out, etc.)
► But, it gains information and reduces
mismatch costs.
► But now, they can increase their
commissions by doing so.
► Everybody is better off!
► They have the information and ability to
fix demand.
► So they order based on their best

estimate.
DISCOVERY DRIVEN PLANNING: XIAOMI
► Launch new products in small batches

► Incorporate customer feedback
► Scale-up production
► Drawbacks and risks?
THE COMMON THREAD
Split Decisions: One Shot Bets→ Gradual bets
INNOVATION TOOLKIT

Delay Decisions
Resequence
Split Decisions
MANAGING RISK
Flexible/Reactive
KEY LESSONS
► Splitting Decisions can reduce the consequences of getting bets wrong
► Right way to go to low cost manufacturing base, while keeping Local production
(Reactive Capacity)
► Advance Purchase Discounts: Split Consumer Purchases
► Kickstarter/IndieGogo: Advance Purchase
► Splitting decisions is a powerful Business Model Innovation Template
VALUE CHAIN INNOVATION
ZARA VS. MARKS & SPENCER

THE FASHION INDUSTRY
Design Sourcing Mfg. Distribution Sales Discount
Example
Retail Price: $35

Cost: $3 (production, shipping)
Margin: 91%
Firm Profitability: 5%!!!!
Grease Jacket from Costume Gallery’s
2008 Catalogue
While Margins are high for fashion goods, but firm profitability is Very Low! Why?
• Demand for such products is driven by fickle trends and is notoriously hard to predict
• Consequently, firms end up with a lot of excess inventory which must be discounted. At the same time,
many customers don’t find what they want.
ZARA-2 INSEAD
NEWSVENDOR QUIZ
► Three Newsvendors face the following Demand Distributions. On average who will
make the most money? (All Newsvendor act optimally and have the same underage
and overage cost structure)
Demand Distribution
20% 20% 20% 20% 20%
17 18 19 20 21
33% 33% 33%
18 19 20
100%
19
ZARA-3 INSEAD
ZARA’S FAST FASHION BUSINESS MODEL
ZARA-4 INSEAD
RELATIVE POSITIONING
Price +
Benetton
Conventional high street fashion
Gap
Fashion - Fashion +
ZARA
H&M
Price -
ZARA-5 INSEAD
LET’S UNDERSTAND ZARA
► Compare and contrast how each of the above steps (the business process) is different at ZARA
as compared to a regular apparel retailer (think: clothing section at Marks and Spencer, high-
end clothing). Think over and above the market they are targeting.
ZARA-6 INSEAD
CONVENTIONAL BUSINESS MODEL IN FASHION RETAIL
HOW LONG IS THE ENTIRE CYCLE?
HOW MANY SELLING SEASONS?
WHO MAKES THE DECISIONS OR ORDERING AT EACH STAGE?

BASED ON WHAT INFORMATION
ZARA-7 INSEAD
MARKS & SPENCER: CUSTOMER PROPOSITION AND BUSINESS PROCESS
Business Process
► Design Team defines detailed cloth specifications one year before store delivery.
► Merchandisers decide prices and quantities well before start of selling season.
► Stylists focus on quality improvements of traditional styles.
► Quality: High conformance to specification, quality of stitching, etc.
► Many Distributed Warehouses
► Seven Weeks of Inventory
Customer Proposition
► High “Quality” garments, traditional styles, plus sizes.
Clock-Speed
16 month cycle
ZARA-8 INSEAD
CONVENTIONAL BUSINESS MODEL IN FASHION RETAIL
HOW LONG IS THE ENTIRE CYCLE?
HOW MANY SELLING SEASONS?
WHO MAKES THE DECISIONS OR ORDERING AT EACH STAGE?

BASED ON WHAT INFORMATION
WHAT ELSE IS DIFFERENT? WHAT DO YOU NEED TO SUPPORT THE

ABOVE DIFFERENCES?
ZARA-9 INSEAD
ZARA’S MARGIN AND INVENTORY TURNS
14
12
10
Inventory Turnover
4 Benetton
Inditex
H&M
2 GAP
0
- 0.10 0.20 0.30 0.40 0.50 0.60
Gross Margin
2003-2006 data
ZARA-10 INSEAD
PRODUCT FLOW AT ZARA/INDITEX
Retail Stores
Fabric and Inditex Sub- Inditex Ironing Spain
Components Factory: contractor: Factory: &
Cutting Sewing Finishing QA Arteixo DC
Portugal
Middle East
Outsourced Finished Goods:
Unironed Apyl
Zaragoza
DC
Rest of
Europe
Outsourced Finished Goods:
Ironed
Tempe
DC Americas
Footwear and Leather Asia

goods
Source: Phyllis Chu, 2005
ZARA-11 INSEAD
SOURCING & PRODUCTION
Sourcing
Comditel S.A. -Barcelona~40%
► Fabric -dyeing, patterning, and finishing of
► Spain, Portugal, Italy, Germany greige fabric for all of Inditex's chains
--Dying and printing process takes 1
► Glass beads, sequins, crystals, week
other components
► China
► Leather shoes, bags, garments Inditex Asia, Ltd Hong Kong
► Italy, Turkey, Morocco --Synthetics and fashion fabrics
► Silk Zara Asia, Ltd from Hong Kong
S--ynthetics and fashion fabrics
► India, Italy
Source: Zara: El Modelo de Negocio de Inditex, Fernando Fabrega 2004
Production
► 60-80% of all of Inditex’s products produced in Europe
► 20 Inditex factories (capital intensive/automation) and >400 subcontractors (mostly
in Galicia and Portugal, manual sewing ops)
► 20-40% produced in Asia, Africa, Latin America (Basic Items/ Finished Goods)
► More price sensitive than time sensitive
► Commitments made approximately 6 months prior to season
Source: HBR Rapid Fire Fulfillment & Zara: El Modelo de Negocio de Inditex, Fernando Fabrega 2004
ZARA-12 INSEAD
ORDERING AND STORE OPS
► Store managers use Casio PDA’s to place orders 2x per week to

distribution centers
► Locus of decision making (store, DC, central DC, corporate SC
planning team?)
► Garments are received floor ready

► With tags, ironed, and on hangers
► Items unsold within 2-3 weeks are sent out of the store
► Deliveries are made 2x weekly to stores

► Europe within 24 hours of order placement
► Americas within 48 hours of order placement
► Japan within 72 hours of order placement
Source: HBR Zara Fast Fashion, Rapid Fire Fullfillment ,Zara: El Modelo de Negocio de Inditex, Fernando Fabrega 2004
ZARA-13 INSEAD
ZARA: CUSTOMER PROPOSITION & BUSINESS PROCESS
Business Process
► Design Team develops platform models but holds off finalizing detailed design
► Single production. Production often falls short. New styles pick up unsatisfied demand.
► Internal Raw Material finishing capabilities, allowing for last minute changes
► Production is consolidated in Zara Industrial Areas (close to markets- Mexico, Spain)
► All new products every month
► Word of Mouth Advertising. Location, Location and Location!
Customer Proposition
► Trendy Stylish products, very affordable prices. Buy new styles often, high turnover.
Clock-Speed
21-29 day cycle
ZARA-14 INSEAD
CAPACITY BUFFERS & FLEXIBILITY
► Extra Capacity at most

stages
► DC’s operate at less than
50% capacity
► Factories only run for
one shift in routine
times
► Extra capacity to achieve
rapid response
► Extra capacity to meet
demand at peak periods
Source: HBR Rapid Fire Fulfillment
ZARA-15 INSEAD
ZARA VS. M&S: KEY BUSINESS PROCESS DIFFERENCES
Marks & Spencer Zara
Quantity Decision made 1 year before sales Quantity Decisions made days before sales
Multiple, Distributed Warehouses Centralized Warehouse
Platform designs are frozen and flexible capacity

Detailed Designs are frozen well in advance is booked. Detailed Product differentiation
decision is postponed.
Demand is satisfied , leftover inventory is heavily Full Demand is often unsatisfied, Customers
discounted substitute other products.
Distant Production Location, Long lead times Close Production Location, Short lead times
7 weeks of Inventory Few days of inventory
ZARA-16 INSEAD
FORECASTING TRUMPET AND FASTER REPLENISHMENT
Demand Forecasts for short time horizons are usually more precise
ZARA-17 INSEAD
ILLUSTRATIVE EXAMPLE: VALUE OF TIME (INFORMATION)
GRAPHS INDICATE FORECAST ACCURACY, CLOSENESS TO 45O LINE INDICATES HIGHER ACCURACY
A Year Before Demand 3 Months Before Demand 2 Weeks before Demand

0.025 0.025
0.025
0.02 0.02 0.02
0.015 0.015 0.015
0.01 0.01 0.01
0.005 0.005 0.005
0 0 0
0 50 100 150 200 0 50 100 150 200
0 50 100 150
Demand Forecasts become better with time as more information becomes available
Firm 1 orders 1 year ahead of Demand.

It’s forecast is D ~ N(Mean, Variance1). Compute Newsvendor Quantity and Profit
Firm 2 orders 1 month ahead of Demand
It’s forecast is D~ N(Mean, Variance2), where Variance2 < Variance1. Compute NV Qty & Profit.
Mismatch Costs will be higher for Firm 1, Newsvendor profits for Firm 2 will be higher!
Delaying Ordering Decisions reduces demand-supply mismatch, and increases profits
ZARA-18 INSEAD
AGGREGATING DEMAND FROM TWO MARKETS (GEEK ALERT)
Market 1 Market 2
DEMAND ~ NORMAL (Μ, σ) DEMAND ~ NORMAL (Μ, σ)
Combined demand for the two markets ?

► The mean demand will be 2*μ, but the combined standard deviation depends on the correlation between the two variables
► Correlation refers to how one random variable’s outcome tends to be related to another random variable’s outcome.
20 20
18 20
18 18
16 16 16
14 14 14
12 12 12
10 10 10
8 8 8
6 6 6
4 4 4
2 2 2
0 0 0
0 5 10 15 20 0 5 10 15 20 0 5 10 15 20
NEGATIVE CORRELATION NO CORRELATION / INDEPENDENCE POSITIVE CORRELATION
Normal (2µ, σ (2(1+ρ))1/2)

Random demand for two products (x-axis is product 1, y-axis is product 2). In scenario 1 (left graph) the correlation is –0.9, in scenario 2 (middle graph) the correlation is -0 and in scenario
3 (the right graph) the correlation is 0.90. In all scenarios demand is Normally distributed for each product with mean 10 and standard deviation 3.
ZARA-19 INSEAD
ILLUSTRATIVE EXAMPLE: VALUE OF CENTRALIZED WAREHOUSE (LOCATION POOLING)
Consider two supply chain designs
Decentralized Warehouses
Demand from Territory 1: Normal (µ, σ)
This is the effective demand for W1. Compute NV Profits
W1 Territory 1
Demand from Territory 2: Normal (µ, σ)

W2 Territory 2 This is the effective demand for W2. Compute NV Profits
Centralized Warehouses
Demand for Warehouse is sum of Demand from Territory 1 and
Territory 1 Territory 2.
W Effective Demand: Normal (2µ, σ (2(1+ρ))1/2), where ρ is the
Territory 2
correlation between the two markets.
Compute NV Profit for Warehouse.
The Variance in Pooled Demand is less than 2x the variance in original demands!
Having a centralized warehouse, reduces effective demand variance and increases profits
ZARA-20 INSEAD
ILLUSTRATIVE EXAMPLE: VALUE OF PLATFORM DESIGNS (DELAYED DIFFERENTIATION)
Early Differentiation Process†
Dyeing Knitting

Knitting Dyeing
► In an early differentiation process, dyeing and knitting quantities have to be decided on the basis of individual
demand for colors.
► In a delayed differentiation process, knitting quantities can be decided on basis of total demand.
► Total Demand has lower variance than the sum of variances of individual demand. This will reduce total mismatch
costs and increase profits
Delayed Differentiation also leads to pooled demand, and reduces mismatch costs!
Anand K. S. and K. Girotra, “The Strategic Perils of Delayed Differentiation,” Management Science, 53:5, May 2007, pp. 697-712.
ZARA-21 INSEAD
DELAYED DIFFERENTIATION: OTHER EXAMPLES
Paint Industry
Early Differentiation Process
color pigments, retail sales

paint mixing,
packaging
color pigments retail sales,

paint mixing,
packaging
ZARA-22 INSEAD
FASHION INDUSTRY: SUBSTITUTABILITY, DISCOUNTS & CONSUMER BEHAVIOR
► Consider the underage and overage costs of the newsvendor model (In all previous
examples, we considered effects of the demand distribution)
► In fashion goods, customers can be encouraged to substitute. Thus, if one product is

out of stock, customers may purchase another product. Zara recognized this and
modified its Newsvendor logic to consider a smaller cost of lost sales (cost of
underage). Thus, it stocks less.
► Stocking less reduces the amount of inventory that has to be discounted. Further, in
principle, Zara has very limited sales
► This further modifies customer behavior to buy early rather than wait for the sale.
Customers can also be fashion forward this way.
► All this increases Zara’s total profits!
Overage and underage costs can be modified by changing customer behavior
ZARA-23 INSEAD
ZARA’S SUCCESS
THE CHALLENGE FOR ZARA TODAY: COMPETITORS ARE CATCHING UP
► The Dilemma of game changing innovators.

► Zara continues to relentlessly drive further
efficiencies from its newsvendor game.
ZARA-24 INSEAD
REDUCING DEMAND UNCERTAINITY (SUBSCRIPTION, PREDICTION, RECOMMENDATION)
ZARA-25 Source: StitchFix INSEAD

REMEMBER CAPACITY BUFFERS & FLEXIBILITY
► Extra Capacity at most

stages
► DC’s operate at less than
50% capacity
► Factories only run for
one shift in routine
times
► Extra capacity to achieve
rapid response
► Extra capacity to meet
demand at peak periods
Source: HBR Rapid Fire Fulfillment
ZARA-26 INSEAD
SMALL BATCH MANUFACTURING & CAPACITY BUFFERS
Through a Marketplace (More on this in Session 12)
ZARA-27 INSEAD
KEY LESSONS
► Faster replenishment cycles allow one to use better forecasts and reduces information risk and
inefficiency. Also delayed decisions
► Pooling Demand from various locations (Centralized Warehouse) reduces variance in demand
► Delaying Differentiation of products (platform designs) allows one to bet on joint demand for
different product variants, which has lower variance
► Fewer discounts can change customer behavior and encourage immediate purchases and
substitution.
From Cost Focus to Risk Focus – Reduce Information Risks by Speeding-up
► Business models are hard to copy; their success is dependent on many BMI working together
ZARA-28 INSEAD
ALIGNING INCENTIVES
AND
SOURCING STRATEGIES
RISK!
INFORMATION RISK: DECISIONS MADE WITH POOR INFORMATION
SESSIONS 3-7:
NEWSVENDOR MODEL, FORECASTING, FLEXIBLE CAPACITY,
“FAST FASHION” BUSINESS MODELS
INCENTIVE RISK: DECISIONS MADE WITH SELF INTEREST (AS OPPOSED TO TEAM INTEREST)
SESSIONS 9-10:
MANAGING SUPPLIER RELATIONSHIPS, SOURCING,
SC CONTRACTING
MANAGING INFORMATION & INCENTIVE ALIGNMENT RISKS IS KEY TO SUCCESS
JIRO-2 © VILLE SATOPÄÄ INSEAD

JIRO’S VALUE CHAIN
Suppliers Internal Customer

Processes Value
► What are the key features of how Jiro structures and runs his
entire value chain?
RISKS
SCALEUP POTENTIAL?

SOURCING

SOURCING-CONVENTIONAL
Analyze Requirements Identify Evaluate Negotiate Manage
Estimate Identify suppliers Evaluate potential With a selected few Contract

requirements for in the market suppliers on management
Long term vs.
procurement different attributes
Develop transactional Performance
Classify items to be RFI/RFQ/RFP relationship monitoring
procured and
Retrospective spend
sourcing strategy
analysis
Category spend
analysis
JIRO-6 © VILLE SATOPÄÄ 6

INSEAD
SOURCING: CHANGING RELATIONSHIP STRUCTURES
Customer Service
Customer Service
Distribution
Distribution
Operations
Marketing
Marketing
Finance
Finance
Operations
MIS
MIS
Sales Buyers
Traditional Butterflies Buyer-Seller Relationship
Source: David F. Pyke. M. Eric Johnson. Tuck School of Business at Dartmouth. April 8, 2002. Working Paper
Adapted by Wang 2005, MIT-Zaragoza MS Thesis

SOURCING: CHANGING RELATIONSHIP STRUCTURES
Marketing Marketing
Distribution Distribution
Finance Finance
Customer Service
Customer Service
Operations
Operations
MIS MIS
Sales
Buyer
New Diamonds Buyer-Seller Relationship
Source: David F. Pyke. M. Eric Johnson. Tuck School of Business at Dartmouth. April 8, 2002. Working Paper
Adapted by Wang 2005, MIT-Zaragoza MS Thesis

How many suppliers ?

INSEAD
LARGE SUPPLIER BASE ?
► More suppliers more buying power

► “Leverage buying power” mantra
► Zero-sum game ?
► Increased number of suppliers broadens the range of choices
► Risk diversification

INSEAD
SMALL SUPPLIER BASE ?
► Interacting with each supplier entails a coordination cost

► More transactions mean higher transaction cost
► Supplier relationships become transactional
► Lower non-contractible investments by suppliers, such as quality, information sharing, design help
and innovation
► Reducing the number of suppliers
► increases their ex post bargaining power
► increases their ex ante incentives to make non-contractible relationship-specific investments
► Suppliers become "partners" who can reasonably expect to share in the gains from investment,
instead of "contractors“
► Procurement cost vs. Goodwill cost tradeoff
► DMS problem for near end of life cycle products

INSEAD
ONE SIZE FITS ALL ?
► Product category dependent decision on number of suppliers

► Factors to be considered:
► Strategic value of the procured component in the final product
► Level of maturity in the market
► Innovation vs. cost dynamics

INSEAD
Supplier Selection

INSEAD
TYPICAL ATTRIBUTES IN SUPPLIER SELECTION
► Product quality
► Lead time
► Delivery reliability
► Product costs
► Cost transparency
► Commitment to sustainability
► Technical capabilities of the supplier
► Service after the sale
► Financial strength of the supplier
► Location of the supplier
► Development capability
Note: not a comprehensive list

INSEAD
SELECT BASED ON MINIMUM TOTAL COST OF OWNERSHIP
Easier to
Identify Unit
Price
Freight Duties
Costs
Planning
Warehousing
Purchasing
& Inventory
Quality Control
Lead Time Poor Quality Late Delivery

Impact Costs Costs
Harder to Identify,
Measure and Relate
Field Failures Service General Admin.
to Purchase
Adapted from: Lisa Ellram, Total Cost of Ownership, CAPS, 1993

THE TOYOTA WAY

“KEIRETSU” APPROACHES TO SUPPLIER MANAGEMENT IN THE AUTOMOTIVE INDUSTRY
Keiretsu approach:
► Learn every part of suppliers’
business.
► Maintain joint ventures and
stakes in suppliers.
► Develop suppliers’ technical
and innovation capabilities.
► Supervise and benchmark
suppliers.

TOYOTA SUPPLIER RELATIONSHIP BLUEPRINT
► Understand how suppliers work before commitment

► Learn about suppliers by giving them small orders with set targets
► “Toyota invests significant time and effort—often three to five years—in exploring a
relationship with a supplier before signing a first contract.” (Getting to Win-Win: How Toyota
Creates and Sustains Best-Practice Supplier Relationships, BCG report)
► Add flexibility to long term relationship

► Develop two to three long-term suppliers for each component
http://www.autonews.com/article/20160208/OEM10/302089968/dual-sourcing-dilemma-hits-japan-again
► Treat suppliers fairly and take responsibility for their development and growth
► “All performance-improvement programs are supported by highly qualified Toyota engineers
who work as consultants—often on fast-track management career paths. Significantly, any
cost savings that suppliers achieve do not transfer directly into lower prices to Toyota. And
Toyota typically shares cost savings with the supplier during the next annual cost review.”
(Getting to Win-Win: How Toyota Creates and Sustains Best-Practice Supplier Relationships, BCG report)

BOSE JIT II
► Better Sound through Research!

► High quality, innovative audio products
► Bose JIT II
► Long term relationships with suppliers
► Information sharing by integrated IT systems and implants

BOEING AND NORDAM: INNOVATING TOGETHER
► Five years of joint testing and

development of a new window frame for
the Dreamliner
► "We’re proud to be an innovator for the

787 window frames," NORDAM CEO
Meredith Siegfried said. "It’s gratifying
to help Boeing achieve performance
targets for this aircraft, as well as
distinguish ourselves as a leader in
manufacturing advanced-composite
structures.”
► 20,000 window frames developed so far

SOURCING IKEA RUGS: JOINT IMPROVEMENT ACTIVITIES AND SUPERVISION
Align tactical response to firm's strategy:

► Only abandoned existing suppliers if they were
unwilling to address child labor
”create a better everyday
life for the many people”
► Addressed root causes: "Alternative Learning
Centers" together with UNICEF to help children
bridge into traditional schools. Over $200million
invested
Marianne Barner
Child labor in rug industry report

ECOVADIS: TRANSFORMING SUPPLIER RIVALRY INTO OPPORTUNITY
► Ecovadis is a sustainability rating agency (>50mm euros in revenue per year) founded
by Sylvain Guyoton (MBA 02D) and Pierre-Francois Thaler (MBA 99D)
► Targets procurement officers in large corporations (Coca-Cola, Adecco, Heineken, J&J,

L’Oreal, etc)
► Suppliers must provide “proof” for various sustainability criteria
► Used by buyers as a way to promote compliance, competition and transparency

among supplier base.
► Has rated over 10,000 companies

ALTERNATIVE SOURCING STRATEGIES
► Buy from Spot Market or Short Term Contracts

► Source product by RFP or from a spot market.
► Compare, evaluate and select between alternate suppliers on a continual basis.
► E-procurement, Procurement Auctions, Spot Market purchases
► Internet enabled price discovery mechanisms have made this much more common
► Common in US Manufacturing
► Buy (Long term relationships)

► Designate a supplier. Source from the same supplier again and again.
► Contract terms may vary with time.
► Relationship specific investments are often made.
► Cross ownerships, integrated ERP systems, embedded employees.
► Common in Japanese Manufacturing.
► Make (Vertical Integration)

► Produce inputs in house
► A captive production facility
► Examples: Oil Refineries with dedicated oil fields; Pit-head Thermal Power Plants

COVID CAPSULE
► VACCINE MANUFACTURING and SOURCING

► VIALS,
► LIPIDS,
► LIPID NANO PARTICLE,
► OTHER EXCIPIENTS,
► ADJUVANTS
► ANITGEN PRODUCTION
► FILL & FINISH
► For each what would be your sourcing strategy?

► Buy using Short Term Contracts
► Buy by building Long term relationships
► Make (Vertical Integration)

MAKE
This is a vertical integration approach
► Easier to ensure highest quality within organizational boundaries (+)

► Timely and reliable supplies are easier to ensure (+)
► Market forecasts and other Information flows are seamless (+)
► Both the “supplying” and “buying” unit are working towards the same goal (+)
► The supplying unit has the full incentive to invest in capacity for new products with
unproven demand (+).
► The supplying unit has incentives to invest in producing custom parts (+)
► There is less potential for information leakage to competitors. (+)
► There are lower economies of scale: These increase prices/costs and decrease
investments in R&D (for the component), ability to learn.
► Benefits of Focus are lost. Firms tend to become very diffused and unfocused in their
activities.

BUY-SHORT TERM CONTRACTS
In a one-shot interaction, a supplier has the incentives to opportunistically cheat the
buyer on all items that are not covered in their contract or are legally enforceable.
Thus,
► Quality verification is hard for most products. For commoditized or standardized products, it may
be possible (-)
► Timely delivery may be hard to ensure (-)
► Market forecasts and information sharing is often not possible (-)
► Incentives are not aligned. Each firm, the buyer and the supplier want to maximize their own
profits at the expense of the other (-)
► Suppliers have no incentives to invest in capacity for parts for new products with as yet unproven
demand (-)
► Suppliers have no incentive to invest in capacity for custom parts (-)
► Proprietary Information may leak from the supplier to other competitors (-)
► Full benefits of economies of scale. In cost and innovation for supplied components (+)
► Each supplier can be a highly efficient focused firm (+)
► In each period, one can buy at the cheapest price (+)
► Reputational concerns limit the harm to the buyer (+)
In an one-shot interaction, if a supplier can gain by harming the buyer, they would do so!

BUY-LONG TERM RELATIONSHIPS
► If both the supplier and buyer believe they have a profitable long-term relationships,
then they are inclined to not cheat one another; else either could break off the
relationship and the other would loose all future benefits.
► A short-term contract is like a prisoner’s dilemma; whereas a long-term contract is a
repeated prisoner’s dilemma (where one can induce full cooperation)
► Now one can expect for most reasonable demands to be met:
► Easier to ensure higher quality (as incentives to cheat are smaller) (+)
► Timely and reliable supplies are ensured (+)
► Market forecasts and other Information flows can be shared without fear of leakage
(+)
► The supplying unit has higher incentives to invest in capacity for new products with
unproven demand (+).
► The supplying unit has incentives to invest in producing custom parts (+)
► There are the full benefits of the economies of scale (+)
► Each supplier can be a highly efficient focused firm (+)
► One is stuck with a supplier (-)

WHEN DO LONG TERM RELATIONSHIPS WORK?
► These contracts can achieve the best of both worlds! But there are certain conditions
to make this all work.
► Both parties should believe the relationship is going to last for the long-term.
► Both parties should find the collaboration sufficiently rewarding.
► Both parties should be interested in maximizing long-term gains.
► Requiring both parties to make relationship-specific up-front investments is a
good way to ensure these
► One problem with these relationships: Suppose both firms believe it is a long-term
relationship and make relation-specific investments. Now both firms are stuck with
each other (holdup). They could demand unreasonable terms which the other party
will find preferable to meet than have to incur the cost of setting up a new long-term
relationship (Renegotiation). This can be prevented as long as both firms have made
enough comparable relation-specific investments.
► Long Term Relationship if executed correctly, can get all the benefits of buying and
mitigate most of the downsides.

KEY LESSONS
► Information and incentive risks are key to understanding a business model (and
making it work)
► Incentives issues in SC sourcing can be minimized by aligning decision horizons
► Each raw material, product or service that needs to be sourced should be matched
with an appropriate sourcing strategy

ALIGNING INCENTIVES
ALIGNING PAYMENT STRUCTURE

CONTRACTS
SERVICIZATION
INCENTIVE RISKS
Incentive Risks
in Supply Chains
Sourcing Strategies Incentive Alignment

Jiro: aligning incentives by aligning via Contracts
horizons
Aligning incentives by sharing risk
and aligning objectives
INCENTIVES -2
INSEAD
DOUBLE MARGINALIZATION
► Even if players don’t cheat, self

interested behavior can lower profits in
the supply chain
► Phenomenon is known as “double

marginalization”
INCENTIVES -3
INSEAD
NOT “SHANGHAI SURPRISE” AGAIN!!
► How many times have you gone to your nearby video rental store and found
that copies of the movie title you are looking for have all been rented out ?
► You wonder why the rental store does not stock more copies of the new
release movies
► Is it because the DVDs are expensive to stock.
► The overstocking risk is high?
► Or is that the rental stores knows that you will rent something else if you didn’t
find what you are looking for and hence they will not lose their margin. The “ If I
don't find Coke I am easily willing to settle for Pepsi” argument.
► The understocking risk is too low?
► Let us think through the cost economics of such a problem more carefully
using a simple example
INCENTIVES -4 4
INSEAD
BLOCKBUSTER
► Video rental Supply Chain
Blockbuster Customer Demand

$3/tape 0.12
Movie Studio
$65/tape rental 0.10
(Producer) 0.08
Probability
0.06
0.04
0.02
0.00
40000
44000
48000
52000
56000
60000
64000
68000
72000
76000
80000
84000
88000
Daily output (thousands of barrels)
► Supply chain faces information risk (uncertain customer demand)

► Break-even point for Blockbuster ~ 22 rentals x $3/tape rental
► “It was simply too expensive to stock enough copies of every movie the customer
requested… [30%] of people who walked into Blockbuster stores were walking out with
nothing”, Sumner Redstone, Viacom CEO
► Lack of SC coordination : Blockbuster faces entirety of information risk alone
Source: https://insight.kellogg.northwestern.edu/article/supply_chain_coordination_with_revenue_sharing_contracts
INCENTIVES -5
INSEAD
ALIGNING INCENTIVES: DOUBLE MARGINALIZATION
r $/unit
s $/unit .
.
+ .
Customers
Manufacturer = Retailer
Production Cost: c $/unit Demand: D units
Sale Price: r $/unit
Salvage value: s $/unit
Underage Cost 𝐶𝑢 = 𝑟 − 𝑐
Overage Cost 𝐶𝑜 = 𝑐 − 𝑠
INCENTIVES -6
INSEAD
r $/unit
.
w $/unit
.
.
Manufacturer Retailer Customers
Production Cost: c $/unit Sale Price: r $/unit

Salvage value: s $/unit Demand: D units
Wholesale Price: w $/unit
Underage Cost 𝐶𝑢 = 𝑟 − 𝑤 As w increases, Cu ↓
Overage Cost 𝐶𝑜 = 𝑤 − 𝑠 As w increases, Co ↑
INCENTIVES -7
INSEAD
► Centralized optimal order ► Decentralized optimal order

𝐶𝑢 𝑟−𝑐 𝐶𝑢 𝑟−𝑤
Pr 𝐷 ≤ 𝑄𝑐∗ = = Pr 𝐷 ≤ 𝑄∗ = =
𝐶𝑢 +𝐶𝑜 𝑟−𝑠 𝐶𝑢 +𝐶𝑜 𝑟−𝑠
As w increases, Cu ↓
𝐶𝑢 = 𝑟 − 𝑐 𝐶𝑢 = 𝑟 − 𝑤
𝐶𝑜 = 𝑐 − 𝑠 𝐶𝑜 = 𝑤 − 𝑠
As w increases, Co ↑
Since 𝑤 ≥ 𝑐, underage is reduced and we have that 𝑄 ∗ ≤ 𝑄𝑐∗
INCENTIVES -8
INSEAD
DOUBLE MARGINALIZATION: WHAT IS GOING ON?
► Retailer faces all the demand risk
► The higher the margin of the manufacturer, the less the retailer orders
► In general, if every firm chooses to maximize its own expected profit, the result is
lower sales, and lower total profit compared to the SC’s maximum profit.
How can double marginalization be mitigated?
INCENTIVES -9
INSEAD
REVENUE SHARING
► Pioneered revenue sharing contracts in

the 1990s
► Half the revenues from renting a movie

go to the movie studio
► Revenue sharing contract: a fraction of retailer revenues (salvage

and sales) is transferred to manufacturer
► Manufacturer now also faces demand risk
► Who is better off with a revenue sharing contract?
INCENTIVES -10
INSEAD
TERMS VARY
• $ 0.00 upfront fees
• 65/35% split (you keep 65%) vs $1.05 minimum transaction fee for full term
• Sell on 29th day (VHS cost = $2.00 vs 35% / DVD cost = $3.50 vs 35% – can sell 50%)
(VHS/DVD)
• 6 month lease term (cost = $0.50) if you want to purchase
• $0.00 up-front fee
• Sell on 41st day (VHS cost = $2.40 can sell all but one)
Source: Rentrak Website
INCENTIVES -11
INSEAD
TERMS VARY
• Sell on 28th day (VHS cost = $2.00 vs 39% / DVD cost = $3.00 vs 39%)
(VHS/DVD)
• $0.00 up-front fee
• Sell on 31st day (VHS cost = $1.70 DVD Cost = $ 3.35)
INCENTIVES -12
INSEAD
REVENUE SHARING CONTRACT
► The revenue sharing contract transfers part of the demand risk from retailer to
manufacturer
► But since manufacturer has low cost of production, it is better at handling this demand risk
than retailer (Blockbuster) on their own
► Risk is being transferred to the supply chain party that can best handle it
► “Size of the pie grows”

► Contracts are not a zero-sum game
► Both parties can be better off from a well-designed contract
INCENTIVES -13
INSEAD
ORDER MINIMUMS
To illustrate typical structure only. A detailed ordering matrix can be obtained from Rentrak website for each studio
INCENTIVES -14 14
INSEAD
FURTHER DISCUSSION
► If we claim that with revenue sharing the video stores will stock the optimal
quantities, why did the studios have to specify order minimum quantities in
contract structure?
INCENTIVES -15 15
INSEAD
16
REVENUE SHARING VS. FIXED PRICE
A titles
B titles
C titles
Source: Mortimer 2002, Working Paper
INCENTIVES -16
INSEAD
BUYBACK CONTRACTS
► Manufacturer commits to buying unsold goods from retailer

► Reduces overage costs
► Like revenue sharing contracts, buyback contracts can achieve supply chain alignment
► Part of demand risk shifts to manufacturer, who can bear it best
► Common in book printing industry
► When to implement revenue sharing versus buyback contracts?
► Revenue sharing contracts

► Require a process for monitoring revenues
► Long term relationship
► Buyback contracts
► Require an effective reverse logistics system
► May increase logistics costs
► What are the retailer incentives when selling a mix of products under regular wholesale contracts
and under revenue sharing/buyback contracts?
INCENTIVES -17
INSEAD
CONTRACTS: NOT JUST ABOUT COSTS!!!
► There are many types of contracts

► Revenue sharing
► Buyback
► Trade credits (temporal alignment)
► Quantity flexibility
► Price protection
► When negotiating contracts with a supplier, customer, or any other service provider,
don’t think just about costs. Consider also how the contract is shifting risk between the
parties.
INCENTIVES -18
INSEAD
SERVICIZATION AT ROLLS ROYCE : POWER BY THE HOUR
► A typical relationship: Time & Materials Contract (pay per repair)

► Incentive problems:
► The supplier (Rolls Royce) wants more repairs
► The customer (airline) faces all the risks associated with engine breakdowns and wants fewer
repairs
► The customer’s costs are unpredictable
► The customer is forced to buy spare parts while what it wants is the working engine
► The alternative: Power-By-The-Hour contract

► Pay per flying hour
► Ensure predictable costs
► Both parties have the same incentives
► Rolls-Royce takes over the risk of break-downs…
► …but it is better positioned to handle it!
INCENTIVES -19
INSEAD
SERVICIZATION
THE INNOVATION
► Rather than sell the product, Sell the service that the customer cares about…
► Offer “Power by the hour” Style contracts; lease instead of sale
► Increases Revenues and Increases Risk Exposure, Risk is borne by Party that can best
manage it.
WORKS BEST WHEN

► Customers do not want to own assets, but only want service
► Customers don’t want to bear risk of asset productivity
► You can bear the risk better– Risk Pooling, Private Information, Aligned Incentives
INCENTIVES -20
INSEAD
SERVICIZATION AND SUSTAINABILITY: REDUCING USE OF MATERIALS
► Dupont and Ford UK: paying for painted car instead of paint
► Resulted in less paint use
► Also resulted in higher integration and paint technology innovation
INCENTIVES -21
INSEAD
SERVICIZATION AND SUSTAINABILITY: INCREASING ENERGY ADOPTION
► M-Kopa sells solar systems as a service in Kenya

► Powers over 600,000 homes in East Africa
► Installs the panel and has a GSM + Mobile payment system
INCENTIVES -22
INSEAD
RISK SHARING IN THE PHARMACEUTICAL INDUSTRY
► Outcome-based (value-based) pricing for new drugs

► Tie price charged per treatment to outcome: if the drug does not achieve desired outcome,
payer pays nothing, or receives a rebate
► Pros: drug manufacturer bears outcome risk
► Cons: how to disentangle effect of drug from other confounding factors?
Source: https://www.fastcompany.com/40461214/how-novartis-is-defending-
the-record-475000-price-of-its-pioneering-gene-therapy-cancer-drug-car-t-kymriah
Source: https://news.aetna.com/
news-releases/aetna-and-merck-sign-a-
unique-value-based-contract-for-januvia-and-janumet/
INCENTIVES -23
INSEAD
COVID CAPSULE
► Manufacturer of therapeutics and vaccines for COVID-19 may under-invest in

manufacturing capacity
► Cost of underage?
► Cost of overage?
► Who bears the most significant cost of underage?

► Who bears the cost of overage?
► Who is best positioned to bear the cost of underage?

► What contracts could that party used to align the incentives?
INCENTIVES -24
INSEAD
KEY LESSONS
► Supply chain can function inefficiently because of incentive mis-alignment
► Supply chain contracts can shift risk to the party best equipped to handle it
► SC alignment is NOT a zero-sum game, all parties can be better off!
► Servicization (“x-as-a-service”) can help reduce costs but, more importantly, help both
seller and buyer share risks and align objectives. Works when:
► Customers do not want to own assets, but only want service
► Customers don’t want to bear risk of asset productivity
► You can bear the risk better– Risk Pooling, Private Information, Aligned Incentives
INCENTIVES -25
INSEAD
INTERMEDIARIES/MARKETPLACES
HOW PLATFORMS ARE CHANGING THE WORLD!

THE MARKETPLACE THAT STARTED IT ALL
SEARCH, PRICE DISCOVERY
INTERMEDIARIES-2 INSEAD
Sellers Buyers
Product A 50% Product A
50%
50%
Product B 50% Product B
STATISTICAL POOLING
Sellers Buyers
Fraction of buyers/sellers matched
90
Fraction Matched (%)

80
70
60
50
0 10 20 30 40 50
Size of Market
► Marketplace advantage #1: Statistical Pooling increases the number of matches
ON DEMAND SERVICES AND TWO SIDED MARKETPLACES
ONE SIDED & TWO SIDED MARKETS
► Homogenous set of network actors = side of market
Network effects ► A same-side effect for each

► Increasing returns to side, i.e., preference
scale in consumption regarding number of other
(network effects) users on own side
► A market exhibits ► A cross-side effect in each
network effects when direction, i.e., preference
regarding number of users
the value of an
on other side
additional transaction is
► Each effect can be positive or
higher when more units
negative
change hands,
everything else being
equal
TWO-SIDED MARKETS AND PLATFORMS
Taxi Seekers Rochet & Tirole, 2004 Taxi Drivers
Job seekers Job Providers

Same-Side
Network
Effect
Two-Sided Market User Set 2 Seller
User Set 1
Buyer Platform
Patients Cross-Side Network Effect Doctors
Readers Publishers
Large and Growing Share of Global Economy. 60 of the world’s 100 largest companies earn most of their revenue
from platform-mediated networks
CHALLENGES TO TWO SIDED MARKETPLACES
Uber’s Biggest Danger Is Its Business Model, Not

Bad PR
Helsinki’s Uber for Buses Is Stuck in First Gear
Startups Want to Be the Next Airbnb, Uber
Pitfalls of Building the “Uber for XYZ”
► Challenges
► Oversimplification is common
► Over- or underestimate strength of network effects
► Price to network’s sides as if they were separate markets
► Failing to scale & the Penguin problem
► Competition among marketplaces (Uber vs Lyft)- Switching costs
► Regulation
WHY IS SCALING HARD?
► Higher probability of
ride request accepted.
► Reduced waiting time
► Less time waiting for

Passengers
► Higher Revenues
No side wants to join without the other

Chicken and Egg problem
GROWTH PILLS AND ACCELERATED GROWTH STRATEGIES
Incentivize Riders
► Encourages more Riders
to join
► Encourages more Drivers
to join
Incentivize Drivers
► Encourages more Drivers
to join
► Encourages more Riders
to join
Effectiveness of Incentives function of same side response and cross-externality
Which side to focus on more first?
Customer/supplier engagement model, acquisition & retention
INTERMEDIATION REDUCING SEARCH COSTS: PLAY MATCHMAKER!
THE INNOVATION
► One Stop shop to match two distinct customer segments
► Create Value by reducing Search and trans costs, Share value created
► Reduce Costs, Risk of not finding a partner, Increase Exposure to Opportunism Risks
WORKS BEST WHEN

► Disaggregated Buyers, Sellers
► High Search Costs
► Low Potential for Opportunism
COVID CAPSULE
Buyers : Hospital/Health Systems
Sellers : Small to mid sized PPE

manufacturers
Which side to focus on more first?
INTERMEDIARIES & BUSINESS MODEL INNOVATION: LI & FUNG LIMITED
120
Sales (HK$ B)
80
40
E. Belavina, and Girotra, K., “The Relational Advantages of Intermediation”, Forthcoming at Management Science
A HARD CHOICE: FLEXIBILITY VERSUS COMMITMENT
SHORT TERM RELATIONSHIPS LONG TERM RELATIONSHIPS
► Flexibility to choose best party, each period ► Relation Specific investments

► No relation specific investments ► Stuck with one party
Pooling risks for both sides
HOW TO BREAK THE TRADE-OFF?
Intermediary
► Intermediaries relieve the trade-off by allowing for dynamic buyer-supplier matching while keep
all relationships intact. The benefits of relationships flow through a 3 tier system and there is
additional flexibility
PLATFORMS FOR MATCHING AND LIMITING OPPORTUNISM
THE INNOVATION
► Match two distinct customer segments
► Create Reputation Systems, and Long-Term relationships with both segments
► Create Value by reducing Search costs and Limiting Opportunism Risks
► Reduce Costs, Risk of not finding a partner, Exposure to Opportunism Risks
WORKS BEST WHEN

► Disaggregated Buyers, Sellers; Buyer and Sellers with long-term perspective
► High Search Costs
► High Potential for Opportunism
SUMMARY
► Successful marketplaces/intermediaries
► Increase the number of matches through statistical pooling
► Secure transactions
► Allow for price discovery
► Reduce transaction costs
► Combine benefits of flexibility and long term relationships
► Network effects (distinct from supply side economies of scale)
► Two sided marketplaces- Cannot price as two separate markers
► Which side to incentivize for early customer acquisition requires careful
consideration
► Read the Fore-Firestone case for next time!
THE CO$T OF QUALITY
FORD-FIRESTONE
QUALITY MANAGEMENT:
TACTICAL
FORD FIRESTONE-2 ©
INSEAD
VILLE SATOPÄÄ
WHAT IS QUALITY?
Good consistency, On target,

but not on target but poor consistency
hit the target on average (unbiased)

High Quality = and
with good consistency (low variance)
FORD FIRESTONE-3 ©
INSEAD
VILLE SATOPÄÄ
SOURCES OF VARIANCE
► Assignable Cause Variance vs. Common Cause (or Random) Variance
Common Cause Variation (low level)
Common Cause Variation (high level)
Assignable Cause Variation
► Identify and mitigate any assignable cause variance

► Common cause variance is what defines quality-capability of the process.
FORD FIRESTONE-4 ©
INSEAD
VILLE SATOPÄÄ
MEASURING PROCESS CAPABILITIES
 
min acceptable Design tolerance width max acceptable
FORD FIRESTONE-5 ©
INSEAD
VILLE SATOPÄÄ
MEASURING PROCESS CAPABILITIES
 
3 3
min acceptable Design tolerance width max acceptable
FORD FIRESTONE-6 ©
INSEAD
VILLE SATOPÄÄ
WHAT DOES PROCESS CAPABILITY MEAN IN TERMS OF DEFECT RATES?
Probability of a Defects per

Tolerance Limits
Defect Million
±1σ 0.317 317,000
±2σ 0.0455 45,500
±3σ 0.0027 2,700
±4σ 0.0001 63
±5σ 0.0000006 0.6
±6σ 2x 10-9 0.002
► Numbers are computed using the normal distribution.

► Probability of a draw outside the tolerance limits is given by the normal
distribution table.
FORD FIRESTONE-7 ©
INSEAD
VILLE SATOPÄÄ
DEPENDENT COMPONENTS
► Consider a product made of 1200 components

► The percentage of defect-free products is
Sigma P(No Defect)

±3σ 0.99731200 = 4%
±4.5σ 0.9999961200 = 99.6%
±6σ 0.9999999981200 = 99.9998%
FORD FIRESTONE-8 ©
INSEAD
VILLE SATOPÄÄ
COVID CAPSULE
QUALITY OF TEST KITS, N95 MASKS
LOT TESTING?
PRE-SHIPMENT INSPECTION?
QUALITY MANAGEMENT SYSTEM & CERTIFICATION?
HOW TO INCREASE COSTS OF POOR QUALITY FOR SUPPLIER?
FORD FIRESTONE-9 ©
INSEAD
VILLE SATOPÄÄ
QUALITY MANAGEMENT:
STRATEGIC
FORD FIRESTONE-10 ©
INSEAD
VILLE SATOPÄÄ
FORD-FIRESTONE
INSEAD
VILLE SATOPÄÄ
CASE OVERVIEW
► Set in the Fall of 2000;
► Against of backdrop of confusing information, as claims against Ford and

Firestone mount following a TV segment, we are asked to assess whether
the problem is critical enough to warrant a recall.
► As the case develops, it becomes clear that the problem is possibly with
one make of tire, made during 1996, and one plant. It is still unclear what
would have happened if the recall had not taken place.
► It is still amazing how difficult it was and still is to trace the root cause of
the problem until one realizes that the problem has several layers, and
very small numbers are involved.
► How can one manage these situations? What principles are available to
managers?
INSEAD
VILLE SATOPÄÄ
INSEAD
VILLE SATOPÄÄ
BACKGROUND: 1999
► Ford:
► Profitable;
► Most admired corporation;
► Fatalities: 1 in 100 million pkm1;
► Other SUVs: 1.3 in 100 mil. pkm fatalities;
► Fewer rollovers relative to vehicle population;
► Firestone:
► Global Market leader
► Profitable
► Indy tradition.
► “NHTSA received 46 complaints about Firestone tires. Over the same
period, the agency logged 970 complaints about Goodyear tires and 725
gripes about Michelins.”
1 A passenger-kilometer (pkm) is the product of the distance a vehicle travels times the number of occupants travelling that distance.
INSEAD
VILLE SATOPÄÄ
WHAT WOULD YOU HAVE DONE? WHEN TO RECOGNIZE?
► 1989-1990 (design stage), new tire (1996)
► 1997: Claims(?), Venezuela
► 1998: 21 cases, Saudi Arabia
► 1999: Ford memo to Firestone
► 2000 February TV Show
► 2000 May: 90 complaints + 4 deaths
► August 9, 2000 issued recall notice for 6.5 million tires (went up to 13 million
out of 47 million made in the 1990s)
► 2000 September: 2200 complaints + 103 deaths + 400 injuries
► Perspective: 3.6 million Explorers

► (0.00286% deaths, 0.06% complaints = 2.86/100,0000, 60/100,000)
Coincidentally Firestone said there was a problem with 2/100,000
tires made in Decatur plant
INSEAD
VILLE SATOPÄÄ
MAGNITUDE OF CRISIS 2000
Some of the recalled models (not just Explorers).
► RECALL COST
► 6.5 million tires $50 each ($325 mil) + legal costs + liability + lost profits
= $800 million to $1 billion
► Later data shows that these numbers were exceeded.
► Brand Image
► Most profitable segment for Ford
► 40% loss of sales for Bridgestone
► Etc.
Was this a real crisis?
INSEAD
VILLE SATOPÄÄ
20
40
60
80
0
100
120
140
1/5/1998 160
5/5/1998
9/5/1998
1/5/1999
FORD FIRESTONE-17
5/5/1999
9/5/1999
1/5/2000
5/5/2000
9/5/2000
1/5/2001
5/5/2001
9/5/2001
1/5/2002
5/5/2002
9/5/2002
1/5/2003
5/5/2003
9/5/2003
1/5/2004
5/5/2004
9/5/2004
©
1/5/2005
5/5/2005
9/5/2005
VILLE SATOPÄÄ
1/5/2006
5/5/2006
9/5/2006
1/5/2007
5/5/2007
9/5/2007
1/5/2008
5/5/2008
9/5/2008
1/5/2009
5/5/2009
9/5/2009
1/5/2010
5/5/2010
Ford
9/5/2010
1/5/2011
5/5/2011
9/5/2011
1/5/2012
Toyota
INSEAD
TREAD SEPARATION: INDUSTRY WIDE PROBLEM?
Manufacturer Recalled Fatalities Injuries Complaints

Tires
Firestone 6,500,000 150 500 3000
Goodyear No Recall 15 125 -----
Continental – 160,000 0 0 ------

General
Many tire makers have problems. So why bother?
INSEAD
VILLE SATOPÄÄ
EVALUATE: PARETO CHARTS
Percentage of defects
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
Cause 1 Cause 2 Cause 3 Cause 4 Cause 5
Pareto charts are used to graphically summarize the relative importance of various causes of
problems. The charts are based on the Pareto principle, which states that when several factors
affect a situation, few factors will account for most of the impact (80-20 rule).
INSEAD
VILLE SATOPÄÄ
WHO’S RESPONSIBLE?
Firestone argues that Ford …But Ford insists that

Explorers are more accident-prone Firestone tires are more likely
than Ford Rangers pickups to fail on Ford Explorers than
equipped with the same tires… any other tire brand.
1183
1456
PROPERTY DAMAGE AND EXPLORER TREAD

PERSONAL INJURY CLAIMS SEPARATION CLAIMS
FORD EXPLORER 1995-1997
FORD RANGER
285 250
3 2
With Firestone With Wilderness With Firestone ATX and With Goodyear
ATX AT Wilderness AT
Source: Bridgestone/Firestone, Inc. Source: Ford Motor Co.
INSEAD
VILLE SATOPÄÄ
EVALUATE: WHAT CONTRIBUTES THE MOST TO IT?
Not all of these are valid Pareto charts. Why?
INSEAD
VILLE SATOPÄÄ
MORE RED FLAGS?
INSEAD
VILLE SATOPÄÄ
Source: Krueger and Mas (2004)
INSEAD
VILLE SATOPÄÄ
TOYOTA’S SUDDEN ACCELERATION FIASCO – WHAT WILL BE THE FINAL COST?
2004 First claims (139) of sudden unintended acceleration.
9/26/2007 Toyota recalls 55,000 vehicles over floormats
2008 Two accidents involving Camry 2005, two people killed. First lawsuits.
8/28/2009 Four people killed when Lexus ES350 suddenly accelerates
9/29/2009 Toyota recalls additional 3.8M vehicles over floormats (largest Toyota recall ever)
11/3/2009 ABC News investigates Toyota Sudden Unintended Acceleration
11/6/2009 NHTSA asserts floormats do not address underlying defect
11/25/2009 Permanent solution for recalled floormats announced: reshape, replace pedals
1/12/2010 Toyota announces brake override systems for all vehicles by end of 2010
1/21/2010 Toyota recalls 2.3M vehicles over sticking accelerator pedals
1/26/2010 Toyota halts sales of eight models recalled over sticking accelerator pedal
1/28/2010 Additional 1.1M vehicles added to floormat recall
1/29/2010 Pedal recall expands to include 1.8M vehicles in Europe
1/31/2010 Class action lawsuit filed over recalled Toyotas
2/2/2010 Toyota sales slide 16%
7/13/2010 Early tests pin Toyota accidents on drivers (WSJ)
11/19/2010 Toyota settles lawsuit in fatal crash that launched recall saga
02/22/2011 NASA and NHTSA inquiry: no electronic faults in Toyota cars. Accelerator pedal
entrapments remains a problem. Toyota recalls an additional 2.17 million vehicles
INSEAD
VILLE SATOPÄÄ
http://abcnews.go.com/Blotter/toyota-pay-12b-hiding-deadly-unintended-acceleration/story?id=22972214
INSEAD
VILLE SATOPÄÄ
http://www.businessinsider.sg/r-toyota-to-recall-362000-camry-other-models-globally-2014-
11/#.VGVk4vnF9hE
INSEAD
VILLE SATOPÄÄ
60 MILLION CAR BOMBS: TAKATA'S AIR BAG CRISIS
http://www.bloomberg.com/news/features/2016-06-02/sixty-million-car-bombs-inside-takata-s-air-bag-crisis
INSEAD
VILLE SATOPÄÄ
OTHER INDUSTRIES: CHIPOTLE
INSEAD
VILLE SATOPÄÄ
SAMSUNG GALAXY NOTE 7: THE IMPORTANCE OF QUICKLY FINDING ROOT CAUSES
INSEAD
VILLE SATOPÄÄ
CROWD-SOURCING QUALITY CONTROL
INSEAD
VILLE SATOPÄÄ
RECAP
► There are tactical and strategic levels to quality

management
► Crises can and will happen, need processes to respond quickly and search
for the needle in the haystack
► Beware of early warnings
► Response processes should be in place before crisis happens
► Don’t let the data lie to you
► No silver bullets, think long and hard about your specific situation
INSEAD
VILLE SATOPÄÄ
NEXT TIME
► Wrap-up and Summary 10 min

► Innovation Exercise 3-minute presentations
INSEAD
VILLE SATOPÄÄ

All Session Merged Yadav

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

All Session Merged Yadav

Uploaded by

Copyright:

Available Formats

BUSINESS MODEL INNOVATION

& HERNIAS, HEARTS AND DIAPERS

Analyze and create superior business models that disrupt

MATH CREATES PRECISENESS OF THOUGHT AND BUILDS CHARACTER

Strictly no food in the classroom

Drinks must be in closed containers

Put mask back on after eating/drinking

Questions? Go to inse.ad/covid-19 / email prevent.coronavirus2019@insead.edu

► Read the Outline. Full Information. Clearly Defined Rules. No

WHO: DECISION MAKER WHEN: DECISION TIMING

WHO: DECISION MAKER WHEN: DECISION TIMING

Clear offering for customers Dependent on one market, risky, no diversification

Challenges in finding, retaining and motivating

Focused strategies place constraints on growth, innovation and risk-diversification

High Unit Cost

Rigid Well Defined

Matching the Product and Process is Vital

Job shop type kitchen.

Each dish is slightly

Multiple service Single service class

► Focused vs. Diversified

► Focused strategies deliver superior operational performance (lower variety risks)

► Focused strategies expose firms to market risk

WHO: DECISION MAKER WHEN: DECISION TIMING

QUESTIONS FOR YOU TO REFLECT ON

Pick your favorite Industry-Market

Are there focused & diversified players in this industry?

If yes, Who are they and what is the relative profitability?

Is there an opportunity for a focused player to enter the market?

BOTTLENECKS & CAPACITY MANAGEMENT

WHAT WE LEARN FROM THE GOAL

THE GOAL: 2-2

THE GOAL: 2-4

Welfare Capitalism Vertical Integration

THE GOAL: 2-6

Ford Rouge plant

THE GOAL: 2-7

Source: Detroit News/ Sent to me by Joseph Butler ZLOG Class of 2006

THE GOAL: 2-8 8

Source: Presentation by G. Long, UPS

THE GOAL: 2-9 9

THE GOAL: 2-11

THE GOAL: 2-12 12

Managing for the Suppliers: From Competition to

Session 9: WHY Relationships? * Extra: Read “Standing on the Shoulders of

THE GOAL: 2-14

Adapted from: John Kay - NSF Workshop

THE GOAL: 2-15

Platform to Match Supply with Demand

THE GOAL: 2-18

THE GOAL: 2-20

THROUGHPUT: The Rate at which the System Generates Revenue/Sales

OPERATING EXPENSE: The Rate at which the System Generates Costs

INVENTORY: The level of capital invested in system

►Money costs money

► Equity: Dilution of shareholding

$$ Going in the System $$ Coming from the System

THE GOAL: 2-21

Net Operating Profit after Taxes (NOPAT) = Throughput – Operating Expenses †

THE GOAL: 2-22

TWO KINDS OF INEFFICIENCIES

THE GOAL: 2-23

Step I Step II Step III

Service Time 2 Minutes/Unit 3 Minutes/Unit 1 Minute/Unit