Innovations that Will Change Supply Chain Planning During Your Career Optilon Supply Chain Conference September 4, 2014 Joseph Shamir, CEO, ToolsGroup
Innovations that Will Change
Supply Chain Planning
During Your Career
Optilon Supply Chain Conference
September 4, 2014
Joseph Shamir, CEO, ToolsGroup
Powerfully SimpleCompany Confidential All Right Reserved2
Five Innovations …..
In a dynamic world, leveraging additional data will
allow us to understand demand, despite its volatility1
Powerfully SimpleCompany Confidential All Right Reserved3
Stable distribution networks
Limited product range and
NPI
“continuous” advertising
Limited promotions
What has changed for Demand Forecasting
Traditional time series approach worked
SUPPLY CHAIN
BE
FO
RE
AF
TE
R
DEMAND FORECASTING
Dynamic demand allocation
Internet drives:
• intense demand shaping,
• extended of product
offering and NPI
• accelerated trends due to
social media Time series approach doesn’t work anymore
due to the high volatility and the high
relevance of market variables
Powerfully SimpleCompany Confidential All Right Reserved44 Company Confidential All Right Reserved
Understanding demandProcessing related data closer to the consumer
VENDOR
VENDOR DCs
RETAILER DCs
RETAILER Channel LEVEL (Hyper, Super, Discount,etc)
POS DATA
RETAILER Depots
Powerfully SimpleCompany Confidential All Right Reserved5 Company Confidential All Rights Reserved
Leveraging Downstream and Marketing Data
Powerfully SimpleCompany Confidential All Right Reserved66 Company Confidential All Right Reserved
Why Demand Forecasting should be automated
Today’s Demand:
• impacted by high number of
marketing variables and indicators
• managed as close as possible to the
end-customer
• Must be calculated at high frequency
Low Attributes Number
Low Frequency
High Attributes Number
Low Frequency
High Attributes Number
High Frequency
web Timely a reliable mid/long-term
forecast for S&OP collaboration
Robust short term forecast to
drive a responsive SC.
Scale + frequency
Automation to provide:
Su
pp
ly C
hain
Resp
on
siv
en
ess
Mu
lti -Ch
an
nel in
ten
sifie
d M
ark
etin
g
Powerfully SimpleCompany Confidential All Right Reserved7
Five Innovations …..
The “long tail” will continue to get longer in most industries – but
again data will address the problem
In a dynamic world, leveraging additional data will allow us to
understand demand, despite its volatility1
2
Powerfully SimpleCompany Confidential All Right Reserved8
Wayfair Example of long tail in e-commerce
Powerfully SimpleCompany Confidential All Right Reserved9 Company Confidential All Rights Reserved
0
20
40
60
80
week 1 week 2 week 3 week 4 week 5 week 6
Also the control frequency increases the long tail
0
200
400
600
800
1000
1200
week 1 week 2 week 3 week 4 week 5 week 6
What You See:
What Your Supply Chain Sees:
Weekly forecast by Product and Region (Thousand Island dressing, US North-west)
Daily Requirements by SKU and Warehouse (SKU 034512, Tacoma warehouse), colors are individual customer orders
Powerfully SimpleCompany Confidential All Right Reserved11
Line orders: example of data that should be used
Two “identical”
demand streams …
… requiring twice the safety stock on the left to provide the same level
of service.
… yield two different
forecasts of
uncertainty…
Powerfully SimpleCompany Confidential All Right Reserved12
Fast-Moving Item
Powerfully SimpleCompany Confidential All Right Reserved13
Slow-Moving Item
Powerfully SimpleCompany Confidential All Right Reserved14
Increased Efficiency as their Business Grows Exponentially
-50%
50%
150%
250%
350%
2012 2013
Program Growth vs Stock Out
Program Growth Stock Out
0
2
4
6
8
10Inventory Turns
2012 - BEFORE 2013 - AFTER
0%
20%
40%
60%
80%
100%Forecast Error (MAPE)
2012 - BEFORE 2013 - AFTER
0%
1%
2%
3%
4%
5%Unhealthy Inventory
2012 - BEFORE 2013 - AFTER
Case study: e-commerce
Powerfully SimpleCompany Confidential All Right Reserved15
Andrew Page: A leading aftermarket distributor in the UK
SO99+
• 2 tier network
• 13 hubs, 67+28 depots
• 384k items supplied, about 100k stocked.
• Daily replenishment orders for 20m sku/l’s
• calculated nightly
• SO99+ automatically decides to stock,
• cluster or de-cluster
Results & Benefits
• Increased growth rate
• Increased profitability
SO99+
Powerfully SimpleCompany Confidential All Right Reserved16
Five Innovations …..
New technologies, such as machine learning, will also enhance
forecasting and demand management
The “long tail” will continue to get longer in most industries – but
again data will address the problem
In a dynamic world, leveraging additional data will allow us to
understand demand, despite its volatility1
2
3
Powerfully SimpleCompany Confidential All Right Reserved17
Statistical forecast1
Seasonality, calendarand daily sale patterns
2
External variables modeling & market intelligence
3
6
7
Trade Promotion
Media event effect4
Special Actions
5New Product Introduction
Daily baseline
De
ma
nd
In
sig
ht In
cre
asin
gImproved Forecasting from Demand Analytics
Powerfully SimpleCompany Confidential All Right Reserved18
Promotions
&
Pricing
Media Event
Forecasting
Catalogue &
Advertising
Web, Mobile
& Flyers
Customer
Profiling
&
Loyalty
Demand
New Product
Introduction
Web Traffic
&
Sentiment
Web & Mobile
Daily Demand Forecast and What-If Analysis
Demand Planning – Modeling Marketing actions require new
technologies: “Machine Learning Technology”
Machine Learning and Big Data drive Demand Shaping
Powerfully SimpleCompany Confidential All Right Reserved1919 Company Confidential All Right Reserved
Machine Learning
Engine
Embeded Machine Learning Technology in SO99+
Market, Product, Demand Dataset
Demand Modeling
SPECIALACTIONS
EXTERNAL VARIABLES
LAUNCHPROFILES
Historical Demand Profiles
NPI Variables
Consumer Attributes
Web Traffic & Indicators
Media Events
Promotion Variables
New Product
Intro
Trends & Sentiments
SEASON-ALITY
Baseline Definition
Future Demand Projection
Historical Demand and Causal Factors
Powerfully SimpleCompany Confidential All Right Reserved20The Quest for Demand Management Excellence @ DANONE ITALY
Contact: [email protected] 20
DEMAND SHAPING “NEW NORMAL”Results AND Performance
DEMAND SHAPING
KPIs (PROMO)
“TRADITIONAL”
SUPPLY CHAIN KPIs
…BEYOND
NUMBERS…
Powerfully SimpleCompany Confidential All Right Reserved21
Five Innovations …..
Supply chains are becoming so dynamic that even tactical models
will become time-phased
New technologies, such as machine learning, will also enhance
forecasting and demand management
The “long tail” will continue to get longer in most industries – but
again data will address the problem2
3
In a dynamic world, leveraging additional data will allow us to
understand demand, despite its volatility1
4
Powerfully SimpleCompany Confidential All Right Reserved22
Time-Phased Supply Chain of bathing-suits
Sourced from China Sourced from
Mexico
Product Reallocation within
the network
Multi-Channel
Metropolitan areasIncreased demand in holiday
locations Outlets
Alternative Sourcing
Pre
SeasonDuring
Season
Late
Season
Dynamic Distribution
Powerfully SimpleCompany Confidential All Right Reserved23
Supply chain evolves
Production and
Distribution are
tailored to serve local
demand from local
warehouses and
factories
Standard demand
(i.e. Winter)
Winter season
Powerfully SimpleCompany Confidential All Right Reserved24
Supply chain evolves
To satisfy seasonal
demand peak, the
Croatian market is also
served by Italy in the
period june-august, with
implications on lead-time
and replenishment policy
Summer season
Exceptional demand
(i.e. Summer)
Powerfully SimpleCompany Confidential All Right Reserved25
Five Innovations …..
Supply Chain Planning and execution will converge, with many
benefits, including reduced latency
Supply chains are becoming so dynamic that even tactical models
will become time-phased
New technologies, such as machine learning, will also enhance
forecasting and demand management
The “long tail” will continue to get longer in most industries – but
again data will address the problem2
3
1
4
In a dynamic world, leveraging additional data will allow us to
understand demand, despite its volatility
5
Powerfully SimpleCompany Confidential All Right Reserved26
Drive an Integrated End-to-End Demand Signal
and Replenishment Response across the Supply Chain
Processing a
Stochastic Demand Signal
Across the Entire Network
Powerfully SimpleCompany Confidential All Right Reserved27
Costa Express
Powerfully SimpleCompany Confidential All Right Reserved28
Case Study
Costa Express
• 20% reduction in field inventory
• 50% fewer delivery refusals
• Centralized stock holding locations reduced from nine to one
• 30% reduction in annual logistics operating costs
Powerfully SimpleCompany Confidential All Right Reserved2929 Company Confidential All Rights Reserved
Dynamic Replenishment across the Operational Backbone
INVENTORY MODEL
DEMAND MODELS
REPLENISHMENT TUNNELS
Truck Load
Building
Route
Optimization
Production
Scheduling
Supply Chain
Planning
Capacity
Planning
Warehouse
Management
Powerfully SimpleCompany Confidential All Right Reserved30
• Market-driven demand analytics
and supply chain optimization
• Business-focused, technology-
enabled
• Armed with precise modeling for
deep insight into demand
signals, inventory behaviors and
supply chain volatility
• Leverages data to improve
forecast accuracy, mitigate
volatility and increase supply
chain certainty
• Optimized replenishment
ToolsGroup - Accelerating Business Performance
Powerfully SimpleCompany Confidential All Right Reserved31 Company Confidential All Rights Reserved
The Convergence of Supply Chain Planning and Execution
MULTI-ECHELON
DEMAND
MODELING
DEMAND
SENSINGPROMO, MEDIA,
NPI MODELING
ERP
SCP
MULTI-ECHELON
INVENTORY
OPTIMIZATION
REPLENISHMENT
TUNNELS
TRANSPORTATION
TUNNELS
Single Model
Market-Driven Demand > Optimization > Process > Exceptions