how brick and mortar retailers can beat amazon at its own game
TRANSCRIPT
Reta
il’s B
IG S
how
2017
| #
nrf1
7Re
tail’
s BIG
Sho
w 20
17 |
#nr
f17
How Brick and Mortar Retailers Can Beat Amazon at Its Own
GameVIVEK FARIAS, CTO, Celect
JOHN ANDREWS, CEO, Celect
Reta
il’s B
IG S
how
2017
| #
nrf1
7Re
tail’
s BIG
Sho
w 20
17 |
#nr
f17
JOHN ANDREWSCEO, Celect
VP Product, Oracle CommerceVP Product & Marketing, Endeca
VIVEK FARIASCo-Founder & CTO, Celect
Robert N. Noyce Professor, MIT Sloan
PhD in EE, Stanford University
• Who we are• Predictive analytics SaaS
platform for retail• Based in Boston, MA• Venture-backed, technology
out of MIT• Awards & Recognitions• MIT Computer Science and
Artificial Intelligence (CSAIL) Top 50 Greatest Innovations
Reta
il’s B
IG S
how
2017
| #
nrf1
7Re
tail’
s BIG
Sho
w 20
17 |
#nr
f17
What is Amazon’s Game? A network of DCs and last-mile logistics partners that work well at scale.
Distant fulfillment center
Sortation center
Delivery via Carrier or Amazon Flex
Customer
City storefront Delivery via Amazon Flex
Customer
Local fulfillment center
Delivery via Carrier
Customer
1.1 Billion Orders per Year
Reta
il’s B
IG S
how
2017
| #
nrf1
7Re
tail’
s BIG
Sho
w 20
17 |
#nr
f17
What is Amazon’s Game? • RISK POOLING• Demand for large
chunks of population served out of a relatively small set of Fulfillment Centers• Minimizes unsold
inventory
Indiana FC target area
Reta
il’s B
IG S
how
2017
| #
nrf1
7Re
tail’
s BIG
Sho
w 20
17 |
#nr
f17
You are Not Amazon
Which isn’t a bad thing.
Here’s two reasons why…
Reta
il’s B
IG S
how
2017
| #
nrf1
7Re
tail’
s BIG
Sho
w 20
17 |
#nr
f17
Reason #1You are the expert in your domain
Reta
il’s B
IG S
how
2017
| #
nrf1
7Re
tail’
s BIG
Sho
w 20
17 |
#nr
f17
Reason #2You have a physical presence
• 42% of in-store customers showroom, we must accept this reality.
• Store location puts you much closer to the customer.
Locations of Home Depot & Lowes in the Tri-state Area
Source: http://www.planetizen.com/
Reta
il’s B
IG S
how
2017
| #
nrf1
7Re
tail’
s BIG
Sho
w 20
17 |
#nr
f17
Leveraging these Advantages is HardYou have a hot new line of graphic t-shirts launching next season. Which scenario would you prefer?
Pooled DemandAn average demand of 3,000 units spread across 3 stores
ORDiffused Demand
An average demand of 3,000 units spread across 1,000 stores
Reta
il’s B
IG S
how
2017
| #
nrf1
7Re
tail’
s BIG
Sho
w 20
17 |
#nr
f17
Leveraging these Advantages is HardYou’re stuck with Diffused Demand and one of two things is happening:
• Inventory isn’t where you need it, causing stock outs.
• Too much inventory was bought, resulting in markdowns.
Reta
il’s B
IG S
how
2017
| #
nrf1
7Re
tail’
s BIG
Sho
w 20
17 |
#nr
f17
The Solution: Virtual Pooling
Transforming an environment with Diffused Demandinto one with Virtually Pooled Demand.
Reta
il’s B
IG S
how
2017
| #
nrf1
7Re
tail’
s BIG
Sho
w 20
17 |
#nr
f17
The Solution: Virtual PoolingRequires two key ingredients:
1. Predictive Analytics
2. Real-Time Optimization
Reta
il’s B
IG S
how
2017
| #
nrf1
7Re
tail’
s BIG
Sho
w 20
17 |
#nr
f17
Virtual Pooling with a Typical Order Management System (OMS)
1,000 UNITSAVAILABLE
A SIMPLE IN-STORE PURCHASE
WALKS INTO STORE
TO BUY A TV
PURCHASED
999 UNITSAVAILABLE
Reta
il’s B
IG S
how
2017
| #
nrf1
7Re
tail’
s BIG
Sho
w 20
17 |
#nr
f17
Virtual Pooling with a Typical OMS
AN ONLINE PURCHASE
OMS
STORE 1
WAREHOUSE
STORE 2
CUSTOMER ORDERS A TV
ONLINE
• Where do we fulfill the order from?
• A store? A warehouse?
• Which one and how do we avoid a split shipment?
-1
-1
-1 Short time to customer
High weeks of supply
Long time to customerHigh weeks of supply
Short time to customerLow weeks of supply
Reta
il’s B
IG S
how
2017
| #
nrf1
7Re
tail’
s BIG
Sho
w 20
17 |
#nr
f17
Virtual Pooling: 4 Pieces of the Puzzle1. Shipping Cost• Closer is better• Aggregated orders are better than
split orders• Will typically conflict with other
critical objectives
Reta
il’s B
IG S
how
2017
| #
nrf1
7Re
tail’
s BIG
Sho
w 20
17 |
#nr
f17
Virtual Pooling: 4 Pieces of the Puzzle2. Throughput• Process as many orders as possible• Limited processing capacity at a
stores• Can raise shipping costs
Reta
il’s B
IG S
how
2017
| #
nrf1
7Re
tail’
s BIG
Sho
w 20
17 |
#nr
f17
Virtual Pooling: 4 Pieces of the Puzzle3. Delay• Minimize the time it takes to get
to a customer• Can raise shipping costs• Conflicts with the ability to satisfy
in-store demand
Reta
il’s B
IG S
how
2017
| #
nrf1
7Re
tail’
s BIG
Sho
w 20
17 |
#nr
f17
Virtual Pooling: 4 Pieces of the Puzzle4. Average Weeks-of-Supply• Ship out of locations with many
weeks-of-supply• Related: Ship out onesies• Speeds up inventory turns and
maximizes full price sell-through• Conflicts with shipping costs
Reta
il’s B
IG S
how
2017
| #
nrf1
7Re
tail’
s BIG
Sho
w 20
17 |
#nr
f17
Virtual Pooling: 4 Pieces of the Puzzle
Average Weeks of Supply
Shipp
ing
Cost
Throughput Delay
Reta
il’s B
IG S
how
2017
| #
nrf1
7Re
tail’
s BIG
Sho
w 20
17 |
#nr
f17
Is Virtual Pooling Really Possible with a Typical OMS?Issue #1: Typical OMS is purely rules driven.
Issue #2: Works well on a few high priority objectives, but doesn’t scale well beyond that.
Issue #3: There’s no way of ‘sacrificing now’ for a future gain.
Reta
il’s B
IG S
how
2017
| #
nrf1
7Re
tail’
s BIG
Sho
w 20
17 |
#nr
f17
It’s Usually One Extreme or Another
WEE
KS O
F SU
PPLY
(IN
VENT
ORY)
THROUGHPUT
There’s a balance between the extremes – to maximize inventory turns and utilization.
OMS RULE:Maximize
Throughput
OMS RULE:Maximize Weeks of Supply
Reta
il’s B
IG S
how
2017
| #
nrf1
7Re
tail’
s BIG
Sho
w 20
17 |
#nr
f17
WEE
KS O
F SU
PPLY
(IN
VENT
ORY)
Virtual Pooling with a Predictive OMS
8%
5%
Real-time OptimizationTrue Demand across all channels
THROUGHPUT
O
Reta
il’s B
IG S
how
2017
| #
nrf1
7Re
tail’
s BIG
Sho
w 20
17 |
#nr
f17
Case Study: Vertical Fashion Retailer• Goal: Optimize Order Fulfillment with respect to the
following parameters:• Throughput / Units Shipped: Maximize utilization of
network capacity• Shipping Cost: Reduce shipping cost (ship closer and
avoid splitting• Onesies Shipped: Increase fulfillment of returned units not
part of original store assortment• Weeks of Supply: Maximize turns• Average Order Delay: Increase customer satisfaction
Reta
il’s B
IG S
how
2017
| #
nrf1
7Re
tail’
s BIG
Sho
w 20
17 |
#nr
f17
Case Study: Vertical Fashion Retailer
Representative Day
Status Quo Celect % Diff.
Throughput (units) 1,171 1,307 11.6%Unit Shipping Cost $5.05 $4.61 -8.8%Onesies Shipped 549 777 41.6%Weeks of Supply 7.6 17.9 135%
Average delay 0.047 days -0.15 days -0.2 days
Comparative Results
Reta
il’s B
IG S
how
2017
| #
nrf1
7Re
tail’
s BIG
Sho
w 20
17 |
#nr
f17
What should you take away?1. The key to Amazon’s success is better Risk Pooling2. Brick and mortar retailers have fundamentally
Diffused Demand3. Modern predictive analytics can transform this
demand and create Virtually Pooled Demand
Reta
il’s B
IG S
how
2017
| #
nrf1
7Re
tail’
s BIG
Sho
w 20
17 |
#nr
f17
Predictive Analytics for Retail
Reta
il’s B
IG S
how
2017
| #
nrf1
7Re
tail’
s BIG
Sho
w 20
17 |
#nr
f17
The Most Fundamental Task in Retail
The Right Product
The Right Place
The Right Person
The Right Price
The Right Time
Goal: Avoid Stock-outs and Markdowns Solution: True Demand Prediction
Reta
il’s B
IG S
how
2017
| #
nrf1
7Re
tail’
s BIG
Sho
w 20
17 |
#nr
f17
Understand True Demand
A Choice Model tells you what a customer would prefer to buy when given the choice.
Today, you only know what a customer bought
✗✔
✗
Reta
il’s B
IG S
how
2017
| #
nrf1
7Re
tail’
s BIG
Sho
w 20
17 |
#nr
f17
Celect Optimization Platform
AssortmentOptimization
Build robust assortments
specifically optimized for the foot traffic in each individual store
Predictive Personalizatio
nReal-time online
recommendations based on individual
customer preferences
Markdown OptimizationOptimize markdowns
while maximizing conversions and
revenues
CELECT OPTIMIZATION PLATFORM
FulfillmentOptimization
Fulfill from stores based on customer demand,
without negatively impacting store
assortments
Reta
il’s B
IG S
how
2017
| #
nrf1
7Re
tail’
s BIG
Sho
w 20
17 |
#nr
f17
Questions?Email us: [email protected]
www.celect.com