Omni-Channel Revenue Management Order Fulfillment and Pricing – Two Case Studies Dr. Markus Ettl Commerce Advanced Analytics IBM Research

Supply Chain Management Symposium University of Pittsburgh Nov 4, 2016

BRICK & MORTAR RETAILERS ARE BECOMING MORE AGILE TO COMPETE WITH AMAZON, EBAY, ALIBABA, …

30%

$3B

5x-10x

18c

Year-over-year eCommerce sales growth

Cyber Monday eCommerce sales (2015)

Peak-to-off-peak loads

Fulfillment cost per USD of eCommerce revenue

DEMAND IS SHIFTING FROM STORES TO ONLINE … WORST DURING PEAK

Today 

Online sales growing rapidly, straining capacity



Shipping costs rising dramatically to meet customer expectations



Markdowns substantially impacting revenue and margin

What if you could … 

Fully leverage your stores as part of the network to increase capacity



Save on shipping costs while still meeting customer SLAs



Intelligently source slow moving inventory and avoid markdowns

MANY CHALLENGES OF OMNI-CHANNEL RETAILING

PRICING 

Cross-channel fulfillment



Price-based channel substitution



FULFILLMENT   

Emerging services (1 hour) Peaky demand High cost of e-Com fulfillment

New competitor every minute

RETURNS 

50% Retailers offer free returns



2015 Christmas: 30% returns



Hidden impact on margins

Omni-Channel Price Optimization

Fulfillment Optimization

Returns Optimization

Predict omni-channel inventory (SKUxNode)

Total cost to serve optimization

Personalized Returns: Causal Analysis

Intelligently price slow moving inventory

Shipping Optimization

In Progress Returns Avoidance

Estimate cross-channel price elasticities

Capacity and Execution Optimization

Reverse Logistics Optimization

Better manage markdown campaigns

Inventory Optimization

Recovery Optimization

Omni-Channel Fulfillment Optimization: Maximizing profit by understanding cross-channel inventory Deliver the perfect order every time with intelligent fulfillment. By streamlining the order management process, using a single view of orders and inventory across the entire fulfillment network, customers can order and receive from any channel, get a committed fulfillment promise and track the order status.

THE ORDER OPTIMIZER ANALYZES TRADE-OFFS BETWEEN CONFLICTING BUSINESS GOALS •

Order #: 1636732462314

•

2015-11-26

•

Zip code: 48456

•

FUR BUTTON BOOT (2 x $49.99)

•

Lightweight Zig Zag Loop ($12.99)

•

Total Basket: $112.97

Distance based sourcing: Ship from Midwest Distribution Center Shipping Cost: $5.22

Order: $112.97

Zig Zag Loop Price: $12.99 Node: 1097 Shipping: $5.22 Expedite : $0.00

Fur Button Boot Price: $49.99 Fur Button Boot Price: $49.99

Multi-objective sourcing model Shipping cost

Markdown savings

Time to Customer

Node: 1097 Shipping : $5.22 Expedite : $0.00 Order: $112.97

Loyalty

Labor

Fill rate

Combinatorial optimization selects the lowest total cost to serve option

Node: 953 Shipping : $5.22 Expedite : $0.00

Zig Zag Loop Price: $12.99

Fur Button Boot Price: $49.99 Fur Button Boot Price: $49.99 Markdown savings: $10.00

Margin improvement on this order: $4.78

BALANCING COMPETING OBJECTIVES OF SUPPLY CHAIN AGILITY AND COST–TO–SERVE

Business objectives for omni-channel fulfillment Fulfillment Capacity

Maximize agility to fulfill

Ship from store

Markdowns

Speed of fulfillment

Pick-up in store

Transportation

Ship direct

Inventory

Fulfillment Options

Same day delivery

Labor

Minimize cost to serve

1,000+ nodes (DC’s, retail stores, 3PL, Darkstore)

Fulfillment Balance Deliver orders faster and differentiate by customer loyalty

Improve customer sat

Utilize inventory at the most profitable price point

Avoid markdowns

Manage cost-per package and packages-per-order

Reduce fulfillment costs

Improve agility to fulfill digital orders during peak demand

Increase capacity

WHY IS ORDER FULFILLMENT CHALLENGING?

Complexity 500 msecs / order call back SLA 500K order lines / hour throughput

Up to 30,000 decision variables Up to 5,000 constraints 160 parallel optimization threads Live feeds flow every 1 min Live feeds flow every 30 mins Live feeds flow on-demand

31B+ possible combinations for 5 item order in a 100 node fulfillment network with 5 carrier service options.

44TB of historical data in cloud

CASE STUDY: A LARGE DEPARTMENT STORE RETAIL CHAIN

9% lower shipping cost per order

10% Less expedited shipping costs during peak

2.5% less packages per order 9

WHERE DID THE ORDER OPTIMIZER PERFORM BETTER ?

Omni-Channel Price Optimization: Recommending prices based on cross-channel demand and competitor pricing Understand product and competitive elasticities, and most important competitors in each product segment Understand where products should be priced identically across channels Strategically update prices based on competitors, inventory, vendor costs, …

CONSUMER CHOICES IN AN OMNI-CHANNEL RETAIL ENVIRONMENT

Challenges 

Retailers must ensure that pricing and inventory decisions seamlessly follow a customer across channels, maximizing the purchase decision at every touch point.



Transparent competitor prices affect consumer purchasing behavior



Consumer’s willingness to buy depends on the type of item, strength of competition, perception of value and brand loyalty.

OMNI-CHANNEL FULFILLMENT OPPORTUNITIES ARE OFTEN IGNORED IN PRICING DECISIONS

AN EXAMPLE FROM A LARGE CONSUMER ELECTRONICS RETAILER

Online prices follow brick prices during markdown campaigns

Online orders fulfilled by retail stores (ship-from-store)

Increasing stock-outs in retail stores Online market share spikes during clearance

GEO-SPATIAL VOLUME OF SALES IN ALL CHANNELS AND CHANNEL SHARES

35K, 4%

Sales volume, Online share

44K, 10%

57K, 11%

70K, 9%

32K, 7%

Partitioning of online sales is capturing location-specific channel demand

IMPACT OF CHANNEL PRICES ON RETAILER’S STORE AND ONLINE SALES

Percentage decrease in channel sales for each 1% decrease in price

Store price

.com price

Amazon price

Store sales

-0.6%

0.8%

0.8%

.com sales

2.9%

-4.9%

2.1%

*Average elasticity to final prices across entire selling season

If Amazon lowers prices by 1%, retailer’s store sales drop by 0.8% and .com sales drop by 2.1%.

OMNI-CHANNEL INTEGRATED PRICING AND INVENTORY OPTIMIZATION PROBLEM

We developed a tractable MIP-based reformulation that exploits the structure of attraction demand models Revenue (includes salvage) Shipping costs

Pick-up in store sales Ship-from-store variables

Allocation variables

Online sales less than demand Online sales less than inventory + SFS

Brick sales less than demand and inventory - SFS

Markdown prices and business rules

Reference: P. Harsha, S. Subramanian, J. Uichanco (2016). Omni-channel revenue management through integrated pricing and fulfillment planning. IBM Research Report. (submitted for publication).

RESULTS FROM OMNI-CHANNEL PRICING PILOT WITH A LARGE ELECTRONICS RETAILER

7% Historical

Gain in markdown revenue capture

9 points Pilot Margin erosion avoidance

Live plans

23% Reduction in unsold clearance inventory

SOME MORE INSIGHTS

Shallower store markdowns when combined with lower e-commerce clearance prices often avoids “margin erosion” and can lead to high profitability of markdown campaigns

Inventory partitioning helps retailers direct e-Commerce demand towards stores with slow-moving inventory to increase sell-through

Omni-channel pricing leads to the highest impact for products with a solid online presence.