32. ML in retail / commerce
Primary goals for every retailer
• Increase BRAND equity
• Drive top line sales / growth
• Increase Margin
• Reduce operational costs
• Deliver a more consistent, timely experience to delight customers
33. • Dynamic Pricing
• Pricing Optimization
• Digital Store Assistant
• Next Best Action / Offer
• Propensity to Act
• Relevance to Consumer
• Chat & Voice Assistance
• Location Selection
• Forecasting & Optimization (Profit, Rev
enue, Supply Chain, Demand)
• Real Time Personalization (Product rec
ommendations, content / messaging, Li
fetime Value, concierge)
• Pricing Optimization
• Emotion Recognition
• Bot Creator Authoring
• Personalization Platform
• Conversation Outcomes
• Case Automation
• Threat Vector Identification
• AI-enabled Toys & Games
• Product Placement
• Shopper Intelligence
• ML Image Tagging
U s e C a s e s
ML Retail Use Cases
34. • Consumer Insights
• Identity Management
• Order Management and Fulfillment
• Brand Loyalty / Campaign Management
/ Marketing attribution
• Customer Care / Customer Call sentime
nt analysis
• Virtual merchant / markdown cadence /
merchandise buy planning
• CC bad debt, fraud, and optimization
• Counterfeit product detection
• Gray market selling
• Customer quality check (manufacturing
/ receiving / shipping)
• Traffic in store analysis (mood, mercha
ndise hot spotting, planograms, dwell ti
mes, traffic in the stores, etc)
• Loss prevention
• Procure to Pay audit
• ML for IT Ops (notification reduction)
• Returns Optimization
U s e C a s e s
ML Retail Use Cases
35. …determine which ML Use Case will have the highest value
Complexity
Business Value
personalization
forecast replenishment
churn prediction dynamic pricing
digital store assistants
talent optimization
a/b testing
product recommendation
merchandising hotspot
store footprint selection
spill detection
frictionless shopping
chat bot
sentiment analysis
real-time bidding platforms
predictive maintenance
position matching
new product development
loss prevention
predictive customer care
Image/video recognition
predictive maintenance
programmatic media buying
return analysis
Consumer Insights
Operational & Inventory Insights
Post-sales Insights
service root cause analysis
cross-sell up-sell
margin analysis
assortment planning
36.
37. Amazon Forecast Amazon Personalize
AWS Retail ML | Experiment With Amazon ML Services
Experiment with speed, using AI Services based on the same technology used at Amazon.com
38. ML Roadmap (function and phase)
1. Video Analytics
6. Foresee Survey Text Analysis
10. Data Capture (customer attributes)
2. CC Optimization
3. Demand Forecasting (EU POC)
4. In-stock LIN Merge
9. Size Scaling
Apr-19 May-19 Jun-19
5. ClickStream (Phase 1)
7. Sentiment Analysis
8. Image recognition (Master Product Data)
Corporate
Retail
Digital
Value Chain
POV/Case Study
Tech Feas/POC
Execute
Deploy
1. Video Analysis – status update lorem ipsum
2. CC Optimization – status update lorem ipsum
3. Demand Forecasting – status update lorem ipsum
4. In-stock LIN Merge – status update lorem ipsum
5. Clickstream – status update lorem ipsum
6. PFS Foresee Survey Data (text analysis) – status update lorem ipsum
7. Sentiment Analysis – status update lorem ipsum
8. Image Rekognition – status update lorem ipsum
9. Product / Size scaling – status update lorem ipsum
10. Data Capture – status update lorem ipsum
Executive Summary
39. Personalization
O p t i o n 1 O p t i o n 2 O p t i o n 3
Amazon
Personalize
Amazon
SageMaker
Pre-built notebook
s with Built-in high
-performance algo
rithms
One-click
Training with mo
del Tuning
One-click
Deployment and Fully
managed hosting aut
o-scaling
Amazon
SageMaker
Built on 20 years of operating
Personalization at scale acro
ss segments, geos, and indu
stries