NRF 2019: Retail's Big Show
Sunil Suram, Director of AI and Big Data, Tango
Pranav Tyagi, President and CEO, Tango
Mark Zygmontowicz, SVP, Business Development, Tango
7. Retail Disruption is Here
Shifting consumer shopping behavior fueled by
technology is disrupting the retail industry and retail real
estate.
Delivery Technologies
‐ 3rd party service partners (customer information)
‐ Autonomous vehicles
Ordering/eCommerce Technologies
‐ Online ordering
‐ Self-serve kiosks
Effects on Real Estate Product
‐ Commissary kitchens
‐ Stores as points of distribution
‐ Optimizing balance of physical retail vs ecommerce
8. Retailers are aggressively investing in digital and other
technologies to stay ahead.
Sephora
Heavy investment in
Color IQ technology –
augmented reality tools
like Visual Artist which
lets customers try on a
given look or style.
Ulta
Invested in e-commerce
platform Spruce and
acquired AI and
augmented reality
startups QM Scientific
and GlamST.
9. Walmart expands its robot
program to 50 stores across the
country.
The autonomous robots
go up and down the
aisles, scanning for out-
of-stock items, incorrect
prices, and wrong or
missing labels.
10. Digital ordering and delivery is
expected to reach $200 billion
over the next four years.
That’s a bigger share than
drive-thru all together and
it’s happening in a period
that is 10 times shorter
than the history of the
drive-thru.
Digital Delivery Driving Industry-wide Shifts in Restaurant Real Estate; Forbes, 11/21/18
11. As digital and delivery
continue to grow, expect a
major shift in how
restaurants approach their
real estate footprints.
Already according to NPD
Group, nearly 50% of
dinners purchased from a
restaurant are consumed
offsite.
That impact will
include smaller
footprints,
nontraditional
footprints and the
emergence of
commissary-only
locations that focus
on delivering and
catering orders.
Digital Delivery Driving Industry-wide Shifts in Restaurant Real Estate; Forbes, 11/21/18
13. Potential Impact
as a Percentage
of Sales
3.2% - 5.7%
Estimated
impact of
artificial
intelligence by
industry,
function and
business
problem.
Retail - $800
Billion or 3.2% -
17. Computational Infrastructure
Infrastructure speed, availability, and sheer scale
has enabled bolder algorithms to tackle more
ambitious problems
Big Data
Collecting every piece of data possible has
brought about a high demand for solutions that
go beyond simple statistical analysis of data
According to Accenture, 85% of executives plan to invest extensively in AI in the next three years, and in the same
time period, Forrester estimates that businesses using AI will “steal” $1.2 trillion from companies that don’t.
Why the Time for AI is Now
18. Fewer than 20% of
retailers have deployed
artificial intelligence.
What’s holding them
back?
The belief that they lack
the data, skills, tools,
culture and leadership
required.
AI – A Retail Growth Opportunity Hiding in Plain Sight; Accenture, 06/01/18
19. Barriers to Adoption of AI
Technology
Rapidly changing
and very different
than traditional
enterprise IT
requirements and
systems
Data
Massive and
Unstructured
Skills
Different and may
not exist in most IT
organizations
Methodologies
Far more complex
and often originate
from non-traditional
disciplines
20. Required Skillsets & Human Infrastructure
Data Scientist
vs.
Data Analyst
Ability to handle abstract
concepts and unstructured
data vs. solving a specific
industry problem with
structured data.
Deeper understanding of
theory and programming vs.
using off the shelf tools
(always evolving) and basic
interpretation skills.
ML Engineers
vs.
Application Users
Success requires a different
skillset with a different
mindset.
Recruiting is critical and
talent is scarce and in high
demand.
Revolutionary
vs.
Evolutionary
21. Required Technology Infrastructure
Scalable
Able to handle vast
datasets and changing
usage demands
Adaptable
Different algorithms by
situation
Reusable across
customers / problems
Changing datasets
Interpretable
& Deployable
Not a black box
Easily deployable
Platform agnostic
Infrastructure agnostic
Secure &
Compliant
Sensitive datasets
Secure and rest /
transit
GDPR compliant
22. Stacked Models Are Superior but Require
Advanced Skill Sets and Industry Specific
Experience
• Fundamental Techniques
• Lasso
• Elastic Net
• Neural Network
• Random Forest
• XGBoost
• Etc.
• Models are Stacked and Pre-
Processors Applied
• Deep Model Interpretability –
Influence on Target Based on
Primary Features
23. Lower Machine Learning Complexity
Higher Interpretability
Higher Machine Learning Complexi
Lower Interpretability
Stacking of
models and
algorithms
drives/
improves
accuracy but
reduces
interpretabilit
y
24. Lower Machine Learning Complexity
Higher Interpretability
Higher Machine Learning Complexi
Lower Interpretability
Balancing
complexity
and
interpretabili
ty results in
accurate
AND
interpretable
models.
27. Brick and Mortar Stores
#1 Source of
Revenue
$4.14 Billion
in revenue
$158 Million
in annual rent
expense (1)
$161 Million
in CAPEX for new &
existing restaurants
#1 Use of
Capital
#2 or #3
Operating Expense
$36.0 Billion
in revenue
$521 Million
in annual rent
expense (2)
$770 Million
in CAPEX for new &
existing stores2,427 Stores (1) Panera Bread Company FY16 10-K
(2) Loblaw Companies Limited 2017 Annual Report
2,088 Restaurants
AI/ML Has Broad Applicability for Retailers but
Brick and Mortar Real Estate has the Greatest
Value Potential
28. Brick and Mortar Stores
#1 Source of
Revenue
$4.14 Billion
in revenue
$158 Million
in annual rent
expense (1)
$161 Million
in CAPEX for new &
existing restaurants
#1 Use of
Capital
#2 or #3
Operating Expense
$36.0 Billion
in revenue
$521 Million
in annual rent
expense (2)
$770 Million
in CAPEX for new &
existing stores
(1) Panera Bread Company FY16 10-K
(2) Loblaw Companies Limited 2017 Annual Report
2,088 Restaurants
2,427 Stores
AI/ML Has Broad Applicability for Retailers but Brick
and Mortar Real Estate has the Greatest Value
Potential
32. Market Planning
&
Site Selection
Deal Management
Market Planning
Site Analysis
Site Identification
Trade areas define the puzzle pieces that
make up market strategy
34. Massive Datasets That Change at Various Frequen
Location MatrixStore & Transaction Customer Profiles
Brand EquitySite Variables Crime
DemographicSocial Media Affinity
Cell Phone
Competition
PsychographicDaytime PopTraffic Trade Areas
Demographic Data
Over 3000 variables for 54MM Zip+4
Geospatial Data
Over 1 Billion records
Customer Data
50-100 million records analyzed
Merchandize Data
120+ product categories w/1000s
SKUs
36. Existing stores: 7,390
Competitor points: 241,220
Seed points: 120,339
Demo variables: 300
Psychographic Segments: 73
Site characteristic variables: 150
Customer transactions: 39,334,367
CBSAs: 940
Block groups: 220,334
Tango analyzes over 1.7 trillion
inputs just to create the surface
to run the optimization, for
which there are 1.071509e+301
possible combinations
Dunkin Brands Case Study
37. Brick and Mortar Stores
#1 Source of
Revenue
$4.14 Billion
in revenue
$158 Million
in annual rent
expense (1)
$161 Million
in CAPEX for new &
existing restaurants
#1 Use of
Capital
#2 or #3
Operating Expense
$36.0 Billion
in revenue
$521 Million
in annual rent
expense (2)
$770 Million
in CAPEX for new &
existing stores
(1) Panera Bread Company FY16 10-K
(2) Loblaw Companies Limited 2017 Annual Report
2,088 Restaurants
2,427 Stores
AI/ML Has Broad Applicability for Retailers but Brick
and Mortar Real Estate has the Greatest Value
Potential
38. EARLY STAGE CONTRUCTION AI
Project Schedule
Optimizers
Consider millions of
alternatives for project
delivery and continuously
enhance overall project
planning.
Image
Recognition
Image recognition and
classification can assess
video data collected on work
sites to identify unsafe
worker behavior and
aggregate data to inform
future training and education
priorities
Enhanced
Analytics
Enhanced analytics
platforms can collect and
analyze data from sensors to
understand signals and
patterns to deploy real-time
solutions, cut costs, prioritize
preventative maintenance
and prevent unplanned
downtime.
Source: Artificial Intelligence: Construction technology’s next frontier; McKinsey, April 2018
40. Brick and Mortar Stores
#1 Source of
Revenue
$4.14 Billion
in revenue
$158 Million
in annual rent
expense (1)
$161 Million
in CAPEX for new &
existing restaurants
#1 Use of
Capital
#2 or #3
Operating Expense
$36.0 Billion
in revenue
$521 Million
in annual rent
expense (2)
$770 Million
in CAPEX for new &
existing stores
(1) Panera Bread Company FY16 10-K
(2) Loblaw Companies Limited 2017 Annual Report
2,088 Restaurants
2,427 Stores
AI/ML Has Broad Applicability for Retailers but Brick
and Mortar Real Estate has the Greatest Value
Potential
41. RETAIL LEASE AI OPPORTUNITIES
CAM
Reconciliations
Analyze
common area
maintenance
charges to
identify lease
contract
violations and
anomalies.
Lease
Negotiations
Mine market,
landlord and
deal data to
determine rent
thresholds and
deal proforma
optimization.
Deal
Matching
Set criteria such
as site
characteristics,
demographic
requirements,
geographic
parameters, etc.
and be notified
when a deal
matches criteria.
Lease
Renewals
Analyze lease
renewal options
relative to sales
and profit
forecasts to fuel
renewal
decisions and/or
renegotiation
tactics and
thresholds.
Lease
Abstraction
Use of machine
learning to
identify, extract
and transform
lease data from
contracts.