1) The document discusses using online and offline customer data from a large retail company with 76 stores in Spain to improve the company's digital marketing performance.
2) The analysis clusters Spanish zip codes based on distance to nearest store and customer data to develop targeted digital marketing strategies for different areas.
3) Reviews of stores from Google are also analyzed to help identify strengths and weaknesses of different store locations.
2. BigML, Inc
Online and Offline World in the
Retail Sector
Friends or Enemies?
Miguel Blanco
Data Scientist, Good Rebels
Team work: Mar Castaño & Sergio Vázquez
2
3. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels 3
Who are we?
Good Rebels is a digital
strategy and creative company.
We help our clients to innovate at the intersection of
people, brands & technology.
We engage customers daily with our unique
Smart Social methodology.
4. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels 4
4
OFFICES
MEXICO CITY
60
CLIENTS
30
CAPABILITIES
9M€
TURNOVER
120
PEOPLE
BARCELONA
MADRID
BRIGHTON
+
++
Who are we?
5. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels 5
PERFORMANCE
INTELLIGENCE
Developing deep insights
and strategy that help our
clients put people at the
heart of everything they do.
EXPERIENCE
Creating culturally relevant
products, services and
consumer experiences.
Delivering and
measuring real business
impact.
Accelerating change,
empowering people and
transforming organizations.
ENABLEMENT
I E
P E
CREATIVITYDATA
TECHNOLOGY
PROJECT
MANAGEMENT
Who are we?
6. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels
•Big company in the retail sector.
•76 stores in Spain.
•Powerful e-commerce.
•It buys and sells products to its customers.
6
Our case
7. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels
How to improve the e-commerce
performance through digital marketing using
offline information
7
Our question
8. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels
•E-commerce:
•Info about the client (location).
•Info about the transaction (price).
•Info about the purchased item.
•Where the item is stored.
8
Our data
•Stores:
•Info about the visits to the store.
•Info about the opinion of the
store (Google Reviews).
•Location.
•Zipcodes level.
•Info about the location.
•Population.
•Average spend per home.
•Average income per home.
•Background.
Online world
Offline world
External data: geographic and
demographic
9. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels 9
Our tools
Python
Tableau
Geographic
information system
(GIS)
BigML
10. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels 10
Our tools … as a kitchen
Python
Tableau
Geographic
information system
(GIS)
BigML
11. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels 11
Our project
To cluster the territory to optimize
the digital marketing strategy
12. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels 12
Data exploration
Around 12k zipcodes
in Spain.
Sorry Canaries.
45M people
13. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels 13
Data exploration
Disparity in zipcodes sizes.
But we focus on the
background.
14. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels 14
Data exploration
Disparity in zipcodes sizes.
But we focus on the
background.
15. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels 15
Data exploration
With GIS, we have
calculated the distance to
the nearest store
16. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels 16
Data exploration
With GIS, we have the
distance to the
nearest store
17. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels 17
Data exploration
We focus on the
assumption of 20 km
as a limit
18. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels 18
Data exploration
Around 25M people
have a store in less
than 20 km.
19. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels 19
Data exploration
20. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels 20
Data exploration
21. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels 21
Data exploration
22. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels 22
Data exploration
23. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels 23
First discovery
Awareness problem
in places far away from a store
24. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels 24
First action
First and easy segmentation by
distance to the nearest store
25. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels
Two groups
25
Population Customers
Total estimated income E-commerce bill
26. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels 26
We attack each group separately
Places far away
from the store
Places near
the store
Info about the place.
The e-commerce.
Demographic.
Info about the place.
The e-commerce.
Demographic.
+
Info about the store
Two groups
27. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels 27
First big group
Let’s begin with those
far from the store
28. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels
Places far from a store
28
29. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels 29
Places far from a store
30. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels 30
Places far from a store
31. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels 31
Places far from a store
32. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels 32
Places far from a store
33. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels 33
Places far from a store
34. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels 34
Places far from a store
35. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels 35
Population Customers
Total estimated income E-commerce bill
Two groups, back again
36. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels 36
Places near the store
Let’s move to those that are
near the store
Places near
the store
Info about the place.
The e-commerce.
Demographic.
+
Info about the store
37. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels 37
Places near the store
What about the store?
We’ll use the
Google Reviews
38. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels 38
Places near the store
More than 10k reviews
From the 76 stores
Rating from 1 to 5
And text
39. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels 39
Places near the store
40. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels 40
Places near the store
41. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels 41
Places near the store
42. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels 42
Places near the store
43. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels 43
Places near the store
44. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels 44
Places near the store
45. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels
Were we talking about zipcodes or stores?
45
Places near the store
46. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels 46
Places near the store
47. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels 47
Places near the store
48. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels 48
Places near the store
49. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels 49
Places near the store
50. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels 50
Places near the store
51. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels
Conclussions
•Awareness problem in zipcodes far from the store.
•Info to improve campaigns by location.
•Mechanisms to cross-selling between e-
commerce and store.
•Associate territory characteristics to customers
(to e-mail strategy).
•Reviews analyzed to help stores to identify
strengths and weakness.
51
52. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels
Take aways
•All machine learning projects need work.
•You need to work hard on the data: sources, format, …
•You need expertise on the field.
•You need the basics on math and stats.
• But the most difficult part can be simplified.
52
53. BigML, Inc #MLSEV: Online and Offline World in the Retail Sector: Friends or Enemies? / Good Rebels
Questions?
Now…
or to miguel.blanco@goodrebels.com
53