3. Agenda
• Myths Shattered by marketing analytics.
• Implementation of Analytics.
• Darden resources on Marketing Analytics.
4. Myths Shattered by Marketing Analytics
I. Marketing is a fixed cost
II. Coupons are a short-term promotional vehicle
III. Target Customers who are responsive
IV. Competition’s loyalty program decreases customer retention
V. Soft metrics are not valuable in predicting customer value
VI. Traditional media (TV) is not dead
5. Old World New World
Marketing is a fixed cost Marketing can be variable, test and learn
Coupons are a short term promotional
vehicle
Customized coupons can build longer
term brand value
Target customers who are more
responsive to offers
Target customers who are more valuable
even if they are less responsive
Competition’s loyalty programs
decreases retention
Spatial agglomeration is amplified by
mobile devices, co-opetition not
competition
Soft Metrics are not valuable for
predicting customer value
Harness information from all data
sources, customer attitudes, online
chatter etc.
TV creates brand awareness and is all-powerful
TV is still powerful, but it enables other
media; email, paid search etc.
6. Myth I. Marketing is a Fixed Cost
Venkatesan, Rajkumar, and Paul Farris, “Transformation of
Marketing in the Ohio Art Company (A+B)”,
UVA-M-0833
7. Betty Spaghetty TV Experiment
June-July, 2007
Sales
Units
Color
Crazy
Go
Go
Glam
Test
Arizona
163
206
Control
California
30
112
Be#y
Spaghe#y
was
supported
with
the
2M
adver6sing
campaign
in
2007
holiday
season
13. Myth II. Coupons are a promotional
vehicle
Venkatesan, Rajkumar, and Paul W. Farris. "Measuring and managing returns from
retailer-customized coupon campaigns." Journal of marketing 76, no. 1 (2012): 76-94.
Venkatesan, Rajkumar, and Paul W. Farris (2012), “Unused Coupons Still Payoff“
Harvard Business Review, May.
14. Data – Quasi Experimental Design
Q1
Q4
Q8
Purchase History
FSI Coupons
Retailer Discounts
Retailer Matching
Feature/Display
Purchase
History
FSI
Coupons
+
Targeted
Coupons
Retailer
Discounts
Retailer
Matching
Feature/Display
Q1
Q8
Purchase History
FSI Coupons
Retailer Discounts
Retailer Matching
Feature/Display
N
=
1,584
N
=
952
15. 37%
Exploratory Results
5% 6%
46%
38%
60%
17%
57%
34%
70%
60%
50%
40%
30%
20%
10%
0%
Number of Customers in
Group
Growth In Trip Spending
Per Customer
Growth in Total Group
Trip Spending
Did Not Receive Received But Did not Redeem Received and Redeemed
17. Myth III. Target Customers Responsive
to Offers
Bodily, Sam, Rajkumar Venkatesan, and Gerry Yemen,
“Dunia Finance LLC”,
Darden Business Publishing Case Study, UVA-M-0842
18. dunia
is
born
in
the
midst
of
an
unprecedented
period
of
global
macroeconomic
stress,
as
a
highly
pedigreed
enterprise
Mubadala
• Government of Abu Dhabi
• $53B assets
• Industries: Aerospace, Oil & Gas,
Healthcare, Information and
Telecom, Financial services
Temasek Holdings
• Government of Singapore
• $152B portfolio
• Industries: Financial services,
telecom, media & tech, transport,
real estate, energy, lifesciences
19. Achieved a lot since launch…
• 2012 First Half Net Profits of AED 29.1 Million, up 61% vs. Full Year 2011
• Deposit balances up 84% vs. H1 2011 to AED 313 Million
• Broke-even in third year of operation, ahead of plan
20. Customer centricity: 360° view of the customer
Call center
Branches
ATMs
Internet
• Shows complete relationship details of the customer
• Active lead management facilitating x-sell and relationship deepening
• Allow customer access through diversified channels set
21. Making customer centricity a reality: Cross-sell
Cross_product Penetration Grid
Cards
Unsecured
Loans Auto Loans Investment Insurance
Revolving
Credit
Banc-assurance
Cards 100%
Unsecured Loans 100%
Auto Loans 100%
Investment 100%
Insurance 100%
Revolving Credit 100%
Bancassurance 100%
Cross Sell Principles:
• List 1: All auto loan
customers with mid
size+ new cars
• List 2: All
preterminated auto
loans
• List 3: All auto loans
booked in last 2
months
Cross-sell discipline: Drive the penetration matrix every month and identify opportunities
List down all possible product pairs, determine the channel and generate the list
Day 1 cross sell: Each new customer should come with multiple products
On-going cross sell through CRM: For example, each auto loan customer would be
contacted for a card at 3rd month and investment at 4th month (unless cross sold day 1)
Use of statistical propensity models for better targeting
Track: Products per customer and profit per customer
Objective: Address customer’s additional product needs, so as to
maximize our products / customer ratio.
22. Responsive Customers Are not Necessarily
Profitable
High Profits Low Profits
High
Propensity to
Respond
Very Good
Targets (18%)
Reduce
Marketing
Spend (34%)
Low
Propensity to
Respond
Invest Until
Marketing
Spend <
Customer
Return
(30%)
No Investment
(18%)
23. Myth IV. Competition’s loyalty program
decreases customer retention
Rajkumar Venkatesan (2014),
“Cardagin: Local Mobile Rewards,”
Darden Business Publishing Case Study, M-0825
Pancras, Joseph, Rajkumar Venkatesan, and Bin Li,
“Returns from customizing mobile loyalty programs,”
Working Paper, Darden GSB.
24. The Market
Served
Available
Market
online
&
mobile
spending
(2)
$11.1
billion
(1) Source:
VSS
Communications.
2009
figure.
(2) Source:
BIA/Kelsey.
2011
figures.
Total
Addressable
Market
local
advertising
spending
(2)
$132
billion
Target
Market
Loyalty
spending
(1)
$2.19
billion
26. Case Study: Shenandoah Joe’s
• Three location coffee shop in Charlottesville, VA
• Launched in April 2012
Month 1 Month 4
5.1 monthly transactions per member 9.4
$22.84 monthly revenue per member $47.22
$4.46 average spend per member $5.02
“Cardagin has turned our occasional customers into regulars and compelled regulars to visit the
shop more often than before.”
Shenandoah Joe’s Management
27. Case Study: Calvino Café
• A family-owned, single location coffee shop
• Empirical results:
– More than 1,500 transactions and $10,000 recorded during
first four months on Cardagin
– Approximate ROI of 450% in first four months
• Customer Testimonial:
– “Previously, there were numerous customers whose names
we did not know. Now, we’re learning everyone’s names
because their names come up on Cardagin.” - Katie, owner
28. Consumer Graph
Visits: 73
Spend: $371
John
member id: 5453
Visits: 1
Spend: $51
Visits: 3
Spend: $95
Visits: 1
Spend: $2
Visits: 22
Spend: $269
Visits: 1
Spend: $10
Visits: 2
Spend: $8
• John frequents 9 participating businesses in Charlottesville
• Information inferred from Cardagin:
Visits: 1
Spend: $43
Visits: 1
Spend: $456
– John spends most of his time in two Charlottesville neighborhoods
– John has relatively high disposable income given his merchant visits and purchase history
29. Spatial aspects of Mobile Coupons
Spatial Map of
Retailers on Cardagin
Network in
Charlottesville
31. Value of Information From Mobile Loyalty Program
Network
• Estimated maximum net sales per store
– without competitive information = $1194.92
– with competitive information = $443.61
• One additional competitor on the network within a
1 mile radius reduces the
– Number of rewards provided by a retailer by 15%
– The range of rewards by 2 points
32. Myth V. Soft metrics are not useful for
predicting customer value
Venkatesan, Rajkumar, Werner Reinartz, and Nalini Ravishankar (2013),
“Role of Attitudes in CLV based Customer Management,”
Marketing Science Institute (MSI) White Paper, 12-107.
Reinartz, Werner, and Rajkumar Venkatesan (2014),
“Track Customer Attitudes to Predict Their Behavior”,
Harvard Business Review Blog, September.
http://blogs.hbr.org/2014/09/track-customer-attitudes-to-predict-their-behaviors/
36. Value
of
A?tudes
in
Customer
Level
Resource
AllocaCon
Percentage Improvement in Maximized Customer Profits compared to
Predicted Customer Profits
All Customers
(n=1161)
Observed Attitudes
(n=553)
Imputed Attitudes
• Average Customer Profits = $2,368 (in 2 months)
• Incremental lift of 18% equals $426 in annual profits per customer
(n=608)
Including
Attitudes
25.0%
(22.7%, 28.8%)
26.2%
(23.9%, 29.5%)
23.9%
(21.2%, 26.4%)
Excluding
Attitudes
7.0%
(5.4%, 8.3%)
8.4%
(5.7%, 9.8%)
5.8%
(3.8%, 7.2%)
37. Myth VI. Traditional Media (TV) is not
dead
Venkatesan, Rajkumar, and Joseph Pancras (2014), “Estimating the
Consumer Purchase Funnel From Aggregate Media Metrics,” Working
Paper, Darden GSB.
38. Context drives device choice
The goal we
want to
accomplish
The time and
day of the
week
The device
capabilities
Our location and
“velocity”
The device we
choose to use at
a particular time
is often driven
by our context:
40. Google’s Attribution Setup
Last Interaction
Last non direct Interaction
Last AdWords Click
First Interaction
Linear
Time Decay
Position Based
41. A
Media
Mix
System
of
Metrics
Units Sold
Email
Impressions
Price
Web Visits
Emails
Paid Search
Clicks
TV
Facebook,
Mobile
Paid Search
Spend
Facebook Clicks
TV
Paid Search,
Mobile reach
Facebook Spend
Mobile Clicks
TV
Facebook, Paid
Search reach
TV Impressions TV Spend Mobile Spend
-‐
sales
-‐
First
level
media
effects
-‐
Second
level
media
effects
-‐
Media
Spend
42. AJribuCon
Model
Findings
• Sales = f(lagged sales, web visits from search….)
• Webvisits from search = f(lagged webvisits from search,
paid search clicks, mobile
search clicks)
• Paid search clicks = f(lagged paid search clicks, TV
spend, paid search
impressions, display impressions)
43. Myths Shattered by Marketing Analytics
I. Marketing is a fixed cost
II. Coupons are a short-term promotional vehicle
III. Target Customers who are responsive
IV. Competition’s loyalty program decreases customer retention
V. Soft metrics are not valuable in predicting customer value
VI. Traditional media (TV) is not dead
44. Old World New World
Marketing is a fixed cost Marketing can be variable, test and learn
Coupons are a short term promotional
vehicle
Customized coupons can build longer
term brand value
Target customers who are more
responsive to offers
Target customers who are more valuable
even if they are less responsive
Competition’s loyalty programs
decreases retention
Spatial agglomeration is amplified by
mobile devices, co-opetition not
competition
Soft Metrics are not valuable for
predicting customer value
Harness information from all data
sources, customer attitudes, online
chatter etc.
TV creates brand awareness and is all-powerful
TV is still powerful, but it enables other
media; email, paid search etc.
45. Implementation of Marketing Analytics
Organizational Structure
1. What is the function and
process of marketing
analytics?
2. What are the organizational
metrics for resource
allocation?
3. Does the business cycle
match the marketing analytics
cycle?
4. How to foster sales and
marketing collaboration?
Analytics Process
5. How to combine data
and heuristics?
6. Does the language of
marketing analytics
match the language of
the business?
Organizational Change
7. How to develop
effective feedback
loops?
46. Resources on Marketing Analytics
46
Resource Videos and Datasets @
http://dmanalytics.org
47. Strategic Marketing Analytics: Leveraging Big Data
Monday,
November
10,
2014
Tuesday,
November
11,
2014
Wednesday,
November
12,
2014
Thursday,
November
13,
2014
7:00
-‐
8:00
am
7:00
-‐
8:00
am
Con6nental
Breakfast
Con6nental
Breakfast
8:00
-‐
Noon
8:00
-‐
noon
Resource
AllocaCon
Framework
II
Pricing
AnalyCcs
ImplemenCng
AnalyCcs
System
of
Metrics
Conjoint,
Willingness
to
Pay,
Tradeoffs
Apply
the
alloca>on
framework,
telling
a
story
Allocator
SimulaCon
Regression Workshop
12:00
-‐
1:00
pm
12:00
-‐
1:00
pm
12:00
-‐
1:00
pm
Boxed
Lunch
Lunch
Lunch
Lunch
1:00
-‐
5:00
pm
1:00
-‐
4:00
pm
1:00
-‐
4:00
pm
Resource
AllocaCon
Framework
I
Digital
AnalyCcs
Sales
Force
AnalyCcs
System
of
Metrics
Experiments,
Paid
Search
Customer
Life>me
Value,
Sales
Pipeline
November 10-13, 2014, Charlottesville, VA