Transcript Lecture 3

Sources and Uses of Marketing
Data
Customer Data
• All sales, promotion, and service activity
relating to a customer.
• Best bets for use in predictive statistical
models.
• Not available in equal measure for every
customer
• More data available for old customers.
• Appropriate measures that use time a
customer has been on file hence required.
Cohort or Enrollment Group
• Groups that contain customers that have
been on file for similar lengths of time.
• Basis for all forecasting systems.
• Used to alert management on changes in
lifespan, and lifetime value.
Other Sources
• Billing status, service interactions, back orders,
product shipment, claims history etc.
• Marketing department internal operations
• Customer classifications
• Response scoring models
• Expected sales
• Marketing Objectives
• Projected customer value
• Expected promotion costs.
Response Data
• Recording a purchase in response to a coded
promotion.
• Example: Multistep lead generation process.
Problems in coding response data
– Transactions occur across multiple channels
– Matching promotions and responses to
appropriate customers.
– For example, in the case of retail promotions
point of sale scanners cannot capture customer
identification.
– Cost minimization in call centers may not allow
promotion and customer codes to be recorded.
– Responses may not be matched at the
individual customer level but at the zip code
level.
Response Attribution
• What if the customer is sent multiple
promotions and he/she responds to one of
them?
• What if the customer passes along the
promotion to someone he knows?
Prospect Data
• People who have been promoted in the past but
have not made a purchase yet.
• Prospect Databases
– Used when there is relatively large variation in potential
customer values.
– Primary applications
• Track promotion history
• Calculate number and type of lists that contain information on
a prospect
• Combine descriptive statistics from internal sources
Prospect Data
• Two-Way Customer Dialogues
– Focus on developing and managing a relationship with
each customer.
– Manage communication across all channels
• Example: Financial Services
– A customer may not be ready to invest currently.
– Keep the communication channel open with the
customer in order to convert the customer at the
appropriate time.
Prospect Data
• All information is potentially important.
• Data gathering is an ongoing process.
– Begins before the first purchase is made.
– Pay careful attention to
• How the customer is contacted?
• When the customer is contacted?, and
• What data can be captured at each stage?
Nontransactional Data Sources
• Data provided directly by individuals about
themselves.
• Third Party vendors.
• Directly supplied data:
– Obtained from lead generation questionnaires,
warranty cards etc.
– Very critical for relationship marketing.
Nontransactional Data Sources
• Directly supplied data consists of three major
types
• Behavioral Data
• Attitudinal Data
• Demographic Data
• Primarily a forte of marketing researchers until
recently.
• Marketing research studies have information on
only a sample of the customers.
• This information is not enough to create
customized, individual level campaigns.
Macro vs Micro level data
• Consider two companies and two customers
• Firms have same shares in both figures but their
customers have different purchase patterns
Firm 1
Customer A
1
Customer B
Firm 2
2
1
2
Firm 1
Customer A
0
Customer B
2
Firm 2
4
0
Nontransactional Data Sources
• Relationship Marketing
– Third party data is so commonly available that it does
not provide a competitive advantage.
– Leverage investments in customer service to collect
individual information during regular business
interactions.
– Advantages:
• Better coverage
• Data directly relevant to marketing objectives, and
• Faster acquisition cycles.
Nontransactional Data Sources
• Relationship Marketing-The Advent of internet
– Lead generation
– Automated brochures provide wealth of product
information and enable collection of e-mail, address
etc.
– Surveys can be posted on the web
• Questions in the survey can be tailored to each
customer.
• Growing evidence that customers are less reluctant
to provide information on web sites.
– Privacy issues need to accounted for.
– If relationships are developed customers are ready to
provide sufficient information.
Example: Insurance Marketers
• Age is the most critical information needed.
• Third Party sources provide unreliable information
and have poor coverage.
• Insert a small survey in initial promotion packets.
– Inquire in the surveys about
• Date of birth,
• Other insurance products customer currently owns, and
• Level of Satisfaction.
Example: Insurance Marketers
• Primary benefits
– Better targeting
– Better mailing efficiency
– Reduced dependence on less accurate data
• Auxiliary benefits
– Eliminate or reduce promotions to those who are not
responding.
– Use survey information to offer additional products.
Using Questionnaires
• Internal customer data does not include
information on willingness to purchase.
• Use a two-step communication strategy.
• First Step:
– Simple, inexpensive attitude and behavior survey
• Second Step:
– Expensive brochures that contain product information
and special offers.
• People who respond in the first step but not the
second provide information for relationship
marketing.
Survey Data: Assigning Customers
to Segments
• Segments: Small relatively similar pockets of customers.
• Customers within a segment are similar to each other and
differ from customers in other segments.
• Issues:
– Confirm that segments exist
– Determine attitudes and characteristics of each segment.
– Design cost-effective ways to assign individuals to appropriate
segments.
Survey Data: Assigning Customers
to Segments
• Use survey responses to identify characteristics of
segments.
• Characteristics useful in designing customized
campaigns.
• Responses may be available only from a sample of
customers.
• Very expensive to send surveys to all the
customers in the database.
Survey Data: Assigning Customers
to Segments
• Relate survey data to internal customer data.
• Use statistical models to infer segments
membership based on
– Internal data, and
– Relation between internal data and survey
responses.
• Response rate depends on the relation
between an organization and its customers.
Profiling: Assigning Customers to
Segments
• Ways to create customer profiles
- RFM
Based on behavior
-Product affinity
- Demographics
- Cluster or lifestyle coding Based on attitudes,
demographics, lifestyle
Profiling: Assigning Customers to
Segments
• Classification by product affinity
- Affinity starts from customer’s perspective
- Use Cross-Buying rates.
-This is done by cross-tabulating purchasers
of one product against purchasers of another
product
Profiling:Cross-Buying rates
between A and B
A
No row
Yes row
Total row
B-No
B-Yes
Total
268431
96.99%
27023
68.47%
295456
93.43%
8328
3.01%
12444
31.53%
20772
6.57%
276759
100%
39467
100%
316228
100%
Profiling:Affinity Matrix showing
likelihoods of purchase
Prod A
Prod B
Prod C
ProdD
Prod A
eq
10.5
2.4
4.5
Prod B
10.5
eq
9
1.1
Prod C
2.4
9
eq
3
Prod D
4.5
1.1
3
eq
Third Party Sources
• Primarily demographic, attitudinal, lifestyle
and financial data.
• Available at the zip code and census tract
level.
• Census tract (or block) level is a finer
classification but is more expensive and
requires additional statistical techniques.
Third Party Sources
• Zip code used when number of customers or
prospects is large (> 100,000).
• Zip code data can be overlaid with purchase
data for profiling purposes.
• Major Products:
•
•
•
•
ClusterPlus (First Data Solutions)
PRIZM (Claritas)
MicroVision (National Decision Systems)
Mosaic (Experian).
Third Party Sources
• Data is primarily averaged at the zip code level.
• Based on the premise that
– “ Birds of the same feather….”
• Issues:
– Possibility of outdated information.
– Results in promoting to the wrong people.
– Useful only when any form of prospect or customer
information is unavailable.
National Databases: File Enhancement
• Nearly total coverage of US households.
• Attitudinal Data
– Contains information on general opinions, and
perceptions of the people.
– Useful when launching new products/services.
• Lifestyle Data
– Provides information on personal interests, and leisure
time activities.
– Result of combining geo-demographic and market
research data.
– Example: Claritas (geo demographic)
+ Simmons (Market Research)
National Databases: File Enhancement
• Lifestyle Data (Continued)
– Improves the reach of print and electronic
media.
– Representative strategies for use:
• List profiling.
• Use the lifestyle characteristics for only customers
with the highest priority.
• Apply profiles to prospect files.
• Used as a guideline for obtaining other lists.
National Databases: File Enhancement
• Financial Data
– Largest providers – Experian, and Transunion.
– Data on credit card purchases, installment loans, applications for
credit, and payment history.
– Marketers can send their house lists to financial data providers.
– The financial data providers then provide a profile of their best
customers.
– Information at segment level not individual level.
– Then prospect list can be used to send promotions to prospects that
match profiles of best customers.
National Databases: File Enhancement
• Demographic Data
– Available at the household or individual level.
– When certain data (e.g., age) is unavailable
– A reasonable inference can be made for a majority of the
individuals.
– Multiple sources:
• Motor Vehicle Registrations (Polk)
• Telephone and City Directory (First Data Solutions and
Metromail)
– Values that are available are accurate and are not
summaries at the Zip Code Level.