Ryan Murphy and I share an introductory analysis of the CLV of a national credit union. It includes an exploratory analysis of the data set of over 60,000 accounts and how demographic and other factors play into the profitability of our calculated customer clusters.
2. The Customer Lifetime Value (CLV) metric
is important for businesses across many
industries. It is a metric that allows
companies to save money, market more
strategically, and drive greater amounts
of sales — all because it can inform an
organization of the customers that have
the highest potential.
Context
3. While the calculation for CLV is relatively
straightforward in many contexts, it can
be difficult to determine for financial
institutions such as banks or credit
unions.
Take for example a savings account:
banks must pay interest to the depositor,
making it a cost; banks also rely on these
savings to use when distributing loans,
which generate revenues for the
institution.
Key Problem
4. The data being used in this project
measure the personal financial activity of
over 600,000 customers with data
spanning six months from March through
August.
Data Set
7. Questions
Question 1.
How valuable is the customer right now?
Question 2.
How valuable will the customer be
for the remainder of the relationship with the institution?
Question 3.
How can we describe the highest-valued customers
and subsequently pursue them?
8. How valuable is
the customer
right now?
Current Customer Value to the Bank
● Difference in generated revenues
and incurred costs
○ Spread from interest-bearing
accounts
○ Interest earned from loans,
credit cards, and debit card
transactions
○ Costs of services such as branch
visits or using ATMs
● Values are annualized
9. How valuable will
the customer be?
Customer Lifetime Value
● Forecasted estimate of Current
Customer Value to the Bank
○ Estimated 92% retention rate
○ 6% discount rate
○ Assumed death at age 80
○ Customer acquisition cost
10. How do we
describe these
customers?
Lifetime Value Groups
● Two-step clustering into five groups
○ Cluster #1 comprises the least
profitable customers including
all negative-CLV
○ Cluster #5 comprises the most
profitable customers
■ 75% correlation between
totals in mortgage interest
and CLV
13. We recommend that Navy Federal Credit
Union do the following:
● Utilize RFM approach
○ Adjusted RAND Index
● Evaluate additional variables
Recommendations
14. Suggested Variables
● Greater timeline of data
● Specific age of customer
● Geographic area
● Terms of mortgages
● Customer acquisition costs
● Retention rate
● Default rates and refinancing on loans
● Number of dependents
Additional
Variables
Presentation notes.
When will they pay off loans to determine decrease of revenue generation
Churn is how many people you lose.
Make a variable that says that?
Or we can estimate an age when usefulness runs out.
Determine profitability by category—when have mortgages and auto loans and educational loans been paid off?
Presentation notes.
Take into consideration whether this is reflective of the entire customer base
Presentation notes.
Presentation notes.
Not including interest still to be paid.
The 92% retention rate is based on industry averages across both commercial banking and related industries such as insurance.
6% discount rate is based on the 10-year treasury yield with a built-in risk-premium.
According to the World Health Organization, American males die at about age 79 while females die at about age 81.
The 92% retention rate is based on industry averages across both commercial banking and related industries such as insurance.
6% discount rate is based on the 10-year treasury yield with a built-in risk-premium.
According to the World Health Organization, American males die at about age 79 while females die at about age 81.
Presentation notes.
Presentation notes.
Specific age of the customer to provide a more granular analysis of …
Number of kids
Greater timeline of data, especially because consumer spending trends differ at various points of the year.
Geographic area, taking into consideration greater property values by location.
Customer acquisition costs, for a more complete post–CLV analysis to make cost-benefit determination.
Retention rate for churn analysis (predictions of future churn)
Specific age of the customer to provide a more granular analysis of …
Greater timeline of data, especially because consumer spending trends differ at various points of the year.
Geographic area, taking into consideration greater property values by location.
Customer acquisition costs, for a more complete post–CLV analysis to make cost-benefit determination.
Retention rate for churn analysis (predictions of future churn)
Number of kids
Presentation notes.
What kinds of marketing implications are there for this sort of analytics?
Supplement: things to think about with these implications?
RFM over CLV
“You end up with more questions than answers.”