As any
other business, banking relies heavily on its customers. Is it possible to
predict if any customer is about to leave? Well, starting from the fact that
keeping customers under your umbrella costs less than acquiring new ones we are
in a good path. Acquiring new customers cost from 5 to 15 times more than
retaining current ones.
Let´s suppose
we have Loui who all Mondays goes to the ATM for his customary $100, all of a
sudden Loui transactions begin to be scarce or inexistent. Loui is about to leave
but we do not know why. And the worst is that we barely notice about Loui among
all our customers. How important is to predict customers who move off from our
bank? Which kind of data do we need to analyze?
Our mission
is to predict customer’s departure before it actually happens to retain him.
Data that comes to our help include: number of contracted products, savings
evolution or account movement, occupation, claims, demographic data, etc.
Other
source is the one that the bank has collected from lost customers, who had left
the bank. Behavioral patterns are there.
Moreover,
there are different kind of “break ups” to design and build maps of departure,
profile of customers who leave, most common channels that originate such
response from our customers, geographic zones, products, and related data.
What do we
obtain from such analysis?
1.
To
know which customers’ segment should be
care and kept.
2.
Plans
and strategies to respond before customer relinquishment. We can know with
anticipation when, who and why.
3.
Increase
customer’s satisfaction and fidelity by knowing his needs in a better way. consumption
|
Data
|
How to analyze
|
|
·
External Sources that have information about demographics, employment
status and current personal status.
·
Social Networks
|
·
Fusion and analysis of structured and unstructured data
·
Interactive data visualization
|
|
·
Customer value that it is used as index of acquiring customers who
potentially can leave
|
|
|
·
Statistical external sources with expenditures/payments information.
|
