One of the first
data warehousing projects I worked on was for a bank that wanted
to calculate customer profitability. We had millions of
customers, with a range of different account usage patterns,
balances and accounts held, other products bought through our
life and pensions arms, and length of time with the bank. Our
project built a data warehouse that brought all the customer and
account information together, and calculated the annual profit
(or loss) the bank made from each customer.
The point of all
this was to identify those customers who cost the bank money, or
made very little profit contribution, and to either make them
profitable, or to encourage them to use lower cost channels
(ATMs). In addition, those customers who made a high
contribution to profit (and in fact a small proportion of
customers made a disproportionate contribution to profit) were
then actively focused on, given their own personal account
managers, and identified as customers we had to keep. Building a
customer profitability model allowed us to segment our customers
into distinct groups, encourage the lower contributing ones to
either incur less costs for us or move up in to a higher
contributing segment, and letting us focus more attention on
those customers who made the most money for us.
On this theme, SFGate.com (via
Slashdot) are running an article titled
"The customer is always right? Not anymore" which looks at
how BestBuy is handling
customer profitability. According to the introduction:
"So much for the customer always being right.
Some retailers are deciding that the customer can be very,
very wrong -- as in unprofitable. And some, including Best Buy
Co. Inc., are discriminating between profitable customers and
shoppers they lose money on.
Like a customer who ties up a salesworker but never buys
anything, or who buys only during big sales. Or one who files
for a rebate, then returns the item.
"That would be directly equivalent to somebody going to an ATM
and getting money out without putting any in," Brad Anderson,
Best Buy's chief executive, said in a recent interview. "Those
customers, they're smart, and they're costing us money."
Anderson said Best Buy was tightening its rebate policies in
the case of customers who abuse the privilege, but declined to
say what else his company was doing to discourage its most
"What we're trying to do is not eliminate those customers, but
just diminish the number of offers we make to them," Anderson
Interestingly, BestBuy don't ever 'fire' these customers,
instead aiming to make them profitable instead. The article does
go into examples though where 'firing' customers does happen,
with one consultant for another company quoted as saying:
""Then there are those customers that are just evil
customers ... fundamentally they're out to cheat us," Selden
said in a telephone interview. "It's not a large number of
customers, but they can have a material impact on a business."
Take a look if you get a chance. It's a good article that
illustrates one of the real-world applications for data
warehousing and data mining, and one that applies to banking as
well as retailing.