strategy+business is published by the global management consulting firm Booz & Company
 
or, sign in with:
strategy and business


Click here to list all strategy+business/Knowledge@Wharton articles.

Published: 12/04/02
Page 1 2 3 4 5 6 7 All
 

The Customer Profitability Conundrum: When to Love 'Em or Leave 'Em

He also says that customers who tend to be unprofitable may not be as bad as they may look, and the ones that appear profitable may not be as good as they appear.

“Any time you rank customers, you’ll find that the extremes tend to regress toward the mean and become more average over time. If you have an incentive or rewards program for good customers, you’ll find that, eventually, they may not be as good as they used to be. This is especially true when you’re talking about very sporadic behavior, as we are here. If a person applies for a bank loan only once in a while, how can the bank know if the customer is abandoning the bank or just has no need for the bank’s services at the moment?”

Fader acknowledges that companies have difficulty analyzing their customer bases at the level of the individual person because each customer behaves randomly. However, it is a long-accepted practice to make predictions about groups of people. Actuaries do it all the time. They can forecast accurately what percentage of a group of people will perish in auto accidents, but cannot name the individuals who will die.

“Customers who tend to be unprofitable may not be as bad as they may look, and the ones that appear profitable may not be as good as they appear.”
“No matter how much data you have on a customer, it’s hard to capture everything that’s going on,” Fader says. “But when you aggregate a bunch of people together it is possible to make very accurate statements about the cohort as a whole. You can, for example, study a group and say with confidence that 25% of these customers will buy products twice from us this year. That’s the way a customer base should be analyzed. I worry about clients or vendors who sell or use CRM services that claim to be able to pinpoint customers and their behavior. It’s voodoo.”

Fader argues that data mining “is wonderful for the original reasons it was developed, like credit card scoring to determine who should get cards. Data mining can help you figure out what makes group A different from B. There are lots of applications for that framework. But it doesn’t work well where behavior is evolving. Data mining is pretty darn limited when you move outside of a static snapshot.”

It is more difficult in B2B relationships than in B2C to lump customers together, and analyze them on that basis, Fader says. However, the benefits of one-to-one marketing can be greater in B2B relationships where the monetary value of each customer is much larger than in B2C.

Don’t Shoot from the Hip
Fader’s opinions represent the minority view. Dallas-Feeney, the Booz Allen consultant in New York, says managing customers can be done and done well, as long as it is not misunderstood. It involves, he says, more than just firing customers.

“None of our clients would say, ‘OK, your analysis tells me I should exit this many customers, so let’s get rid of them immediately.’ Every one of those customers -- with the obvious exceptions of bad credit — will be presented with a set of incentives so that a company can say, ‘Hey, if we can make this relationship better, we’ll keep you. But we can’t be successful unless you do the following.’”

The overarching philosophy is that every unprofitable customer should get a chance to become a better customer, with the sales force being given the responsibility of turning those customers around. But some customers will remain unprofitable no matter what steps a company takes to change its cost structure.

Page 1 2 3 4 5 6 7 All