Fire your bad customers.

That piece of advice has become widely accepted in recent years as companies have sought to manage their relationships with customers in more sophisticated ways. The rationale for this idea is clear-cut: Low-value customers — such as the ones who hardly spend any money on your services or products yet tie up your company’s phone lines with questions and complaints — end up costing more money than they provide. So why not jettison them and focus your customer-relationship efforts on more profitable individuals? Or, as an alternative, why not at least try to increase the worth of the low-value customers to your firm? If a firm has only valuable customers, the thinking goes, its profitability and shareholder value should increase.

It all sounds quite rational, and many corporations have jumped on the bandwagon. But a new study by two Wharton marketing professors, Jagmohan Raju and Z. John Zhang, and Wharton doctoral student Upender Subramanian, cautions that firing low-value customers may actually decrease firm profits and that trying to increase the value of these customers may be counterproductive.

The notion that firing unprofitable customers is a smart thing to do has emerged out of the broad acceptance of a practice usually referred to as Customer Relationship Management (CRM). With CRM, firms often use information technology to quantify the value of individual customers and provide better privileges, discounts or other inducements to customers identified as having high-value. In their study, Raju and Zhang have coined the term Customer Value-based Management (CVM) to describe this central component of CRM. These customer analyses have often shown that a small proportion of customers contribute to a large percentage of profits, and that many customers are unprofitable.

Financial institutions are perhaps best known for treating low-value customers differently from good ones. For instance, bad customers at Fidelity Investments are made to wait longer in queues to have their calls taken by call centers, according to examples cited in the study. But many other types of firms have embraced CRM and are giving low-value customers the cold shoulder. Continental Airlines e-mails only its high-value customers, apologizing for flight delays and compensating them with frequent-flier miles. At Harrah’s, room rates range from zero to $199 per night, depending on customer value. Some firms fire customers outright. In July 2007, CNN reported that Sprint had dropped about 1,000 customers who were calling the customer-care center too frequently — 40 to 50 times more than the average customer every month over an extended period.

In the study, “Customer Value-based Management: Competitive Implications,” Zhang, Raju and Subramanian break ground by analyzing CVM in the context of a competitive environment. The researchers acknowledge that firing bad customers may make some sense in industries where there is little or no competition. If a firm treats all customers equally, the argument goes, not only does the company waste resources on attracting and retaining unprofitable customers, it also under-serves profitable customers, who may become unhappy and leave.

Targets for Poachers

However, for the overwhelming majority of companies operating in a competitive environment, firing low-value customers can be counterproductive, the researchers conclude. The key reason: Companies that rid themselves of low-value customers — or take steps to turn low-value customers into high-value ones — leave themselves open to successful poaching by competitors. If the competition knows that you have fired many or all of your low-value customers, they are likely to intensify their efforts to take your remaining customers away from you because they now know that all, or most, of those remaining customers are of the high-value variety.

“Over time, companies have acquired a lot of capabilities in processing customer information,” Zhang says. “They have all sorts of analytics to do data mining and to figure out how to use that data. One thing companies have done is to figure out who are their profitable customers, and they have concluded that firing low-value customers is a good idea. The problem, however, is that while this idea seems to make sense, it only makes sense in situations where there is no competition, which is very unlike the real world. Our paper looks at how CVM affects companies competing with one another.”

“What our analysis tells us is companies make money, in part, by confusing their competitors about their customers,” Raju says. “If you make your customer base transparent by firing your low-value customers, competitors will hit you hard because you will be left with customers of one type.’

Instead of firing unprofitable customers, some companies have tried to turn them into high-value customers by giving them inducements to change their behavior, such as teaching them to spend more or to use low-cost support channels. But the Wharton researchers found that this idea is also wrongheaded. “If you make low-value customers more valuable, this can also be counter-productive because it also encourages your competitors to poach more intensely,” Raju says.

So what is the proper way to manage relationships with low- and high-value customers? “Our research finds that a better approach is to improve the quality of your high-end customers at the same time that you keep your low-end customers, but you should find other, cheaper, ways to manage the low-value customers, such as encouraging them to use automated phone-response systems or the Internet or offering minimal discounts or other benefits,” says Raju “You have to keep your competition confused about who your good and bad customers are.” 

CVM has enjoyed significant support amongst corporations, researchers and others because its logic seems so compelling. But CVM, once adopted, has often proved disappointing. Studies have shown, for instance, that the retail banking industry, while investing billions of dollars in CVM, has been unenthusiastic about the results to the bottom line, according to the Wharton paper.

“One reason why actual results differ from expected outcomes could be that, hitherto, researchers and industry experts have by and large looked at firms in isolation without considering competitive reactions,” the Wharton scholars write. In their paper, the researchers provide the first theoretical analysis of CVM practices when CVM capabilities are potentially available to all firms in an industry. The researchers set up a mathematical model and applied game theory to see how two competing firms, each with the same size base of customers called ‘Good’ and ‘Poor’, would compete for customers by offering various inducements. Among other things, the model assumes that the firms have access to the same CVM technology, that the firms are equally efficient in offering inducements and that each firm can identify its customers. 

Another finding: Firms in an industry may become worse off as CVM becomes more affordable. Hence, they have an incentive to self-regulate their ability to collect or use customer information. “In some cases, CVM can do damage to an industry,” notes Zhang. “Say you and I are competitors. We both have good information and we continue to poach each other’s customers. This is high-tech marketing warfare. If the cost of CVM increases, it’s not necessarily a bad thing. It’s like when armies fight each other with high-cost ammunition: When the cost increases, both sides have less of it, and fighting subsides. But if the cost of ammunition drops, the armies have more ammunition and fighting intensifies. So there’s an incentive for companies to get together in industries and agree to use certain kinds of information.”

CVM vs. Targeted Pricing

The Wharton researchers stress that it is important understand that CVM is different from another concept that has taken root in many companies in recent years — targeted pricing. With targeted pricing, firms differentiate between customers based on their willingness to pay and they charge a higher price to those who are relatively price insensitive. In this respect, a high-value customer is one who can bear a higher price. Put another way, a high-value customer is treated poorly. By contrast, under CVM a customer may be of high value due to other characteristics, such as the kinds of goods purchased, the number of times a product is returned to the seller, and the number of times the customer requests customer support. Hence, under CVM, a high-value customer would typically receive lower prices or better service than a low-value customer.

The researchers say that, in future studies, they will continue to explore CVM. They want to analyze such topics as how customer value can be more accurately measured, how it can be enhanced, and in which industries could CVM prove most valuable. In the meantime, they say their new study should help convince firms to reconsider the notion that firing bad customers is a smart decision.

“What we’d like readers to take away from our paper is that just ‘cleaning up’ your customer base is not good enough,” Raju says. “You should focus on good customers and try to improve their quality and not just try to get rid of the bad ones. Firms should find cheaper ways to keep low-value customers because they are confusing your competition to your advantage and there’s a chance someday that they will become good customers.”

Customer Value-based Management: Competitive Implications