Selling Power Magazine Article

When a company's customer base shrinks, it's time to reinvest, rethink, and revisit the retention strategy that's no longer working. Customer retention, sometimes referred to as preventing customer churn, is especially important when you sell a subscription service or a product with recurrent transactions. But holding on to customers can be expensive, especially in highly competitive markets. In addition, as markets change, customer retention strategies must adapt.  


While customer retention may not be simple, the first step is spotting potential defectors, because it's more cost effective to focus on them. But recognize that some customers are going to drop out no matter what you do. Next, understand how much value you can expect from customers if they stay on board. This sets your retention budget for keeping them. 


These challenges are tougher if you sell a high volume of low-dollar business. Personal conversations are too expensive in these markets. You have to use data and smart analytics to inform your retention decisions.


Hierarchies  


MIT professor Michael Braun has developed what he calls "a hierarchical competing risk model" to do just that. "Competing risk models are often used when observations include both the time and cause of an event, as in medical research, for example," he explains. 


Braun's model addresses which customers may choose to terminate your services and why. It distinguishes the customer churn that can be influenced by something your firm can do (e.g., improve service, cut prices, or upgrade value) from uncontrollable churn due to customer relocation, change in economic circumstances, death, or bankruptcy. 


Braun used data from a land-based telecommunications provider to estimate the incremental customer value a firm can earn by delaying its controllable churn while recognizing that some churn is unpreventable. The approach can be used to focus retention on customers you can keep.  


"We used past customer behavior to determine churn due to controllable factors," Braun explains. "The method also yields a measure of the return on investment of retention activities."  


The Braun model is directly applicable to contractual sales, such as those in communications, cable television, health clubs, or other services that are membership-based, even direct-debit charitable donations. Its basic principles could be adapted to any recurrent sales to regular customers. 


The telecommunications provider had plenty of data on its customers, including initial service date, churn date, location, income, age, employment, homeownership, and education. More than three-quarters of the customers who left gave a reason for discontinuing service: competition, price, dissatisfaction with service, dissatisfaction with product, moving from service area, and so forth. Some customers had been cut off for nonpayment. In the case of this telecom provider, much of the churn turned out to be uncontrollable, a signal that retention efforts must be tightly budgeted to be cost effective.


Allocation, Allocation, Allocation 


Braun, who worked in telecom before teaching at MIT, says allocation of retention resources is a very common challenge: "It's important to know what you can and cannot control. There are costs associated with any activity. And you must know the return on investment you want to achieve."  


These are straightforward business principles. The hard part is applying them. To do so, you must recognize and allow for the uncontrollable churn or risk reducing the return on investment on whatever you do to hold on to the rest. Uncontrollable churn will likely differ by market segment and length of time a customer's purchased from your firm. 


Although the model was applied to a retail telecom market, Braun says it is "absolutely relevant to B2B." Subscription-based B2B sales are most appropriate to the model, but any regular transactions that suddenly stop or sharply decline are good candidates. 


"It can work for a repeat-purchase model,  as in the purchase of office supplies, for example, that stops or is interrupted for a while or goes below its natural rate," Braun explains. A customer's going out of business is uncontrollable, for example, but you might control whether that customer moves his or her business to a competitor. 


The model's math would differ for regular, rather than subscription, sales. But the vital decisions are still in whom you should invest and how much. Braun's approach can be used to tailor retention strategy and timing to individual customers.


Another implication of Braun's approach is the need for constant investment in customer acquisition. "You must be constantly acquiring new customers, because some are going to drop off no matter what you do," he stresses. 


Estimates of how much you will spend holding on to even controllable churn can inform your new customer strategy. Some customers may be prevented from leaving or just delayed a while but at a higher cost than obtaining new customers. Furthermore, recognition of retention costs and inevitable churn is necessary to properly predict the value of each new customer. 


Application of Braun's model does require data about or estimates of the reasons for churn. It is best to get this data from customers when they cancel a service, but sales reps can also make informed judgments. Additionally, companies can use customer surveys to uncover the reasons behind churn. "The better the data you have on the reasons for churn," explains Braun, "the more accurate the model will be." 


Many companies have struggled with the retention puzzles Braun's model addresses. Business-intelligence tools address the problem by tapping into massive customer (continued on page 2)

databases. But Braun says it is risky to rely on a "black-box approach." He thinks cutting-edge statistical and modeling techniques are the best approach, and these are more likely to come out of academic institutions than from hard-pressed corporate IT departments. He continues, "You have to blend an understanding of business with statistical expertise."


The Number One Objective

Companies that make high-dollar B2B sales often rely on relationships to determine retention strategy. Whirlpool, for example, has both ultimate customers, who buy and use its appliances, and trade customers, the distributors and other firms that sell Whirlpool products directly to consumers. 


Trade customers are Whirlpool's regular, steady buyers. Whirlpool's success - or failure - depends on them. Sam Abdelnour, Whirlpool's vice president of North American sales, says that, although all customers are valued, the days of treating all trade customers alike with the same service at the same cost are long gone. His company devotes substantial effort to modeling the expected volume, margin, and profitability of each trade customer. These models are used both to plan the business and estimate a cost of service that Whirlpool can afford to support each segment of its distribution network. 


"We use these estimates to calculate how much human resources and other investments we can make in each customer segment," Abdelnour explains. "For example, we might make an on-site sales call to one twice a week but to another only once a month. And for some, we must tell them we have a Web solution they can use." 


Whirlpool never loses a trade customer completely, unless that customer goes out of business. "[Customers] never say they will stop carrying our brand," Abdelnour says. "It is all about how much they carry and the volume they sell." 


If trade customers feel that Whirlpool is not delivering satisfactory quality or service on either existing products or new launches, they may decide to move some business away from Whirlpool to a competitor. But Whirlpool has such a close relationship with its trade customers that this seldom comes as a surprise. 


"We don't model or do algorithms to determine who we might lose," Abdelnour says. "We look at the numbers, but we generally hear from them when they are having problems." 


Moreover, trade customers always give Whirlpool an opportunity to retain the customary volume and profit. Then it's decision time. "If we might lose some business, we will invest more in the customer," Abdelnour explains. "But it is always give and get. [The increased investment] will have strings attached. We will say, ‘Here is how we can help you, but here is what we want to get.'" 


William Lintz is senior director of sales in Asia Pacific for RadiSys. The company sells $350 million a year in hardware and software to telecom, medical, industrial, aerospace, and defense companies. Typical transactions range from $100,000 to more than $1 million, and there are some continuing subscription sales, as well. 


"Retention is definitely our number one objective," Lintz stresses. "You want to hang on to customers you already have before you find new ones, because finding new ones is a lot tougher." 


In this business, if something goes wrong with a product and the provider does not address the issue properly, then that's a big threat to retention. "Customers understand that there will be problems," Lintz notes, "but the way to keep a customer is to address the issue immediately and with all your resources." 


Several times, RadiSys has been brought in to troubleshoot a system, even if the problem was caused by another supplier's equipment. Solving these problems, no matter who is at fault, keeps customers happy and loyal. 


Rivals pose another retention risk. "The competitor will come in with a lower price and try to disrupt our business," Lintz says. "The client gets excited and fears paying too much." 


In this case, the client will always alert RadiSys to the less-expensive offer, and Lintz can usually figure out both the competitor and its value proposition. But effective defense has to start long before the rival's pitch.  


"The solution is to be continually selling your value so customers know what they are paying for and that it is benefiting them," explains Lintz. "You have to continually protect your turf."


Lintz says another proactive retention tactic is to constantly build closer contacts with each client. "Fortunately, our CEO likes to make telephone calls, and our senior VP likes to, as well. This reinforces the value of our product. Customers want to have good relationships, as well as good products."

Profile of Customers in MIT Telecom Churn Study

CHURN TYPE	NUMBER OF CUSTOMERS	% OF TOTAL CUSTOMERS
Did not churn	35,736	73
Controllable	1,154	2
Uncontrollable	2,498	5
Nonpayment/abuse	6,293	13
Missing	3,012	6
TOTAL	48,693	100

Source: Copyright, Modeling Customer Lifetimes with Multiple Causes of Churn, Braun and Schweidel (braun@mit.edu).