Learning from Customer Defection

Frederick Reichheld

Corporations lose half their customers every five years, and that executives are "shocked" by the statistic because they do not measure customer defection and make little effort to prevent them.

Defection is significant because it pertains to the stream of revenue for the company, the cost of retaining a customer is significantly higher than attracting a new one. Failure to be attentive to defection leaves a company blind to the problems in its business practices that, if addressed, would improve the company's long-term performance.

Some of the reason companies don't pay attention to defection are listed:

Loyalty and Profit

Generally, the longer a customer remains with a company, the greater their worth. It's suggested that reducing defection by 5% can in some cases double annual profits. Among CEOs, there is a notion that loyalty is important, but the company is often focused on the short-term cash flow and profit.

What keeps customers loyal is "the value they receive." When a company focuses on short-term profit at the cost of value creation, they can gain a short-term boost in their performance, but at the cost of long-term gain.

Most of the traditional financial and marketing metrics are geared toward short-term phenomena, comparing monthly or quarterly data, which encourages short-term rather than strategic thinking.

Searching for Failure

The contemporary climate of business in the USA is success-driven. We seek things that have a dramatic short-term impact and seek to apply the principles in other situations, assuming they are repeatable.

From a career perspective, little regard is given to the manager who discovers ways to avoid failure. There is no award or even appreciation for those whose action prevents something negative from occurring (because it never occurred, there's no "proof" that their action prevented it).

The study of failure is recognized in safety. The example is given of the airline industry, whose safety record far exceeds six-sigma standards. This arises because, in the wake of any crash or malfunction, intense attention is paid to discover what went wrong, and measures are taken to address the defect.

A quote from Warren Buffet on investing: to outperform the market, it would be difficult to identify the best performing companies. It would be far easier, and success far more likely, if you bought a market portfolio and avoided the worst performers.

Defining Defection

Defections can easily be spotted when there is a complete shift: a customer who closes and account or stops buying altogether. It's harder to detect when a customer shifts some purchases to another vendor.

The author presents a case-study of a company that had very few complete defections. When they examined the numbers more closely, they discovered that there were many customers who had decreased their purchases. Interviewing these "partial defectors" uncovered a lot of complaints, and addressing these problems led to significant bottom-line benefits.

Core Customers

Core customers can be identified as those who are most profitable and loyal: they spend more money, pay bills more promptly, require less service. Another factor is how well your products suit their specific needs. Finally, there are those who are worth more your firm than they are to your competitors - as the ones valued by competitors will be wooed and are more likely to defect.

While there is a notion to serve and satisfy all customers equally, allocating resources to non-core customers has a lower return. A better performance will come from preferential treatment to the segments that will be most loyal in the long term, even if it means neglecting those who may be profitable in the short term. Some companies waste a lot of resources trying to maintain the loyalty of customers who are likely to leave, and thus represent the lowest long-term value.

The key, then, is to place the greatest emphasis on customers who will give you the greatest share of their lifetime business, and focus on them. That is, to consider not the immediate sale, nor the next quarter, but the lifetime value, in deciding how much to spend or ensuring their interests are served.

Root Cause Analysis

The author suggest asking "why" at least five times to get to the root of a failure: a product was returned (why) because a connector came loose (why) because a plug was out of tolerance (why) because a machine failed (why) because maintenance wasn't done on schedule (why) because there is an attendance problem in that department. Even at that point, there's probably a few more levels of "why" before the problem can be identified.

With customers, it's generally not so simple: a customer may not defect until they have had three or four disappointing events - and the "last straw" may not have been the decisive factor.

The author also speaks of causes unrelated to business practices, such as major life events - a career move, relocation, marriage, birth, divorce, or death, which causes the customer to re-evaluate their choices. The likelihood of defection increases, from double to quadruple its normal rate, when the customer is in such a situation. This represents an opportunity for a competitor, and a time for vigilance for a company that wishes to retain the customer.

In examining both internal and external causes, a company will discover instances in which customers are likely to defect (these are called "doors" in the "customer corridor") and determine what can be done to prevent defection at these instances.

The author dismisses focus groups: they tend to be people speaking about hypothetical situations, and engaging in "groupthink" as they attempt to define a plausible, logical, and even admirable motivation for their choices.

The author proposes with a senior management group and a sampling of front-line personnel. Then, identify key defectors. Consider:

  1. Their previous purchasing behavior should reflect a purchasing pattern (sufficient quantity and frequency of orders)
  2. Their demographic characteristics (a clear match to your most profitable customer profiles)
  3. Ensure they have defected (they have switched to a competitor, not stopped using the product altogether)
  4. That they are customers who would have been worth keeping (some customers, even regular ones, are high-maintenance)

Once that's done, assign each participant 10 to 25 defectors to call. A telephone conversation with a former customer can be awkward and uncomfortable, but there is no better way to get valid information. And while executives might prefer to delegate this task, it has the greatest impact if they do this first-hand, as there is a tendency to be skeptical of information provided by subordinates.

Getting the Right People to Learn the Right Lessons

Market research is limited to traditional marketing concerns - setting prices, designing packaging, and conducting promotional campaigns - which are most often aimed at appealing to new customers. In the rare cases where a company seeks to learn about existing customers, the reports are generally limited in scope, and the information is provided to senior management, and seldom makes its way to the employees with customer-facing roles in a timely manner. Even then, research that is unflattering is generally suppressed to protect the company's image and employee morale.

The "root cause" interviewing process (above) overcomes some of these weaknesses: both executives and front-line employees get direct feedback from the former customers (the information is not interpreted or filtered)

In companies driven by metrics, a thorough analysis of the financial impact of defection, which draws a direct correlation between defection and performance, can be compelling, and the establishment of metrics that monitor defection (as well as typing rewards to reducing defection) can provide incentive to address the problems.

It's suggested that metrics and incentives based on new business lead to dysfunctional behavior: a manger whose bonus is based on increasing sales and bringing in new customers will undertake actions that do exactly that, even to the detriment of current customers, even if the "new" customers have a short churn rate (which is actually in his interests, because they can be "new" customers again, counted toward the next initiative).

While maintaining an existing customer base lacks glamour, even minor differences make a difference. The author uses a sports metaphor: a baseball fan recognizes that there's a major difference between a .280 hitter and a .320 hitter, even though it's only a 4% edge.

Likewise, a minor improvement in defection has a major impact on the bottom line. In one example, an insurance company found that a 1% reduction in defection increased commissions by 20%. In another, it was illustrated to auto dealerships the potential increase in income if customer loyalty were addressed.

(EN: These examples involve individuals whose income is directly impacted. A dealership owner can see the correlation, but the sales force are paid on new sales, and the service department often receives no bonus at all even though their performance has a major impact on retention. To those who hold the purse-strings, it's easier to convince them of the impact of actions that increase income, but harder to sell the notion of actions that prevent a decrease: it's arguable how much business might have been lost if something was not done.)

Making Failure Analysis Permanent

For a business that is new to the concept of retention, measuring share-of-wallet rather than raw sales figures can help drive customer retention. For a business that already has a churn rate, a reduction in churn can be measured and its financial impact predicted. (EN: but for a business with high share-of-wallet and a low churn-rate, it is more difficult to assess in a convincing way).

A few case studies:

The author mention's Pareto's concept of the "fruitful error" that is bursting with the seeds of its own correction, implying that the most successful companies, to which customers are most loyal, are those who are most attuned to feedback, and most aggressive in solving problems.

(EN: The article concludes here. A few sidebars are presented afterward)

Sidebar: The Satisfaction Trap

The author suggests that satisfaction surveys are imperfect: a company with a high defection rate can also conduct a satisfaction survey that has overwhelmingly positive results.

First problem: surveys measure what customers say about their level of satisfaction, which often misses the elements of their customer experience that wins their loyalty. The example is given in the automotive industry, where most dealers send a satisfaction survey after a purchase. The dealer surveys show a 90% satisfaction ratio - meanwhile, repurchase rates hover around 30%

Second problem: the surveys are often poorly conceived and executed, by a company who is motivated in advance to generate numbers that prove that they're doing well, or a research firm that wants to provide good news to their client, and the survey is "rigged" accordingly. (EN: and when a company brags about its customer satisfaction, my sense is that it's a desperate attempt to convince their customers that they really are satisfied when, in truth, they are not.)

Third problem: even when the survey is designed to be objective, it is generally designed to inquire into specific areas that the designer felt are critical to customer service, which may miss factors that are important, and the questions are too high-level (e.g., how satisfied are you with customer support?") to provide meaningful detail.

Fourth problem: rewards and incentives are based on the survey scores, decoupled from metrics of customer loyalty and profits. Fundamentally, they don't care what the reality is, so long as the scores are high. A few examples from the automotive industry are a dealership that offered a free auto detailing to customers who returned surveys marked "very satisfied" in all categories; another dealership provided a sample survey form showing how to answer the questions; salesmen routinely ask customers to give them the highest scores in all categories (even using tactics such as suggesting they will lose their job if the ratings are low).

Fifth problem: surveys measure a one-time activity (such as purchasing a new car) rather than the customer's long-term experience of owning and using the product, which is more critical to their decision to repurchase.

Sixth problem: few companies have the systems in place to determine the value of their customers and segment the market accordingly. Hence, all customers are treated equal in terms of satisfaction scores, and the company responds equally to any inbound complaint, regardless of whether it comes from a loyal and high-value customer or a short-term parasitic one.

There's a not about "customer complaints" as being better, but likewise misdirected. The example is given of a bank manager who heard, and reacted to, customer complaints about the length of time spent waiting in line for a teller, whereas the most profitable customers do most of their business by phone, internet, and ATM. So addressing the problem at considerable cost, by hiring more tellers, decreased complaints but did not have an effect on the bottom line.

While the author does not completely dismiss surveying, he indicates that companies should not "rely solely" upon them. The primary metric is lifetime value: the frequency and volume of purchases, share of wallet, length of engagement. Even in industries where purchase frequency is low (such as car dealerships), there is behavior to be measured between purchases that can help gauge satisfaction (such as how much of the service and repair business is done in the dealer's shop).

A quote from the general manager of Lexus: "The only meaningful measure of customer satisfaction is repurchase loyalty."

Sidebar: Discovering the Causes of Defection

The author presents a "case study" of Everybank, but concedes that it's a fictitious company composed of several such banks (EN: as such, this borders on a fiction story rather than a case study, though the incidents portrayed may be based on fact, they are likely to be contrived.)

The typical regional commercial bank has about two million customers and an annual defection rate of 20% - which translates to a loss of 400,000 customers per year and customer tenure of five years. To make matters worse, many of the customers who were leaving were among the core customers (the 20% of accounts that generated 80% of profits).

Other banks offer better rates, and had done so for quite a long time, but customers didn't shop around or pay much attention to competitive advertisements until there was a disastrous marketing campaign, which cast a broad net to attract new credit card customers. Those who were banking customers were solicited and subsequently refused felt irritated enough to shop around.

Follow-up questioning led to other sources of irritation: interests rates were low on savings and high on loans, there were high fees on ATM transactions, conveniently located branches were closed, customers "liked" employees who are no longer with the company, there were errors in their statements, their complaints had been ignored or dismissed by the bank's phone service reps.

Questioning about their new bank indicated that the rates were better, and the customers valued the consolidated statement. When asked if the new bank charges lower ATM fees, customers indicated "To be honest, I'm not sure."

Other problems discovered were the tendency of salesmen to steer customers toward products that were inappropriate for their income level, the marketing department's failure to consider payment history on loans when sending offers to customers, the collections department's brusque tactics with customers who were late on payment but had significant CDs, and the fact that a person who called in would get a different person each time ad would be bounced around to get to an employee who was authorized to serve them.

Examining the bank's attrition levels, it was seen that customers with a particular kind of money-market account tended to leave less frequently and customers with three or more different products had the lowest attrition rate. Conversely, the customers with the lowest retention rate were those who bought a single product after a promotional campaign (a high-yield CD, a credit card with a low introductory rate) and who left after the promotional period had ended and the product returned to its normal rate.

As a result, there were a myriad of small problems to be addressed to retain customers, the cost of doing so was far less than advertising campaigns to attract replacements, and would have a greater impact to the long-term performance of the bank.

(EN: The author goes on to mention the efforts to address these problems, but solutions are self-evident in the descriptions of the problems and, being as this is a hypothetical story rather than an actual case-study, there's little credibility in the details.)