Linking Customer Loyalty to Growth

To most senior managers, growth is the engine of prosperity and success. Growing companies flourish; shrinking companies die. Good managers understand that the road to growth runs through customers — not just attracting new customers, but holding on to the ones you have, motivating them to spend more and getting them to recommend your products and services to the people they know. However, it is one thing to believe that customers are the driving force for profitable growth. It’s altogether different to know how to measure and manage the customer relationship effectively. Using the right customer metrics is essential to assessing and monitoring how companies deliver for customers and determining customers’ new and unmet needs.

Most companies do a relatively poor job of managing their relationship with their customers. It isn’t that they don’t care, but rarely do they have any insightful information they can act upon to make the relationship more valuable. As a result, understanding how customers perceive the relationship and anticipating what they will do is typically no more reliable than reading tea leaves.

In a world where managers are looking to analytics to help clarify their most critical decisions, this presents a challenge: How do managers measure how customers really feel and what they are likely to do? More importantly, what impact can this information have on the business?

Most companies lack good information about their customers — for example, most do not have good customer databases and, of those that do, almost none tie customer survey information to customer behavior information. Therefore, it is not surprising that growth is often so unpredictable, leaving managers scrambling for useful ways to measure their customer relationships so they can predict how customers will behave — and how successful their business will be in the market.

In recent years, researchers have advanced a number of customer metrics to illustrate the connections between customer behavior and growth. In the harsh reality of the marketplace, however, these efforts have generated more smoke than heat. The best metrics have shown only modest correlations to growth, and none of them have shown themselves to be universally effective across all competitive environments. But the failures and weaknesses of existing metrics have not discouraged company managers from adopting new ones.

About the Research >>

Searching for a Light in the Dark

The initial development of forward-looking customer metrics rested on attempts to understand why people buy. The underlying logic was relatively straightforward: If customers preferred a company, managers assumed there was an increased probability that they would become loyal. Simplistically, most of the differences among these types of metrics revolved around the meaning of “prefer” (for example, whether it meant perceived quality or value).

Historically, getting customers to select one offering over another was a matter of marketing. Marketing had brandcentric objectives — getting consumers to view the company’s products and services favorably in comparison with those of competitors. Marketers tended to focus on the aspects of brand image that were under their control.

The drawback of this strategy is that not everything is a marketing problem. In fact, most of the issues that affect customer loyalty do not fit neatly into the marketer’s standard tool kit of the four Ps (product, price, promotion and place/distribution). This problem became abundantly clear during the 1970s and ’80s. At a time when Total Quality Management was becoming important, managers realized that the product or service was only part of a broader customer experience, and that the relationship between companies and customers was complex. For example, a product couldn’t be separated from what it felt like to own it (everything from learning about it to purchasing it to getting it repaired). Satisfaction meant feeling good about the overall experience. This gave rise to a new category of metrics designed to monitor customer experience.

The first experience-oriented metric to gain a significant foothold with companies was focused on service quality. Specifically, it allowed companies to survey customers about their ownership experience. The most prominent tool in this category was the SERVQUAL scale, developed in the mid-1980s by A. Parasuraman, Valarie Zeithaml and Leonard Berry.1 It examined multiple dimensions of service quality in terms of five core constructs:

• Tangibles: the appearance of physical facilities, equipment, personnel and communications materials
• Reliability: the company’s ability to perform the promised service dependably and accurately
• Responsiveness: the company’s willingness to help customers and to provide prompt service
• Assurance: the knowledge and courtesy of employees and their ability to convey trust and confidence
• Empathy: the degree to which the company offered caring, individualized attention

However, measuring service quality did not provide managers with the insights they were hoping for. Managers want information that can guide specific actions. But SERVQUAL was difficult to implement. Although the dimensions it looks at are sound, they don’t point managers to specific action. Rather, performance on each dimension represents the aggregate influence of the entire company. In other words, a dimension like empathy isn’t the responsibility of a single individual or process. Everyone is responsible, which makes it impossible to assign meaningful ownership for its improvement.

Beginning in the 1990s, in large part because of the difficulty with service quality management, many companies shifted their attention to customer retention. Research conducted by the Technical Assistance Research Program in Washington, D.C., found that the primary reason that customers left companies was dissatisfaction.2 Dissatisfied customers hampered a company’s ability to attract new customers; specifically, the research found that a dissatisfied customer told nine people about his bad experience. In fact, research has shown that dissatisfied customers are more likely to speak about their experiences than are satisfied customers.3 These and other similar findings provided a tremendous push for companies to learn how to manage dissatisfaction and complaining behavior.

Managers and researchers found that controlling dissatisfaction was not sufficient. In fact, few customers, even those abandoning one company for another, classified themselves as dissatisfied. The research focus shifted once again — to measuring very high levels of satisfaction, frequently referred to by managers as “delight.”4 Unlike other measures such as service quality or brand preference, customer satisfaction is relatively straightforward to understand, generic and therefore universally applicable. Customer satisfaction is the metric managers use most widely to gauge customer loyalty. Even without a precise definition of the term, satisfaction is clearly understood by customers, and its meaning is easy to communicate to managers.5

A great deal of research has gone into investigating the impact of customer satisfaction on customer behavior and business results. The results have been consistently positive. High levels of customer satisfaction have been linked to customer retention, increased share of spending and positive word of mouth — the primary behaviors associated with customer loyalty.6 Additionally, most studies have found high customer satisfaction to link to improved financial performance.7

But measuring customer satisfaction has problems as well. To begin with, the linkage between satisfaction and customer behavior and positive financial outcomes has tended to be modest. What’s more, managers often have difficulty acting on the information they obtain. In particular, how much does satisfaction have to improve for it to have any impact? Satisfaction metrics don’t behave like other numbers — their relationship to behavior tends to be nonlinear. As a result, big changes in customer behavior frequently don’t happen until satisfaction reaches a threshold, either very low (dissatisfied) or very high (delighted). This makes it difficult for managers to determine where to allocate investments in improved satisfaction or how to project a return on their investments.

As a result, managers have been clamoring for better loyalty metrics that can help them predict customer behavior and company performance. Books such as Customer Satisfaction Is Worthless, Customer Loyalty Is Priceless, by sales and service expert Jeffrey Gitomer, are indicative of the widespread frustration. In the past few years, a number of consultants and researchers have stepped forward with alternative metrics.

One of the most prominent metrics to appear is an approach called customer value analysis. Introduced in 1994 by strategy consultant Bradley Gale, CVA was supposed to demonstrate the link between what customers perceived to be the value of a company’s offering and its market share. Gale argued that customers selected companies to do business with based on the perceived value they received vis-à-vis similar offerings from competitors. Gale’s premise — that customers decide to purchase (and repurchase) products and services based on their perceived value — resonated with corporate managers. First, it is true that customers buy based on the relative value they believe a product or service provides (although defining “value” is not as simple as it sounds). Second, Gale’s methodology for calculating CVA was based on a seemingly objective, mathematical formula for determining how well a compan was doing in customers’ minds at delivering value relative to competitors.

Consequently, a number of respected companies, including AT&T, DuPont, Johnson & Johnson and Pitney Bowes, used CVA to measure their competitive positions and to help generate growth. But for a number of reasons, mostly having to do with the accuracy of the calculation itself, CVA fell out of favor. Managers found that simple, insignificant changes in the data could result in dramatically different results.8 In the end, CVA’s reported linkage to market performance did not hold in most settings.

Connecting “Buzz” to Growth

Unlike earlier methodologies, today’s most popular metric, the Net Promoter Score, focuses not on quality, satisfaction or value, but on how customer word of mouth — both negative and positive — can advance growth. It’s based on the idea that word of mouth can be managed to create a buzz around a brand or a product, and that by keeping close track of the NPS, companies can develop ways to become more successful. The methodology, developed by Bain & Company Inc. consultant Fred Reichheld, was introduced in 2003 in an article published in the Harvard Business Review.9 It described how companies could determine future success by studying customer responses to a single survey question: “How likely is it that you would recommend this company to a friend or colleague?” Reichheld, who published several books on loyalty during the 1990s, maintained that NPS was the most accurate gauge of customer loyalty and the “single most reliable indicator of a company’s ability to grow.”10

The approach works like this: Customers who rate the likelihood that they will recommend the company as high (at 9 or 10 on scale of 0-10) are classified as “promoters”; those who rate the likelihood as low (6 or below) are classified as “detractors.” The score is calculated based on the difference: the percentage of promoters minus the percentage of detractors.

Clearly, one of NPS’s most attractive features is its simplicity. One doesn’t need a Ph.D. in statistics to do the calculation. With NPS, companies can eliminate the need for long surveys and perhaps even reduce their research budgets. And the core idea that the more promoters you have, the brighter your future — seems to ring true. But the simplicity does not account for all of NPS’s appeal. What seems to draw the most attention from corporate managers is Reichheld’s claim about a link between NPS and growth. The Harvard Business Review article presents data that purports to demonstrate a strong relationship between high NPS and high growth rates relative to competitors. The benefits claimed are striking:

• Having the highest NPS in an industry results in growth rates averaging 2.5 times greater than that of competitors.
• Each 12-point increase in NPS corresponds on average to a doubling of a company’s rate of growth.
• Using additional metrics alongside NPS offers insignificant predictive advantage. (Reichheld argues that everything you need to know to predict growth can be explained with NPS; he goes so far as to assert that other survey-based metrics, such as customer satisfaction, have no link to growth at all.)

The idea that there could be a simple metric that ties current customer feedback to growth has resonated with managers. Among the many companies that have embraced NPS are eBay, Four Seasons Hotels, Intuit and GE. Some CEOs are even citing their Net Promoter Scores during conference calls with analysts as signs of their company’s ability to grow.

Testing the Net Promoter Metric

We set out to test Reichheld’s claims about NPS. In the absence of peer-reviewed, longitudinal studies about the methodology, we attempted to verify the results by replicating as closely as possible the methodology Reichheld used in his reported research.11 Does NPS really help companies predict customers’ future loyalty behaviors? Does it link to growth? Is it really better than other commonly used metrics? (See “About the Research.”)

The first thing to do was understand exactly how the research into Net Promoter was conducted. The research reported by Reichheld and Satmetrix Systems Inc. (a research company and coinventor of Net Promoter) was conducted in two parts: a microlevel investigation into the relationship between NPS and future customer behaviors and a macrolevel investigation into the relationship between NPS and company growth. Any attempt to validate the claims attributed to Net Promoter needed to follow the same process. To this end, we conducted a series of investigations into how different survey-based metrics tie to individual customer behavior and ultimately to growth.12

Net Promoter and Customer Behavior Determining the relationship between Net Promoter and how customers will behave in the future requires examining customers over time to understand their attitudes and subsequent behaviors. We conducted a two-year study of more than 8,000 customers of companies in retail banking, mass merchant retail and Internet services, monitoring individual customer ratings on common satisfaction and loyalty metrics. The study’s second year also tracked behaviors associated with customer loyalty, in particular customer retention, share of spending and recommendation behaviors. To be consistent with Reichheld’s methodology, we then examined the correlations among the various survey measures, including the Net Promoter question, and customers’ purchasing and recommending behaviors.

The linkage between every customer perception metric and subsequent customer behavior turned out to be modest at best. NPS’s ability to explain customer behavior ranged from 0% to 20% (meaning that 80% or more of the differences in customer loyalty behaviors was caused by something other than what we were measuring). Furthermore, no metric was well correlated to all of the behaviors associated with customer loyalty. In fact, regardless of the industry being looked at, different loyalty metrics worked better at explaining different customer behaviors. Specifically:

• The best predictor of share of spending was past share of spending, not recommend intention.
• The best predictor of retention was repurchase intention, not recommend intention.
• The best predictor of recommendation was recommend intention.

In other words, the foundation upon which Net Promoter is built — recommend intention — is clearly not the ultimate question for gauging loyalty.

These findings point to one of the inherent problems of using a single metric to gauge customer loyalty: The behaviors associated with customer loyalty are varied and distinct from one another. There are different motivators and market factors that influence customers’ willingness and ability to engage in loyalty behaviors, including inertia, delight and sunk costs. More importantly, the behaviors — increased spending, retention and word of mouth — contribute differently to a company’s ultimate success. Continuing to be a customer is not the same thing as increasing (or decreasing) the amount of money someone spends on a company’ products. And both of these behaviors are different from recommending a company or a product to friends. To be sure, each of these behaviors is associated with customer loyalty, but all loyal customers do not necessarily manifest all of these behaviors. In fact, they typically do not. We found that customer behavior models containing multiple variables outperformed models based on responses to Reichheld’s single survey question consistently and significantly.

Recommend intention, the criteria upon which Net Promoter is based, is presented as the “ultimate question” and the single best predictor across customers’ future loyalty behaviors. Our research did not corroborate this.13

Net Promoter and Growth Although we did not find a strong link between NPS and customers’ loyalty-based behaviors, we are cognizant of the fact that most managers are not drawn to NPS for its ability to predict customer behavior. Rather, they are impressed by its claims about a linkage to growth. To investigate this relationship, we examined data from more than 15,000 consumers and 21 companies from five industries over multiple years.14 We appended revenue growth rates to the data files of each company under investigation. We then looked at the correlation between company growth and several common satisfaction and loyalty metrics, including Net Promoter, using the same methodology reported by Reichheld.15

Surprisingly, none of the metrics were good predictors of revenue growth in any of the industries we analyzed. Virtually none were significantly positive over time, meaning that using the methodology Reichheld used, no metric could be considered a good predictor. Even when ignoring statistical significance (the likelihood that the correlations occurred by chance), Net Promoter was the best predictor in only two out of 19 cases.

Without access to the raw data used in the original analysis, we can only speculate as to the differences between our results and those reported by Reichheld. However, we did have an opportunity to compare Net Promoter to one specific metric that Reichheld reported he examined and found not to have a link to growth: the American Customer Satisfaction Index. Reported quarterly since 1994, the ACSI is based on U.S. data from more than 200 companies in 43 industries. In his original article, in a subsequent book and in industry presentations, Reichheld dismisses the ACSI for lacking a connection to company growth.16

In his book, Reichheld presents charts for six industries to demonstrate the relationship between Net Promoter and company growth (four industries in the United States, one in the United Kingdom and one in Korea).17 Three of the U.S. industries (airlines, life insurance and computers) are also tracked by the ACSI. To present a fair comparison between the ACSI and Net Promoter, we replicated the data. Data showing the relationship between Net Promoter scores and growth were reconstructed based on their scatter plots. We compared Reichheld’s data by substituting the average ACSI scores for the Net Promoter scores for the same time period.

The most obvious initial finding was the strong similarity between the NPS graphs and the ACSI graphs. (See “Comparing Net Promoter to Other Metrics.”) Given Reichheld’s assertions that ACSI showed no correlation to growth, sharper differences had been expected. Moreover, in two of the three cases, the ACSI-growth relationship as measured by the R-square came out higher than that for Net Promoter-growth.18 In the third case, the R-square for Net Promoter was higher although, as noted earlier, the charts were very similar.

We are not suggesting, however, that Net Promoter and the ACSI are effectively equivalent. The industries Reichheld selected for the comparisons were meant to demonstrate the strength of the NPS-growth relationship, so the Net Promoter data was positively biased in favor of NPS. One thing is certain: There are numerous peer-reviewed scientific studies showing the relationship between the ACSI and company financial performance.19 To date, there are none for Net Promoter.

Finding a Path Forward

What are managers hoping to gain from customer metrics? At a fundamental level, they want a realistic and reliable way to monitor and improve performance that is based on solid data. However, according to our research, many of the basic assumptions about NPS are flawed. As a result, managers who are guided by NPS may develop unrealistic views about performance, value and shareholder wealth, leading them to misallocate resources.

Although Net Promoter may be simple to implement and easy to communicate, and may help remind companies of the importance of customer loyalty, the claims about the research supporting the metric are focused on growth. Specifically, Net Promoter has been touted as the “single most reliable indicator of a company’s ability to grow.” Based on our research, it is difficult to imagine a scenario in which Net Promoter could be called the superior metric.

Reichheld has done researchers and managers a service by championing the importance of customer loyalty. Moreover, Net Promoter highlights the need for metrics that are easy to understand and advance the fundamental goal of managers: to build successful organizations. The ability of NPS to gain a dedicated following in a short period of time underscores how eager managers are for tools that can help them enhance customer loyalty.

We are skeptical that there is a “silver bullet” that definitively links customer loyalty to company growth — or that one will be forthcoming. There are myriad dimensions affecting customer loyalty and how it impacts customer behavior and profitability. When simple solutions are sought to such complex problems, the answers are often technically correct in a narrow sense — but substantially wrong. Net Promoter, like any measure of customer intentions, is inherently unreliable. Customer intentions may point in the right direction for some behaviors, but they will never provide all of the information needed to understand the complete picture.

So what is a manager to do? How do you manage something that has many different dimensions, and how do you encourage people in your organization to act on information that is complicated and often contradictory? First, managers need to be realistic. They must accept the fact that metrics are tools that can assist them in making decisions, but metrics don’t make the decisions — they are only guidelines. A metric can’t reduce complex, multifaceted constructs to one or two dimensions — and if it does, there’s a good chance it will ignore one or more important aspects of the equation.

Second, managers must be willing to do whatever level of analysis is required to understand their customers and their particular market opportunities. Just as there are significant risks in doing too little (for example, pinning your hopes on one “number” that promises too much), there are risks in doing too much. Managers need to balance their desire for simplicity with their need for accuracy. This is one of the toughest challenges managers face. If it were easy, the debate over which metrics pack the most punch would be a lot less spirited.