Customer Loyalty Surprises and Myths

A Loyalty Builders White Paper

You could have knocked us over with a feather.

There’s nothing new about the desire to keep your customers from defecting to competitors. But a lot more emphasis and new thinking have been devoted to the subject in the past few years. Loyalty Builders is part of that intense new focus.

When Loyalty Builders was founded in 1999, we believed, with others, that loyalty was an important component of business success. As we strove for a more granular look at loyalty, we were surprised again and again by what our research uncovered. These surprises were profound and shaped our thinking and our business strategy. They have and continue to make our technology stronger.

The first surprise came as we tried to measure customer loyalty. We discovered that business people were using “loyalty” and “satisfaction” interchangeably, making no distinction between the two. What companies had been doing was simply measuring customer satisfaction and assuming that satisfied customers were also loyal.

Satisfaction is fairly easy to measure, which perhaps is why companies were doing so much of it. You ask the customer how much they like (fill in your product name), whether they would recommend you to a colleague or friend and whether they would buy from you again. Add the scores, give them a customer satisfaction rating, and presto, you’ve measured customer satisfaction.

What we realized early on is that satisfaction and loyalty are very different beasts. Satisfaction is an attitude reflecting past experience, typically recent past transaction; you ask people what they think of the transaction and score them on their responses. What we saw as we analyzed our clients’ customer data was that customers can be completely satisfied, but given even a small incentive will still jump to a competitor. Think of the thousands who change their telephone supplier when they get that dinnertime telemarketing phone call offering them two silk purses and a month of free calls for switching.

Loyalty on the other hand is less affected by those small incentives and is thus a more reliable indicator of future behavior. Loyalty is a mental construct, like beauty or intelligence. But how to measure a construct is far from obvious. People have fought for ages over how to measure beauty. Measuring intelligence has provoked its own share of controversy. In our quest for a reliable way to measure loyalty, we encountered some surprising things.

Surprise #1. CONSTRUCTS CAN BE MEASURED

We know loyalty when we see it. It is not an opinion. It is a commitment that is real. Percy Bridgman, the well-known philosopher of science, believes that if something is real, it has a quantity. If it has a quantity, it can be measured.

Bridgeman says. “Things are defined by the 'operations' (methods) you use to measure them. They are called 'operational' definitions. For example, 'intelligence' is defined in terms of the scores on tests you use to measure a person's IQ, 'temperature' is defined in terms of the readings you get from the instrument you use to measure heat, etc.”

In a commercial environment, we define loyalty as a customer’s intent to purchase again. If we can measure future purchase intentions then we have an opportunity to test the value of all we do to promote our business. Working on the tested assumption that past behavior is a reliable predictor of future behavior, Loyalty Builders constructed a model that uses past transactions to score a customer’s loyalty.

Surprise #2. CUSTOMER BEHAVIOR DOESN'T FOLLOW A "NORMAL," BELL-SHAPED CURVE

Our loyalty model produced a further surprise when we began scoring customer populations. We expected to see a spectrum of loyalty scores resembling a bell-shaped curve; a so-called “normal distribution.” That is, we expected to see some small number of customers with low scores, a similarly sized small number with high loyalty scores, and a large number of more or less average customers in the fat middle of the curve.

What we discovered instead was a relatively small group of customers with high loyalty as expected, but an unexpectedly large number of low-loyalty customers at the bottom end of the curve. What we discovered in between were groups of customers clustered around several intermediate values on the loyalty score spectrum. The bell-shaped curve we’d expected was nowhere in sight.

We suspected that these clustered customers would have similar buying characteristics, and that proved to be the case. These clusters graphically showed us that while it's easy to identify a company’s best and worst customers it’s far more difficult to distinguish variations among the large numbers of customers in between. Our mathematics actually revealed interesting new behavior patterns and suggested ways a company can get more revenue from customers in this middle area.

Surprise #3. NO SINGLE METRIC IS A RELIABLE MEASURE OF A CUSTOMER’S LOYALTY

When Loyalty Builders went looking for ways to differentiate customers according to their loyalty, we began with what we now call the “traditional” metrics: how long has it been since their last purchase, how often have they purchased, how much the customer spent? These are usually called R, F, and M, for Recency, Frequency, and Monetary value. RFM is still a widely used method which scores customers on the basis of these traditional metrics. (See the Loyalty Builders white paper comparing RFM to our techniques.)

In spite of the fact it was in general use, we realized that RFM has severe limitations and really doesn't work very well. Just because a metric (for example frequency) is easy to label and easy to measure doesn't make it one that adequately differentiates customers on a loyalty scale. Loyalty is a complex construct that doesn't follow a simplistic equation. In scientific language, this is a non-linear variant of the eigenvalues problem, where the task is to find the variables that do separate the customers cleanly into disparate groups. It took Loyalty Builders over a year to solve this problem. We found that common linear mathematical processes like regression and RFM do not produce accurate results. Now our regular deliveries to clients include our own proprietary metrics that produce more reliable, descriptive and useful analyses of customer populations.

Surprise #4. SEGMENTATION BASED ON ATTITUDES DOES NOT REFLECT OR PREDICT CUSTOMER BEHAVIOR

Once we began sending our clients account-level data with loyalty scores for each customer, we also began to segment customers into Loyalty Groups according to their scores. Some clients had been segmenting customers according to demographics or satisfaction scores. We were asked to compare segmentations based on satisfaction (an attitudinal measure) with actual customer behavior.

In these comparisons we could find no positive correlation between customer satisfaction scores and their actual future purchases; satisfaction, it turned out, was not a predictor of loyalty.

Upon investigation and reflection, we should not have been surprised. Researchers have known for a long time that survey respondents tend to say one thing and do another. Responses on satisfaction surveys don't correlate with actual purchase behavior. We call this The Myth of Customer Satisfaction and published an article about it in Strategy+Business, a publication of Booz Allen Hamilton.

SURPRISE #5: LOYALTY SCORES ARE A MANAGEMENT TOOL

Segmentation based on quantitative loyalty measurements reveals startling and previously unrecognized patterns of customer behavior, as well as uniquely useful information about the company as whole.

Looking at collections of customers aggregated by their Loyalty Groups yields new information about the company at the "50,000 feet" level. Loyalty segmentation reduces the noise level in the data, revealing differences in behavior between customers in different loyalty groups. The new information shows up when we cross-tab typical business variables by Loyalty Group and one of our proprietary metrics. And, regardless of the size of the customer set, loyalty analyses let the viewer drill down and get account level information after the big picture is in place.

There are undoubtedly more surprises for us out there, waiting to show up. As more and more companies switch their focus from simply selling products and services to retaining the customers they serve, we will all discover new patterns of customer behavior. Loyalty Builders is very interested in hearing about what you've learned, and what surprises you've encountered. Please contact us and tell us about your experiences.