A Rebuttal: "Loyalty Really Isn't All That Simple, Restrictive"
William D. NealMarketing News
14 August 2000
To the Editor of Marketing News.
I would like to thank Randall Brandt for responding to my 5 June Marketing News article on the measurement of customer loyalty, "When Measuring Loyalty Satisfactorily, Don't Measure Customer Satisfaction." When I wrote the article, I was not aware that Burke Customer Satisfaction Associates claimed authorship of those famous three questions, but I am pleased they have chosen to defend their usage.
In his letter to the editor, Dr. Brandt disagrees with most all of my positions and refers the reader to a more detailed paper available on Burke's website. The printed letter was a bit sketchy, so I have chosen to respond to Dr. Brandt's more detailed article.
My position is that I believe that the three survey questions so often used to measure customer loyalty – overall satisfaction, intent to repurchase/renew, and intent to recommend – are really measuring the same construct - overall satisfaction. As most of us now know, customer satisfaction is a necessary but not sufficient condition of customer loyalty.
Furthermore, I attempt to make the case that the more generalized and encompassing definition of loyalty is as a behavior. It is the concrete, bottom line metric that has a direct impact on corporate profitability and marketing efficiency.
Finally, I make the case that behavioral loyalty is driven by perceived value (as Dr. Brandt correctly modified). Value can be defined as the perceived benefits received from the brand less the cost to get those benefits. Benefits have two components – the set of tangible benefits delivered by the product or service itself and the set of intangible benefits delivered by the brand name – it's brand equity.
In his paper Dr. Brandt states that the definition of loyalty as a behavior has two critical flaws – it ignores the attitudinal component of loyalty and it overlooks the distinction between lead measures and lag measures. I disagree.
The behavioral definition of loyalty does indeed encompass the attitudinal component, but it also recognizes that there are many non-attitudinal components that can have a greater effect on loyalty and, thus, repurchase. Customers are behaviorally loyal for many reasons, attitudinal loyalty being one of them. But also, customers may be behaviorally loyal because of convenience, accessibility, lack of choice, high search and evaluation costs, high switching costs, and a myriad of other reasons. Likewise, an attitudinally loyal customer may not be a behaviorally loyal customer for many of the same reasons. Again, the bottom line issue is behavioral loyalty – how many customers repurchased.
The motivation for repurchase, or the lack of repurchase, must consider all elements of the behavior, not just the attitudinal component. To consider only the attitudinal element can be very misleading and may point the marketer to make the wrong investments. Indeed, our research in competitive categories clearly demonstrates that of those customers who show high behavioral loyalty, very small percentages are attitudinally loyal. Most are either optimizing the tangible benefit set or are declining to engage in repeated search and evaluation activities. We believe that it is extremely important for marketers to know all the elements that are driving repeat purchase as well as those that are barring repeat purchase. Concentrating on just the attitudinal components provides only part of the answer.
Dr. Brandt then goes on to cite the Jones and Sasser article, "Why Satisfied Customer Defect" published in the Harvard Business Review (November/December 1995) concerning lack of choice issues. This paper also influenced me and it helped convince me that there was more to loyalty than just attitude. Indeed, rightly or wrongly, many firms still use high switching costs, monopolistic activities, and other switching barriers to assure a higher level of repeat purchase and behavioral loyalty – just as Jones and Sasser discovered. Regardless, Frederick Reichheld gave us a more expansive and behaviorally oriented understanding of customer loyalty in his book, The Loyalty Effect (1996). In my opinion, Reichheld's book provides the more appropriate and supportable view of customer loyalty.
As far as the lead verses lag indicators, I assume that Dr. Brandt is implying that the three survey questions are a lead indicator and that actual behavioral measurement is a lag measure. He is right and I agree that both lead and lag (e.g., validation) indicators are necessary for good marketing management. However, well-designed and well-executed trade-off experiments are a far better predictor of future behavior and represent the superior lead indicator of customer loyalty, especially when compared to stated intent survey questions. I made that point in my original article. We regularly use trade-off experiments to predict preference and actual choice. The behavioral validation of these techniques is overwhelming compared to stated intent survey questions.
Furthermore, trade-off experiments provide direct, in-depth diagnostics for improving loyalty and repeat purchase. Trade-off experiments allow the marketer to more clearly understand the drivers of choice and repeat purchase at the individual level and to segment their customers based on preference structures. Interestingly, individual preference structures derived from trade-off models readily identify respondents who are attitudinally loyal – they are those respondents who have a very high utility for a single brand. Furthermore, these trade-off models also readily identify customers who are loyal to a brand set – something the three-question survey protocol cannot easily uncover.
Well-designed and well-executed trade-off experiments can also uncover the non-attitudinal (i.e., behavioral) elements that drive loyalty (or the lack thereof), such as location convenience, switching costs, search and evaluation costs, and so forth. In other words, these models allow the marketer to understand the broader context of customer loyalty and integrate attitudinal, behavioral, and environmental elements in the same protocol.
Now, to the key issue of intercorrelation and redundancy between the three measures – overall satisfaction, intent to repurchase, and intent to recommend. Dr. Brandt argues the high correlation between the measures, which he admits, does not translate into high redundancy.
His analytical procedure, as I understand it, is to measure each response on (usually) a five-point scale. If the respondent gave the highest value (top box) on all three measures, that respondent is designated a "secure" customer. Anything less, relegates the customer to "satisfied," "vulnerable," or "dissatisfied" in accordance with some unrevealed algorithm. I assume a "secure" customer is equivalent to a "loyal" customer. Dr. Brandt then goes on to cite two studies where his classification discriminated between retention and sales growth.
I contend that the three measures are redundant measures of satisfaction and are no different than asking, "how satisfied are you," "how REALLY satisfied are you," and "how REALLY, REALLY satisfied are you." Said differently, Dr. Brandt is asking us to believe that a customer will treat all three questions as independent measures of how they feel about the producer's performance. The fact that they may be totally satisfied with the producer's performance has little to do with whether they would re-purchase the brand or recommend it to others. I am very skeptical of that position and I believe that the three questions are simply measuring different points on the same continuum.
I have seen the correlations between these three measures range from 0.75 to as high as 0.95. Assuming there is no measurement error or response error (which is tough to assume) that leaves very little independent variation from which to draw conclusions about the differences in response patterns. Indeed, if we perform a reliability test on the three measures, the change in Cronbach's Alpha statistic is so small that anyone one of the questions will adequately represent the construct or scale of overall satisfaction.
I believe the effect of the Burke procedure, as described by Dr. Brandt, is that the number of secure (loyal) customers is considerably overstated mainly because their model narrows the focus to three redundant measures of overall satisfaction.
In a recent study by SDR Consulting, we used a trade-off model to derive customers' preference structures and validated those with a choice experiment. We also asked the three satisfaction questions advocated by Dr. Brandt, using a 0-10 point scale. The results were strikingly different. The SDR model revealed that 4% of customers were brand loyal. That is, their utility for brand name of the client would offset reasonable competitor advantages in price and performance levels.
Dr. Brandt's methodology showed that 35% of customers were brand loyal (i.e. secure) – they gave all three measures a top 2-box score on a 0-10 point scale. More Precisely, 49.6% were top 2-box for overall satisfaction, 44.6% for repurchase, and 46.9% for recommend. Said differently, of the 475 (out of 798) who gave a top 2-box score for any of the three items, 58.3% gave a top 2-box score for all of them. That sure does not look like a set of independent measures to me.
To verify that, the data were subjected to a principal components analysis. Only one significant factor could be found. The three questions loaded on that principal component at the following levels: 0.86, 0.89, 0.92. That's about as redundant as you can get in survey data among a large sample of customers.
Yet, when it is all said and done, Burke, SDR, and all the others exploring these issues are asking the same question: how do I measure and manage customer loyalty? Where we differ is in our philosophical approach. I'll stick to my guns – I believe behavioral loyalty is the bottom line issue and attitudinal loyalty is only a part of the whole picture. Furthermore, I would ask you, the reader, to more deeply explore why your loyal customers defect. Some of it may be due to shifts in attitude, but I'll lay odds that the majority switch because of improved competitive offerings, changes in needs sought, or changes in the marketplace environment.
I'm pleased that Dr. Brandt and I will have the opportunity to debate this issue further at the AMA's upcoming Attitude and Behavior Conference in January.