Satisfaction be Damned, Value Drives Loyalty
William D. NealPresentation to the ARF Week of Workshops
For over a decade American business has been focused on customer satisfaction as a way to become customer-focused and improve customer loyalty and thus profitability. The assumption has been that the more satisfied a customer is, the more loyal they will be. Loyal customers consume fewer marketing and sales resources, buy more, and buy more often from the organization that has gained the customer's loyalty.
Well, that's not quite right.
Customer satisfaction measurement and tracking are fine methods for tracking process and product performance and for providing a quantitative feedback loop for process improvement. However, customer satisfaction has to little to do with customer loyalty.
In a November 1995 Harvard Business Review article Thomas Jones and Earl Sasser explained, at least partly, "Why Satisfied Customers Defect." A year later Fred Richheld published "The Loyalty Effect" where he demonstrated that satisfaction was not nearly enough to insure loyalty. Then Brad Gayle in Managing Customer Value told us that we need to transition from customer satisfaction to customer value and customer loyalty to improve customer retention. Unfortunately, these seminal works were long on theory and antecedents and short on measurement and methods.
Regardless, we researchers and managers of customer relationships have generally failed to deliver large numbers of loyal customers to our clients by using the typical customer satisfaction measurement systems in place today.
Customer loyalty is the proportion of times a purchaser chooses the same product or service in a specific category compared to the total number of purchases made by the purchaser in that category, under the condition that other acceptable products or services are conveniently available in that category. Customer loyalty is a behavior. It is measured as a proportion.
Customer satisfaction is the attitude resulting from what customers think should happen (expectations) interacting with what customers think did happen (performance perceptions.) Customer satisfaction is an attitude and it is typically measured using some sort of attitudinal scale.
We have been attempting to use satisfaction, an attitude, to predict customer loyalty, a behavior. That has not worked well in the past and it probably won't work any better in the future.
We know that satisfied customers, even highly satisfied ones, often switch brands and suppliers. The observed relationship between stated satisfaction and repeat purchase is very weak to non-existent. The degree of satisfaction, once past a particular threshhold, cannot reliably predict repeat purchase.
However, the relationship between stated dissatisfaction and defection is very strong. This suggests that "satisfaction" is not simply a continuum from very dissatisfied to very satisfied, but rather two different constructs.
The degree of dissatisfaction with a product or service can be used to accurately predict whether that product or service remains within a purchaser's consideration set. And or course, once you're out of the consideration set, you're out of the game. Your product or service cannot be chosen for purchase if it's not considered.
Maintaining a minimum acceptable degree of satisfaction only keeps the product or service in the purchaser's consideration set. The degree of satisfaction, once it has passed an acceptance threshold, does not predict which of a set of competing items will be chosen at the next purchase opportunity.
Said more bluntly, increasing levels of satisfaction beyond the acceptance level does not result in a proportionate increase in share of choice.
Furthermore, purchaser answers to our typical behavioral intent questions cannot reliably predict future purchase intent at the individual level. Again, you can predict consideration set (or the lack thereof) to some extent with self-explicated intent questions, but you cannot accurately predict actual purchase at the individual level.
So, what does predict customer choice and loyalty?
The simple and accurate answer is VALUE.
Value predicts choice, thus loyalty. Buyers who are considering a purchase in a particular product or service category scan their product/service options and develop a consideration set. Within the consideration set, they develop a hierarchy of products based on their assessment of value. They then choose the product at the top of their value hierarchy, if available. This may be a conscious, cognitive process or a subconscious process with some emotional elements.
If one can accurately measure a purchaser's relative value structure for a product or service category, then one can accurately predict that purchaser's choice among a set of competing products/services or brands in that category.
Value can be measured. There are several competing models for doing so. Srinivasan, Kamakura and Russell, while at Vanderbilt University, hypothesized the model I like best. Basically this brand value model has three elements - price, the bundle of tangible deliverables (product/service attributes) and the bundle of intangible attributes (imagery drivers), collectively called the brand's equity as shown in Figure 1.
Thus, we can view the three key elements of value as price, product/service deliverables, and brand equity. Each element and sub-element of this model can be viewed as having a weight. The set of weights for an individual purchaser is their preference structure, or more accurately, their value structure. See Figure 2.
Each purchaser has a unique value equation for each product or service category in which they have some experience. The value equation provides the buyer with a preference structure for making a choice among a set of competing products or services. Rational buyers choose the best value. Thus, value drives choice. If we know a purchaser's value equation we can very accurately predict their choice among a set of competing products/services in a category.
There are several methods for deriving respondent weights in this model. Srinivasan and Park illustrate one method in a May 1994 JMR article. We prefer an alternative measurement system that combines conjoint and choice experiments to derive the weights in a four-step interviewing process. Although this presentation does not allow me enough time to go into detail, the four steps are as follow:
Step 1 is a brand-price conjoint exercise of all relevant brands in the category. This is done to get an accurate estimate of the relative utility of price.
Step 2 is a brand-price-attributes conjoint exercise, which includes all of the relevant tangible attributes. During analysis, price from step one is merged into the results from step 2 and all utilities are re-scaled.
Step 3 is a discrete choice experiment whose design is driven from the results of step 2 and is individualized for each respondent. We use this task to confirm the conjoint results and to estimate a switching barrier. The switching barrier utility is integrated into the value structure of each respondent.
Step 4 is measurement of a set of relevant image drivers on a 0-10 scale for each brand. These ratings are then ridge-regressed against the derived brand equity from step 2. The beta coefficients from the regression, or a similar model, reveal the key drivers that are most associated with each brand's derived brand equity.
Once this value structure is derived for each respondent, it is easy to predict shares of choice, given each product's brand name, price and tangible attributes, using a standard conjoint-type simulation procedure. Furthermore, rich diagnostics are available to show how to improve product value, thus increasing share of choice.
Recognizing that within a particular product or service category, purchaser value structures may vary widely, it is highly recommended that share of choice and choice improvement diagnostics be reported by benefit segment (in which value structures are more homogeneous.)
Well-constructed value models accurately predict choice. Value models can be a key tool for developing a deeper understanding of customers, how they make choices, and what will keep them loyal.
In summary, the key to greatly improving loyalty – the number of repeat purchasers - is to get into purchasers' consideration set, then making sure you have the best relative value among the products in that consideration set for the largest number of purchasers.
American businesses are still spending hundreds of millions of dollars annually to measure and manage customer satisfaction, believing it will improve loyalty and profitability. Incremental improvements in customer satisfaction may improve consideration, but there is overwhelming evidence that it does not improve loyalty. Value does that.
If you are measuring and reporting satisfaction, but not value, I think you may be in the right woods but barking at the wrong tree.