For What It’s Worth
Interest in, and concern about, the measurement of customer satisfaction has continued to rise. Whether the push is driven by internal or external champions, top management is being exhorted to establish a process for quantitatively determining the level of satisfaction among the firm’s customers. Marketing research firms have even created full divisions to exploit the demand.
However, once a system is set up, marketing executives often get caught up in chasing the metric. They launch a seemingly endless series of programs to improve the scores on customer satisfaction. Assets are committed, prices changed, communications adjusted, changes in scores are duly noted, and rewards or sanctions are distributed. By the sheer act of measuring and reporting customer satisfaction, its importance is elevated — in some cases to the level of a strategic outcome. A recent national survey of executives found that “information about customer satisfaction is highly valued by the largest percentage of executives, even more than the traditional management gauges of financial performance and operating efficiency.”
With the growing amount of time and resources that have been going into the measurement of customer satisfaction, it is not surprising that questions are being raised about whether these commitments are truly worthwhile. What is the financial value of a one-point improvement in a satisfaction score? Some studies have been able to show evidence of benefits to business operating results that can accrue from a high level of customer satisfaction. Yet, in an article published just last year, it was noted that less than 30% of managers reported being confident that their firms’ customer satisfaction levels were showing economic value. Only 2% claimed any ability to measure the financial impacts that resulted from any customer satisfaction initiatives.
The Roots of the Problem
This weak record in linking satisfaction to financial performance can be at least partially explained by the fuzziness of the customer satisfaction construct. Each company tends to have its own definition, using different inputs and procedures to operationalize the term. Thus, a finding of no correlation between a company’s satisfaction scores and its profitability may simply be due to poor satisfaction measures being used. The development of the American Customer Satisfaction Index may help the efforts to link satisfaction to financial performance, since there is now one standardized measure available for benchmarking.
Actually, one could argue that true customer-satisfaction tracking demands more, not fewer, customized measures, since each customer may consider different inputs and/or weight them differently. However, any implementation of customized measures could actually confound the efforts to link satisfaction to performance measures across multiple companies.
A second complication stems from the fact that the economic benefits of customer satisfaction, as one author put it, “are still lost inside the shades of traditional accounting” since the benefits are not necessarily turned into profit within the immediate accounting period Any improvement in measured satisfaction could be seen as having three types of impacts: a) improved feelings among customers; b) immediate financial impacts such as greater sales volume; c) longer-term financial impacts such as increasing sales in years to come. To the degree that the last category is significant, the correlation between current satisfaction and profit-type measures will be distorted.
Finally, satisfaction computations most often give equal weight to all customers, regardless of their relative profitability to the firm. Hence, in an effort to raise satisfaction, the firm may take actions that could end up worsening profitability. For example, those patrons who only buy a few items in a store want the checkout lines to be short. Adding dedicated checkout stands for these customers will likely bring about an increase in their satisfaction. However, if this means fewer checkout stands for other customers, it may very well drive away those with large profitable baskets, thus generating a “double-whammy” of increasing costs to serve while lowering average check size.
A Not-So-Modest Proposal
As Peter Drucker is often quoted as saying, “The only profit center is the customer.” In this spirit, I believe that we can treat the pursuit of customer satisfaction as we do any other profit-driven investment — that is, assess it in terms of its net present value (NPV) and/or return on assets employed (ROA). To get to this ability requires a few adjustments in our thinking. First let’s return for a moment to why firms pursue customer satisfaction.
The pursuit of customer satisfaction is based on the belief that satisfied customers are “worth more” to the firm. That is, they are widely expected to:
- be retained longer and in greater numbers
- buy more goods
- cost less to serve
- be willing to pay slightly higher prices
- respond faster to promotional efforts
- refer others, thus helping reduce the cost of acquiring new customers
- suggest and evaluate new products and revenue streams
In fact, for some firms, measures of the above activities are directly included in the satisfaction index.
For firms to be able to look at customer satisfaction in NPV or ROA terms, the key necessary tool would seem to be customer lifetime value (CLV). Long used by the direct mail industry among others, this indicator directly assesses the financial value of each individual customer. Under the CLV models, a customer represents a stream of future revenues that depend on the time frame that he/she is retained and the dollar rates of purchase per period. For each future period, these revenues are reduced by the total costs of acquiring, retaining, and fulfilling the customer. To this is added the additional cash flows that come from such derived sources as secondary purchases and referrals of other potential customers. In many cases, rather than doing a customer-specific calculation, average CLV’s are computed for customer segments.
Once we have calculated the average CLVi, or net present value of each customer, then the total NPV of the firm at any time would be
where CLVs* is the average CLV across all customers in a segment and Ns is the total number of customers in that segment.
So, rather than continuing to define customer satisfaction as an index of multiple inputs, what is proposed here is that we split the inputs into two streams. The first stream, containing opinions on specific quality and service levels, will provide input for changes in marketing activities – product/service features, ad themes, etc. The second stream should directly provide all the inputs to the CLV calculations. Examples of the types of questions that provide direct connection to the NPV are:
- What are the chances that you will purchase again in the next “x” months?”
- How much will you likely buy in the next “x” months?
- Of your category purchases in the next “x” months, what percent will be from us as your provider?
- If our brand were not available, what would you do?
- How likely would you be to refer an associate to our firm?
There are several other adjustments that follow the refocusing from a customer satisfaction index to the CLV approach. Most customer satisfaction programs are content to take measures on an infrequent basis. However, the CLV, like any measure of future intentions, is temporal. Customers’ future purchases, their probability of making a successful referral, and the other components of CLV will change with their experiences and state of mind. Consequently, CLV should be considered as a variable that needs to be tracked, constantly fluctuating in value as opinions and intentions change.
Then too, we will be asking customers about their intentions. That means there is still the need to connect reported intentions with ultimate behaviors, and to track the objective measures of customer experience such as response and service times, completion and error rates, etc. The inputs to our CLV calculations should not be the raw reported intentions, but adjusted values that account for customer tendencies to misstate their realized rates. For example:
Expected Value of the retention rate = a function of (reported intent to repeat, historic repeat rate for different intentions levels, the firm’s achieved performance on objective satisfier dimensions).
The focus on consumer intentions will also have to be expanded to include potential customers as well as active ones. While satisfaction among active patrons has some impact on the number of customers that will be acquired, many will arrive at our doors without a referral. For a given desired target group, sampling will have to be done and measures established that will help forecast these numbers of new customers. Thus, the firm will need to include questions such as:
- “Have you heard of our firm/brand/product?”
- “What are the chances that you will visit our store in the next week?”
Once we have set up the procedures for sampling the inputs, we can then connect any proposed change in the marketing variables under control of the firm to the corresponding impacts they are expected to have on CLV factors. For example, a considered price change should impact rates of current and future purchase, retention and referral rates, as well as the rate of acquisition of new customers. Knowing the resources required to implement the change, we can then calculate the new NPV as well as an associated ROA.
Summary and Conclusions
The process proposed here may seem like a tall order, rife with the trepidations inherent with using subjectively estimated inputs. However, we have developed some degree of comfort in doing this type of analysis for the decisions we make about property, plant, and equipment investments. Our efforts to more fully understand the ultimate source of all revenue, our customers, deserves no less. Moreover, unless we adopt this kind of thinking, the previously identified limitations of current customer-satisfaction-measurement procedures virtually preclude any chance that these measures will be able to be linked to strategic financial performance on any widespread basis.
A word of caution is in order here. A Sears team which succeeded in building a system similar to that described here noted that while any retailer could copy the Sears measures, this company might still fail to achieve any demonstrated financial results “…because the mechanics of the system are not in themselves enough to make it work.” They go on to remind us that management must be fully aligned around the modeling effort, and that there must be a deep sense of ownership among the employees who must implement it.
If we are to make progress in tying customer satisfaction to strategic performance, firms must commit to the following:
- Formulate a customer-centric revenue model – recasting the P&L into customer-specific categories
- Restate customer satisfaction to include all Customer Lifetime Value components
- Connect marketing actions to each facet of the Customer Lifetime Value
- Consider shifting the organization from SBU’s (Strategic Business Units) to SCu’s, that is, Strategic Customer Units.
 J. H. Lingle & W. A. Schiemann ,”From Balanced Scorecard to Strategic Gauges:Iis Measurement Worth It?” Management Review, 85 (1996): 56-62.
 E. W. Anderson, C. Fornell, & D. R. Lehmann, “Customer Satisfaction, Market Share, and Profitability: Findings from Sweden, Journal of Marketing, 58, (1994): 53 – 66; C. Fornell, M. D. Johnson, E. W. Anderson, J. Cha, & Br. Everitt-Bryant. “The American Customer Satisfaction Index: Nature, Purpose, and Findings,” Journal of Marketing, 60 (1996): 7 – 18; R. T. Rust, A. J. Zahorik, & T. L. Keiningham, “Return on Quality (ROQ: Making Service Quality Financially Accountable,” Journal of Marketing, 59 (1995): 70 – 88; A. Westlund & C. Fornell, “Customer Satisfaction Measurements and Its Relationship to Productivity Analysis,” Proceedings of the 8th World Productivity Congress Stockholm (1993).
 M. M. Andre & P. M. Saraiva, “Approches of Portuguese Companies for Relating Customer Satisfaction with Business Results, Total Quality Management, 11 (2000): 929-941.
 Fornell, Johnson, Anderson, Cha & Everitt-Bryant, 1996.
 Lingle & Schiemann, 1996.
 Rust, Zahorik & Keiningham, 1995.