Guest article by G. Huang and K. Sudhir, winners of the SERVSIG Best Service Article Award.

We are honored to receive the SERVSIG’s 2022 Best Service Paper Award for our 2021 paper in Management Science, “The Causal Effect of Satisfaction on Loyalty.” This award is particularly gratifying as the paper’s research question and analysis was driven by a real-world problem at American Express. The paper did not tradeoff rigor and relevance, but the RIGOR WAS CRITICAL FOR ITS RELEVANCE!

Standard ways of measuring the value of effective customer service systematically underestimate the power of a positive interaction between customer and representative.

The paper started as a managerial challenge from Vikram Nangia, then EVP of Customer Relationships at Amex (currently SVP at Comcast) to us at the Yale Center for Customer Insights (YCCI). He said: “Help our customer service organization gain respect with senior leadership at Amex…we don’t want to be treated as a cost-center, whose costs need to be minimized.” 
He even framed the research question: “Can you show RIGOROUSLY with the customer satisfaction survey data that we already collect (and spend millions on) that the service organization adds value to the brand?”

What was remarkable about the question was that senior management at Amex intuitively understood that “correlation is not causation.” They understood that correlation between customer satisfaction and positive “recommendations to friend” in the survey data could simply be an artifact of overall satisfaction with the American Express brand. Senior management intuitively understood “common methods bias” in surveys. That is what they wanted us to avoid with rigorous analysis.

Two main insights

After much thinking, we arrived at our first methodological insight: In many service settings, we can take advantage of the natural randomness in which service agent provides service to estimate the causal effect of satisfaction on loyalty. The idea is general, and applies in many service settings: hotels, restaurants, fast food chains, call centers, banks etc.

The second managerially relevant insight is that most customer satisfaction research in companies and academia, underestimate the value of service satisfaction on customer loyalty and lifetime value. This is because for the same level of service quality, customers can differ in their satisfaction rating. This is a form of measurement error and leads to the underestimation. With our method, we show that the true effect on customer retention and lifetime value is double the naive estimate. 
This insight is important not just for Amex because the naïve regression analysis is common at most companies collecting such service satisfaction and “recommend to friend” scores. The underestimation will lead to underinvestment in customer service. We recommend that organizations redo their analysis using our easy to use method so that they can get a more accurate measure of the value of service satisfaction for their organizations. 

With the higher and more accurate estimates, firms may find it profitable to increase customer satisfaction by improving the skill level of its customer-service workforce. They may want to expand the use of elite or specialized representatives, who currently tend to be reserved for only high-value customers; these same teams could be used to tackle difficult customer calls. Companies could also create stronger incentives to improve customer service. 


Overall, we hope the paper not only offers a practical approach to evaluate and understand the benefits of investments in customer satisfaction, but also helps improve the customer experience at many firms.
The paper also suggests that many academic papers may have underestimated the effects of service satisfaction on customer loyalty due to this measurement error problem. We hope new papers will correct for this problem, and future meta analyses include a control for whether the measurement error issue has been addressed in papers.  

At Amex, our results led to not just more resources for customer service, but marketing began to use customer service for brand building and cross-selling. For us, it led to more projects and field experiments at Amex and a research agenda! We are currently writing multiple papers around how and why combining cross-service with cross-selling within the same service person greatly improves cross-selling efficiency. Interestingly, now due to cross-selling, customer service is no longer seen a cost-center, but seen as a revenue driver—the original managerial problem we sought to address!

We wanted to share this story with our academic community to highlight that managerially relevant questions from practice can lead to creative and rigorous solutions from academia!  We hope it assures our academic colleagues that our care and effort in rigorously measuring causal effects is valued by senior management, impacts strategic decision-making, and even leads to changes in organizational and job design.

Guofang Huang
Assistant Professor of Management
Krannert School of Management, Purdue University

K. Sudhir
Professor of Private Enterprise and Management, Professor of Marketing
Yale School of Management, Yale University

Picture credit: Karine Avetisyan.