Skip to Content

A Better Way to Measure the Impact of Customer Service Satisfaction on Loyalty

Thursday, September 30, 2021

Customer Service Representative

You just got off the phone with a customer service representative at a credit card company. The rep was friendly and helpful, making you feel good about your relationship with the company. A minute later, your phone rings. It's an automated recording asking if you'd like to participate in a survey about your customer service experience.

Such surveys are an increasingly common way for companies to measure customer service satisfaction. They want to find out if your customer service experience will help them retain you as a customer.

"They try to use all kinds of survey methods to measure the quality of service, but there are some inherent problems in measuring the impact of service quality on customer loyalty if you use the survey method," says Guofang Huang, assistant professor of marketing in Purdue University's Krannert School of Management.

The survey method typically introduces multiple sources of bias that distort the relationship between service satisfaction and loyalty.

Huang and his co-researcher, K. Sudhir of Yale School of Management, have proposed an instrumental-variable (IV) approach that addresses these sources of bias and more accurately estimates the effect of customer service satisfaction on customer loyalty.

The researchers used the IV approach to analyze survey data from a large credit card-issuing company, sharing their results in a paper entitled "The Causal Effect of Service Satisfaction on Customer Loyalty," forthcoming in Management Science. Their findings suggest that many companies may have "grossly underestimated" the return on investment in customer service.

This underestimation can be attributed to the multiple sources of bias that survey methods introduce, including common methods, attenuation and omitted variables.

Common methods bias may arise, for example, when two customers who have received identical service are asked to provide ratings on a scale of 1 to 10 for satisfaction and willingness to recommend (i.e. loyalty intent). One customer may provide more generous ratings than the other, despite receiving the same quality of service, which creates an inflated correlation between service quality and loyalty.

Attenuation bias is introduced because the survey does not provide a perfect measure of a customer's views. If a customer wants to give a rating of "8.5," the only choices are 8 and 9, and the customer is forced to pick a rating that's not perfectly accurate.

Common methods bias tends to cause overestimation of the relationship between service satisfaction and loyalty, while attenuation bias tends to cause underestimation of this relationship.

If only one of these biases exists, the estimate may still be informative, Huang says, but when they're both present, the picture is muddled considerably.

"Once you have these two sources of biases, then companies don't really know where they are," he says.

Another source of bias, omitted variables, occurs because the surveys don't measure every factor that might affect a customer's loyalty. Even if a credit card company's customer service is poor, a customer may remain loyal because, for example, the company offers a very low APR (annual percentage rate).

"If you combine all these sources of bias together, you really need a technique to help you get rid of them," Huang says.

The IV approach does this by taking advantage of randomness – the random way in which service employees are assigned to customers. When a service rep has finished helping one customer, the rep helps the next customer waiting in the queue, and the availability of the reps is independent of the customers in the queue.

Using survey data that identified which customer service representative assisted each customer, the researchers were able to estimate the skill level of each rep. Because the reps were assigned to customers randomly and independent of various factors (including a customer's credit score, credit limit and tenure with the company), the researchers were able to isolate the impact of service quality (and thus satisfaction) on loyalty.

"Whether you were handled by a good or bad service agent has nothing to do with, say, whether you are satisfied with your APR or not," Huang says. "That part of your reporting of satisfaction has nothing to do with other factors."

The researchers measured loyalty in two ways: whether a customer was willing to recommend the company to a friend, as well as whether the customer actually stayed with the company.

They found that not only have companies underestimated the impact of service satisfaction on loyalty, this impact is even greater for more difficult or important calls.

The IV approach can be applied to many other settings, such as banks and hospitals, where service quality is random, Huang says.

"If you walk into urgent care, which doctor serves you really depends on the availability of the doctors," he says. "We're definitely interested in how doctors' service affects our satisfaction with medical services, or our loyalty. If their healthcare service quality is higher, how does that affect our loyalty to a healthcare organization?"

By Melvin Durai

 

Faculty Research