Using Affective Displays to Predict Customer Satisfaction
In this article, Shelly Ashtar reflects on her longstanding interest in service-related work and how it connects to her research interest in customer satisfaction. Ashtar explores this topic with collaborators Galit B. Yom-Tov, Anat Rafaeli and Jochen Wirtz in “Affect-as-Information: Customer and Employee Affective Displays as Expeditious Predictors of Customer Satisfaction,” in the Journal of Service Research.
My interest in service-related interactions started during my first job as an adult, as a customer service representative at a telecommunications company. The most challenging aspect of this role involved ensuring customer satisfaction, especially in situations where customers encountered technical issues with their phones or high bills. Encountering diverse customer behaviors and various employee strategies, increased my fascination in such interactions, which led me to pursue academic studies in Organizational Psychology.
During my PhD studies, I had the opportunity to learn from, and collaborate with, esteemed scholars as professors Rafaeli, Yom-Tov, and Wirtz. The fruitful collaboration with scholars with expertise in different perspectives of service-related areas enriched the learning, discussions, and eventually, the development of this work.
In the current research, we had the opportunity to work with real-life large-scale data of service interactions and to examine our hypotheses in an innovative manner. Through the examination of over 20,000 service interactions, comprising more than 300,000 customer and employee messages, we showcased a detailed method for employing sentiment analysis to provide rapid insights on customer satisfaction with service interactions.
Our analysis reveals that in addition to the importance of overall customer affective displays and in alignment with theory, their peak and end displays are indicative of their satisfaction. Also, as customers might not fully display their felt affect due to regulation strategies, we demonstrate that the displays of the employees can serve as additional cues of customers’ satisfaction. Lastly, we show that these effects are more pronounced in service failure than non-failure encounters.
In today’s fast-paced world, the ability to gain automatic insights into customer sentiments is a gold mine for researchers. Additionally, the ability to gain these swift insights with no friction, is an asset also for businesses. The methods we apply in this work can be leveraged by businesses to expedite the acquisition of valuable insights on the quality of their service interactions. Moreover, these methods can facilitate rapid reactions to customers dissatisfaction by employees and managers, during or immediately after the initial service interactions.
Apart from monitoring customer satisfaction, our findings carry significant implications for the management and training of frontline employees. Traditionally, service companies focus on tracking operational indicators like employee response time and number of turns till case resolution. Our research demonstrates that although these features undeniably play vital roles in service interactions, evaluating affective displays within interactions serves as a superior predictor of customer satisfaction, especially in critical instances like service outcome failures.
Thus, to create an ideal customer experience, employees should track customer affective displays throughout the entire interaction. Achieving this could involve incorporating automatically generated visual cues, such as real-time and user-friendly “emotion thermometers,” into employee dashboards. Additionally, as part of front-line employees training, we recommend emphasizing displays of positive affect, with particular attention to concluding interactions on a positive note.
We hope that our research will help researchers and managers utilize automatic methods to gain insights on customers’ experience, on a larger scale and with resource conservation.
