Racial bias in customer service
Racial bias in customer service
February 7, 2024 | By Professor Huaxia Rui
In this blog post, Professor Huaxia Rui presents the first large-scale evidence of racial bias in customer service interactions on social media.
As a researcher of social media analytics, I have long been fascinated by how brands respond to customer service complaints in the digital age. Most customer service interactions take place privately, whether on a phone call, through email, or with a chatbot. But when a customer service agent responds to a complaint via social media, that interaction is out there for the entire world to see. This has far-reaching implications for the power balance between customers and brands.
My interest in the topic led to the development of the Twitter Sensor project, a Google Trends-inspired platform that analyzes and visualizes tweets related to customer service complaints. One of the early questions I set out to explore was whether there was evidence of racial bias in how companies respond to complaints.
After collecting and analyzing more than 57,000 social media complaints to major U.S. airlines, here is what I discovered:
1) There is clear evidence of racial bias in how major U.S. airlines respond to customer service complaints on social media.
We know that racial bias permeates every inch of society, and implicit bias is particularly difficult to defeat. Using a variety of analytics techniques, including text mining and facial recognition, my co-authors and I presented quantitative evidence that Black customers are less likely to receive a response when they complain than white customers.
2) This bias is directed toward Black customers.
In our research, we found no statistically significant difference in response rates to Asian, Hispanic, and white customers. For example, agents providing customer service on social media respond, on average, to 49.5% of complaints from white customers and 50.07% of complaints from Asian customers. That percentage drops to 44.78% for Black customers.
3) Bias is entirely based on visual cues.
During the course of our research, we performed a deep learning-based falsification test to see what happened when there was no customer profile picture to reference during a customer service interaction. We found that bias against Black customers completely disappeared when we factored the profile photo out of the equation.
4) In this case, there is a simple fix.
Implicit bias is much more challenging to weed out than explicit bias. Bias is fundamentally etched into the human brain and unlikely to be resolved by a training session or seminar. But in this case, there is an easy solution: conceal all customer profile pictures from employees who are delivering customer service on social media platforms.
5) Liability is a complex question.
Our ability to generalize from small datasets is a key source of human intelligence. Ironically, this ability becomes a liability when it comes to treating customers equally and fairly. And unlike bias that shows up in customer-to-customer (C2C) interactions on platforms like Ebay or Airbnb, implicit bias shows up in business-to-customer (B2C) interactions. If a company’s employees acted in an explicitly discriminatory fashion, the company would be held legally liable. But when discrimination is implicit, the issue becomes less clear-cut.
The Bottom Line
My co-authors and I were surprised to find large-scale evidence of racial bias in such a public forum. Ideally, companies develop their own tools to identify and combat bias against their customers on social media platforms. By doing so, they can improve the effectiveness of their social media engagement while avoiding bad publicity and strengthening brand perception among their customers. And, when they’re lucky, the most powerful course of action can also be a simple one. For major airlines delivering customer service on social media, the solution is right in front of them: remove the profile photo and reap the benefits.
Click here to read the full research paper.
Huaxia Rui is the Xerox Chair of Computer and Information Systems at Simon Business School.
Follow the Dean’s Corner blog for more expert commentary on timely topics in business, economics, policy, and management education. To view other blogs in this series, visit the Dean's Corner Main Page.