How One Simple Design Change Helped Eliminate Racial Bias in Customer Ratings

New research in Nature offers business leaders a powerful, practical way to reduce inequality on their platforms

Based on research by Tristan L. Botelho, Associate Professor of Organizational Behavior at the Yale School of Management; Sora Jun, Assistant Professor of Strategy at the University of Notre Dame’s Mendoza College of Business; Demetrius Humes, doctoral student at the University of Toronto’s Rotman School of Management and Katherine A. DeCelles, Professor of Organizational Behavior and Director of the PhD Program at the University of Toronto’s Rotman School of Management and former Postdoctoral Fellow at the Erb Institute.

When we think about racial discrimination in the workplace, we often imagine hiring decisions, performance evaluations, or promotion pathways. But what about something as small as a customer clicking a star rating?

A new study published in Nature—“Scale dichotomization reduces customer racial discrimination and income inequality”—shows that the way we design feedback systems can have an outsized impact on equity. The research, coauthored by Katherine A. DeCelles, a former Erb Institute Postdoctoral Fellow and now professor at the University of Toronto’s Rotman School of Management, delivers a striking message: the design of rating systems can either reinforce or reduce expressions of racial bias—and changing them can close pay gaps entirely.

The Power of Simplicity

The study focuses on a home-services platform that moved from a traditional five-star rating system to a simpler thumbs-up/thumbs-down scale. Researchers analyzed nearly 70,000 real-world customer evaluations and found that this small shift made a big difference:

• Under the five-star system, non-white workers earned only 91 cents for every dollar earned by white workers.
• After the platform switched to a binary scale, that racial pay gap disappeared entirely.

The simplified scale didn’t just make things more equitable—it neutralized the effect of subtle, often unconscious biases that were creeping into customer ratings.

Why This Matters Now

“Our findings are significant especially today, where DEI is under attack, and many organizations are scaling back efforts for equality,” says DeCelles. “We show how organizational design can result in often unintentional but meaningful disparities.”

At a time when many companies feel pressure to do more with less in their DEI strategies, this research offers something rare: a practical, low-cost intervention that really works.

It’s Not Just the Stars—It’s the Psychology

Perhaps most intriguing, the researchers also found that it’s not necessarily the rating scale itself, but the psychological process that matters most. When participants used a five-star scale as if it were binary—focusing on whether the work was “great” or “not”—the equalizing effects were also produced.

“I think the fact that we can manipulate the psychological process of using a dichotomous rating scale while still actually using a five-star scale is the most interesting discovery,” says DeCelles. “It underscores how the experience of giving feedback is critical, as well as the structure itself.”

Takeaways for Leaders and Designers

For business leaders—especially those in the platform, service, or consumer-facing economy—this study is a wake-up call. Here’s what you can do:

1. Rethink Your Ratings

If your organization uses customer evaluations, consider a simpler binary scale. Even small UX shifts can significantly reduce biased outcomes.

2. Design for Equity

Many organizations invest in training to reduce internal bias. But as DeCelles points out, “firms can’t train or select customers,” making interface design a much more powerful lever than previously assumed.

3. Test for Bias Early

Run A/B tests to explore how different feedback mechanisms impact equity outcomes across groups. The best design is one that works for everyone.

4. Don’t Abandon DEI—Evolve It

This research demonstrates that advancing equity doesn’t always require massive programs or budgets. Sometimes, it starts with thoughtful design decisions backed by good data.