98% of Customers Stay Quiet: Decoding the Extreme Voices Phenomenon

Online reviews are a key source of information for consumers before they make a purchase, guiding their decisions and influencing billions of dollars in sales (Magnani, 2020). However, the reliability of these reviews is often weakened by a common issue known as self-selection bias.

This "Extreme Voices" phenomenon occurs because the collection of reviews is typically dominated by people who had either a fantastic or a terrible experience. The majority of customers, who had a more average experience, tend not to leave a review. This imbalance creates a "J-shaped" pattern in ratings data, where most reviews are clustered at the highest and lowest ends, a trend proven in foundational research by Hu, Pavlou, and Zhang (2009). This analysis explores the reasons for this phenomenon, its impact on the market, and potential solutions.

The "J-Shaped" Pattern in Review Data

The clearest evidence of the "Extreme Voices" phenomenon is the consistent J-shaped curve found in star ratings. This is different from the balanced "bell curve" one would expect if the reviews represented all customers fairly (Hu, Pavlou, & Zhang, 2009). On major platforms, there is a large peak at the highest rating (5 stars) and a smaller, secondary peak at the lowest (1 star). On Yelp, for example, 5-star and 1-star reviews make up over 70% of all ratings, while moderate 3-star reviews account for only 8%.

Researchers identified two main biases that cause this pattern:

• Purchasing Bias: Customers who already have a positive view of a product are more likely to buy it in the first place. This naturally skews the pool of potential reviewers toward a more positive outlook.
• Under-reporting Bias: After a purchase, people with extreme opinions feel more motivated to write a review. Those with moderate experiences often don't have a strong enough reason to share their thoughts, leading them to be underrepresented.

Because this "silent majority" doesn't participate, the available reviews come from a small, unrepresentative portion of the customer base. This makes the average star rating an unreliable indicator of a product's actual quality.

The Psychology of Leaving a Review

The choice to write a review is often driven by strong emotions. Feelings like extreme satisfaction or frustration provide the motivation needed to spend the time and effort (Jebde, 2020). People are also motivated by a desire to help others, express their feelings, or feel part of a community. This is amplified by "negativity bias," the natural human tendency to give more weight to negative information, which increases the impact of critical reviews (Yin, Mitra, & Zhang, 2016).

The design of review platforms can also distort the results. For example, some platforms require users to post reviews under their real names. While intended to promote courtesy, this can create a "chilling effect," where people avoid leaving honest negative feedback because they fear arguments or social backlash (Vaas, 2017). Such policies can also create safety risks for vulnerable individuals who depend on anonymity for their protection (boyd, 2011).

Market Impact and Fake Reviews

The bias in online reviews has significant economic consequences. It can mislead consumers into making poor buying choices and cause money to be spent inefficiently. Since a one-star increase on Yelp can boost a business's revenue by 5-9% (Luca, 2016), this system can create unfair advantages based on who is leaving reviews, not on the actual quality of the product or service.

This problem is made worse by fake or inauthentic reviews. An analysis by Fakespot suggested that around 42% of the Amazon reviews it examined were not genuine. This type of fraud is estimated to influence $152 billion in global spending each year (World Economic Forum, 2021). The challenge is growing with the rise of AI-generated content, which increased by over 279% on Google Reviews between 2019 and 2024.

How to Get More Representative Feedback

To get more balanced feedback, businesses are using new strategies to hear from a wider range of customers. The most effective methods make it much easier to share an opinion. For instance, the platform TruRating uses a single-question feedback system at the point of sale. This approach has been shown to increase customer participation from the typical 1-2% to over 80%. The large volume of data helped one grocery store identify and fix service issues, leading to a 22% increase in service scores and a 5% increase in sales (TruRating, 2024).

Another successful strategy is to use structured feedback systems that ask specific questions. Popular methods include:

• Customer Satisfaction Score (CSAT): Asks about a customer's satisfaction with a single, recent interaction.
• Customer Effort Score (CES): Measures how easy it was for a customer to get their issue resolved.
• Net Promoter Score® (NPS®): Assesses overall loyalty by asking how likely a customer is to recommend the company to others.

By making the feedback process easy and structured, companies can successfully encourage the "silent majority" to share their opinions, providing a more accurate and useful picture of the customer experience.

Conclusion

The "Extreme Voices" phenomenon is a deep-rooted bias in online review platforms where participation is voluntary. The resulting J-shaped distribution of ratings gives a distorted view of customer experiences, which has major effects on consumers and businesses. This bias is worsened by fake reviews and the social dynamics of review platforms. The best solution is not to try to change user behavior, but to redesign how feedback is collected. By using easy, structured, and proactive systems, businesses can lower the barrier for participation and finally hear the valuable and more representative voice of the silent majority.