For the first time, we are informing our community of users and business owners about important updates we’re making to Yelp’s recommendation software. You can learn more about our recommendation software here.
For decades now, online reviews have helped us make informed decisions on what products to buy, where to stay when traveling, who to hire for a home project and where to eat dinner, among many daily consumer decisions. However, deceptive reviews (including fake or paid reviews) can drown out honest insights and even cause consumers harm unless actively mitigated. In recent years, we’ve confronted a rise of review exchange groups, which attempt to flood sites with fraudulent online reviews, making them less reliable for everyone. Review exchange groups exist on various online platforms and facilitate the buying, selling, or exchange of fake reviews.
At Yelp, we’ve spent more than 15 years developing and updating our recommendation software, which showcases the reviews that it determines are most helpful. Starting today, we’re rolling out a new recommendation software update that is programmed to identify reviews likely coming from review exchange groups. Through this new recommendation software update, we estimate approximately 177,000 currently recommended reviews (out of more than 224 million reviews that have been contributed to Yelp to date) will become not recommended. We estimate about 154,000 businesses may see reviews become unrecommended as a result of this update – most businesses impacted by this update will see only one review become unrecommended – which means the software determined those reviews may be connected to review exchange groups. While this behavior is not common, if left unchecked this type of activity can make Yelp less helpful for everyone. This update is not only meant to provide consumers with the most helpful and reliable information, but it also protects businesses that work hard to honestly earn their reviews from bad actors trying to game the system.
Yelp’s recommendation software is our first line of defense to mitigate misinformation on our platform. It’s engineered to provide consumers with the most helpful and reliable reviews, and level the playing field for all businesses. The software does something no human can by regularly analyzing billions of data points from all reviews, reviewers and businesses to evaluate the usefulness and reliability of each review. The software automatically detects reviews that it suspects may be conflicts of interest, solicited, or otherwise less reliable or less useful, and moves them to a business page’s “not recommended” section. Not recommended reviews are still accessible via a link at the bottom of a business’s Yelp page, but they don’t factor into the business’s overall star rating or review count. Our recommendation software does not take into account a business’s advertising status (and never has), so advertisers are not treated more favorably, and non-advertisers are not penalized.
Independent from the recommendation software, we also regularly conduct sting operations, monitor online groups, and reach out to other platforms when we detect suspicious behavior. For example, in 2020 we reported nearly 200 online groups to other platforms to help shut down fake review rings. We also publish Suspicious Review Activity Alerts on Yelp business pages as part of our Consumer Alerts program when we see a large number of reviews from users with possible connections to deceptive review rings. In 2020, we closed down more than 1,200 user accounts associated with review rings we uncovered.
Consumer trust is Yelp’s number one priority. We’re dedicated to providing the most helpful and reliable information to consumers and cracking down on any attempts we find to mislead people on our platform. Combating the spread of misinformation is something we invested in early on, built our platform around, and established systems and processes that continue to scale over time.