Yelp is a user-based reviewing website and app that helps others determine whether or not to visit and spend money at a particular business. Although reviews help determine whether or not to spend money on a specific product or service, this becomes problematic when certain reviews are blocked, censored, or produced with malicious intent. Blocked, censored, and negative reviews are current issues with Yelp.
HOW DOES YELP WORK?
To learn more about how Yelp works, I went to their website. ‘Best to learn from the source,’ I thought.
The about page of the Yelp website states, “Yelp connects people with great local businesses,” and then provides links for employment, investors, news, and trust & safety. I decided to go into the trust & safety page to investigate further what Yelp does.
This page talks about gaining and maintaining trust for the interest of both business owners as well as consumers; to do so, it states, “Yelp works hard to feature content that reflects real experiences that consumers are inspired to share.” It doesn’t go into detail about how it elicits or fact-checks these experiences.
Reading on it states that some software is used to “identify the most reliable content” but doesn’t specify reliability standards. Further down, it says that there are “teams and policies in place to protect our community.” Lastly, it mentions they provide lots of information about businesses, such as health scores, wait times, bathrooms, etc., which makes me wonder where they get this information. I know here in Washington, restaurant health codes are posted, but how does Yelp get them? Are these health codes public records? Or are these basic facts about these businesses reported by consumers?
I wanted to look a little deeper into Yelp and the “recommendation software” that references garnering reliable content. This software reviews every post made, which is expected. It probably flags posts that use insensitive language and posts that feature spelling and grammar irregularities for further consideration. The software looks at many indicators and then promotes those as recommended. This immediately lets me know that not all posts are treated equally. One of the indicators that the software looks for is frequent users. Who are regular users, though? What is the demographic for users of Yelp? The potential user, I imagine, has time to post reviews to Yelp, which means they’re probably not working multiple jobs to feed their own family. This seems to be the case, according to page 2 of their SEC Filing. At least 50% of their users have an earned income of over $100,000. The user also has some technical know-how because Yelp is primarily visited through an application on the phone, which means they’re probably younger. All of these questions about the unknown demographics of Yelp users have me worried that there are conscious or unconscious biases towards businesses. For example, if Yelp users are mostly upper-class, white women, there might be some bias towards male, black-owned companies in urban areas. Specific cultural differences may be misidentified as poor service or rude behavior.
Also, if a user hasn’t started using Yelp and has a bad experience with a company, their post will get flagged by the recommendation software and not shown on the site until the user has multiple posts.
The recommendation software also weeds out conflicts of interest and solicited reviews but doesn’t detail how those are determined. How does Yelp know that I am a friend of a business? Or a competitor? How does it know I was asked by a business to give a review? Also, if Yelp does flag someone, can they contest that decision? None of this is spelled out on their website.
The “teams and policies” in place are primarily around anti-hate and coordinated attempts to impact a business negatively. This part of the website seemed to be more geared towards companies themselves, so I didn’t look too much into it.
So far, I have many questions about Yelp and how it works, not just how the company says it works. The company will say everything works spectacularly, and there are no flaws in its system, but I know that can’t be the case. I have some doubts based upon the demographic information I found about Yelp users and how that might relate to racial and economic biases on the website. Also, how it favors some content creators over others looks pretty suspicious to me. Come back in the future to see other posts looking into the dark side of Yelp.
CONTROVERSY #1 PAY PER REVIEWS
According to some of the court cases waged against Yelp over the years, I've learned that's not always the case. I was already suspicious of the Yelp algorithm picking and choosing which reviews to post already, but what you’re about to read will call more into question with Yelp.
There have been some issues with Yelp in the past. ABC 7 out of San Francisco from 2010 talks about an ad purchase for a good review racket that Yelp was supposedly promoting.
Here’s how many business owners in San Francisco described their interaction. A Yelp representative would request the company purchase ads through Yelp, and in turn, bad reviews for the business would be removed or lowered in the queue of reviews. Some business owners found that they started getting more negative reviews after refusing to purchase ads. This led them to believe that Yelp was targeting their business and extorting ad purchases. Later, it could not be proven beyond coincidence that bad reviews followed declined ad purchases during the trial.
In regards to the court case, Vince Sollitto, the then vice president of corporate communications, said, “we use a complex algorithm to weigh some factors including the votes of the community or the established credibility of the reviewer and a variety of other factors to help us show information and content that’s valuable.” This is true based on my perusal of their website, but Sollitto brings upvotes from the community. This could be an outdated way the review service worked because this lawsuit and quotes are from a decade ago, but it is a problem if user votes were used simultaneously.
This goes back to my previous post when I discussed the demographics of Yelp users. If votes are used, then there is most likely a bias towards proficient English language users. Those who speak or write in a broken or less than proficient English form wouldn’t be upvoted or voted as applicable by others.
Ultimately, the lawsuit was dismissed as the court upheld Yelp’s ability to arrange reviews however it liked, according to VIN News. It seems like, at least from 2014, Yelp can continue this approach to business. Yelp isn’t held accountable for its business practices here. I will continue to research and see if there has been any update to Yelp’s practices. Yelp is still able to organize reviews however it wants, which seems very suspicious and problematic.
CONTROVERSY #2 ASTROTURFING
As mentioned in my first blog post, the Yelp Recommendation Software works in conjunction with the Teams and Policies of Yelp to help provide the most accurate reviews of businesses. Part of this is filtering out possible fraudulent or less frequent reviewers when a series of positive or negative reviews are produced in a concerted effort to affect a business called astroturfing.
Why filter reviews in the first place? There are obvious reasons, such as competitors leaving bad reviews or companies leaving good reviews for themselves. Both occurred after Yelp’s prominence began to rise in the mid-2010s, which led to several lawsuits.
One example, as reported on from ABC News, is from Julian McMillan of McMillan Law Group. He allegedly had employees of his law firm write and post positive reviews of his company. Essentially the positively astroturfed his own business. McMillan took Yelp to court over his treatment, and he believes Yelp is bringing this other lawsuit against him for fake reviews as retaliation. Yelp claims the two issues are separate.
If I’m reading the case result correctly, note I am not knowledgeable about legal matters at all. Aside from this case, there are legitimate cases of companies paying for reviews such as buyYelpreview.com and buyYelpreviews.net, both of which used accounts with high user frequency to post reviews for companies that paid. It appears that these astroturfing companies were shut down. Using previously created accounts or just accounts on Yelp in general, it seems that they were breaching the terms and service contracts signed at the creation of the Yelp accounts.
The last example is of one company astroturfing a competitor. According to ABC News, Samsung paid reviewers to give bad reviews for competitor HTC, an up-and-coming mobile phone, and technology company, in the early 2010s. Although this happened in Taiwan using a Yelp clone, it doesn’t mean the same actions are taking place on the Yelp platform.
Although all of these examples could be considered “pay per review,” which was the title of my previous blog, these companies are paying for posts, not paying to rearrange or hide reviews. These companies might have come into existence because of the unfair practices by Yelp, as many of these court cases appeared after the 2010 court case regarding Yelp company’s unfair approach to reviews on the website and their ability to filter and adjust reviews as it pleases.
CONTROVERSY #3 YELP AND RACE
I stated that there might be implicit bias against non-white businesses based on the demographic information provided by the company’s SEC filing. Today, I finally found some research to back that presumption up.
According to a 2015 study done by Sharon Zukin, Scarlett Lindeman, and Laurie Hurson out of City University in New York, the demographics of yelp are young, 20-year-old childless women who are college graduates make over $60,000. This seems similar to my proposed idea of the Yelp consumer in my first blog post.
The paper quotes some Yelp reviews of a neighborhood whose population is majority people of color from 2015, “They emphasize the danger, alerting readers that to eat in a certain good restaurant, they must be prepared to ‘‘brave the dark alleys of Bed-Stuy.’’ Some even go out of their way to focus on fear: ‘‘Once I stepped out the door, I wanted to go right back inside! The neighborhood is a little scary — and that’s putting it nicely.’’ This leads to some posts using racialized language to describe the neighborhood's atmosphere and nature in black communities.
Although I stated that there might be implicit bias in posts, there seems to be outright bias and racism towards non-white businesses and neighborhoods.
On a related note, this past year, Yelp proposed an idea to flag businesses as racist. One business owner and YouTuber, Louis Rossman, in an October 2020 video, expressed his distrust that Yelp had the good technological algorithm and workforce to check into claims of racism by a business. This was proven by trolling yelp reviewers posting fake reviews about his firm just days after the first video. In the second video, Rossman shows his correspondence with Yelp complaining that a reviewer using George Floyd’s picture as their profile picture, claiming to be from Minnesota, is fake. Yelp’s response is, they don’t believe the profile and review are affected, and it will remain on Rossman’s business’s yelp page.
There is an issue of implicit and explicit bias and racism on the Yelp platform.
REVIEW AND SUGGESTIONS
From day one, there has been something fishy about Yelp. I initially discovered that not all posts are treated equally and that some unknown behind-the-scenes “recommendation software” was being used to filter reviews. Also, based on the demographics of Yelp users, I had some suspicions about implicit bias towards non-white businesses.
I have found that Yelp has been hiding behind its recommendation software to bully companies into paying for reviews to be manipulated throughout my research. I read from the articles in the blog post that upwards of 20% of posts on Yelp may be false. Not only that, but outside companies and people might be manipulating reviews on Yelp through small or large-scale astroturfing.
In addition to reviews being manipulated by Yelp itself and from outside sources, posts and reviewers are biased towards non-white businesses and neighborhoods.
As for suggestions, I’m not sure what all Yelp can do—perhaps being more open about how their algorithm works and how their team's function would be helpful. At least then, people can judge whether to believe what they see on the website. Overall, it makes sense to guide customers to post reviews on your website and other third-party websites such as Google, Clutch, and Upcity, to name a few. This will allow your reviews to be distributed over several platforms and prevent a single platform from bullying you.