Georgios Zervas, an SMG assistant professor of marketing, coauthored a study that should make online review followers tread with caution. Photo by Kalman Zabarsky
If you’re combing Yelp reviews to decide where to eat, you’re likely to be misled or even duped, according to Georgios Zervas, coauthor of a recent study concluding that at least 16 percent of the reviews are fake. The researchers found that the worst offenders are restaurants seeking to offset negative write-ups, that chain restaurants are the least likely to commit review fraud, and that restaurants sometimes take the low ground by posting fraudulent negative reviews for establishments competing for the same customer base.
Although they make efforts to ensure the veracity of their content, sites like Yelp,TripAdvisor, and Angie’s Listremain easy prey, conclude Zervas, a School of Management assistant professor of marketing, and Harvard Business School’s Michael Luca (GRS’11) in their report Fake It Till You Make It: Reputation, Competition, and Yelp Review Fraud, now under review for publication.
The study was cited in a recent Wall Street Journal blogpost prompted by the New York attorney general’s recent Operation Clean Turf initiative uncovering manipulation in the reputation management industry. That investigation led to 19 companies agreeing to stop commissioning fake online reviews and facing fines of $350,000. After its results were reported on additional blogs and business news sites, Zervas’ study triggered a response from Yelp’s vice president for communications and public affairs on the site’s official blog. The study “confirms something we have long known: businesses that don’t have a good reputation online will try to create one by submitting phony reviews,” the response reads. Yelp says its filter algorithm sifts through its more than 42 million reviews, rooting out the 25 percent found to be fakes or submitted by the businesses themselves.
BU Today: Did you discover anything particularly surprising in your study?
Zervas: One thing that was slightly surprising, not so much to me but to most people, is the proportion of suspected fake reviews that Yelp removes—approximately one quarter of all reviews submitted to Yelp are not published. That’s about 10 million reviews.
What are some of the concerns your study raises?
The main concern is for firms like Yelp and TripAdvisor. Platforms that crowd-source reviews rely on the integrity of these reviews, and fraudulent reviews pose a major threat to their trustworthiness. Furthermore, consumers should be concerned that fake reviews are leading them to suboptimal choices, and businesses should be aware that some negative reviews might come from their competitors.
How much of an impact do sites like Yelp have on a business?
My coauthor Michael Luca did a great study on this and found that having an extra star on Yelp causes the revenue of a business to rise by 5 to 10 percent, so there’s a direct connection between Yelp ratings and a business’ bottom line.
How can consumers view these sites more critically?
I think there are many signals on Yelp that consumers can combine to make up their minds. The way I use Yelp is, I read individual reviews, trying to be aware not just of whether they’re fake, but beyond that, whether they come from consumers who are like myself. There are plenty of biases in reviews besides their being fake or real. The other thing I look at is the number of reviews a business has. I have a lot more faith in a business with 3½ stars and 100 reviews than I do in one with 4 stars and just 3 or 4 reviews. That’s common sense. Also, when available, you can use sites, like Expedia, that allow consumers to review a business only once it’s confirmed that they are paying customers.
Are there any data on who is more likely to write a review, those who had a terrible experience or those who had a great one?
Not in our study, but in general people seem more likely to contribute reviews after extreme experiences, either positive or negative.
Are the fake reviews the same as false advertising?
According to the Federal Trade Commission, a review is considered deceptive advertising if any material connection between the reviewer and the business being reviewed is not disclosed in the review. In a recent example of enforcing these regulations, the New York attorney general went after reputation management companies, which produce fake reviews in bulk.
How fine a line is there between a fake restaurant review and one written by a friend at the urging of the restaurant’s owner?
The friendship between the reviewer and the restaurant owner is a material connection—information that third parties reading the review might want to know before they buy a product or a service, and therefore it should be disclosed. So this review is still fraudulent.
How reliable are algorithms for weeding out fake reviews? In lay terms, what are the most prominent giveaways?
The answer is: I don’t know, because I don’t know what they’re doing—it’s something that companies like Yelp would rather keep secret. But from my experience reading many reviews over the past year, I think Yelp does a decent job of catching fake reviews.
Do some surviving reviews stand out as suspect?
Sometimes you might see people who have reviewed one or two businesses, but have zero other reviews, no friends on Yelp, and give glowing five-star reviews to these two businesses—that seems a bit fishy. Also, I remember one case cited in a recent New York Times article: someone wrote a wonderful five-star review for a dentist following a string of really bad one-star reviews; one of these one-star reviews mentioned that the dentist’s drill broke inside the patient’s mouth. Context can provide a lot of clues about what’s fake.
Isn’t this problem likely only to worsen?
Yes, as more consumers are basing their decisions on crowd-sourced reviews, there’s a greater incentive for businesses to cheat. And so far the consequences have been extremely small, though the New York fines are a small step in the right direction.
So what’s the best way to get reliable information?
Beyond online word of mouth, use off-line word of mouth if that’s available. Mine your own social network as much as possible.
Would the system benefit if more people made a habit of reviewing places they’ve patronized?
Yes, the more information we put online, the better the quality of the signal other consumers are going to get. More reviews will enhance the trustworthiness of review platforms. For example, currently there is a lot of self-selection among reviewers, who are usually more likely to write a review for extremely positive or extremely negative experiences. It would be much better if people also wrote about their moderate experiences.
What else can consumers to do keep these forums honest?
Most platforms provide tools on their websites for consumers to flag suspicious reviews. So beyond writing reviews of your own, you can flag a suspicious review, and the Yelp team will take a close look at it.
I know of a place that gets consistently scathing reviews for mistreating customers, especially those with children, yet people swarm there. Do negative reviews ever improve business?
Yes, there are cases when negative reviews might make you take positive action. For example, if I’m someone who doesn’t like children around me, that’s a good place for me to hang out. So, for businesses, there can also be positive value in negative reviews.