April 22, 2016
Reviews are fast becoming a vital part of consumer culture. With so many businesses now moving to utilize online tools, it’s easier than ever for customers to communicate with them and give their feedback on the service and quality of products. Unfortunately, fraudulent reviews, typically written by people working directly or indirectly for the companies whose products are under review, are also on the rise. Is there a way for business owners to weed out these fraudulent reviews and have the reliable, genuine customer feedback left?
Firstly, we must understand that fraudulent reviews come in many different forms. For-hire reviewers, another kind of fraudulent review in that they aren’t a customer’s genuine feelings about a product or business, are a result of digital-age business. From Upwork to Fiverr, specific platforms seek freelancers and taskers to team up with jobs to write and post reviews retailer-specific sites. Unfortunately, racking down on fraudulent reviews continues to pose a challenge to review sites and legal authorities alike.
As humans, we naturally have a problem distinguishing between authentic and fraudulent reviews. A 2011 study on “deceptive opinion spam”, carried out by a team of Cornell University computer scientists and linguists, concluded, “The detection of deceptive opinion spam is well beyond the capabilities of human judges, most of whom perform roughly at-chance.” In addition, attempts to enforce laws that seek to prohibit fraudulent reviews also remain difficult to effectively enforce. But there is hope. A growing body of research suggests that armed with the right algorithm, it is possible to screen out most, if not all, fraudulent reviews.
How Prevalent Are Fraudulent Reviews?
While there is no way to verify how many reviews are fraudulent, the problem is widespread across industries and review sites. In 2013, Yelp admitted that fraudulent reviews may account for over 20% of reviews on their site—up from a mere 6% in 2006.
Amazon has also admitted that their site continues to be plagued by fraudulent reviews, but haven’t released data on the percent of reviews that fall into the fraudulent category. And after coming under intense public scrutiny, TripAdvisor now employs hundreds of people armed with advanced fraud detection tools to help maintain the integrity of its site.
The Other Side of the Problem
Although paying people to write reviews is clearly fraudulent, it would be naïve to assume that paid reviews are the only problem. Indeed, what makes a review fraudulent versus authentic is by no means an easy thing to discern nor is it an entirely new problem.
But in some cases, review cultures are even more deceptive. Remember the 2004 Amazon error that briefly led to all the names behind the site’s anonymous reviews being revealed to the public? The revelation that not only exposed the identities of anonymous reviewers but also revealed that a high percentage of these reviewers were reviewing their own books.
Algorithms and B2B Solutions
In response to the growing prevalence of fraudulent reviews, researchers and private companies alike have responded with a series of algorithmic solutions.
The Cornell team’s study on “disruptive opinion spam” found that by paying closer attention to the specific linguistic markers of the genre—for example, an increased use of the first-person singular in deceptive reviews—one can screen out fraudulent reviews more effectively.
But researchers and crowd-sourced sites like Yelp and Amazon, are not the only people working to find a solution to the rise of fraudulent reviews. We are now witnessing B2B solutions to help companies, including those lacking the resources to hire their own team of full-time fraud detection specialists, to authenticate reviews.
One example of this is the site, CrowdReviews, which has an algorithm that is not only designed to filter out fraudulent reviews but also to reward businesses with a wider variety of reviews. CrowdReview’s algorithm ranks reviews in five categories, including reviewer strength. To determine reviewer strength, they take into whether or not the review was left by a client, potential client, competitor, employee or non-disclosed reviewer and whether or not the reviewer has a LinkedIn profile. As one might expect, reviews left by disclosed reviewers with LinkedIn profiles are worth more than reviews left by non-disclosed reviewers without LinkedIn profiles.
As Jeev Trika, investor and facilitator of CrowdReviews observes, “Reports about fraudulent reviews are understandably leading to heightened skepticism about user reviews, but these reviews remain extremely important.” But as Trika further emphasizes, “We now live in a era, when user reviews can no longer stand on their own, and that’s where CrowdReviews comes in.
Our truly unique ranking algorithm is designed to help restore the integrity of user reviews. That being said, we do not control the rankings, the crowd does.” But is CrowdReviews effective enough to not only weed out reviews left by the Fiverrs currently under attack by Amazon, but also by self-proclaimed pros like Barry?
Asked to review CrowdReviews’s criteria for vetting reviewer strength, Barry agreed that their B2B solutions would make his work increasingly challenging. “I’ve been doing this for five years and every year, it gets a bit harder,” he admitted.
“I’m really hustling now to ensure that my identities can be verified, which is why I have created LinkedIn profiles and even Twitter accounts for each identity. I also have five devices with different IP addresses and use various computer labs on campus to help trick up searches focused on weeding out repeated IPs.”
Time will tell how we evolve to conquor the issue of fraudulent reviews, but as businesses become more tech-savvy, they may uncover natural responses for these issues.
Did you like this article?
Get more delivered to your inbox just like it!