How fake customer reviews work – and don’t –

As more and more consumers turn to e-commerce rather than in-person shopping, it is essential to ensure that these online platforms offer a reliable rating and rating system to maintain consumer confidence. . Unfortunately, recent research suggests that fake reviews are much more common than you might think – and just as difficult to eliminate. Since e-commerce platforms like Amazon factor product ratings into their search ranking algorithms, there is a strong incentive for sellers to manipulate their product ratings by soliciting fake reviews. In this article, the authors shed light on how sellers get hold of fake reviews, the impact of those reviews on short and long term sales performance, and what platforms can do to tackle this pervasive problem. .

Between April and June 2020, the American e-commerce market experienced a record 44.4% growth – and it is likely to continue to thrive as businesses large and small turn to online sales in the face of the pandemic. For e-commerce to work, platforms like Amazon need a rating and rating system that helps consumers make these online shopping decisions with confidence. But because these reviews are usually a big factor in search ranking algorithms and therefore have a big impact on product visibility and sales, these systems often also create powerful incentives for sellers to manipulate the ranking of their products. through fake reviews.

In response to this problem, many platforms have developed automated tools to identify and remove clearly bogus reviews. For example, Yelp uses a proprietary algorithm that filters about 16% of their reviews. But as sellers get more sophisticated in the way they craft these bogus reviews, it’s harder and harder for platforms to weed out them. To better understand the extent of the problem – and what companies can do about it – we conducted a 10 month study explore how fake reviews are generated and how they impact sellers, buyers and platforms.

Through our research, we have discovered a large and flourishing market for fake reviews. One of the most common mechanisms we found to get these reviews was through private Facebook groups: sellers used these groups to recruit people to buy their products and leave a five-star review that sounded authentic, then compensate them through PayPal for the cost of the product, taxes and fees, and in some cases a commission of $ 5-10. We also found that these groups would disappear occasionally, and then were almost immediately replaced by new, similar groups.

We worked with a team of undergraduates from UCLA to infiltrate these markets, observe them, and collect data on the products the sellers were seeking reviews for. At the same time, we collected data on these products from Amazon, including their ratings and reviews, sales rankings, prices and advertising strategies. Because we were able to observe exactly when sellers started and stopped soliciting fake reviews, and compare that activity to sales data for their products, this allowed us to measure the effectiveness of those reviews in increasing the results of the reviews. short and long term sales. .

Our first important conclusion is that while this phenomenon may not be well known, it is extremely widespread. Based on our observations, we estimate that as many as 4.5 million sellers have obtained fake reviews through these Facebook groups in the past year.

Second, fake reviews seem to be the most common for a certain type of product. These products tend to be priced the same as their competition, usually in the $ 15- $ 40 price range, and they usually already have high ratings, with an average rating of 4.4 and an average of 183 reviews. (suggesting that many of these preexisting reviews may also have been wrong). Finally, these products are generally not trademarks and the vast majority of sellers are located in or around Shenzhen, China. While we can’t be sure the reason for this trend, recent changes in Amazon policies to encourage more global sellers have resulted in a significant increase of Chinese manufacturers sell directly on the platform (rather than supplying US companies). These new sellers often have tight margins and little reputations to maintain, creating a host of quality control issues.

All of this suggests that fake reviews are more common than you might think, but is this an effective sales strategy? When it comes to short-term impact, the data is clear: fake reviews are extremely effective. In the two weeks after vendor reviews began recruiting, their product ratings increased by an average of 0.16 stars, and the average number of reviews received by these products per week doubled from five at an average of 10 evaluations per week. While some of these reviews may be organic, that is, from real unpaid customers, the fact that the review jump occurs immediately after sellers start buying fake reviews suggests that this spike is caused by false reviews. This increase in reviews translated into a significant increase in sales, with these products experiencing an average increase of 12.5% ​​in sales.

However, this boost was short-lived. We speculated that companies might use fake reviews to solve the “cold start” problem (ie, high-quality products without reviews have a hard time getting noticed). This would suggest that once they started building a positive reputation through fake reviews, real consumers would start buying these products and leave organic reviews, resulting in a self-sustaining sales process. But instead, we found that the increases in ratings, number of reviews, and sales tended to fade within one to two months. Just eight weeks after sellers stopped buying fake reviews, their product’s average ratings dropped 6.3%, their sales dropped 21.5%, and they started receiving a large number of ratings. one star from unhappy customers. This suggests that these products were in fact substandard, and customers likely felt cheated by the inflated ratings and reviews.

Obviously, these fake reviews cause real problems for buyers. But it’s not just consumers who suffer when sellers use fake reviews. As fake reviews eat away at consumers’ trust in the review system, the platform itself takes a hit as well. So what are platforms like Amazon doing to combat this problem?

In 2019 alone, Amazon spent more than $ 500 million and employed over 8,000 people to reduce fraud and abuse on its platform. And in our study, we found that Amazon actually removed around 40% of these fake reviews, but it took them, on average, over 100 days after posting a fake review to remove it. That’s more than enough time for sellers to take advantage of the short-term increase in sales, and for enough consumers to be misled to generate a surge in reviews from an angry star. And this limited impact is not surprising. After all, despite Amazon’s efforts to combat fake reviews, sellers still find them effective enough to be worth buying.

As sellers become more sophisticated in the way they create fake reviews, online marketplaces like Amazon find themselves in an endless arms race to develop effective ways to identify and remove them – and to remove them. maintain consumer confidence. While there is no clear answer, our research suggests a potential solution: do what we did. Ecommerce platforms should consider partnering with social media platforms like Facebook to better understand how sellers recruit fake reviews, and potentially use that visibility to speed up the process of identifying and removing fake reviews. Finding ways to crack down on this activity would be win-win, as it would reduce illicit activity on social media platforms, increase trust in e-commerce platforms, and enable a more positive and reliable experience for consumers.

Joseph P. Harris