Study: Fake reviews often written by best customers

New research from MIT Sloan professor suggests sites should verify reviewers to boost credibility

July 22, 2014

Professor Duncan Simester

Professor Duncan Simester

Some of a company’s best customers may be writing nasty online reviews about products they didn’t even purchase, new research from MIT Sloan marketing professor Duncan Simester shows.

It is not obvious why customers would decide to lie about purchasing a product, but Simester said they may be acting as “self-appointed brand managers,” using negative reviews to give the company feedback and influence its sales.

“They’re basically browsing around the website, seeing stuff they don’t like, and responding to it,” Simester said.

Other explanations offered include reviewers looking for online social status and influence or reviewers who are upset with some perceived slight by the company whose products they are reviewing.

Simester’s research, “ Reviews Without a Purchase: Low Ratings, Loyal Customers, and Deception,” was published last month in Journal of Marketing Research. The article was written with Eric Anderson, a professor at Kellogg School of Management. It uses review and sales data from an online clothing company that is the sole retailer of its products to determine if a review is backed up by a purchase. It finds that the authors of fraudulent reviews tend to be more frequent purchasers of other products from the company.

Whatever the reason for fake reviews, the small group of fabricators—only about 15 in 1,000 customers write reviews at all and only about 1 in 1,000 write fake reviews—has a real impact on business, Simester said. Online reviews have been shown to have considerable influence on the success of a product, and fake reviews tend to be negative. Twice as many fake reviews as real reviews give a product the lowest rating, Simester and Anderson write.

If consumers catch on that some reviews are fake, Simester said, websites that verify a purchase before allowing a review, like restaurant reservation site OpenTable, may be perceived as more trustworthy. Those that don’t or can’t verify a purchase, like review giant Yelp, may become less trusted.

“It does make you think about their business model,” Simester said. “Suddenly Open Table reviews become more interesting than Yelp’s.” There is middle ground. Amazon, for example, allows reviewers to verify their purchase when writing a review, but does not require the verification.

The article offers some pointers for spotting a fake online review. Fabricated statements tend to be longer, contain details unrelated to the product, use shorter words and multiple exclamation points, and mention family. (“My dad used to take me when were young to the original store down the hill. I also remember when everything was made in America,” one probably fake review reads.) Fake reviews are also less likely to contain words describing the fit or feel a product, Simester and Anderson write.

The data for the research comes from a decade-long partnership with the clothing retailer. Simester, who recently renewed a research contract with the company for another five years, said having a long-term arrangement to study the company has allowed him to run a number of “data-rich” research experiments.