How to Spot Fake Online Reviews

In the movie The Matrix, Morpheus (Laurence Fishburne’s character), the wise leader of the resistance against the machines and captain of the hovercraft Nebuchadnezzar, asks, “What is ‘real’? How do you define ‘real’?” He then goes on to define the concept of reality in the Matrix: “If real is what you can feel, smell, taste, and see, then ‘real’ is simply electrical signals interpreted by your brain.” For users of online reviews and review sites, the question raised by Morpheus applies as well: Which of the reviews are real, which are fake, and how do we tell them apart? Of course, these readers also hope that the reviews they read are much more than electrical signals. Rather, they hope that the reviews represent the opinions of people who have experienced the product or service in question and that the reviewers’ intent is to help other potential customers decide whether or not they want to purchase the product or service.

Increasingly, however, there is a sense that some reviews that one encounters online are fake. They could be promotional reviews—written by company employees, not identified as such, extolling the virtues of their company’s products. Or they could be critical reviews—written by competitors’ employees providing negative feedback about rivals’ products. A recent Travel Weekly article discusses the case of an Accor employee, playing the role of a reviewer on Tripadvisor, who posted positive reviews for Accor properties and negative reviews for competitors’ properties—in this case, the InterContinental Adelaide (a direct competitor of an Accor hotel), which he characterized as “outdated and overpriced.” Such reviews have several negative consequences. First, readers may end up choosing properties that are not the ones best suited to their needs. More importantly, the useful information that can be gleaned from the reviews might be ignored by customers if they feel that reviews are rife with puffery.

The question, then, is: Are fake reviews prevalent, and if so, how do we identify them? Clearly, this is easier said than done, but some recent academic research has attempted to make some headway in answering these questions. A recent study looks at differences in reviews between two online sites for hotels, Tripadvisor and Expedia. The researchers exploited a fundamental difference between these two sites—on Expedia, one needs to have booked at least a one-night stay at a hotel to post a review of that property, whereas on Tripadvisor, anyone can post a review. Thus, fake reviews are more likely to occur on Tripadvisor than on Expedia. Further, the researchers looked at the differences in the nature of reviews across the sites for specific categories of hotels: branded versus independent hotels, properties with small owners versus those with larger owners, etc. To control for differences in the populations that post at the two sites and for differences in the populations that might patronize branded hotels as opposed to independent ones, the researchers worked with the “difference in differences” of reviews between the sites. To do this, they first calculated, for each hotel at each website, the ratio of 5 star (the highest) reviews to total reviews. Then they looked at whether the difference in this ratio for Tripadvisor versus Expedia was higher for independent versus branded chain hotels, whether the difference was higher for hotels that are owned by large owners versus small owners, etc. In this way, they “differenced” away systematic factors specific to the sites and types of hotels.

The study finds that independent hotels with small or single-property owners (the type of hotels the researchers claim have a higher incentive to fake) had a greater share of positive reviews on Tripadvisor relative to Expedia. Additionally, the hotel neighbors of such hotels had more negative reviews on Tripadvisor relative to Expedia. At minimum, the results from the study should caution readers about differences in the nature of reviews that exist across sites.

If fake reviews do exist, how do we tell them apart from real ones? This is an even bigger challenge, although researchers have attempted to address this as well. For example, Myle Ott and his colleagues at Cornell University looked at linguistic differences between reviews known to be real and those known to be false. In this way, one can learn to separate the real reviews from the fake ones. How do they know that a review is fake? The researchers used Amazon’s Mechanical Turk to recruit reviewers to come up with fake reviews for a set of 20 hotel properties. They were then able to look at the patterns in the reviews relative to those in the real reviews posted for those hotels. They find that their method did a better job than humans do in ferreting out fake reviews. Indeed, several companies have sprung up (such as KwikChex) to help companies manage their online reputations by keeping tabs on potentially fake reviews being posted on websites.

One can take the perspective of Cypher (played by Joe Pantoliano), another character from The Matrix, who says: “You know, I know this steak doesn’t exist. I know that when I put it in my mouth, the Matrix is telling my brain that it is juicy and delicious. After nine years, you know what I realize? Ignorance is bliss.” For most of us spending real money making decisions about products or services we buy, understanding the nature of the reviews we base our decisions on is an important step in the buying process. And websites such as Review Skeptic help us figure out whether a review is real or fake.

Pradeep K. Chintagunta is the Joseph T. and Bernice S. Lewis Distinguished Professor of Marketing at Chicago Booth.