I’ve written extensively about the role of online reviews in the past, and in particular the rise of astroturfing, or fake reviews to you and me. This multi-million dollar industry has grown to such an extent that New York attorney general Eric Schneiderman settled cases this week with 19 companies, landing them with a collective fine of $350,000 into their fake activities online.
The review found that thousands of fake reviews were posted onto Yelp, Google et al by people from as far away as Bangladesh and the Philippines. The undercover sting even found a company willing to post false reviews for a company that didn’t even exist.
The reasons such activities remain is that reviews are persistently one of the most influential forms of content read online. This was underlined by a recent piece of research emerging from Wharton. The research looks in particularly at the role homophily and balance play in the power of reviews.
The researchers believe that they have found the key to the most profitable use of user reviews. They found that when companies showcase reviews from a customers social group that includes dissimilar recommendations, this yields the highest return. No reviews at all actually saw sales decline.
They arrived at their finding by asking nearly 200 participants to listen to 10 songs before rating each on a scale of 1-10. Each song chosen was of a similar quality (based upon iTunes ratings).
The participants were then played a list of 18 anonymous songs, with no meta information given for them except their average rating on iTunes. The participants were also told that previous participants had rated the songs from both phases of the experiment, after which they were informed how closely their own ratings matched those of both the third party group and their fellow participants.
They were then asked to choose whether to keep the song or receive $1 in cash after choosing the number of reviews to read from the two groups of reviewers (their peers and the third parties).
The results showed that participants accepted the songs 25% of the time, with 10 people consulted on average before the decision was made. The researchers found that information was mainly gathered from like minded people, but that the disparate opinions were nevertheless very useful.
In short, people want to read reviews from people with similar tastes to theirs, but they are also interested in seeing reviews from people in their social networks who don’t have the same tastes that they do, says the researchers. “Typically, retailers have been thinking about one-to-one relationships,” they add. “What we are trying to say is they should also think about how relationships among different friends might impact the information seeker.”
The researchers found that the optimum display is of 8 reviews, 5 of which should come from people similar to the consumer, with the remaining 3 provided by people with less similar tastes. This display would boost profits by 24% over that of a display containing 8 reviews all from similar people.
Of course, the research didn’t touch on whether the similarity of fake reviews matters or not.Original post