Eric Anderson on Deceptive Product Reviews

Deceptive reviews are back in the news, if they ever left: a year-long investigation by the New York Attorney General’s office into “astroturfing”—or attempting to pass off fake reviews as neutral, third-party ones on sites like Yelp and TripAdvisor—recently left nineteen companies on the hook for over $350,000 in fines.

The ubiquity of deceptive reviews and the growing concern about the damages they inflict come as no surprise to Eric Anderson, a professor of marketing at the Kellogg School of Management.

In a recent study—highlighted in our August issue of Kellogg Insight—Anderson and a colleague investigated reviews posted to a retailer’s website by customers who did not actually purchase the product in question. When compared to reviews that could be linked to real purchases, these deceptive reviews were disproportionately negative.

But while deceptive reviews can be written on behalf of a company or its rivals, Anderson’s study reveals that oftentimes deceptive reviews are written by a firm’s best customers. Why? “They’re such good customers that they now feel like they should be guiding the company,” Anderson told Insight. “And when the company makes mistakes in the eyes of the customer, they want to correct them.”

Given the timeliness of the topic, and the number of comments generated by our earlier article, we wanted to follow up with Anderson. Here’s our conversation, edited for length and clarity:

Kellogg Insight: Your study compared online reviews that could be linked to actual purchases to reviews that could not be linked to actual purchases. How do you know customers didn’t just examine products at retail stores, decide not to buy, and then post bad reviews online?

Anderson: In our study, we looked at customers who don’t live near the stores, and thus couldn’t have in-person access to those products from the stores.

But another, related question is, could these customers have purchased the products on e-Bay? We did some additional analyses that didn’t end up in the paper. For example, we found a product category for this company for which there are zero items available on eBay: underwear.  People don’t buy used underwear on eBay. And yet we still see the same results for underwear! To us, that’s pretty convincing that this has nothing to do with eBay.

For our study, we could be reasonably confident about which reviewers actually purchased the product. But on other platforms, it’s much harder to know what’s going on, because you can’t really link up products with customers very easily. You don’t really know whether they’ve bought anything. This is actually a really interesting problem for manufacturers. A big thing that firms worry about is, well, what are people saying on all these different review sites? Because there’s nothing to stop them from speaking if they didn’t buy the product. The sites are just platforms for talking about your brand and sharing your opinion.

KI: I’m thinking now about Texas State Senator Wendy Davis. For thirteen hours, she filibustered a bill that would have restricted abortion rights—all while wearing a pair of bright pink sneakers. Now the Amazon page for these sneakers has been besieged by mock “reviews” that are about politics, not the shoes. It just goes to show how quickly companies can lose control of the way their brand is depicted.

Anderson: And how far reviewers can push the envelope. There was a guy who used customer reviews to invent a fictitious restaurant in the U.K. He went on TripAdvisor and started writing reviews for this restaurant where you ate on a boat and they would catch you fresh fish. The entire thing was made up. People show up to where the restaurant is supposed to be located, but it doesn’t exist!

KI: One of our readers wondered about the effects of log-in policies. Are log-in policies effective at weeding out deceptive reviews?

Anderson: Yes. Log-in policies are very analogous to verified purchases at Amazon: Amazon verifies whether you buy an item from them. In another study in our paper, we compare the distributions of reviews written by verified buyers and unverified buyers and find that the verified buyers are much more positive. About 10% of the unverified buyers give a product a 1, the lowest rating, but only 5% of the verified buyers do.

KI: Your research finds that deceptive reviews result in revenue losses of 1–2% for firms. But couldn’t this be offset by all of the free publicity? At the end of the day, people are still talking about your brand.

Anderson: Some people believe that all statements about your brand, at some level, raise awareness. So in that sense awareness should be good. But it’s hard to believe lots of negative awareness is good for you.

Then again, think of the recent Miley Cyrus MTV Video Music Awards performance. Did that hurt her brand? It generated tons of negative attributions amongst the Disney crowd, but the amount of awareness she got was ridiculous. She got millions of hits on YouTube. She’s on the front page, and in Jay Leno’s monologue. Her awareness factor is huge, even though, for some people, it’s completely negative. It probably turned off an entire segment of young families! But that’s the trade-off for her.

Photo credit: Health Gauge