Products with more reviews sell more, but that could be sales causing reviews. Researchers have found three ways around the problem, and all three say reviews cause sales.

Same books, two stores

In 2006 Judith Chevalier and Dina Mayzlin compared the same books on Amazon and Barnes & Noble’s website. Anything about a book itself, its author, its marketing, its cover, is the same on both sites, so a change in reviews on one site but not the other isolates the effect of the reviews. An improvement in a book’s reviews raised its relative sales on that site. And the effect was lopsided: a one-star review hurt more than a five-star review helped Published.

Reviews were also overwhelmingly positive: 67% of reviews on Barnes & Noble were five stars, and 53% on Amazon Published. Hold that thought for chapter 5.

Different designs, same answer

• Restaurants. Michael Luca used the way Yelp rounds ratings to the nearest half star: restaurants either side of a rounding line are nearly identical but show different stars. A one-star increase led to 5% to 9% more revenue for independent restaurants, and no measurable change for chains Reported (Harvard Business School working paper).
• Amazon products. Sherry He, Brett Hollenbeck and Davide Proserpio tracked products that bought fake reviews in Facebook groups and used a temporary change in Amazon’s policy to show that rating manipulation had a large causal effect on sales Published. That’s why the market for fake reviews exists. The same paper found what happens next, in chapter 6.

A one-star review costs more than a five-star review earns. Plan your review system around the bad ones.

The first five

Northwestern’s Spiegel Research Center analyzed data from PowerReviews, a review platform, across many retailers. A product with five reviews had a purchase likelihood 270% greater than the same kind of product with none. The benefit of each extra review fell off quickly after the first five, and showing reviews lifted conversion more for higher-priced products Reported (vendor data, analyzed by a university research center).

Treat the exact numbers with care. This is observational data, and products that collect reviews may differ from products that don’t. But it fits the peer-reviewed work: Ewa Maslowska, Edward Malthouse and Vijay Viswanathan found, using data from two retailers, that the rating’s effect on purchase was strongest when a product had many reviews, when the shopper read them, and when the product was expensive Published. For a DTC brand that means the review gap that costs most is the one on new and expensive products.

So coverage beats average: the reviews worth most are the first five on products that have none. Show stars, count and a link near the price and the button; The Honest Test covers the product page.

Do this

• Pull revenue by product with each product’s review count (Appendix A). Mark every product under five reviews. The revenue on that list is the size of your review gap.