The Effect of Word of Mouth on Sales: Online Book Reviews

Sales increase with the number of 'stars' awarded by online reviewers. However, the decrease in sales associated with a negative customer book review is greater than the increase in sales generated by a positive review.

Online booksellers provide customers with reviews and rankings of the books that they offer for sale. In The Effect of Word of Mouth on Sales: Online Book Reviews (NBER Working Paper No. 10148), authors Judith Chevalier and Dina Mayzlin use a snapshot of sales data from two large online booksellers to determine whether customer comments posted on a bookseller's website influence a book's sales by that retailer. After controlling for differences in the price of a book and its promised delivery time across the two sites, they examine whether differences in the reviews across the two sites influence relative sales. They find that sales increase with the number of "stars" awarded by online reviewers. However, the decrease in sales associated with a negative customer book review is greater than the increase in sales generated by a positive review. The results also hold when the researchers use a "differences in differences" approach: that is, changes in the relative enthusiasm of the customers' reviews of a book across the two sites between two points in time are associated with changes in the relative sales of the book at the two sites. The authors conclude that these results are consistent with the idea that "customer word-of-mouth has a causal impact on consumer purchasing behavior." Although their evidence is insufficient to show that a retailer profits from providing customer reviews, they nonetheless conclude that "customers certainly behave as if the fit between customer and book is improved by using reviews to screen purchases."

For this study, the total sample of 2394 book titles is comprised of two sub-samples: a random sample of book titles that were published between 1998 and 2002; and books that appeared on the Publisher's Weekly bestseller lists from 1991 to 2002. Sales of the books at the two sites were inferred from the websites' sales rankings, collected in May and August of 2003. To be included in the sample, a book's most popular format had to be the same at both websites, and its sales had to be above a specified level.

One interesting puzzle is why customers put time and energy into providing these reviews. At one online retailer, 54 percent of the titles in the sample had at least one customer review. At the other, the number was just 13 percent. On average, customer book reviews are quite enthusiastic. Over 50 percent of the reviews on the online websites awarded the highest possible ranking to the book reviewed. Fewer than 10 percent of all reviews assigned the lowest possible ranking.

-- Linda Gorman