The British edition of my book, The Undercover Economist, is out soon, so I won’t be able to keep my eyes off Amazon.co.uk. The function of Amazon is to allow people to buy your book, but it also provides a “sales rank” showing how many books are selling, and it collects readers’ reviews. The sales rank changes every hour and it’s a drug. The reviews, whether good, bad or crushingly indifferent, are nearly as addictive.
The Undercover Economist in me wonders why anyone would write a review and, given that people do write reviews, why anyone would pay attention to one. Writing a review takes time and effort and appears to offer no reward. At the time of writing, about one in 500 buyers of the American edition of my book had left a review. The other 499 hadn’t bothered. This is as economic theory would predict.
That fact has implications for a potential customer trying to interpret the reviews. This is a highly biased sample of reviewers, although it’s not clear quite what the bias is: do people review books when they’re especially pleased or when they’re especially disappointed? Either way, Amazon reviewers are not normal people: one of my reviewers calls himself Drool Clueless – perceptively, I think. Another, Compassionate Conservative, wrote 85 substantial reviews of other books in the 10 days after reviewing mine. That doesn’t seem humanly possible, let alone typical.
One thing is clear enough: since only one in 500 of my readers writes reviews, if I round up 10 friends and persuade them to add five-star reviews, the result should be hard to distinguish from the reviews I would get if I had 5,000 delighted readers.
Frankly, nothing could be easier; the smart buyer will tend to discount positive reviews, knowing that negative reviews are more likely to be genuine. Admittedly, they may not be. I could post a review of Martin Wolf’s book, Why Globalization Works, that says: “This is a terrible book. Read Tim Harford’s brilliant book instead.” But the gains from slating someone else’s book are pretty small in most cases. I’d do better with: “If you liked Tim Harford’s brilliant book, you’ll love this too.” Since bad reviews are less likely to be fictional, if a potential buyer sees a bad review, they should give more weight to it than to a good one.
All this is just armchair reasoning, but it turns out to be correct. A couple of hard-working researchers at the Yale School of Management, Judith Chevalier and Dina Mayzlin, have combined the rankings with the reviews to work out whether reviews do indeed boost sales. Their clever method compares the ranking on Amazon.com with the ranking on the other big online bookseller in the US, Barnes & Noble. Both rankings should go up and down in response to advertising, press coverage and so on, but a good or bad review posted on Amazon should affect sales on Amazon but less or not at all on Barnes and Noble.
It turns out that reviews do make a difference, especially bad ones. Chevalier and Mayzlin reckon that a book with four five-star reviews would drop 20 per cent in the rankings if one of the reviews had been a one-star review instead. Drool Clueless cost me a lot of sales. Although good reviews have less impact, they do boost sales.
So I’m counting on you, loyal readers, to start the buzz. Of course, since these columns are not extracts from the book, you don’t actually know what it will be like. But don’t let that stop you posting a five-star review in eager anticipation.