Tuesday, September 4, 2012

John Locke's Amazon Con

Bestselling self-pub author John Locke
Much has already been said about the NYT article on the "entrepreneur" Todd Rutherford, who built a business selling four and five-star Amazon reviews to self-published authors.  The most widely-discussed revelation of the article is the fact that author John Locke freely admits to purchasing hundreds of reviews from Rutherford at a cost of thousands of dollars.

Most people discussing this article seem to assume that Locke's bogus reviews were a fraud against his readers, and since many people believe they can sniff out bogus reviewers, some readers haven't expressed much concern.  As many people point out, the trustworthy reviews stand out when you read them.  But Locke didn't care about the content of his reviews. He was interested in having a high star rating across a large number of reviews.  He wasn't trying to fool readers; he was trying to fool Amazon's computers.

Over a period of many years, Amazon has collected a great deal of customer purchase information, and used that to construct a powerful sales apparatus.  There are a bunch of "Recommended for you" books on the front page of the site, and there are lots of "people who bought x also bought y" lists it shows you as you peruse the site.  Books that get onto those lists get more sales, and then they get onto various genre bestseller lists, which drives more traffic to the books and boosts the sales further.

Amazon wants to show you stuff it thinks you'll want to buy, including stuff you might not already know about. However, since it sells over 8 million books and millions of other products, its human employees can't put their eyes on every product in its store and decide which ones to recommend, the way a bookseller in your local bookstore might.  Instead, Amazon's computers make predictions about what books or other products might interest you based, in large part, on what other people are buying or talking about (particularly other people whose past sales have demonstrated taste similar to yours).

If Amazon's computers catch a product becoming a nascent trend or a book breaking out, they're designed to recognize that and amplify the sales by showing the book to people who are likely to buy it.  In an environment in which thousands of books are trying to break out of obscurity, that kind of exposure is priceless. So a great way for a self-published author to become popular is to trick Amazon into thinking that he's already popular.

Obviously, if people knew how to do this, everyone would be doing it, so the operation of Amazon's recommendation algorithms is a closely held secret.  But Locke made an expensive bet that he could influence Amazon by buying lots of customer reviews and that bet paid off.  None of the safeguards Amazon had in place thwarted Locke and Rutherford.  The fake reviews came from all over the country, because Rutherford was farming the review work out to freelancers.  The reviewers were all "verified purchasers" because Locke's books only cost $0.99, so he could easily pay an extra buck to each reviewer to buy his e-book (which also spiked his ranking).

Locke had many other mechanisms of self-promotion; he's unquestionably talented at getting his name and his books in front of readers.  If buying fake reviews was what got him favorable placement on the Amazon website, it was probably a much more effective investment of money, in terms of traffic to his book page and overall sales, than purchasing Facebook or Google ads to promote a book.

Phony Amazon reviews are now as cheap as $5 on sites like Fiver.  According to the NYT, Locke paid $20 for each of the 300 he bought from Rutherford. The good news is that, as fake reviews have proliferated, Amazon has changed its recommendation algorithms to reduce the benefit authors can gain from spamming the site with fake reviews. A hundred purchased reviews will no longer boost an author's standing in the recommendation system the way they did for Locke.