James Patterson lashes out at the New York Times bestseller list

Patterson is annoyed that his latest nonfiction book, Walk the Blue Line: No Right, No Left (police tell their true stories to James Patterson), did not make the New York Times bestseller list. According to Circana BookScan data, he says, the book outsold most others featured on the list. He wrote a letter to the New York Times to complain, but they did not run it, so he posted it on social media. It says, in part, “… you suggest your process has statistical rigor. And it simply doesn’t.”

Of course, everyone in the publishing business, especially self-publishing authors, knows that the NYT list is curated and not driven by raw sales data. This is one reason why the disappearance of the USA Today bestseller list was such a blow, as it did not curate and relied on available raw sales figures. On the flip side, BookScan data is also limited in scope, as it only reports print copies sold through retail. BookScan itself says it captures about 85 percent of trade print book sales in the US. Much depends on the type of book; BookScan does not, for example, account for library or direct-to-consumer sales, which can be significant.

Agent Anna Sproul-Latimer commented on Twitter, “It’s true that the NYT weighs titles through a mysterious algorithm that seems to favor indie [bookstore] sales and depresses bulk / Amazon / big-box, which in itself *likely* depresses the ranking of titles heavily skewed toward Red America markets.” She also mentions that BookScan operates via algorithm, but that isn’t the case; it records cash register sales. For ebook sales, BookScan relies on publishers’ own reporting.

The most important point is that it’s exceedingly challenging to get accurate book sales figures, and you should assume any sales figure (if not reported by the publisher directly) has an asterisk. Even then, sales figures can fluctuate due to returns. In 2017, Michael Cader attempted to quantify the entire US market because, he wrote, “All of the core publishing statistics are incomplete in various ways.” (We summarized his findings here and here.)