Authors usually write books that express their inner thoughts and then hope that readers will appreciate their writing. Now, instead of writing from their hearts, authors can write for the bestseller list. Two authors claim to have figured out how to predict which books will land on the bestseller list.
Jodie Archer and Matthew L. Jockers, authors of the soon-to-be-published book The Bestseller Code: Anatomy of The Blockbuster Novel, say they have created a groundbreaking algorithm that has identified the components that grant a book bestseller status. The authors contend that they can predict with 97 percent certainty if a fiction title will hit number one on the New York Times bestseller list, or if it will only hit numbers two through fifteen on the list.
This test for literary success, deemed the “bestseller-ometer,” analyzes the theme, plot, setting, word type, and characters of books to determine how best to appeal to readers.
Features of Literary Success
Archer and Jockers have identified some winning features. For one, a successful novel has two or three central themes dominating around 30 percent of the book. The most popular books in the last thirty years contain a three-act structure and are written in everyday language.
Other factors that seem to resonate with readers are dogs (but not cats), a twenty-eight-year-old heroine and a sex scene or two in every book.
The Bestseller Code is not the first foray into figuring out how to build a better book. In 2014, researchers at Stony Brook University downloaded classic books and used their specially created algorithm to compare the types of words used with each book’s commercial success. In 84 percent of the cases, the algorithm correctly predicted which books would become bestsellers.
According to the Stony Brook research, books containing a high number of nouns and adjectives sell well. Successful books also used a disproportionately high use of the words and and but. Additionally, bestselling books use more verbs that are associated with thought processing, such as recognized and remembered.
When examining less successful books, the researchers found that poor sellers used clichéd words, such as love and extreme/negative words such as breathless and risk. These less successful books also used more verbs and tended to explicitly reference body parts.
Bestseller Prediction: The Future of Publishing?
While many pundits and authors scoff at the ability of machine-driven algorithms to predict bestsellers, some believe that these predictions will drive the future of publishing. Inkitt, a writing platform that invites budding authors to share their stories, has built-in algorithms that analyze reading behavior on the platform to unearth potentially bestselling books. Instead of identifying components that comprise a great book, Inkitt’s algorithms look for books that readers seem to like to read. If many readers on its platform return frequently to the same book, the algorithm recognizes that book as a likely best seller.
Inkitt’s founder, Ali Albazaz, is convinced that predictive data analysis is the way of the future. He may be right, but as many psychologists, marketers, and economists have learned, human behavior can be fickle and difficult to predict.