Moneyball for book publishers: A detailed look at how we read

Andrew Rhomberg wants to be the Billy Beane of the book world.

Beane used analytics to transform baseball, famously recounted in “Moneyball,’’ a book by Michael Lewis. Now Rhomberg wants to use data about people’s reading habits to radically reshape how publishers acquire, edit and market books.

“We still know almost nothing about readers, especially in trade publishing,’’ said Rhomberg, the founder of Jellybooks, a reader analytics company based in London.

While e-books retailers like Amazon, Apple and Barnes & Noble can collect troves of data on their customers’ reading behavior, publishers and writers are still in the dark about what actually happens when readers pick up a book. Do most people devour it in a single sitting, or do half of readers give up after Chapter 2? Are women over 50 more likely to finish the book than young men? Which passages do they highlight, and which do they skip?

Rhomberg’s company is offering publishers the tantalizing prospect of peering over readers’ shoulders. Jellybooks tracks reading behavior the same way Netflix knows what shows you binge-watch and Spotify knows what songs you skip.

Here is how it works: The company gives free e-books to a group of readers, often before publication. Rather than asking readers to write a review, it tells them to click on a link embedded in the e-book that will upload all the information that the device has recorded.

The information shows Jellybooks when people read and for how long, how far they get in a book and how quickly they read, among other details. It resembles how Amazon and Apple, by looking at data stored in e-reading devices and apps, can see how often books are opened and how far into a book readers get.


Jellybooks has run tests on nearly 200 books for seven publishers, one major U.S. publisher, three British publishers and three German houses. Most of the publishers did not want to be identified, to avoid alarming their authors. The company typically gathers reading data from groups of 200 to 600 readers.

Rhomberg recently gave a workshop at Digital Book World, a publishing conference in New York, and some of his findings confirmed the worst fears of publishers and authors.

On average, fewer than half of the books tested were finished by a majority of readers. Most readers typically give up on a book in the early chapters. Women tend to quit after 50 to 100 pages, men after 30 to 50. Only 5 percent of the books Jellybooks tested were completed by more than 75 percent of readers. Sixty percent of books fell into a range where 25 percent to 50 percent of test readers finished them. Business books have surprisingly low completion rates.

How Will Publishers Use the Data?

For the most part, the publishers who are working with Jellybooks are not using the data to radically reshape books to make them more enticing, although they might do that eventually. But some are using the findings to shape their marketing plans.

For example, one European publisher reduced its marketing budget for a book it had paid a lot of money to acquire after learning that 90 percent of readers gave up after only five chapters. A German publisher decided to increase advertising and marketing on a debut crime novel after data showed that nearly 70 percent of readers finished it.


Publishers might also use the data to learn what type of reader a book appeals to, and market it accordingly. One of the novels that Jellybooks tested was written for teenagers but proved surprisingly popular with adults.

Why This Is Scary for Authors

Authors are understandably nervous about how new insights into reading behavior might shape publishers’ editorial decisions. Suppose you are writing a crime series, and readers gave up halfway through the latest installment. Publishers might not want to buy the next one. Or what if readers skip around in your nonfiction book, a common way to read nonfiction? An editor might want to cut the chapters people are skipping, potentially erasing useful context.

And, of course, people who sign up for a free e-book service might not represent the kinds of readers who would seek and pay for a crime novel, or a nonfiction book on a subject that interests them. The sample sizes are relatively small. So the reader data might not represent the reactions of a general audience.

Why This Might Be Troubling for Readers

Jellybooks users consent to having their data tracked in exchange for free books, and must click on a link to send that data to Jellybooks. “It’s absolutely critical to have the user in control, or we face a backlash,’’ Rhomberg said at the conference.

But as the book industry gets more sophisticated about how to measure reading behavior and the practice becomes more widespread, real privacy concerns could emerge. Some big academic publishers already track how students read their e-books, and some e-book subscription services have developed the capacity to measure reader engagement.


Regular e-book readers might not realize that digital retailers are recording and storing the data. A few years ago, California instituted the “reader privacy act,’’ which made it harder for law enforcement agencies to collect information on consumers’ digital reading records from e-book retailers. But most other states have not taken such steps.

But Having Companies Reading Over Your Shoulder Is Probably Inevitable

Analytics are transforming every corner of the media world. Digital publishing cannot be far behind in measuring consumer behavior. As Rhomberg put it bluntly at Digital Book World, sales figures are not enough to gauge a book’s success. “Ask yourself,’’ he said to the audience. “Are you in the business of selling paper?’’

Jump To Comments


This discussion has ended. Please join elsewhere on