Reader Analytics: Not All Authors Want To Know


Your Soul Vs. Data?

When Jellybooks’ Andrew Rhomberg wrote at Digital Book World recently about publishers having a Fear of Data in an age of digital metrics, I thought that taking that issue to authors would be a worthwhile exercise. And I tried it out on some trusted colleagues by making it the “Provocations in Publishing” topic of my column last week at Writer Unboxed: Who’s Afraid of Reader Analytics?

Answer: a lot of readers are fearful of data, just as publishers are.

They may not want to say “afraid.” But skeptical? Hell, yes. And dismissive? Probably too much so, in many cases.

“Can it make us unhappy to hear that someone hated the sentence we thought clever or thinks our characters lack authenticity? Sure. But if the point is to grow, then this is valuable feedback.”
Therese Walsh, author, Writer Unboxed co-founder

We’re not in entirely uncharted territory here. We’ve had some idea for while, at least in broad strokes, of the Amazon Kindle system’s ability to track and register various parts of a reader’s experience — such as where in a book a reader may stop reading a book and which lines he or she highlights.

Rhomberg, whose start-up, Jellybooks, pivoted a while ago from its original book-discoverability storefront-window focus to reader analytics, wrote at Digital Book World:

A DJ will gauge response based on whether people are dancing or not, and a stage actor can measure what the audience thinks based on the applause.

Authors to date [have] only had monthly sales data as a similar barometer, which in many ways can be a very poor indicator. There are professional reviews, but those are the opinions of an elite few. Goodreads may give a broader measure, but those who review books are often at the extremes—people who love a book and people who hate it. The majority of readers never review books.

Reading data provides feedback on those 98 percent, and it can give that feedback before a book is even published by using advance reader copies with tracking software. The data therefore gives authors the opportunity to understand what the audience reaction might be. It might influence them to be even bolder, or it might make them realize they overstretched and pushed too far ahead of their audience.

He’s right, of course, that historically, both writers and their publishers have worked almost entirely on gut instinct. Hunches about which book might sell well and which might not are just that, even now: hunches. And those hunches are notoriously mixed in terms of how well they predict success.

As digital brings more and more capability for actual metrics into view, even artists — at least those who are emotionally secure in their work and not threatened by input — may begin to welcome some better views into what’s happening. And waiting until after the painting is sold, after the film ticket is bought, or after the book is downloaded can be disastrous.

“Reading data…might influence authors to be even bolder, or it might make them realize they overstretched and pushed too far ahead of their audience.”
Andrew Rhomberg, Jellybooks

In 24-hour television news, it’s sometimes said that “you can’t land the plane” to work on things. Programming must continue nonstop, that’s the brand promise. And so tinkering with schedules, formats, on-air lineups, and so on, all must be done “in the air” while “on the air.” You can’t just turn off the lights, put up a “pardon our dust” sign, and fix things.

But the kind of analysis that Rhomberg’s service offers publishers is different. It can happen well in advance of sales efforts.

Publishers, he told us in a subsequent #FutureChat with The Bookseller’s The FutureBook, are engaging Jellybooks to test a book on a focus group of readers as much as 20 weeks ahead of launch. At that point, the plane is not flying yet and a publishing team has ample chances to assess what data an analytical program produces and apply it, if desired.

For example, if a book’s ending takes a beloved character over the cliff and that makes readers livid (anybody want to mention an HBO parallel in Jon Snow’s death in Game of Thrones?), then better to know it ahead.

That does not mean, by the way, that the ending must be changed. If it’s somehow artistically integral to the work, what the early warning device of this kind of reader analytics might provide is a chance for the marketing team to rethink those sunny posters and newspaper ads they’ve been working on. Perhaps there are ways to signal to the readership-to-come that things might get a little serious at the end. Blurbs can be written in such a way as to make dark doings a potential part of the read and road ahead.

As logical as all this might sound, however, in responses to the column at Writer Unboxed, resistance was not unusual.

‘A Depthy View’

Several respondents were generously interested in learning more about the potential of deeper analytics from controlled studies of reader reactions.

Writer Unboxed co-founder and author Therese Walsh, for example, wrote:

We all write from within the comfort of our own bubbles. It’s good to know what the world outside of those bubbles thinks of whatever it is we have developed from within. Can it make us unhappy to hear that someone hated the sentence we thought clever or thinks our characters lack authenticity? Sure. But if the point is to grow, then this is valuable feedback. Once you see trends in feedback, you can use that information to become better at reaching readers with the next efforts.

That’s an enlightened view, and, as Walsh went on to point out, we need a “depthy view” of what’s going on, something that reveals as much rationale as pure reaction.

“At Jellybooks, we recently developed a piece of code called candy.js, which is embedded inside an ebook to track how users actually read. Penguin Random House UK was among our earliest partners in a pilot program of the technology.”
Andrew Rhomberg

Many more writers, however, were dismissive and in ways that might indicate a shrug more of nervous denial than of thoughtful evaluation.

Here are a few comments, excerpted and without names:

The only analytics that I pay attention to is how many readers purchase my books and whether the good reviews outnumber the bad ones by a good margin.

My response to that one is that it’s pretty late in the day to be getting one’s feedback if it’s in the form of reviews, presumably consumer reviews. Wouldn’t having some prior understanding of reactions be helpful?

Here’s another:

For me, reader analytics are like political polls. They give you a sense of what’s working. If I paid too much attention to those dynamics, I wouldn’t be writing.

Hm. I hope that reader votes, at least, when election time comes around, and doesn’t stay home because polls were taken.

And another:

Yes, you might hook a novel onto something that is “current” but unless it is driven by your writing soul and not data, I would question if it would truly be your art.

Fine, but the question isn’t about changing your soul or your art. The question is about finding out how all that artful soul is landing in the world — so you have an idea of whether you’re getting across.

‘100% Completion Rates…Rare Outliers’

In an earlier piece, What Code Is Revealing About Readers, Rhomberg took some heated commentary at the Passive Voice blog. (Where rarely is heard a soft-spoken word unless it’s Passive Guy David Vandagriff, himself, writing.)

Rhomberg had written:

At Jellybooks, we recently developed a piece of code called candy.js, which is embedded inside an ebook to track how users actually read. Penguin Random House UK was among our earliest partners in a pilot program of the technology, and the insights we gathered were fascinating. The question now becomes what story this data tells us and what impact it might have.

The code was only used in ARCs, advanced reader copies provided free of charge by PRH to various reviewers, bloggers, etc. As Rhomberg relates:

Volunteer readers were clearly informed about the program and were asked to share their reading data in exchange for the free book. The technology has never been used in retail copies to date and, if it were, it would be clearly labeled as such. Also the data has to be actively submitted by the reader (by pressing the sync button inside the book) or it stays locked up in the ebook reading app. Few tracking technologes give the user that level of control.

Those are important points, lost, of course, on people determined to be hotheaded. And we have a few such folks in the author corps, no analytics needed to know that.

Personally, I like this comment from the author Mirtika:

I just want Amazon to offer authors that info about reading habits–especially things like where they stop reading…I suspect some authors would happily pay for that, if it’s inexpensive. That’s valuable stuff.

Sensible, and she’s not the first author I’ve heard say they wish they could find out what the Kindle team might know about how readers are reacting to their work.

For the most part, however, the idea of Rhomberg’s Jellybooks reader research — focus groups amped up with digital detection — seem to get a cold shoulder from the creative crowd.

Probably very few writers want to hear Rhomberg tell them:

100% completion rates are extremely unusual outliers.

People aren’t finishing books.

Are we surprised? And the main concern I hear from most authors is that if a digital reader analytics system indicates that readers are becoming confused or put off at one or another point in a book, the system won’t tell anyone why. I’ve asked Rhomberg about this, because it doesn’t really hold water for me. If, say, 35 percent of your readership is stumbling at some point in a story? — it won’t be that hard to figure out why. Rhomberg says that slow, tedious starts to books are the biggest turn-off. He adds that when the data around a problem area is presented to editors, they tend to know exactly what to do with it. When it’s presented to authors? — not so much.

However digital we get,  it seems that we might still be battling the “Oh, to see ourselves as others see us” conundrum. The bravest among us may be the ones willing to view, parse, and evaluate reader analytics with our egos on ice for a bit. I’m interested in knowing what percentage of how many creative types can do that.

Data! We cry for it until it’s about us. Then, no, no, no, we don’t need it, we don’t want it.

It’s a long digital disruption, isn’t it?