Blog Post

Whoever Builds A Good Tool To Help Us Find New Books To Read Will Get Rich

Kevin Smokler is the co-founder/CEO of, an Amazon-funded startup which offers affordable software tools for authors. He is based in San Francisco, and can be found on on Twitter at @weegee.

I can think of almost 20 startups — all with funding — whose primary aim is to improve “book discovery” online, and I’m sure there other companies that I’m not even aware of.

That’s a considerable rush of investment for book publishing, an industry usually wall-flowering at the innovation ball. The entrepreneurial momentum around “discovery”-where publishers suggest unfamiliar but useful content to a user based on that user’s previously stated interests-suggests an important energy and excitement around key aspects of the book business’s future.

If discovery is the next frontier of the content business, music has begun laying out streets and building towns, the book business is in a land rush, and the movie and TV businesses are just starting to pack up their wagons. Music services are clearly at the forefront of cracking the discovery nut. That industry has had discovery as a feature as far back as the LaunchCast Player in 1999, and as the basis of Pandora’s entire business in 2005. (Prepare for another huge shakeup in this zone when Spotify lands in America.) Film and television have Netflix’s collaborative filtering technology (a feature, albeit a core one) and new kid-on-the-block Jinni is billing itself as “Pandora for movies.” But that’s about it.

The visual entertainment space has yet to leverage discovery around new releases in the way that music has with the new Nomis App (it scans your iTunes Library and alerts you when those artists have new albums out) or publishing with the Any New Books email-notification service (Any New Books?). The day I can find out what movies are coming out without sitting through advertisements, or know what in the new fall TV lineup is me, as a fan of Criminal Minds, is the day I move filmed entertainment out of last place. But we’re not there yet.

The rapid adoption of e-books provides a unique opportunity. According to Forrester Research, digital sales of books grew 176 percent in 2009, and they’re predicted to cross $1 billion in sales this year. Meanwhile, the New York Times (NYSE: NYT) recently added an ebook-only bestseller list. An addictive, intuitive discovery tool could tap into this rising tide. Imagine if the next time you’re shopping for an e-book, your favorite retailer, through partnerships with your favorite discovery service, knew not only the last book you read but the last five you read on vacation, how fast you finished them, and if they were better reads for a trip to the beach or a weekend of mountain climbing? It’s safe the say that kind of service would fuel an additional pop in e-book sales.

The race to build this better discovery tool for readers is not only on — it is crowded. For the companies leaning hard on engineering, like Discoverreads and WhatShouldIReadNext, book data from the user (usually beginning with books they’ve already read/liked)) is crunched against preference patterns created by other users and run though a proprietary algorithm. Others services create a social experience around reading itself (BookGlutton with online annotation, Copia with discussion while reading or afterwards), and view discovery as a secondary benefit of participation. More general recommendation services like LivingSocial and GetGlue are hybrids — treating your new favorite book, video game or beer as a set of fertile data but only good for growing recommendations when planted alongside those of your friends and extended social network.

Book discovery is as necessary for the publishing business as it is difficult for the startups to solve. Readers seek out books for four primary reasons: familiarity with the author; interest in the subject; a recommendation from a trusted source; or hearing about it through the media. But readers state their book-reading preferences via dozens of smaller criteria-price and format, genre and setting, length and likeability of the protagonist or point of view.

Hitting a bull’s eye with one of the top four, particularly with fiction, particularly with a debut or lesser-known author, is tough. But toughest of all is to drive sales based on recommendations– recommendations not just from friends but from other sources as well. That’s because recommendations take longer to bubble up, can’t be created by spending more money, and in some cases, never happen at all. People recommend books based on those dozens of smaller criteria/data points above. Any tool that does book discovery well has to take most, if not all, of those criteria into account. Another natural disadvantage for book discovery: It takes more of a user’s time to sample a book than a song or a movie trailer.

To illustrate my point about the shortcomings of discovering as it now exits, I decided to run my favorite book– And the Band Played On by Randy Shilts, the defining political history of the AIDS epidemic–through a few book-discovery services. The tool What Should I Read Next apologetically returned only three results (the first was Ken Kesey’s One Flew Over the Cuckoos Nest) with the explanation: “This is probably because your search title/s don’t appear in many users’ booklists.” The site Discoverreads returned a decent assortment of gay-friendly fiction (Angels in America , A Home at the End of the World, etc.) and investigative journalism (We Wish to Inform You by Philip Gourevitch) but no book younger than 10 years old. GetGlue returned Mr. Shilts other two books and three other titles about AIDS–all logical, if obvious, suggestions. If I were a publisher, I’d have wanted the bestselling Emperor of All Maladies: A Biography of Cancer which came out this past fall, suggested. But that suggestion requires, at minimum, a leap across author, publisher and era.

How much readership data is currently being collected? All of the startup companies have access to elemental book data (author, publisher, category, length) that is collected by one of several competing cataloguing systems (Library of Congress, OPACS, Amazon) and accessible via API. But things gets significantly more challenging as you move away from data that is provided as a matter of course upon a book’s publication and into less standardized indicators of reader preferences. Put simply: how do create a data set around “Books to read following a death in the family?” or “Paperbacks that fit into a back pocket?” Those intangibles are the many minotaurs in the labyrinth.

It is fair for publishers to ask if the current discovery tools are good enough to grow sales – in other words, whether spending precious marketing dollars and time cultivating relationship with new tools will really help sell that many more books. But that is myopic. When it comes to deciding which book to read next, most people today are, at best, uncertain and, at worst, lost. When there isn’t an easy answer to that question, many book consumers decide simply to do something else with that free time–play another game of Angry Birds, or have dinner with friends, or go to bed early- all leisure activates that don’t require lengthy explanations of their value.

Better discovery tools are the key to allowing books to compete with these other activities on an equal footing. When I want to know which song to listen to next, Pandora already knows my tastes and can tell me with one click. It will be a glorious day for readers when, with that same level of sophistication, we have similar tools for books.

This article originally appeared in

12 Responses to “Whoever Builds A Good Tool To Help Us Find New Books To Read Will Get Rich”

  1. Sparechange: World Cat is still, from a design and UI perspective very instutional and customer-unfriendly. When I can find the buttons that say “import your pre-existing book data here” I’ll think its for me as an individual user.

    Lorena: I think your assessment is quite accurate. We’re not quite there yet but on the eve…

  2. That is quite true, Kevin (although the development of virtual/IM reference services is making that less of an issue). I am fascinated by this kind of technology, not being very good at reader’s advisory myself – I’m very interested in your work. Recommender systems are not quite there yet I think because there isn’t quite enough user contribution (consistent tags, etc) to really get a network effect going, but you can see that it’s on its way…The WorldCat Genres project is very exciting too:

  3. Martha Lorini

    Discovery is the important engine when you consider the following:
    1. Big retailers, having put independent booksellers out of business, are on the brink of going out of the book-selling business themselves…
    2. Which means publishers will need to learn how to sell directly to consumers–something they know squat about now, since they ceded that territory to B&N, Borders, Amazon.
    3. In 2009, traditional publishers produced 800m titles…flat with the prior year. Nontraditional publishers (e.g. self publishers) produced 900m, a 181% increase over the prior year. Assuming that even 9/10ths of the self-published books have no commercial aspirations, this is still a vast & growing (in legitimacy as well as volume) list with virtually no way to market broadly.

    The answer is a digital startup I’m developing that uses social & algorithmic & aggregated content to help readers source books (regardless of reading medium) accessible mobiley & online.

  4. Sue,

    All true. I think the solution is a sweet spot of algorithmic data, pre-existing metadata and human intervention in the form of friends recommendations or user tagging. We’re not there yet but the economic energy around the problem tells me we’re close.

  5. Sue Sparks

    Good article. One of the main constraints on recommendation engines is scale which is where Amazon scores, though its recommendations are still fairly off-beam much of the time, and Shelfari is, for some reason, rubbish compared with LibraryThing in spite of being owned by Amazon. I was interested a few months ago when Waterstone’s bookshops in the UK featured favourite reads by famous authors; unfortunately for them, when I went on Amazon for one title, it immediately referenced all the others. Sad as Waterstone’s had done the hard part. As far as e-book discovery goes, a key problem is that the metadata is poor and there is no central, quality-controlled supply of it, unlike the bibliographc data for print books.

  6. Robotech_Master

    A truly great resource for helping people find new books to read has existed for years, and it still beats the pants off of any other in terms of accuracy—even though circumstances have forced it into a relatively slow, low-capacity version of itself for the last few years. was one of the first book recommenders (and e-book stores), but its corporate owners never really recovered from the dot-com crash. Nonetheless, it’s helped me find dozens of great books over the years.

  7. Where I come from, one tool to find good new books (or good older books) is called a reader’s advisory librarian, and you can find them at the public library…(but I’m following these technologies and options with lots of interest since I like to rely on multiple options: Amazon, LibraryThing, and more)

  8. La Taco: I love GoodReads and yes, probably should have mentioned them. Only so much space :)

    Aaric: I certainly don’t think discovery is the only arrestor in book sales. But the statistics are overwhelming in music for example, that increased access leads to increased sales. And fundamentally, I think discovery is about knowing what else is out there and available to someone with your reading tastes when you don’t live next door to a giant bookstore.

  9. Is there really a business here? Assume that people currently read x books/month. If discovery were fixed by one of these new companies, how many books would people read? Is there anyone that is reading only half as many books as he might if discovery were better? It’s not at all clear that discovery is really the arrestor to growing book sales.