Did Random House’s free online book releases affect sales?

john hiltonA doctoral student crunches the numbers to determine whether releasing books through Creative Commons and other methods really does benefit authors and publishers.

John Hilton admits up front that there would be no way to definitively prove one way or another whether releasing a book for free online will help boost print sales of the same title — he called such absolute statements “irresponsible” — but he is nonetheless trying to track down as many hard numbers as humanly possible to try to argue whether there’s a positive effect. Hilton is a doctoral candidate in Instructional Psychology and Technology at Brigham Young University whose interests focus on open education and open access. For his doctoral thesis he’s entering the debate that has reached a crescendo in recent years as Google strives to place millions of books online, all freely available (in some cases, only portions of the books are accessible) to its users.

For his research, he has located, so far, approximately 40 book titles for which publishers have released free online versions at least eight weeks after releasing the printed version. He does not consider books that were released both simultaneously for free online and as print products because then he wouldn’t be able to observe the before and after effects on sales. He then records the Bookscan numbers — which account for about 70% of all US book sales, including those sold at most retailers — for the eight weeks prior to the free release and the eight weeks after.

On March 4 of this year, Random House announced that it would release five books for free through its science fiction portal, all of which came in downloadable PDF files (among other formats). Hilton recorded the before and after book sales and found that “one of the five books has had zero sales in 2009. So no sales before or after the free version. But the other four books all saw significant sales increases after the free versions were released. In total, combined sales of the five books were up 11%. Together they sold 4,633 copies the 8 weeks prior to being released free and 5,155 copies the eight weeks after being released.”

So an uptick of about 500 books. As noted previously, this provides some evidence that there might be a positive effect, but it’s impossible to account for all the other factors that may have come into play during those eight weeks. Perhaps one of the authors was featured by a major media outlet or the publisher launched a signing tour. The possibilities of outside factors are so numerous that it’s almost impossible to create a controlled study on the matter.

“For example, with Random House, if they hadn’t released those books for free, we couldn’t prove that sales wouldn’t have increased 60 percent but because of the free promotion they only increased [11] percent,” Hilton said. “You just can’t come to a firm conclusion because there’s no way to tell. So I think it would be irresponsible to say that we could definitevely show causation, especially in some cases, like with Tor Books, where the numbers show something different.”

Early last year, in an effort to get people to sign up for its e newsletter, Tor Books, the largest science fiction and fantasy publisher, began releasing free ebooks of its titles for those who signed up for its mailing list. Though the platform wasn’t as open as some other free book promotions — in which the publisher releases a book under a Creative Commons license — those who were on the email list received a link to where they could download that week’s free title. Many of the books that were featured in the promotion had been in print for some time (years in some cases) beforehand.

When Hilton pulled up the prior and post eight week numbers, he was astonished at what he found. “When you looked at eight weeks before and after, 20 of the 24 books that Tor gave away saw decreasing sales,” he said. “To me that’s a real puzzle. Random House was four for four with increasing sales, and Tor was 20 for 24 decreasing?”

For the eight weeks prior to the free releases, the Tor books sold 21,824 copies while in the eight weeks after they sold 16,556, a decrease of 5,268, or 26%. Of course many of the books had been out for some time and may have not been available anymore in most retailers. Also, most the books, Hilton said, only showed a difference of a few copies in the two time periods.

“This also doesn’t quantify the fact that Tor received 10,000 or 50,000, or who knows how many people signed up for their newsletter,” he elaborated. “That’s also valuable, that maybe if they said up front that, ‘hey we know that we’re going to lose a few thousand in sales, it’s worth it to us to have this massive email list.”

These are all unknowns, and these unknowns leave enough wiggle room so that proponents of both sides of the argument have plenty of leeway to argue why releasing a book for free — whether it’s through a Creative Commons license or Google Books — has a net benefit or detriment to sales. To try to close the gap on this wiggle room, Hilton will continue trying to find more books to add to his study — he said that finding 50, 100, or even 200 would greatly inform his findings — so that future authors and publishers will have at least some conclusive data before deciding they too would like to make the online book release plunge.

12 Comments

  1. Andrew Karre Says:

    Interesting study. I would love to see this piece of viral conventional wisdom die out on the web, though:

    “He then records the Bookscan numbers — which account for about 70% of all US book sales, including those sold at most retailers — for the eight weeks prior to the free release and the eight weeks after.”

    This is simply not true (despite the near constant repetition in the New York Times). There are lots of categories of book publishing where Bookscan does not report anywhere near 70%. There are major retailers who do not report to Bookscan, Walmart and Sam’s among them.

  2. David Pomerico Says:

    One thing I would be interested in seeing is not only how the five books RH provided for free sold, but how the subsequent books in those series sold (if you notice, all five books are the first in a series). Because that would provide further indication of whether or not there has been positive effects from the free books.

  3. Simon Says:

    Hey David, I actually suggested that to John in an email, he said he would look into it.

  4. Tim Pratt Says:

    I suspect he was looking at the hardcover edition of the Harry Turtledove novel, which didn’t sell any copies in 2009 because it’s not available anymore — the trade paperback did sell some copies, however.

    (As one of the people whose book was given away, I tracked down some numbers myself…)

  5. John Hilton III Says:

    Thanks for the comments and suggestions. I am going to look into the other books in the series.

    I’ve made an updated post at

    http://www.johnhiltoniii.org/update-on-random-house-books/

    that includes the sales of the 5th book that I previously had not found. @Andrew, I believe that Wikipedia gives a fair assessment about Bookscan’s capabilities; I am not aware of a better method to collect data as to the question of how free book releases affect sales.

  6. John Hilton III Says:

    Thanks for the comments and suggestions. I am going to look into the other books in the series.

    I’ve made an updated post at

    http://www.johnhiltoniii.org/update-on-random-house-books/

    that includes the sales of the 5th book that I previously had not found. @Andrew, I concur with your comment. In my blog I linked to an article detailing the capabilities of Bookscan. I believe that Wikipedia also gives a fair assessment about Bookscan’s capabilities; I am not aware of a better method to collect data as to the question of how free book releases affect sales.

  7. worldofhiglet Says:

    Interesting study! So what effect does serialising/podcasting books have on the numbers, and can this even be quantified?

    If the book is available on, say, Podiobooks, then does this increase or decrease the numbers? Scott Sigler and others seems to have had great success by podcasting the book first and then releasing (through a publisher) the print version. There are so many factors involved, though, that I’m not sure you could say with any certainty what effect ‘giving it away’ has had.

  8. moz Says:

    Did your Tor numbers come from Tor or from Bookscan? I wonder if they report to Bookscan (viz, do their books ever really show up in Bookscan? Tor should be able to tell you)

  9. Sra Says:

    You can’t reach a crescendo. The crescendo is the journey, not the destination. You can reach a climax, however.

  10. Simon Says:

    Re: Sra,

    Ha, nitpicking noted :-) Will keep in mind for future articles

  11. Carolyn Jewel Says:

    Actually, for fiction and genre fiction in particular, Bookscan only accounts for 50% of sales. Mass market fiction is sold in outlets that do not report to Bookscan. In addition, Wal-mart does not report to bookscan. Therefore, it’s quite possible that the Tor books, to the extent they were mass market, had sales that were not reported to Bookscan and would therefore seem have lower sales.

    Publishers do not report to Bookscan. It’s way more convoluted that that. The booksellers or shippers (middlemen, bookstores and some but by no means all, outlets) report their sales to Bookscan which then provides their sales reports to publishers.

  12. James Stein Says:

    What was not mentioned was whether any of the variations in book sales actually achieved *statistical significance.*

    Given the comment about the 20/24 books, wherein most books “only showed a difference of a few copies in the two time periods,” I suspect that his data *doesn’t* reach statistical significance.

    Just saying X or Y happens, without testing to determine whether X or Y is reasonably within the realms of chance, is fruitless. Perhaps that was merely something left out of this blog article, but given the comment noted above, I suspect the tests were done and the author preferred to ignore them in favor of keeping his “20 out of 24″ numbers, which make it seem as though there is more heft to the anti-freebook side of the argument.

    (Computer problem; I apologize if this post goes through in duplicate).


Blog Widget by LinkWithin