ae-julyHugh Howey and Data Guy have done it again, producing another interesting report on a sample of data scraped from Amazon. The July report re-runs the numbers for their main chart based on the new data, then branches out into a couple of interesting new measurements—including one that I specifically asked for back in May.

The main measurements show basically incremental change from the last few reports, In terms of daily revenue to authors (taking into account royalty percentages from publisher sales and Amazon revenue percentages from self-published works), Big Five authors take 37% of total Amazon daily revenue from e-books. But for the first time since Howey and Data Guy began their measurements, self-publishing authors are outranking Big Five all by themselves—39% of revenue. (In February, the split was 39% Big Five, 35% self-published.)

Of course, it could just be a quirk of statistics—perhaps if they’d chosen another day, the Big Five sales would outpace independent sales instead. Still, it certainly looks like a trend considering the percentages from the last two charts.

Does DRM affect sales?

I was moved to ask this question back in May, in the comments on the prior Author Earnings report. The idea intrigued Data Guy, and so this next run attempted to collect some figures in that regard. Given that “practically 100%” of Big Five titles have DRM, they had to run the comparison strictly between indie titles that do and don’t have it, as they come out about half and half.

As Howey reports, the statistics seem to suggest that titles without DRM sell twice as many copies as ones with it. This was true at almost all price points, and for the two price points where DRMed titles outsold those without, the difference was accounted for by “3 outlier DRM titles published by only two authors.”

This seems like a welcome conclusion for DRM opponents, but is it the whole story? A commenter posting as “David” did his own analysis on the raw data (which, as with all prior surveys, the Author Earnings people provide). If only a couple of outlying titles are enough to affect the data so profoundly, David wondered, how robust could the results be? When he excluded the top 1% of indie titles, he found “no observable sales impact of DRM” for the bottom 99% of titles.

David explains:

The fundamental problem is that the top 1% of indie titles is only about 300 titles, but those titles earn about 40% of author earnings. This makes “average” earnings-per-title absurdly sensitive to the performance of just a few titles. Thus what appears at first to be a broad-based consumer behavior is in fact just noise, driven by a very low sample of decisions of just a few authors.

Update: However, Data Guy has followed up:

Had a chance to check the effect of outliers. Eliminating the top 1% of all titles as suggested has an unintended consequence of unfairly skewing the averages against non-DRM titles, because there are more non-DRM titles than DRM titles in that top 1%.

A far fairer way to eliminate outliers in each category is to separately eliminate the top 1% of each.

After doing so, we still see the non-DRM titles outearning the DRM titles by 55%.

The remaining Non-DRM-enabled indie titles are also selling better than the remaining DRM-enabled indie titles at every price point.

He also broke it down by publication year, and found the same results all across the board.

That certainly seems to be conclusive: DRM-free titles sell better and earn more. I wonder why that is? I firmly believe most people don’t know or care whether what they’re buying has DRM. All the same, the figures don’t seem to lie. It appears DRM isn’t just annoying, it’s actively bad for business—at least if you’re an indie publisher.

Breakdown by genre

The other interesting new chart in this report divides author earnings up into the genres of Romance; Mystery, Thriller, & Suspense; Nonfiction; Science Fiction & Fantasy; Childrens; and Literary Fiction, and compares what percentage of each genre each type of publisher comprises. The results are instructive: self-publishing rules the roost in romance (66% of earnings) and science fiction/fantasy (56% of earnings). It makes up a quite respectable 23% of mystery/thriller/suspense and 26% of nonfiction, comes in at 21% of childrens but only 13% of literary fiction.

The chart also notes that romance, mystery/thriller/suspense, and nonfiction are the three best-selling categories of e-books; SF sells respectably well but not quite in their league, and childrens and literary fiction are hardly even on the chart. Clearly, if you’re going to go into self-publishing, some genres are more lucrative than others.


While Howey and Data Guy’s conclusions are often more than a little controversial, they represent a lot more data than we’ve ever had available before. It’s great that writers are now able to make more informed decisions on how and where to publish their works.


  1. I wouldn’t be quite so quick to claim *direct* causation for the full sales differential.

    It could easily be that the data is saying that going DRM-free is more common among savvy publishers that know how to effectively promote their titles and have a good understanding of what their readers want, that it is a trait of good, effective publishers that look for every last way of improving visibility and presentation (covers, blurbs, etc), and who understand that doing unemcumbered releases contributes to higher sales because DRM-free titles are by default lendable, TTS-enabled, and easily converted for use on non-Kindle systems, and there is a non-trivial pool of people that just don’t buy DRM’ed titles.

    The full size of the spread, even at the lower 55% level, seems a tad high to be due solely to DRM issues (unless indie title buyers are significantly different than BPH title buyers) but that doesn’t negate the findings that DRM free goes hand in hand with higher sales. I suspect that if all currently DRM’ed indie titles went DRM-free, they *would* see an uptick but not in the 50% range or even half that, simply because there are other things they need to work on.

    In a list of best practices of effective publishers, going DRM-free clearly has a spot high on the list. Which isn’t exactly news, even though many in publishing refuse to acknowledge it. As Eric Flint of BAEN put it, long ago, ebook piracy is the equivalent of minor shoplifting, a cost of doing business, and not worth inconveniencing the paying customers over it. Other industries might have a big issue there but for the typical indie ebook publisher the downsides of DRM clearly outweigh its dubious benefits.

  2. “Of course, it could just be a quirk of statistics—perhaps if they’d chosen another day, the Big Five sales would outpace independent sales instead. ”

    The problem with clearly agenda-driven reports like these is that we don’t know why a particular day was chosen and how that relates to the day before or the day after.

    Choose (deliberately or by chance) a day when a good selection of big-name indies have run Bookbub ads and maybe some other coordinated US newsletter promotions and are getting special in-store promo from Amazon (Countdown for Select, and many big-name indies will be benefitting from the White Glove shoveller programme) and that will heavily skew the results as desired.

    It would be instructive to see Data Guy apply the same techniques to Amazon UK, where there is no comparable promotional newsletter culture.

    Until then it’s a telling point that the man behind the author Earnings report, Hugh Howy, admitted he actually made more money from trad publishing than from Amazon last year.

The TeleRead community values your civil and thoughtful comments. We use a cache, so expect a delay. Problems? E-mail