Hugh Howey is a very successful author and a strong advocate of indie publishing, although he is only a self-publisher himself in terms of eBooks within the United States, having sold print rights to Simon and Schuster and eBooks rights internationally. In 2014 he was famous for two campaigns: being a pro-Amazon cheerleader in their dispute with the Big 5 publisher Hachette and launching Author Earnings. That latter campaign has had less of a public impact, but is the longer lasting one. Author Earnings is a website launched by Hugh Howey and an unnamed statistician referred to as Data Guy. The latter designed software to access sales information from Amazon as Amazon refuse to divulge sales information for eBook sales. Although doubt has been thrown on the quarterly Author Earnings reports from a statistical viewpoint they have become influential in the self-publishing community. The latest report was published today (28th January 2015) just less than a year from the first report on 12th February 2014. During that time only two reports have been done on a retailer other than Amazon when reports were published on Barnes and Noble (aka Nook) in February and July 2014. This leaves the reports very dependent on the data derived at, the only site analysed for the Amazon reports.

The January 2015 Report has a defensive tone to it and challenges anyone to test their statistics by going through a best-seller list at an eBook retailer and comparing it with the Author Earnings reports. It then closes with an appeal to check their dataset (with a broken link) and an encouragement to check for yourself at Barnes and Noble, Apple, or Kobo. As a self-published author and freelance researcher that was a challenge that I could not resist. I was particularly seeking data on what I regarded as spurious claims about ISBN usage in the January 2015 report and as Author Earnings have based all their work on Amazon or Barnes and Noble I chose to go to Kobo for my data. The claim about ISBN data that I regarded as spurious was the central theme of this Author Earnings report. It showed charts claiming that 30% of eBook best-sellers have no ISBN. This is used to counter claims from mainstream publishing pundits that self-publishing is a small portion of the market.

Previous Author Earnings reports have presented data on the self-published segment of the eBook market being very strong in terms of both units sold and dollars earned. On the ISBN claim, however, there is a glaring error in their claims. They seek to extrapolate from ISBN usage on to all eBook retailers, but this misses the fact that ISBNs are of less use for eBooks uploaded to Amazon. Self-publishers upload eBooks to Amazon via the Kindle Direct Publishing service where ISBNs are purely optional. It is similar at Kobo's equivalent Kobo Writing Life, but with one vital difference. If an ISBN is listed at Kobo when an eBook is first uploaded that ISBN becomes the PrimaryID for the Kobo database. At Amazon the database is always sorted via its own unique numbering system, the ASIN. Another difference with Kobo is that like almost the entire eBook industry Kobo uses the industry standard epub format, whereas Amazon uses its proprietary mobi format. That has led to a situation in the UK where the ISBN agency, Neilson, will allow an ISBN to be used for a mobi formatted book, but then refuses to let anyone, including the publisher, see that book. If this issue is raised with Neilson the response is that mobi is a sole retailer format, so they refuse to display them in their systems. This claim by Neilson is not accurate as mobi files can be purchased at Smashwords, Libiro, XinXii, and other retailers.

This stance by Neilson, the fact that epubs cannot be used at Amazon, and the prioritisation of ASIN over ISBN means that self-publishers are far less likely to assign an expensive ISBN to an Amazon eBook than to one at an epub retailer. Therefore the Author Earnings report's extrapolation from data is unwarranted. Further research might show similar levels of non-ISBN use at Barnes and Noble, but until that happens I will have to do a small scale analysis of the Kobo best-seller list.

I should begin by setting out some limitations to my dataset. Kobo automatically shows the site appropriate to your locale, so I am comparing Author Earnings' focus on the United States market with the Kobo best-seller list in the United Kingdom. As I am doing this manually I have only looked at the overall Top Fifty on Kobo, whereas Author Earnings analyses all best-seller lists on, although it excludes any book that earns less than $10 per day. In terms of ISBN usage I was not surprised to discover that 94% of the Top 50 eBooks sold at Kobo have ISBNs. Since this article was first published it has been pointed out that "ISBNs" beginning 123 are Kobo's equivalent of an ASIN. Of the seven self-publishers in the Top Fifty three did not use an ISBN. That is higher than the percentage in the Author Earnings report, but on a statistically questionable sample of less than ten books.

What I was surprised to discover was the way in which this list did not cohere with other claims put forward in the last year by Author Earnings. The Top 50 at Kobo contains just 7 self-published books alongside 14 from small or medium publishers and 29 from Big 5 publishers (HarperCollins, Hachette, Macmillan, Random House Penguin, and Simon and Schuster). This Kobo Top 50 is likely to be sorted by units shifted rather than dollars earned on the basis that the top three are all priced at under one pound. Despite my brief research being based on an overall list that might privilege mainstream publishing, these figures suggest that either the Author Earnings figures or the Kobo Top Fifty are not telling the true story. What is absolutely clear is that Author Earnings is wrong on ISBNs once the dataset moves away from Amazon and onto epub retailers.