Reading The Bookseller‘s weekly list of bestselling ebooks as provided by Bookstat and by actual publishers is always fraught with challenges.

On the one hand we have publishers providing real numbers, but only of course for the books they sell. On the other hand we have Bookstat making educated guesses at sales across all ebooks. And every week the differences between the two, even allowing that the publishers’ numbers are a week behind Bookstat’s, send out mixed messages.

This week The Bookseller shows us the Bookstat top ten:

In stark contrast to the publishers’ list the Bookstat list includes three APub titles, which cannot feature in the publisher charts because Amazon does not provide numbers.

As TNPS has covered on many occasions, the Bookstat numbers claim to be accurate, but there have been many indications the statistics are nowhere near as accurate as Bookstat would have us believe.

In the past Bookstat has managed to count JK Rowling as two separate authors, among many examples of Bookstat’s numbers simply not adding up.

This week an update to The Bookseller‘s report offered further evidence that Bookstat’s numbers are not as reliable as they would have us believe.

Referring to the chart as pasted above, The Bookseller carries this update:

Clarification: Hachette has stated that Eyes of Darkness sold 15,741 digital copies; Twochubbychubs The Cookbook 8,200 and Small Great Things 5,089.

At which point, let’s juxtapose those numbers and see what it means for Bookstat’s credibility.

Small Great Things sold 5,584 copies according to the Bookstat guestimate derived from trying to calculate actual sales from relative chart position. Hachette says the true number was 5,089. That’s an overestimate of 495 sales.

Twochubbychubs The Cookbook, says Data Guy, sold 5,628 ebooks. Hachette says the true number was 8,200. It would seems Data Guy failed to count 2,572 sales.

Eyes of Darkness, guesses Data Guy’s Bookstat, sold 17,685 ebooks in the period monitored. Hachette, the only entity to know the real numbers, says the number of ebooks sold was 15,741.

In other words Data Guy over-guessed the sales of that title to the tune of 1,944 ebooks. And given the title retails at 2.99 GBP it means Bookstat’s revenue guess for that title is asserting the ebook sold for £5,800 more than it actually brought in.

Hachette of course can only comment on the numbers pertaining to its own titles. Other publishers may not wish to publicly dispute the Bookstat numbers one way or another, or may have long since given up bothering with guestimates of this nature.

But given these discrepancies it begs the question how accurate the other numbers are in the Bookstat list and in the wider statistics Bookstat makes available to its paying clients..