Having been out by many thousands in recent weeks, Data Guy’s Bookstat, which purports to track every online sale – print or digital – with pinpoint precision, appears to have reigned in its guestimates for Hachette titles, given Hachette’s inconvenient reporting back to The Bookseller that the Bookstat numbers are wrong.

For those out of the loop, the UK trade journal The Bookseller has for a while now been carrying the top ten ebooks sold in the UK as according to Bookstat, the book-sales counting machine that morphed from the controversial Author Earnings Report.

While Bookstat offers a potentially useful service by collecting figures for ebook sales from Amazon and other ebook services (and publishers including APub) that do not themselves normally report sales numbers, there has always been a big question mark over the purported accuracy of the Bookstat numbers, which are based not on real sales but on estimated sales based on comparative chart rank.

Lately Hachette has taken to reporting back to The Bookseller on its real numbers, making the subtle point that the Bookstat numbers are, well, wrong. The Bookseller in turn reports the “clarification” offered by Hachette and other publishers, but with no further commentary, prompting TNPS to make that point that the Bookstat numbers may not be all that they seem.

Last week, for example, Data Guy overestimated the count on at least (this based just on those publishers that shared the real numbers) five titles to the tune of 14,000 sales too many.

Since when it seems Data Guy has decided to play safe and substantially cut its guestimates for the titles it knows Hachette will publicly dispute.

This week the Bookstat numbers are out of kilter by “only” hundreds rather than thousands, but curiously over-guessing the sales of two titles by a couple of hundred each while under-guessing the sales of another by three hundred.

For the record, this week Data Guy guestimated the ebook sales of Hachette’s The Strawberry Thief at 9,977 units. Hachette says the real number was 9,790. An overestimate of 187.

For Hachette’s Small Great Things Data Guy reported sales of 9,042, compared to the real value of 8,832. An overestimate of 210 sales.

And for The Eye of Darkness Hachette said 5,672 sales had been made. Hachette says it was 5,978. An under-estimate of 307.

Okay, so the differences between the Bookstat guestimates and reality are a lot smaller this time around, but we can safely put that down to Bookstat revising its numbers to fit with what Hachette has previously been telling us. Last week Bookstat numbers had been out of kilter with Hachette numbers by an average of 35.6%, so a simple remedy to take this week’s guestimates and cut them by 35% to give us values closer to the real numbers than we’ve previously seen.

But this still tells us nothing about the accuracy or otherwise of the other titles reported, and leaves us in the dark about the true state of the ebook market in the UK, let alone in the US where Bookstat is based.