If you’re a regular visitor to our website, you’ve probably noticed a recently introduced new feature on our home page, headed ‘Recent Bestsellers’. When you click on this tab, you’re presented with a list of our previous month’s top-12 bestselling titles, listed in order of their sales.

If you’re a thoughtful or inquisitive visitor, you may have wondered what the word ‘bestseller’ means in this context. The answer, it turns out, is that ‘it depends’.

I hasten to say that it doesn’t depend on whimsy or wishful thinking or my subjective judgement. It depends on what you and I understand by the word ‘sales’.

Let me explain. First of all, this is what the list isn’t based on. It doesn’t include any sales we make outside Australia (for instance, in New Zealand), as that would distort the picture we’re aiming to present of our domestic sales rankings. And it doesn’t include direct sales we’ve made to organisations, authors, or individuals, as that would distort the figures for sales that have occurred in the book trade (which are known as ‘trade sales’).

More importantly — and perhaps surprisingly for many readers — the list isn’t based on the number of copies of our books that we’ve ‘sold’ to bookshops. That’s because these aren’t ‘sales’ in the sense understood by ordinary people. These are, rather, invoices for books ordered by bookshops in the hope/expectation that they’ll sell them to the punters who wander into their stores. Any books that aren’t sold are eventually returned (bookshops have to keep them for three months, and usually have up to a year after publication to return them), and the shops are then credited for them.

So, for example, a shop that orders six copies of a new title in March may return three copies in July and one copy the following February, and our initial apparent sale of six becomes a real sale of two. The trouble is, we don’t know in March what our real sales will eventually turn out to be.

That is, we didn’t know until Bookscan came into existence. Bookscan is a private company that tracks trade sales for its subscribers. These are real, weekly, through-the-till sales — not pretend sales, not subject-to-return sales, but bye-bye, have-a-good-life sales. We started subscribing to Bookscan on 1 January 2007, so we now know our real sales. And that’s what our ‘bestellers’ list’ is based on: each month’s sales as reported by Bookscan.

End of story, you might think. I wish it were. The trouble is, Bookscan doesn’t have complete coverage of the book trade: it’s very accurate for large groups (such as Big W) and the chains (such as Angus & Roberston), but it misses out completely on airport bookstores through Newslink outlets (a situation which I’ve heard may soon change), and only has about 25 per cent coverage of the independent bookselling sector. Unfortunately, this sector is where independent publishers like us have their heartland, and where we sell a disproportionately high percentage of our books.

Now, the people behind Bookscan are not stupid, so they deal with the gaps by making allowances for them. This means their ‘sales’ figures are guesstimates, but they’re meant to be highly accurate guesstimates. Booksellers rely on them to keep a check on what’s selling around the country (and what’s not), so it’s a great help to them in making decisions about what to order and re-order. This has been a controversial effect, as it’s led to booksellers reducing their initial orders (and risks), and adopting more of a wait-and-see policy with new releases. This lowers the aggregate initial order for a new title, making it harder for publishers to meet their initial budgets, and anxious about meeting them at all.

Publishers, especially the large ones (who get a comprehensive and sophisticated range of data), also rely on Bookscan to see what titles and genres are selling, where they’re selling, and how their competitors are doing. They also rely on Bookscan to judge whether to reprint successful titles, and to help them decide how many to reprint.

This is where we come to the final twist in the ‘sales’ tale. In the short time I’ve been using Bookscan, I’ve found it frankly useless as an aid to decision-making about reprints. At this point, a case-study is necessary to demonstrate the problem.

The example I have in mind concerns Ghost Plane by Stephen Grey, a superb book that we published on 19 February. (This title is no. 2 on our March bestsellers’ list.) We printed a relatively modest quantity of 2500 copies, as I knew it would be a difficult book to sell with a non-resident author (Stephen is a UK resident, and the book was originally published in late 2006 in the US and the UK). There’d also been a book (which I thought was inferior) on the same subject published in Australia in late 2006, and I was worried that there’d be limited media interest left for our title.

As it happened, a number of things clicked, and we decided suddenly to bring the author out for a quick tour, during which he scored a number of high-profile media interviews. The book was also reviewed favourably and prominently, and it started to sell fast.

Or we thought it did. By 6 March, after several days of heavy ordering from the book trade, there was almost no stock left in our distributor’s warehouse, and I knew we had to reprint. I consulted Bookscan, and its most recent report told me that we’d sold a total of 368 copies by the week ending 24 February. (A few days later, it reported cumulative sales up to 3 March of 482 sales.) So the trade had gobbled up 2500 copies really fast, but Bookscan told me that less than 20 per cent of that total had really been sold.

If I had listened to Bookscan, there would have been no reprint then — or, maybe, ever. Instead, I pressed the button for another 2500 copies. By the time the reprint arrived (very quickly — on 16 March), there were over 900 back-orders logged from the trade. And as I write this blog, there are only 120 copies left from the reprint in the warehouse, and we’ll be out of stock again within days.

But get this: while we’ve now invoiced 5000 copies, Bookscan reports cumulative sales of this title of 1714 copies (as of 31 March). If Bookscan is right, we shouldn’t have reprinted at all; or if we felt we had to, we should have reprinted a much smaller quantity. If the bookshops knew what they were doing (and if we were right by responding to and anticipating their heightened demand), we’ve done the right thing by the book and its author.

Of course, Bookscan may turn out to be correct: all of the reprint and some of the initial print-run may end up being returned. But, intuitively, that doesn’t feel likely. Certainly, if it did happen, not only would I weep and moan, but it would highlight once again just how difficult it is in the book world to know when a sale is really a sale.

I’ve talked about this case with Bookscan, and they’re confident their figures are right. I suppose we’ll all know the answer in a few months’ time. But, in the meantime, there’s a new question: what do we do about another reprint?

And as for our bestsellers’ list, you can be confident that it’s right. Sort of.