One of the frustrations of being a writer (or an editor, or a publisher) is measuring book sales. There are a whole variety of reasons for this, not the least of which is the returns allowance. However, there are some "wet thumb" measurements which can be read like critic's entrails to determine how a book is doing. Specifically, these are Bookscan reports of retail sales of a title, and WorldCat library listings.
I have a tentative working formula for a trade hardback in the sf/f field. Mainspring [ Powell's | Amazon ], for example. This based largely on anecdotal evidence, miscellaneous advice, and my (almost certainly inaccurate) gut feel. It goes thusly:
Hardback sales to date = (Reported Bookscan number X 1.5) + (Reported WorldCat number x 2.5)
Here's my logic: Bookscan reports register sales at most chains and major independents. It does not report Wal-Mart/Sam's Club or airport distribution. (Neither of these is an issue for Mainspring.) It does not report library sales, nor does it report hand sales, convention table sales, or many genre independents of either the brick-and-mortar or online variety. For a book like mine, these sources account for a meaningful number of sales.
Library sales can be imputed to some degree from WorldCat. WorldCat reports the number of library systems a book can be found in. It doesn't report the number of copies available. Not all libraries belong to WorldCat, especially smaller ones. Many larger libraries order multiple copies of books, to have in branches, or simply to meet demand from the main shelf.
(Of course, the hardest part about this for writers is determining your Bookscan and WorldCat numbers in the first place.)
Sales of course only matters in the context of total print run, and returns, and how long the book has been on the market. 2,500 units sold in 90 days on a 4,000 print run is healthy movement. 2,500 units sold in 90 days on 40,000 print run is dead in the water. (There's this whole thing called "velocity" as well, which is another topic.) So it's a moving target — print run minus actual sales gives you some notion of the possible volume of returns. Trend of sales over time lets you predict how many of those possible returns won't be actualized. But without the sales numbers in the first place, you can't even make a wild guess at the rest of it.
What do you think of this logic? Especially you publishing pros out there, how do you feel about this sort of augury?