Is wrong information better than no information?
For many publishers, the answer is yes.
Bookscan is a service that started in 2001, and gathers information from bookstores about what books are sold every week. Not all bookstores report to Bookscan. They have about 13,000 retailers reporting to them from across the country, and then they use the figures from each one to extrapolate about the retailers in each area from whom they don't have sales numbers.
So the sales numbers Bookscan reports are a guess. But on top of that, Bookscan is also only recording the sales of books in bookstores... which does not include any sales direct from the publisher, through school or library accounts, through bookclubs, etc etc etc.
A particular book's Bookscan sales number can often be half what the book's true sales numbers are-- and are sometimes more like a third or even a sixth! Bookscan is not terribly reliable.
Publishers know this, but at the same time, they have no other way to find out how the books at other publishers have sold. So in this case, incomplete information (and who knows how incomplete) is better than no information, at least to a publisher's mind.
"Wow, that's... boring," you say. "What exactly does this have to do with me?"
Well, I'll tell you. When an editor is getting ready to make an offer to you, she'll look up your past published books' Bookscan numbers, and she'll base that offer on those numbers. Modest numbers = modest offer.
Unless. Unless you've included in your past publishing information the real numbers from your royalty statements. Your publisher's numbers are always going to be higher than Bookscan numbers.
So keep track of your sales, huh? And let the acquiring editor know how your books have really done, because she'd rather raise confidence at the publisher with your past sales, and she'd rather pay you more if she can justify it.
And agents? I'm talking to you, too.