We had a catalog, which was basically the electronic data files supplied by two distributors, banged together into a Very Simple Database (was it a relational database? No. No, it was not a relational database) and updated in an extremely crude fashion. Because the electronic data files were never intended to be viewed by customers, but only by staff at bookstores who ordered books that were then sold to customers, they were Full of Errors. The novel thing that this little bookstore did was radically simplify a process that already existed at virtually every bookstore in existence: you could place an order for anything that the distributor carried, or that the bookseller had otherwise become aware of. Did we use Bowker data? No. No, we did not. That's a whole story in and of itself.
In Seattle, there were two bookstores where it was worth your while to try to order an obscure book: University Books and Elliot Bay books. By far, the better option was the former, but once I was gainfully employed, I tended to use the latter because it was walking distance from my place of employment (prior to working for that little online bookstore). When they failed, I'd go to University Books. Sometimes they would tell me, hey, we're sorry, we tried, but they don't respond. Here's the publisher information; you can try if you like.
So when that little bookstore came into existence (before I went to work for them), I tried one of the Difficult Books to Acquire and they produced it lickety split. So fast I jumped when I had a chance to interview for a job working for them and I mentioned the experience. Did that little bookstore have that book in stock when I placed that order? Absolutely not. No. It did not. Was I pleased with the rapid fulfillment of a book that previous determined efforts over a period of months to acquire had not resulted in success?
Oh, yes I was.
Fast forward a few months.
That little bookstore had the most amazing business model in terms of cash flow. Because books tend to sit, unmoving, and increasingly dogeared from customer fondling at bricks and mortar bookstores, publishers offer generous financial terms to bookstores who are willing to stock their warez. That little bookstore had this great thing going: they didn't order the book until they had a customer that wanted it. They collected payment when they shipped it. And then they didn't have to pay the publisher for weeks -- they could do whatever they liked with that money in the mean time. Even better, publishers _loved_ this model when it came to returns, because there weren't any, and a sales channel that never returns a thing in an industry that returns double digit percentages of products -- in a good year -- well, my friends, that is a Miracle. (Also, the warez did not suffer as much customer depredation.)
There was just one thing: the onesies and twosies orders kinda drove the publishers nuts, especially when that little bookstore would order 3 copies on Monday, 5 copies on Tuesday, 2 copies on Thursday, 10 copies on Friday, etc. Why not just order 20 copies on the following Monday? they wanted to know, and negotiations ensued. (Obvs, it was not so simple, because many of the small orders went through the distributor rather than direct to the publisher.) You have to understand, those publishers were accustomed to orders from bricks and mortar chains which consolidated and pre-ordered pallets of books. Their fulfillment departments were not set up for this one and two and three stuff. I will leave the problems associated with a workflow aimed at pallets being pointed at individual items as an exercise for the reader (to ignore. And please continue to ignore just how amazingly difficult it is to get onesie/twosie order fulfillment right, because that's how come I have spare time to write stuff like this.).
And now, in the course of a dispute between Amazon and Hachette, Amazon says the following:
"For titles with no stock on hand, customers can still place an order at which time we order the inventory from Hachette -- availability on those titles is dependent on how long it takes Hachette to fill the orders we place. Once the inventory arrives, we ship it to the customer promptly."
I know that business model. That business model destroys publishing house fulfillment departments.
Returning once again to the mists of internet time (1997, give or take): a variety of commentators who investigated how this little bookstore did business said odd things, specifically, how did it manage to stock millions of titles? Oh, wait. They don't. Hey! That's cheating. Or words to that effect. (Meanwhile, with unlimited access to modifying that catalog, I spent a lot of time fantasizing about putting completely fake titles into it, just to see who would find them and order them. Which I never did. *sigh*)
Here is Scott Turow on the subject of this business model in its current incarnation as part of a negotiation between Hachette and Amazon (http://www.washingtonpost.com/business/economy/amazon-said-to-play-hardball-in-book-contract-talks-with-publishing-house-hachette/2014/05/16/cdd40854-dc62-11e3-8009-71de85b9c527_story.html):
""What kind of entity in a competitive market would willfully drive customers into the arms of its competitors unless it believes it doesn’t really have any competitors?” Turow said. “Can you imagine Best Buy refusing to deliver for a period of weeks what’s available from its competitors? But Amazon behaves as though they’re the only game in town. And increasingly they are. It’s a head-scratcher why anyone with regulatory authority would tolerate it. If this is not an example of untoward power, I don’t know what is.”"
So that's novel. If you decide not to sell something in a situation where your customer might go somewhere else, that is maybe evidence that there _isn't_ some other place to go to acquire it, otherwise you wouldn't do it. And, therefore, regulators should get involved. To require Amazon to fulfill orders faster? You know, you could probably get the FTC involved if Amazon was misrepresenting the amount of time it would take to fulfill an order, and you could prove that and demonstrate harm and intend to defraud and blah blah bleeping blah. But for just being slow? *blink* I kinda want to see that court case, actually. We've _had_ court cases (they involved railroads and occurred in the 19th century) that addressed the question of whether a company could be required to continue to engage in a business in which it was unable to make money. (The answer was no. Duh. Altho apparently it wasn't such a duh thing at the time. Go figure.)
There are two groups of people who LOVE the customer orders it, retailer orders it, supplier sends it, retailer sends it while collecting payment business model: people who read balance sheets (yeah, I know that's not you) and customers. The customers would be a lot less impressed if they had to place their order with a harried cashier, in a mall (or worse, in a Lifestyle Center, accessible only by car, surrounded by enormous, yet strangely always full, parking lots), during business hours; since they can do it at 3 a.m. in their PJs, they are as happy as a pig in slop. Everyone else hates it.
ETA: Okay, because I can't leave it alone. Amazon is not engaged in a merger, nor is it working in collusion with another firm, so the standard you have to apply to go after them as an antitrust regulator is difficult to achieve in the first place. Further, all the cases for decades have revolved around the pricing effects on the end-use consumer -- and Amazon's negotiating position is favorable for them. Attempts to re-orient antitrust law, especially in monopsony cases, to protect sellers from a powerful buyer determined to force prices down have, er, not succeeded. Altho I am not a lawyer, so hey, what do I know.