As happens with so many new things, a precise definition of print-on-demand hasn’t quite settled down yet.

When it first came up, in the early nineties, it meant reprinting a small quantity, more or less equal to the number of backorders you might have on hand. A refined version of it might involve reprinting a standard quantity, maybe 25 copies, and reordering the same quantity when your inventory dropped to say 10. We talked about doing this at that time, but as far as I know nobody actually made the leap until well into this century. This approach to “print-on-demand” is now fairly well established, and is perhaps properly named ultra-short-run printing. At Oxford University Press we call it autoreplenishment, which is quite accurately descriptive. The computer is programmed to send out a reprint order for a predetermined quantity automatically via EDI whenever the inventory level drops below a set level. No human intervention is needed till the books are being put into the carton for shipment to the warehouse.

As may be obvious, this sort of stock replenishment became possible only as a result of the development of digital printing. When digital print engines were first adapted to book manufacturing early in the nineties (IBT of Troy, NY was the pioneer, in conjunction with Bridgeport National Bindery, a library repair bindery) we were working (happily) with a minimum order quantity of 25. The idea that you could print as few as 25 books, hardback or paperback, was revolutionary. Before that we all tended to print at least 500 copies, and keep our fingers crossed that they would eventually sell. You might not bind up all the F&Gs, but almost inevitably much of what we printed ended up as waste, or written down and resident in the warehouse for long years.

The industry has now evolved to its logical conclusion: the ability to print and bind, economically, a single copy of a book, hardback or paperback. What we have to call true print-on-demand means that the book is not printed till after it has been sold. The publisher keeps no inventory. When someone orders the book, the order is sent via EDI to the digital printer, who makes the book and ships it to the customer, with a packing slip or an invoice looking like it comes from the publisher enclosed in the box. This is the ultimate in efficiency for slow-selling titles. The so-called long tail just got infinitely long: MD (manufactured on demand) is the new OP. To be able to manufacture your back list economically one book at a time frees you from these agonizing decisions about the future of a book — reprint what is obviously too many, or declare it out of print and forgo such additional sales as may be out there. Now no book need ever become unavailable.

“Economically” is the key word. If you print 1000 copies of a book at $2.75 a copy, you lay out $2750 before you’ve been paid for any sales. The same book might cost you $5 as a one-off print-on-demand job, but you don’t have to print it till after you have sold it. If in two years you only sell 600 of the books you printed you’ve still invested the $2750. With true print on demand, you will have spent  $3000, but you won’t have any inventory sitting in the warehouse and clogging up your accounts. If you have only sold 500 you will actually have invested $250 less with print on demand. Many publishers are now aiming towards turning their inventory two times a year, which would imply printing for 6 months of sales only. Fixation on unit cost will inevitably lead a publisher into queer street. And traditionally unit cost was what we cared about. Now we just have to forget it. A unit cost of $5 is better than one of $2.75 if you end up wasting the last 400 copies of the run you did to achieve that cost. There are already one or two publishers who only make their books available via print on demand.

One of the hopes around print on demand was that it would lead to an elimination of returns. If we print the book specially for you, we should perhaps be able to assume that you really want it. Although one or two publishers have made their POD books non-returnable, the generality have not been able to make this leap. We just can’t get away from the idea that if we want someone to buy our book, we have to be able to get it into their hands first. So we still supply print on demand books to bookstores, and accept them for return. Now many print on demand books are rather expensive to make — a multi-volume set can have a unit cost of $2000 or more. If we take a return on a print on demand book we have, by definition, no location in the warehouse into which to put it — so it gets wasted. This makes nobody happy, although it is actually better to face that unpleasant cost rather than never be able to supply the set at all because it has had to go out of print (as long as you sell some sets that do stick!). However, a system adjustment can make the POD book returnable. You have to maintain an inventory location for the title, and program the computer to check if there’s any stock on hand before it flicks into it’s automatic process for EDI-ing the order to the digital printer. This is a bit like low-volume autoreplenishment — the book will only be ordered if you are unable to fill the order from such stock on hand as has been accumulated through the returns process.