Print-on-Demand in the Book Supply Chain

Kenneth M Brooks, Jr

Print-on-Demand (POD) is an area of publishing where there is potential for significant short-term revenue, but still a good amount of confusion and a number of open issues, particularly in the incipient area of distributed, or "in-store" POD. Distributed POD is showing a great deal of promise in the few locations where it is installed. But there are still questions to be resolved, namely:

• What is the consumer experience? How is POD best integrated into the retail shopping experience? Traffic patterns in store are highly variable over the course of a day and a year. What happens to waiting time for a customer's book?
• How good is the quality of reproduction? How necessary and desirable is standardization?
• What are the economics of one-off production? In what circumstances is it feasible?
• Books are low volume merchandise at store level. How big does the digital library need to be to achieve sufficient volume?
• Doesn't central as opposed to in-store POD provide better economics at any scale? Why not use central POD for everything?
• When would it be more effective for publishers to replenish their warehouses using POD than to use either distributor or point-of-sale POD?
• Who pays for the increased cost of production? What's fair?
• How does the book get printed? How is it delivered? How secure is the content?

Economics of Print-on-Demand:  a new cost curve

The economics of traditional printing are driven by the length of a print run. As the length goes up, cost per unit comes down. Traditional offset equipment is very expensive, difficult to set up and hard to operate. This leads to a relatively high cost per unit for short print runs — and requires a large team of individuals with very specialized skills — found, naturally, at printers. This high cost of short runs is driven primarily by the set up cost of the press and bindery equipment: changing massive rolls of paper, changing the printing plates on the press, and tinkering with trim and spine size adjustments on the binding line. All of these constitute the cost of make-readies — getting the job ready to run through the machines, not the actual running of the machines themselves.

With intelligent management, many of the factors that create the steep curve can be minimized in digital printing:

• The need for plate changes is entirely eliminated.
• Through good production scheduling and standardization, paper changes can be minimized, as can changes required by different trim sizes.
• Self-adjusting and software-driven trimmers and binders can largely eliminate make-ready effort. Such binders and trimmers exist already and more are under development.
• Workflow integration can eliminate much of the effort of inputting orders, billing, coordinating the production and assembly of the book block and cover, and shipping the title.

Figure 1:  The evolving cost curve

The result cost curve is flat — higher for the long run than traditional printing because of toner and paper waste but much cheaper for shorter runs even down to runs as small as one. The equivalency point between POD and traditional printing is about 800 units:  above 800 units, traditional printing is more cost effective and below, print-on-demand is more effective.

Volume

Volume is also important in digital printing. This is where the distinction between “big iron” POD and “small iron” POD becomes important. "Big iron" POD in these circumstances means POD using equipment such as the IBM InfoPrint 4000 and the Xerox Docutech family [1] . “Small iron” POD is the world of cluster printing and book vending machines. Cluster printing utilizes multiple standard copy machines, both black & white and color, linked together with binding and trimming equipment in a loosely coupled production cell [2] . The book vending machine concept ties these components together tightly using mechanical transport devices — they are often enclosed in a single housing that looks like, well, a book vending machine [3] !

Fully-loaded cost per unit under each of these approaches varies considerably with total volume. A big-iron system can go for $3 million fully installed with costs fully allocated, where a small-iron system can go for $40,000 to 60,000. Because the capacities vary so greatly, however, the ultimate fully-allocated cost per unit is lower for the larger equipment when it is fully utilized. The equivalency point between a cluster approach and a big iron approach is in the neighborhood of 10,000 units per month. Below 150 books per month no investment in printing equipment at today's prices can really be justified — there just isn't enough revenue from such a low volume to justify the equipment.

Considering both length of print run and volume per-month a “map” develops showing the combinations of length of print run and monthly volume that lead to selections of the various printing technologies. There are three regions: high print runs require traditional offset to achieve minimum overall cost, regardless of monthly volumes. For lower print runs, lower monthly volume favors small iron POD due to the low investment required and higher volume favors big iron POD to minimize fully allocated cost per unit.

Figure 2:  Printing Technology Alternatives

The current title base isn't very exciting.

We've seen the importance of volume to overall economics. Volume is a function of the number of titles and the demand at that location. Assuming that there is some sales-based rationale for the selection of exactly which titles are made available in a location, and that retailer, wholesaler and publisher efforts to sell any particular title remain unchanged, the interesting variable is the number of titles that are in a suitable format for POD.

As of December 2001 there are at most 50,000 titles digitized for POD and the number that have any commercial attractiveness at any level of the supply chain is probably closer to 35,000. This is not terribly exciting at either the wholesale or retail levels. At the wholesale level 20,000 titles was a few month ago generating in the neighborhood of 50,000 units per month in single-copy (as opposed to short-run restocking) orders at Lightning Source through Ingram, the largest book wholesaler in the US. If production is moved to the store level, this aggregating effect breaks down rapidly: the same title base at an average retail location (not a superstore) would result in 10 units per month. This suggests that achieving a reasonable target of 150 units per month would require a title base of 300,000 [4] titles of roughly equivalent commercial appeal to those carried by Lightning Source. This is clearly not practical in the short term, although there are interesting ways to get around this problem by perhaps adding print of local newssheets in a coffee shop environment, or academic course packs in a university bookstore environment. In other words, there are ways to boost the volume being driven through the POD machinery without necessarily having to increase book title availability by a significant amount.

So what about service?

Beyond volume another key factor in POD adoption is service. Service consists of many factors, but two of them are critical and manageable. They are lead-time — the time from when an order is placed until the book is delivered — and integration with normal customer ordering patterns.

Long and unpredictable lead-times are two key reasons to stock a title at any level of the supply chain: retail, wholesaler or publisher, versus ordering (or printing) the title when demand materializes. 

At the retail level this means the unit must be on hand when the customer wants it: it's a widely bandied-about truism in the industry that the overwhelming majority of customers won't approach a bookseller if the book isn't in stock. This lost-sale scenario has its corollary at the wholesaler as well: if the title isn't in stock, the ordering retailer will go elsewhere. The wholesaler faces an added penalty that too great a percentage of non-stocks will cause it to fall in the retailers' shopping sequence, which is often automated by “order cascading” software that is used in most stores. Finally if the title isn't available at the publisher, most retail and wholesale ordering systems are set up to cancel the order.

Print-on-demand can apparently address this lead-time issue by providing instant (or quick) availability of the title, but solution isn't actually that simple. If capacity on the POD machine isn't managed carefully, a “sequential backlog” problem can develop in which there are so many orders backed up awaiting printing that the lead time grows and grows — just as traffic jams occur during rush hours around bridges and tunnels. If these traffic jams get too severe and lead time lengthens past what the customer has been led to expect, there is a large risk of alienating the customer even more than what would have happened if he'd been told to go elsewhere, or wait a few days, to begin with.

Another key area of service is how well POD fulfillment is integrated into the ordering patterns at every level of the supply chain.

Integration with ordering patterns

This effect is most obvious at the retail level where it's important to integrate the printing into the retail shopping experience so that it's not a turn-off for the consumer. Most consumers are accustomed to looking for a book on the shelf, finding it and then going to the checkout line to pay for it. If they don't know the exact book they're looking for, they will browse the bookshelves, select something interesting and again go to the checkout line. Only occasionally will they ask a bookseller for assistance if they can't find a title. This surfaces two problems for distributed POD:  How does a consumer find a book that isn't there? And then, were the consumer to find it, how do they buy the book without significantly extending their time in the store or requiring the bookseller to accept the exposure of carrying a good number of printed, but unclaimed titles?

The first question — that of how to find the title — is usually addressed through some sort of on-line catalog in the store. The larger chains are all experimenting with in-store information kiosks. It would be very easy in such an environment to list POD titles in the kiosk catalog with a notice that they can be printed while the customer waits. It is even better if store inventory is integrated into the kiosk information so that the customer can easily determine if the title is available in the store. Clearly this catalog solution works better in an online retail environment in which the customer already must use the catalog to make selections.

This brings us to the second problem — integrating purchasing with the remaining items the customer wants. One way to accomplish this is via entry of a credit card number while the customer is still accessing the catalog — similar to a normal shopping cart procedure at an on-line retailer. The book is paid for and starts printing while the customer completes his remaining purchases. This may seem awkward to the customer (he has to present his credit card multiple times) and doesn't work for those customers wishing to pay in cash. The second alternative is to print a receipt at the catalog to be taken to the checkout where the title is paid for. Once the sale has been completed, printing is triggered automatically and the customer is told to either wait for the book to finish (5-10 minutes for most equipment if there is not backlog) or to come back to pick it up. Clearly the book could be sent to the customer if he doesn't want to wait, but that could be more effectively done from a central facility.

Integration in the wholesaler environment means that the title is shown as “available” when a retailer inquires even if there is no stock currently printed. In the early days of POD there was no status to indicate that the title was available on demand so retailers cancelled orders for which there was not stock on hand at the wholesaler. Ingram addressed this problem at the time by printing several copies of each book carried by Lighting Source and keeping them on the shelves. As the title base grows, however, this solution has become increasingly impractical. It has led to the development of a new inventory code indicating on-demand availability and having both booksellers and bookseller systems aware of what it means. Once such an approach is adopted it becomes a matter of creating a separate zone in the warehouse where orders are directed for printing as opposed to picking. POD titles then join traditional books at the shipping dock for dispatch to the customer.

At the publisher there are several possible approaches. If the publisher has sufficient volume he may opt to install a POD facility in his warehouse or have one run on-site by a printer offering such services. Units can then be either printed in response to customer demand or printed in short runs to create stock in anticipation of demand. The publisher can also contract with an external POD supplier such as Lighting Source to “drop ship” an order that the publisher receives.

A more common approach utilized by many university presses, which tend to have a large number of slow moving titles, is to use the services of a POD printer to do short runs to restock the shelves of their warehouse. This has many advantages including a seamless integration with the current ordering process — after all it uses the current process. It also minimizes the amount of freight paid since books will generally move in larger quantities this way and it eliminates any negotiation issues around title pricing or discount and retail discrimination against POD titles. The sole disadvantage is the cost of carrying the inventory — something that should have been addressed in the printing decision to begin with.

Other Supply Chain Issues

The existing pBook supply chain moves hardcopies of books from printers to publishers to wholesalers to retailers to consumers and then between consumer as pass-along copies. Naturally they can also skip steps, go backward as returns, and move sideways in the channel as needed. Although the supply chain is usually envisioned as a single linkage between steps, each link is actually multiple transactions. A credit sale, in fact, can consist of as many as six transactions, including:

• Issuing a purchase order
• Generating an ASN
• Shipping the book
• Invoicing for the book
• Generating a payment
• Generating a remittance advice

These are then joined by various reconciling activities including receiving of the shipment and a 3-way match before issuing payment on the receiving end and cash application on the part of the shipper. This same series of transactions is repeated all the way down the line to the (finally!) simpler cash sale made by the retailer to the consumer.

And this doesn't even take into account the whole picture. Between the publisher and wholesaler and between the wholesaler and retailer there is also a returns transaction loop that encompasses a variety of transactions. In a well-mannered returns environment there are:

• Return authorization requests
• Return authorizations
• Return shipments
• Credit memos

More often than not, though, large retailers tends to dispense with the returns authorization process and substitutes a deduction from the payment as the first transaction in the returns process.

All of these transactions lead to considerable cost. Beyond the transportation costs that are incurred with every move - either forward or backward - there are huge overhead costs involved in processing all of these transactions.

POD done at a publisher eliminates the transactions between the printer and publisher - but don't offer significant reduction in overall processing load. It is really when POD is done at a wholesaler that the picture at the upper end of the supply chain is significantly altered: there is no longer a connection between the publisher and printer (although strictly they are still there - they're just handled now by the wholesaler), and between the wholesaler and publisher there is only the reporting of sales and receipt and application of cash payments. Even COGS and inventory accounting is simplified (presuming the publisher's systems are up to the task!). The essence of this is that from an accounting perspective the transaction looks more like self-reported subrights income than a traditional book sale. Essentially we're turning a credit sale into a cash sale or even a consignment sale with no inventory at risk.

If wholesaler-based POD simplifies life for the publisher, the in-store model simplifies the entire supply chain. In the in-store POD supply chain the files can reside anywhere, but most likely at an off-site repository. The purchase by the consumer initiates a substantially reduced chain of linked transactions that results in sales reports and payments flowing back up the chain with vastly reduced overhead.

Implications for POD

Publisher

Publishers will find POD valuable on many fronts. There are several categories of work that are natural candidates for POD, most prominently galleys, ARCs, and end-of-lifecycle or slow-moving titles. As the decision to move titles into POD becomes an integral part of the reprint/OP/OSI decision and happen as a matter of course, a publisher is increasingly likely to have sufficient volume to justify investment (either in-house or outsourced) in POD.

In moving to POD the publisher has to choose among three co-existing scenarios for using it:

• Should POD be done in-house or outsourced?
• Should POD be used to replenish stock or to fulfill demand as it arrives?
• Should POD capability be moved down the supply chain to distributors or retailers or should it remain within the publisher's own control?

In-house or outsourced production is almost entirely a matter of scale and business focus. As the number of titles a publisher places in POD increases, it becomes increasingly reasonable to move production inside. The savings in freight and manufacturing margin must be balanced against the cost of setting up and operating the facility. Even if it makes sense from an economic perspective, the cost to business focus must also be assessed.

Make-to-stock versus make-to-order balances the economics of holding inventory versus fulfilling to demand. The systems and business disruption cost of integrating a POD operation into the normal operation of a distribution center must also be considered. If there is sufficient title base it is likely that make-to-order will gradually be started on an opportunistic basis from a stable volume platform of print-to-stock.

The choice as to whether to address POD at the level of the publisher or push it down the supply chain will be influenced by channel discount structure, the source of the majority of orders, extra manufacturing margin paid versus what it would cost to handle the printing internally and the amount of control exerted over the content downstream. A small publisher that gets the majority of his orders through direct contact on his web site or through direct mail might reasonably opt for a drop ship fulfillment model (see next section), while one that gets the bulk of orders through a large wholesaler such as Ingram or Baker & Taylor might opt for a distribution model (see next section). The publisher's objective is to get maximum distribution and he shouldn't make decisions that will compromise that goal.

Finally, it is likely that the very largest publishers with global reach will need to place printing operations internationally to increase service and reduce freight. It is likely that these publishers will utilize a variety of outsourced suppliers to handle this printing until volume internationally is large enough in a particular country or region to justify their own operation.

Publishers are likely to start with outsourced POD to replenish stock (short run model). They will move to dedicated in-house production operated by printers as scale develops. They will then begin to print to demand opportunistically as the title base expands, although they are unlikely to end up with a substantial amount of single unit printing in even the mid-term.

Wholesaler

Wholesalers have less demand for any particular title than publishers do. If they can justify printing to stock, it's only on a limited basis. Most of the printing by wholesalers will be done on-demand in units of one. Here printing should be done in response to demand or, once the number of POD titles gets large, the operation will pay a very high price in terms of inventory investment and wholesalers have less margin on each sale to cover those costs than publishers do.

There are three popular models for wholesalers including:

• Short run. The short run model is used by publishers to replenish their own warehous stock, using the wholesaler as a printer. Manufacting and shipping is charged.
• Drop ship. The drop ship model is when books are printed on demand and shipped to destinations specified by either the publisher or retailer. Manufaturing and shipping are paid but the book ends up with the customer versus in the publisher's warehouse. This is basically a direct sales fulfillment model.
• Distribution model — The distribution model is simply allowing orders that come in through a wholesaler to be printed and fulfilled through that wholesaler. Since the publisher is using the wholesaler as a wholesaler, a wholesale margin will be paid along with the manufcturing cost of the book. Generally no freight is charged.

The last two of these models require a significant title base and to be linked into a substantial order stream to be practical. Consequently the size of the wholesaler is very important — it is no accident that the very largest wholesaler in the world was the first to experiment with this concept.

Retailer

Both the publisher and wholesale environments are candidates for POD of a more central variety. To the extent that a retailer is large enough to resemble either, a central POD operation can also be justified. The retail store environment, however, is clearly the most difficult environment in which to pursue POD today. Retail locations have the lowest demand of all for any particular title — except for the very hottest bestsellers, demand at a retail location for any particular title is purely statistical — most books sell too slowly to actually have a valid “rate of sale” at any retail location. This, by the way, is the cause of the terrible inventory turns seen at most booksellers.

In a retail location printing must be done exclusively in response to demand or there will be a very large inventory price to pay. The title base today is not adequate to generate a high probability of a customer finding any particular title he may be seeking. One large retail chain that was interested in piloting in-store POD was considering addressing this problem by making the POD title a novelty item, allowing the customer to personalize a book by adding a dedication, photos, etc. to any number of available public domain titles. Ultimately the decision was that there were plenty of attractive pre-printed and leather-bound books, calendars, and other items available so that the personalized POD books would end up being seen as an extraneous item and neither drive sales nor contribute to the environment of the store. The idea was temporarily shelved.

Service issues predominate at the retail level including the time that a customer must wait for the book to print and how well the experience of ordering and waiting is integrated into the retail shopping experience. There is a somewhat hackneyed suggestion that the consumer could be  “getting a cup of coffee while the book prints”, but this is not the usual customer experience of shopping for books. Further, as the number of customers seeking to print books at the same time increases, one cup of coffee can easily turn into three or more and the experience begins to disappoint.

How about the printer?

At printers there is a traditional mindset on make-ready that is very difficult to overcome. This mindset is heavily oriented toward cost-control and generating print runs of significant length to gain economies of scale. In order to overcome the need for these long print runs, technology must be implemented that reduces the fixed costs of taking, setting up for, printing and delivering an order to practically nothing and then requires shifting from a “job shop” to an “assembly line” (or flow-shop) approach to production.

The economics also work against a printer — when a retailer, wholesaler, or publisher utilizes POD they have access to margin from pricing and selling of the title that a printer does not [5] — creating an easier ability to pay for the installation and operation of equipment.

Printers do, however, have the skill to operate print-on-demand in a high quality and effective manner — something that is hard to develop and maintain in a publishing, wholesaling or retail environment. One retail chain for example was seriously considering having printing personnel available for picking orders and helping on the shipping dock during peak seasons, neglecting the differences in skill and cost and the ability to attract trained printing personnel. Probably the best way for a printer to operate is to manage outsourced operations for publishers or distributors either on-site in the publisher's location or off-site via drop-ship.

Conclusion:  everyone benefits

Print-on-demand has a very beneficial effect on participants at all stages of the supply chain. Publishers are realizing that they're missing sales and having to revert rights because books are going out of print — this needn't be so. Wholesalers are realizing that they can be generating revenue streams for themselves and their publishers through increased assortment without paying the inventory price to do it. Large retailers are implementing warehouse-based POD (similar to wholesalers) and with increasing title base and focus on the right titles, in-store POD is growing ever more practical. There are even models developing that are a hybrid of all of the above approaches, for example “network POD” in which printing operations of varying sizes are linked with distributed repositories and distributed catalogs. Acceptance of POD business models is growing at all levels of the existing book supply chain and will only increase over time.

---