Abstract
A book that has not yet been printed can appear to be a book without risk. Print on demand changes that appearance by moving manufacture closer to an order: fewer copies may have to be made before demand is observed, and a title may remain available without a conventional pre-printed warehouse stack. Those are meaningful changes. They are not the same as making publishing consequence-free. A demand-printed title still depends on usable files, production settings, accounts or suppliers, order routing, delivery, payment, policy, channel access, and a reader’s ability to find and choose it. If it is sold into a returnable retail route, a copy printed after an order may still generate a later return consequence.
This paper teaches readers to treat print on demand as a risk-timing and risk-location change, not as a general verdict on publishing independence. It introduces publisher risk redistribution as a way to ask where a named risk sits before and after a production change, and it offers the Publisher Risk Redistribution Map, an original Wexley teaching instrument for examining claims that a print model “removes risk.” The paper does not calculate a publisher’s finances or advise a production choice. Its aim is narrower and more useful: to make an unqualified claim about “no inventory” answerable through the lifecycle of one book.
The book that has not been printed
A small press announces a new title. The title can be ordered, but no stack of copies has yet been manufactured and placed in a warehouse. The press has not guessed a quantity, paid to make that quantity, watched copies sit in storage, or faced the specific problem of unsold pre-printed stock in precisely the same way that a speculative batch model can create it.
That difference is real. Ingram’s current description of its Lightning Source service, for example, presents a particular sequence: a publisher uploads a title; the title is made available through the company’s distribution network; a customer orders online or in a store; the service prints the needed book; and the publisher chooses whether the copy is shipped to the publisher or the customer.[1] The sequence moves a manufacturing trigger closer to an order.
But look again at the supposedly absent book. It already requires a viable file. It requires some route by which a reader or intermediary can encounter and place an order. It depends on a service arrangement, production and quality conditions, a payment relationship, information that lets the book be recognized, and a way to fulfill the resulting request. Its risk has not disappeared; some of it has not yet been triggered, some of it has moved, and some of it remains elsewhere.
This is the paper’s first teaching question:
Which risk has been reduced because no copy was made before demand was observed, and which risks have simply changed their time, holder, or form?
The question matters because print on demand is routinely allowed to mean too many things at once. It can refer to a production technology, a method of keeping a backlist available, a service relationship, a retail/fulfillment path, or a general promise that smaller publishers can avoid the hazards of a conventional print run. Those are related but not identical propositions.
A book may be printed after order without being easily discoverable. It may be technically printable while depending on a platform or supplier. It may be returnable under a channel policy. It may be timely for one order and unsuitable for another. It may avoid one kind of storage commitment but require durable files, staff attention, quality checks, customer communication, and service continuity. The central error is not believing that print on demand can help. The error is saying that because a warehouse risk has changed, publishing risk has vanished.
Print on demand is a timing shift, not a single outcome
A careful history begins by refusing a single-invention story. Suzanne Wilson-Higgins’s 2017 book on print on demand for academic books describes a viable commercial model as emerging from the convergence of online bookselling, digital printing, digital document-workflow management, and computerized small-parcel logistics.[2] The accessible publisher description is a useful limit as well as a useful history. A demand-printed book does not materialize because one printer exists. Its possible route joins files, ordering, manufacturing, and physical movement.
That convergence can change availability in substantial ways. A 2012 account from NYU’s publishing program described a particular Espresso Book Machine at McNally Jackson as producing fully bound paperback books in minutes. The article connected the machine with a networked title database, uses involving out-of-print books, and self-publishing activity in a bookstore setting.[3] The point is not that every bookstore acquired such a machine or that the installation proved universally profitable. It is that the place and time of manufacture could be rearranged. A title could become a copy nearer to a reader’s request rather than only through a prior large batch sent into the chain.
A 2025 qualitative study of Spanish university presses makes a parallel point in a different setting. It examines print on demand as a response to specialized or unpredictable academic-book demand, especially where storage and overproduction create difficult conditions for titles with smaller or intermittent audiences.[4] The study’s setting matters. It does not prove that every trade publisher, self-published author, or commercial frontlist title should use demand printing. It does demonstrate why a university press might see a new relation between an uncertain future reader and a copy not yet manufactured.
The change can be named more exactly through five terms.
| Term | Meaning in this paper | What it does not mean |
|---|---|---|
| Risk reduction | A named exposure becomes lower under stated conditions. | That all other exposures disappear. |
| Risk transfer | A stated service, account, policy, or party carries a consequence differently. | That responsibility or agency has vanished. |
| Risk deferral | A consequence becomes relevant later in the book’s lifecycle. | That it will never arise. |
| Risk transformation | An exposure changes kind: stock risk may become a file, quality, service, payment, or channel condition. | That the new condition is automatically worse or better. |
| Risk concentration | A consequence that was distributed elsewhere rests more heavily with a smaller organization. | Proof that demand printing is inherently harmful. |
The terms are deliberately practical. They let a reader say, for example, “This route may reduce the publisher’s exposure to unsold copies manufactured before demand, but it may defer manufacture until order and leave returnability, discovery, service, quality, payment, and delivery conditions active.” That sentence is slower than “POD is risk-free.” It is also capable of being tested.
The 2022 Publishing Research Quarterly study by Rachel Done, Rylee Warner, and Rachel Noorda shows why the slower sentence is needed. Drawing on qualitative research developed with Portland State University, the Independent Book Publishers Association, and PubWest, the study treats offset economies of scale, POD costs and four-color quality, warehousing, returns, distribution practices, and supply-chain disruption as distinct issues.[5] It does not offer a universal prescription. Its recommendations include using POD for proof copies and gap runs, reducing print runs, and promoting books more effectively to minimize returns.[5] A production decision, a promotion problem, a return condition, and a delivery issue may affect each other, but they are not one problem with one solution.
Two order paths for one constructed title
To see the difference, follow one invented book: The Orchard at Low Tide, a 176-page black-and-white scholarly/niche paperback from the fictional Fieldglass Press. It has been edited, designed, and prepared for print. The press, author, book, and reader in this example are constructed. No price, print run, margin, discount, sales volume, provider, or delivery time is being asserted.
The same title can take two different logistical shapes.
Path A: speculative stock. Before enough demand is observed, Fieldglass authorizes a batch. Copies are manufactured, paid for under some arrangement, stored or entered into a distribution route, and later offered to bookstores or readers. The batch can make immediate stock available, but it may also create a specific exposure: copies exist before their eventual buyers do. If sales do not match the anticipated pathway, the press has to manage the consequences of stock, reprinting, delivery, returns, and cash timing through its actual agreements.
Path B: demand-printed. Fieldglass prepares a usable digital file and an order-to-manufacture route. A later order triggers production of a requested copy. The press may avoid having the same amount of pre-printed stock waiting for a future buyer. But the title has not become free of operational conditions. Its file must remain usable. An order must travel through a service and production process. The copy must meet a relevant quality condition, be delivered through a stated route, be visible enough for a buyer or bookseller to request it, and be governed by the account, payment, and return conditions of that route.
The comparison is not a scorecard. It is a risk movement sheet.
| Risk object | Path A: a possible trigger | Path B: a possible trigger | Where the risk may sit | What a real case would need to show |
|---|---|---|---|---|
| Unsold pre-printed stock | A batch is manufactured before enough demand is known. | A copy may not exist before order in the same way. | Publisher, printer, warehouse, financing or distribution arrangement. | Print records, stock reports, invoices, agreements, sales data. |
| File and production readiness | A file is needed before the batch is manufactured. | A file must remain usable whenever an order is triggered. | Publisher, production workflow, service arrangement. | Version records, preflight logs, production specifications, service documentation. |
| Quality and format fit | A batch’s materials and production choices become fixed for the run. | A demand route may have its own format, color, quality, and technical conditions. | Publisher and relevant production service. | Proofs, specifications, complaints, quality records, terms. |
| Order and delivery | Copies must move from stock through a delivery/retail path. | An order must be routed, printed, and fulfilled through a stated path. | Distributor, retailer, fulfillment service, publisher, customer. | Order logs, shipping records, channel agreements, customer communications. |
| Demand and discovery | Existing stock still needs readers, booksellers, or institutions to choose it. | A printable title still needs someone to encounter and order it. | Publisher, audience routes, retailers, platforms, cultural institutions. | Catalog records, marketing records, channel data, reader/reviewer evidence. |
| Returnability and payment | A returnable stock route may generate later financial consequences. | A returnable POD order can also create a later return consequence. | Retail policy, provider terms, publisher account, payment cycle. | Contracts, policies, statements, return records, qualified review. |
The point of this sheet is not that Path A has all the risk and Path B has none. The first row identifies a real difference: a demand-printed path can change the exposure created by a batch made before demand. The last rows identify the error in stopping there. A reader may encounter an available title only if information, channel, and attention conditions hold. A physical copy may still move through policies and accounts that are not settled by its production trigger. And a small press may find that it has exchanged a warehouse decision for another kind of service, file, quality, payment, or communication responsibility.
This is why the language of redistribution is more accurate than a general language of liberation. It does not deny gains. It asks where the consequence has gone.
The risks after the click
An order is not the end of a POD analysis. It is a moment in a lifecycle. The useful question is not simply “Was the book printed after an order?” but “What becomes live before availability, at order, at delivery, after a sale, and after a policy change?”
Before public availability: files, capability, and service conditions
A title can be unprinted and still require an operational environment. Ingram’s public workflow begins with a publisher uploading titles.[1] The Spanish university-press study identifies advanced technologies and training as relevant to print quality and digital-process integration; it also notes perceived quality, visibility, and market acceptance as continuing barriers in some segments.[4] These are not universal obstacles, and the paper does not quantify them. They are reminders that a digital source file and a production route are conditions of availability, not mere background details.
Rick Anderson’s 2011 early-adopter account of an Espresso Book Machine makes the point vividly, though only for one university-library setting. He described quick output and a case in which an obscure work could be printed for a faculty member, while also reporting warm-up time, technical failures, unresolved color capability, limitations in current-content availability, weak search and metadata, staff effort, and early-adopter expense.[6] The historical lesson is not that early machines define current POD. It is that production speed does not exhaust the conditions of a book’s public usefulness. A copy can be made quickly and still be hard to find, technically constrained, or expensive in staff attention.
At order: a manufacturing trigger is not a whole supply chain
In a demand-printed route, an order triggers a new sequence. Ingram’s description names customer order, print, and shipment; Wilson-Higgins’s framing makes the larger dependence visible by joining printing to online bookselling, workflow, and parcel logistics.[1][2] The event of manufacture may be later, but the book still needs a legible order path and a physical route after printing.
This is where a common false contrast appears. “Traditional” is treated as a long supply chain and POD as no supply chain. The better comparison is between different supply-chain configurations. One model may make and hold more copies earlier; another may connect a file to a later manufacturing and fulfillment event. Both require materials, information, arrangements, and people. Neither guarantees that every title is easy to buy, every destination is equally served, or every reader can locate the right book.
At delivery: fulfillment remains a publishing question
The Done, Warner, and Noorda study treats delivery, returns, supply-chain disruption, promotion, and printing as connected but distinct concerns.[5] Its discussion of pre-orders shows the same logic. Pre-orders may help a publisher size production and anticipate delivery, but they do not transform uncertain demand into a certainty. They provide one kind of evidence before a later production and shipping decision.
For the constructed Fieldglass title, the question is therefore not whether POD “solves delivery.” It is: What route exists from a reader’s or bookseller’s request to the finished copy, and which party controls the corresponding service, timing, communication, and exception handling? That question can be answered only with the facts of a real arrangement. It cannot be answered by the term POD alone.
After a retail order: returnability can remain active
A return-status example makes the distinction especially clear. IngramSpark’s 2021 explanation of its own terms says that POD and returnability can coexist. In the service described, a copy may be printed when ordered by a bookstore and later returned if the title is designated returnable and the bookstore does not sell it.[7] The provider also describes return-related charges, compensation timing, and continuing return rights for a period after certain status changes.[7]
This paper does not reproduce the provider’s fees, interpret an agreement, or advise anyone to select a return option. The point is analytical. A copy printed after a retailer’s order is not necessarily a copy whose commercial consequence ends at manufacture. A production trigger has moved, while a channel’s return policy can create a later condition. That is a simple instance of risk deferral and transfer: pre-printed stock exposure may be different, but return and payment exposure may be activated later through a particular policy relationship.
No reader should infer from one provider’s page that every POD service, country, retailer, or title behaves in the same way. The case is useful precisely because it is bounded. It makes visible the category of question a publisher or researcher must ask: When the title is returnable, who carries what consequence if a copy comes back after the reader did not buy it?
Across the lifecycle: availability depends on more than a machine
POD’s strengths are real. The NYU and university-press accounts show why a demand-printed path can support out-of-print, niche, backlist, specialized, or intermittent-demand work in particular settings.[3][4] But continued availability does not rest only on the ability to print one copy. It can depend on a maintained file, accepted production specifications, a continuing service relationship, order and metadata pathways, a channel’s willingness to carry or process the title, and a reader’s capacity to discover and choose it.
This is not a WEXLEY-WP-042 analysis of metadata handoffs, nor is it a WEXLEY-WP-041 action plan for a contemporary self-publishing stack. Here, discoverability and service dependency are included for one reason: they are risk locations in a production lifecycle. A book that cannot be discovered may not generate demand. A file that cannot be processed cannot become a copy. A route with uncertain policy or payment conditions cannot be declared risk-free merely because the warehouse is smaller.
The Publisher Risk Redistribution Map
The Publisher Risk Redistribution Map is an original Wexley teaching instrument. It is for readers who encounter a statement such as “POD removes the risk of publishing” and need to turn it into a claim that evidence can support or limit. It is not a financial forecast, a production-method recommendation, a provider comparison, a return-policy calculator, legal counsel, or a substitute for reading a real contract with qualified help.
| Map field | The reader asks | A proportionate output | A warning sign |
|---|---|---|---|
| 1. Risk object | What could be lost, delayed, degraded, or made unavailable? | “Unsold pre-printed copies.” | “Risk” with no named object. |
| 2. Lifecycle trigger | At what event does it arise? | “Before demand is observed” or “when a return is processed.” | Treating every consequence as if it arises at printing. |
| 3. Earlier holder/location | Who or what carried it in the earlier route? | “A publisher’s stock and storage arrangement.” | Inventing a universal traditional model. |
| 4. New holder/location | Where does it sit in the demand-printed route? | “A later order trigger plus a service/account/channel relation.” | Saying the risk has evaporated. |
| 5. Change type | Is the exposure reduced, transferred, deferred, transformed, concentrated, or unchanged? | “Pre-printed stock exposure is reduced; return exposure may remain in this named route.” | Treating a shift as an automatic improvement. |
| 6. Evidence required | What would substantiate a real claim? | “Terms, invoices, order data, return records, service logs, or agreements.” | Inferring business practice from a tool’s existence. |
| 7. Non-solution | Which adjacent problem does this map not solve? | “Reader discovery, editorial quality, rights, or total demand.” | Letting a production choice become a total publishing verdict. |
| 8. Escalation question | When does the issue require other expertise? | “A contract, policy, financial, rights, or service question needs appropriate review.” | Treating a teaching map as professional advice. |
Apply the Map first to the invented Fieldglass title’s potential pre-printed stock. In Path A, the risk object is stock made before enough demand is observed. The trigger is the batch decision. In Path B, the same object may be reduced or deferred because a copy is manufactured later. The map does not claim that the Fieldglass title will make money, reach readers, avoid returns, or maintain quality. It records only a changed relationship between a future order and a physical copy.
Now apply it to a possible return in a named retail route. The risk object is not “POD.” It is a later return-related and payment consequence. The trigger is a copy’s return under stated terms. In a particular provider arrangement, the new holder may include the publisher account and the retail/service policy relation.[7] The non-solution is just as important: mapping return exposure does not decide whether the title will be stocked, reviewed, discovered, or cherished by its readers.
The Map turns a technological slogan into an evidence question. If someone says a service has “eliminated inventory risk,” the reader asks: Which inventory? Before which event? For whom? Under what terms? What later condition replaced or remained beside it? If someone says a demand-printed book has “no risk,” the reader asks for the file, order, service, delivery, channel, return, payment, quality, and discoverability conditions that sentence has skipped.
A diagnosis, not a verdict
Print on demand can change publishing in ways worth naming accurately. It can reduce the need to make a large speculative batch before enough demand is known. In documented cases, it has helped make particular niche, academic, out-of-print, and locally requested books available through different manufacturing timings and locations.[3][4] That is a real capacity, not a myth to be dismissed.
The risk-accounting language does not diminish that capacity. It protects it from overclaim. A book made after order can still depend on maintained files, technical fit, service relationships, order data, payment, delivery, channel policies, reader discovery, quality expectations, and return conditions. Those exposures may be lower, higher, shared, deferred, or simply different according to a real route. Their presence does not prove that POD has failed. Their absence cannot be assumed from the words on demand.
The final calibration is therefore straightforward.
| Claim version | Editorial judgment |
|---|---|
| “POD makes publishing risk-free.” | Too broad. It treats a changed manufacturing trigger as a total solution. |
| “POD eliminates inventory risk.” | Incomplete. It does not name the inventory, holder, timing, route, or adjacent conditions. |
| “In a stated production and channel route, demand printing can reduce or defer exposure to speculative pre-printed stock; it does not by itself resolve demand, return, file, quality, service, payment, delivery, visibility, or policy conditions.” | A proportionate claim. It names the changed risk and leaves the remaining questions available for evidence. |
A demand-printed book is not a book without risk. It is a book whose risks must be read in a different order.
Boundary statement
This paper does not give financial, legal, tax, pricing, inventory, return-policy, or service-selection advice. It does not calculate print costs, margins, returns, sales, delivery performance, environmental impact, or market concentration. It does not rank providers or declare one publication route superior. The constructed Fieldglass Press scenario contains no claim about a real publisher. The paper teaches a bounded diagnostic method: name the risk, locate its trigger, identify its holder, specify how it has moved, and state what the production choice does not solve.
References
[1]: Ingram Content Group, “Print-On-Demand.”
[2]: Wilson-Higgins, Suzanne. The Impact of Print-On-Demand on Academic Books. Chandos/Elsevier, 2017. Google Books record.
[3]: NYU M.S. in Publishing: Digital and Print Media, “McNally Jackson’s Espresso Book Machine: Write it. Print it. Now!” (2012).
[4]: Magadán-Díaz, Marta, and Jesús I. Rivas-García. “The Impact of Print-on-Demand on Spanish University Presses.” Learned Publishing 38 (2025): e1658.
[5]: Done, Rachel, Rylee Warner, and Rachel Noorda. “Publishing Distribution Practices: New Insights About Eco-Friendly Publishing, Sustainable Printing and Returns, and Cost-Effective Delivery in the U.S.” Publishing Research Quarterly 38 (2022): 364–381.
[6]: Anderson, Rick. “The Good, the Bad, and the Sexy: Our Espresso Book Machine Experience.” The Scholarly Kitchen, Society for Scholarly Publishing (2011).
[7]: IngramSpark Staff. “Understanding Book Returns with IngramSpark” (2021).