The choice between XML First and XML Last typesetting is the single most consequential workflow decision a journal or book publisher makes. It determines your production costs, your turnaround times, your ability to deliver multiple formats simultaneously, and how much rework you absorb every time your template changes.
Most publishers make this decision based on vendor recommendations, industry convention, or — most commonly — inertia. This guide gives you the framework to make it deliberately, based on your actual content type and commercial requirements.
What does "XML First" actually mean?
In an XML First workflow, the manuscript is structured as XML before typesetting begins. The XML is the authoritative source. From that single source, the production system generates print-ready PDF, ePub3, HTML, and structured database records simultaneously — in a single automated pass after the XML is validated and approved.
The typeset page is rendered directly from the XML using stylesheets (XSLT or CSS). Corrections are made to the XML, not to the InDesign file — so every format automatically reflects the correction. There is no format-specific rework.
What does "XML Last" mean?
In an XML Last workflow, the compositor typesets the content in InDesign or QuarkXPress first, producing a print-ready PDF. XML is then extracted from the typeset file using scripts and export tools — often with significant manual cleaning required afterwards.
The InDesign file is the authoritative source. Digital formats are downstream products derived from the print layout. Corrections made to the InDesign file do not automatically propagate to the XML or digital formats — each must be updated separately.
The honest cost comparison
XML First — higher setup, lower ongoing cost
XML First requires a significant upfront investment: XML schema definition, stylesheet development, compositor retraining, and QA process redesign. For a journal programme, this setup typically takes 8–16 weeks and represents a meaningful capital cost.
But the ongoing per-article cost is substantially lower — and drops further as volume increases. At 200+ articles per year, XML First is almost always cheaper per article than XML Last. The automation removes entire manual stages from the workflow.
XML Last — lower setup, higher ongoing cost
XML Last can begin immediately with existing InDesign skills and templates. There is no schema investment, no stylesheet development, and no compositor retraining. It's the right choice when you need to start production now, or when your volume doesn't justify the XML First setup cost.
The ongoing cost is higher per article because XML extraction is a semi-manual process that requires cleanup, validation, and QA. Every correction must be applied in multiple places. Template changes require InDesign file updates across all affected articles.
The break-even point between XML First and XML Last setup cost is typically reached somewhere between 80 and 150 articles per year, depending on content complexity. Below that, XML Last is usually more cost-effective. Above it, XML First almost always wins.
Content complexity matters
XML First delivers its greatest efficiency advantages on content with consistent, predictable structure — standard journal articles, textbook chapters, reference works with defined schemas. The more consistent the structure, the more the automation can do without human intervention.
Highly variable content — trade books, illustrated titles, design-led educational materials — is harder to automate through XML First because the layout decisions are content-dependent. XML Last, or Non-XML composition, may be more appropriate even at high volumes for this content type.
The hybrid approach
Many publishers run both workflows simultaneously — XML First for their journal programme (high volume, consistent structure) and XML Last or Non-XML for their book programme (lower volume, variable design). This is usually the right answer for publishers with both journals and books in production.
What Daintra recommends
- Under 80 articles/year: XML Last. The setup cost of XML First won't be recovered.
- 80–200 articles/year: Evaluate carefully. Content complexity and multi-format requirements tip the balance.
- Over 200 articles/year: XML First. The per-article savings accumulate quickly at volume.
- Trade books only: Non-XML InDesign composition is usually most cost-effective regardless of volume.
- Accessibility or repository compliance required: XML First significantly simplifies EPUB Accessibility 1.1 and PubMed Central compliance.
