Free Back-Year DOIs Will Not Clean the Archive for You

For years, the quietest archive problem in journal publishing has been the stack of older content that everyone agrees should be easier to cite, discover, preserve, and connect, but nobody quite wants to pay to process. The issue is rarely one missing DOI. It is a decade of PDFs with thin landing pages, inherited titles with partial ownership records, special issues from another platform, supplements parked outside the article workflow, and metadata that was good enough for a website but not good enough for the research nexus.

Crossref just changed the economics of that backlog. On August 13, 2026, Crossref announced that, effective January 1, 2027, it will reduce most current-year Content Registration fees and remove fees for all back-year records: https://www.crossref.org/blog/fee-update-lowering-and-removing-doi-registration-fees-in-january-2027/. Crossref defines back-year records as works produced more than two full years ago; in 2027, that means content published in 2024 or earlier.

That is not just a billing update. For journals, societies, repositories, archives, and university publishers, it creates a short planning window. The cost barrier to registering old records is falling away, but the stewardship barrier remains. A free DOI registration can still create a bad public record if the journal cannot prove what the item is, who is responsible for it, whether it already has a DOI, and where the DOI should resolve.

The Discount Is Not The Work

Crossref is explicit that the larger change is the removal of fees for back-year records. The announcement says this matters for repositories, archives, national libraries, high-volume low-budget collections, archival digitisation projects, and members absorbing another organisation's unregistered backlist. That is exactly where the operational risk sits.

Backfile registration is not a checkout cart. It is an editorial and metadata exercise. Before a journal registers a 2011 article, staff need to know whether the article is the version of record, whether a previous publisher already registered it, whether the journal changed owner or host, whether the landing page is stable, whether article titles and author names match the published PDF, and whether corrections, retractions, expressions of concern, or special-issue context should travel with the record.

The fee change makes that work easier to justify. It does not make the work optional. If anything, removing the per-record price may tempt teams to batch-register first and reconcile later. That would be the wrong order. Old records can be messy precisely because nobody has looked at them as records for years.

Build A Backfile Ledger Before January

The first useful artifact is not an XML deposit. It is a backfile ledger. List the content sets the journal might register: volumes, issues, article types, supplements, editorials, book reviews, conference abstracts, corrections, withdrawn items, and hosted archives from title transfers. Then mark each set with basic status: already registered, probably unregistered, uncertain, excluded, or needs rights review.

That ledger should separate bibliographic completeness from stewardship confidence. A spreadsheet may have clean titles, dates, and author names while the rights trail is poor. A platform export may include DOIs for current content but leave early-online articles, legacy supplements, or merged titles ambiguous. A digitized archive may have excellent PDFs but weak article-level metadata. Treat those as different problems, not as one generic backlog.

Crossref's August 12 system-rebuild announcement is relevant here because it shows where the infrastructure is headed: separate services for registration, reports, authentication, billing, matching, metadata enrichment, and DOI resolution work rather than one tightly coupled legacy core: https://www.crossref.org/blog/investing-4.9-million-of-our-surplus-in-rebuilding-the-crossref-system/. Journals should read that as a signal. Metadata flows are becoming more visible, more reportable, and more connected. A weak archive record will have more places to fail.

The Hard Cases Are Usually Ownership Cases

Back-year content often crosses institutional memory. A society moved from one publisher to another. A university press merged two journal sites. A platform migration dropped old issue URLs. A sponsored Crossref arrangement changed hands. A journal title split, changed name, or restarted volume numbering after a dormant period. Those situations are not edge cases for archive work; they are the work.

This is where Crossref's draft Member Practices matter. Crossref opened consultation on August 10, 2026 and invited feedback by October 9, 2026, with final Member Practices planned for December 2026: https://www.crossref.org/blog/share-your-feedback-on-the-new-crossref-member-practices/. The draft says members should register records only where they are the title owner or appointed publisher, should not create duplicate DOI records, and should continue using existing DOIs when ownership changes.

That principle is especially important once back-year registration becomes free. A low-friction archive project must not become a duplicate-DOI project. If an acquired title already has DOI records under a previous member prefix, the fix is transfer and metadata maintenance, not re-registration. NISO's Transfer Code of Practice Version 5.0, published in March 2026, also frames title transfer as a continuity problem for access and publishing workflows: https://www.niso.org/publications/rp-24-2026-transfer. Archive registration plans should sit inside that continuity discipline.

A Landing Page Is Part Of The Archive

The most visible failure in old DOI work is a DOI that resolves to a thin, confusing, or unstable page. For current articles, staff usually know where the live page belongs. For backfiles, the page may have been generated by a migration script, preserved as a PDF-only download, or reconstructed from a print issue. The DOI record should not be better organized than the page it points to.

For each archive set, test the landing-page rule before registration: a reader should be able to identify the journal, article title, authors, publication date, volume and issue where applicable, DOI display, article access route, and any correction or withdrawal notice. If a publisher cannot support stable article pages for a content set yet, that set is not ready for bulk registration. The January 2027 fee change is a reason to start now, not a reason to skip page quality.

The same applies to metadata relationships. Crossref says more back-year records create more material for matching citations, funding, affiliations, versions, and other relationships. That promise depends on deposits being specific enough to match. Minimal records may produce identifiers, but richer records make the archive useful to readers, librarians, discovery systems, funders, and integrity reviewers.

Do Not Let Finance Own The Plan Alone

Because this is a fee announcement, it will naturally land with finance, operations, or platform teams. Journal leaders should broaden the room. Editorial staff know which supplements are sensitive. Production knows where article data drifted during migrations. Librarians know where access continuity has been fragile. Society officers may know title history that is not documented in platform fields. Research integrity staff know which older items carry corrections or contested authorship.

A good plan asks these groups for bounded decisions. Which archive sets are in scope for 2027? Which must be held back until title ownership is confirmed? Which old article types should receive article-level DOI records, and which should remain issue-level or excluded? Which records need Crossmark, license, reference, funding, ORCID, ROR, or relation metadata? Which landing pages need remediation before deposit?

This is also a budgeting opportunity. The registration fee may be zero for qualifying old records, but staff time, metadata cleanup, platform fixes, XML generation, quality assurance, and post-deposit monitoring still cost money. Put those costs into the 2027 plan now. Otherwise, the free-record announcement will produce an unfunded mandate inside the editorial office.

Practical Takeaway For Journal Leaders

Before January 1, select one completed volume from 2024 or earlier and run a full archive-registration rehearsal. Confirm title stewardship, search for existing DOI records, inspect article landing pages, compare metadata against the published files, identify correction or withdrawal status, decide what relationships to include, and produce a test deposit package or vendor brief. Record every manual decision needed to get from archive inventory to trustworthy registration.

If the rehearsal is smooth, the journal has a credible backfile program. If it depends on one staff member's memory, unverified assumptions about old publishers, or landing pages that do not identify the item cleanly, fix that first. Crossref has made back-year registration easier to afford. Journals still have to make it worth trusting.