Implementation pitfalls

Best practices and common pitfalls

The most-common implementation pitfalls in publisher CRediT adoption, with sourced corrections. The list is drawn from the peer-reviewed critique literature and from CRediT Standing Committee implementation reviews.

CRediT works. It is widely adopted across the scholarly publishing ecosystem. But the Hosseini et al. study in Learned Publishing (2026), the parallel Scientometrics paper on disciplinary differences, and the wider critique literature converge on a small set of recurring implementation pitfalls that determine whether a publisher’s CRediT support actually carries through the metadata chain to the downstream consumers it is supposed to feed. The list below sets out the recurring pitfalls and the corresponding good practice.

The companion implementation scorecard formalises the good-practice checks into a self-assessment rubric. The submission systems guide and the JATS XML implementation guide cover the technical detail.

Pitfall 01

Capturing CRediT as narrative only

What goes wrong: The submission form asks authors to paste a free-text Author Contributions paragraph into a textarea field. The paragraph is preserved in the manuscript; nothing is captured as structured metadata.
Good practice: Present a controlled-vocabulary role picker that emits structured selections. The Hosseini et al. study in Learned Publishing (2026) documents this gap as the single largest cause of patchy downstream CRediT availability.

Sources

Hosseini et al. (2026) — Learned Publishing

Pitfall 02

Excluding acknowledged contributors

What goes wrong: CRediT is restricted to named authors. The medical writer who drafted the manuscript, the statistician who ran the analysis and the laboratory technician who ran the assays are listed in the Acknowledgements without role attribution.
Good practice: Allow CRediT-style role attribution for acknowledged contributors, either through a structured Acknowledgements field or through an editorial workflow that captures the contributor and the role explicitly. The Wikipedia editorial summary of CRediT criticisms identifies the acknowledged-contributor exclusion as the most-cited limitation of current adoption.

Sources

Wikipedia: Contributor Roles Taxonomy — criticism section

Pitfall 03

Treating CRediT as a substitute for ICMJE

What goes wrong: Author guidance suggests that CRediT replaces the ICMJE Vancouver criteria for authorship qualification. Reviewers and editors then attempt to use the CRediT statement to decide who should be on the author line.
Good practice: CRediT records what people did; ICMJE decides who qualifies as an author. The two are complementary. Author guidance should reference both explicitly: ICMJE for authorship qualification, CRediT for contribution recording. The ICMJE Recommendations document is explicit on this point.

Sources

ICMJE — Defining the Role of Authors and Contributors

Pitfall 04

Listing a generative AI tool as a contributor

What goes wrong: A submitted manuscript lists ChatGPT or another large-language-model tool as a co-author or as holding a CRediT role.
Good practice: AI tools cannot be authors and cannot hold CRediT roles. ICMJE, COPE, Nature Portfolio and NEJM are unambiguous on this. Editorial workflow should reject AI-tool authorship and AI-tool role attribution; AI use is disclosed separately in a Methods or AI Disclosure section. The CASRAI AI disclosure guide documents the canonical wording.

Sources

Pitfall 05

Emitting CRediT without canonical NISO URIs

What goes wrong: The published JATS XML contains the role names (“Conceptualization”, “Methodology”) but no vocab-term-identifier attribute pointing at the canonical NISO URI.
Good practice: Always include the vocab-term-identifier attribute with the canonical URL from casrai.org/credit. Downstream consumers rely on the URI to disambiguate role labels across vocabularies and to support multilingual deployments.

Sources

Pitfall 06

Stopping at JATS without depositing to Crossref

What goes wrong: The published article XML carries CRediT role attributes; the Crossref deposit profile is configured to the older schema and omits the contributor_role element.
Good practice: Configure the Crossref deposit profile to use schema 5.5 or higher, with the contributor_role element populated. Without the deposit step, institutional CRIS systems and bibliometric services cannot ingest the role data, even if it is present in the published XML.

Sources

Crossref deposit schema 5.5

Pitfall 07

Conflating Methodology with Software

What goes wrong: Editorial guidance asks authors to merge Methodology and Software into a single bucket, on the basis that the line is fuzzy.
Good practice: The line is fuzzy in some cases but the standard distinguishes them deliberately. Methodology covers the design of the approach; Software covers the implementation. Where a contribution genuinely spans both, attribute both roles to the contributor.

Sources

CRediT contributor roles defined

Pitfall 08

Ignoring the degree-of-contribution qualifier

What goes wrong: The submission system captures roles without the optional lead / equal / supporting qualifier; the published statement reads as a flat list of roles per author.
Good practice: Where the publisher's submission system supports it, surface the qualifier and encode it in JATS via the specific-use attribute. The qualifier adds genuine signal — especially in multi-author papers where several authors share a role. Most submission systems support it; most publishers do not require it; the gap is operational, not technical.

Pitfall 09

Treating disciplinary fit as a CRediT problem

What goes wrong: Humanities or qualitative-research papers are forced into the standard 14-role taxonomy without acknowledgement that the role definitions need translation.
Good practice: Recognise that CRediT was shaped by biomedical and STEM contribution patterns and is a forced fit in humanities and some social-science contexts. Publish discipline-specific guidance (the CASRAI qualitative and humanities discipline guides are starting points) and accept a perfunctory statement on genuinely sole-author work.

Sources

Beyond authorship: disciplinary differences — Scientometrics 2026

Putting it together

None of the pitfalls above is fatal. CRediT continues to expand across the scholarly publishing ecosystem. The pitfalls are the live edges of an active standard, and they are exactly what an honest reference site should be highlighting — both to acknowledge the standard’s limits and to surface the open questions that the next decade of CRediT work needs to address.