REF 2029 decouples research outputs from named researchers: institutions submit outputs to a Unit of Assessment rather than to an individual, judged instead on a “substantive link” between the institution and the work. This shifts REF evaluation from researcher performance to institutional research environment, raising the stakes for how contribution is separately evidenced.
Decoupling is the REF 2029 policy mechanism that removes the formal link between a submitted research output and the named staff member who produced it, so that outputs are assessed as belonging to a Unit of Assessment (UoA) rather than to an individual author.
- What does “decoupling” mean under REF 2029?
- How does the substantive link test work?
- How does this differ from REF 2014 and REF 2021?
- What does decoupling signal for evidencing contribution?
- Answer-first questions on REF 2029 outputs
- Implications and outlook for institutions
What does “decoupling” mean under REF 2029?
Under REF 2029, outputs are submitted to a Unit of Assessment without staff details attached to individual pieces of work. No researcher name is carried through the submission record, and no output is presented as belonging to one specific author for assessment purposes.
This reshapes the submitted category itself: what was previously called “outputs” is now Contributions to Knowledge and Understanding (CKU), which carries a 55% weighting in the overall institutional score, according to REF 2029’s official Section 1 overview. The remaining weighting splits between Strategy, People and Research Environment (20%) and Engagement and Impact (roughly 25%).
Practical consequences of decoupling include:
- No minimum or maximum number of outputs required from any individual staff member.
- A recommended (not mandatory) ceiling of five outputs per researcher, reinstated in REF 2029’s December 2025 update after an earlier proposal for no cap at all.
- Eligibility broadened to outputs produced by a wider range of roles, including technicians and research managers, not only conventionally “REF-able” academic staff.
How does the substantive link test work?
Instead of an author-output link, REF 2029 requires institutions to demonstrate a substantive link between the submitting institution and the output. A substantive link is generally established through an eligible employment relationship of at least 12 months at a minimum of 0.2 full-time equivalent (FTE), per REF 2029’s guidance on Contributions to Knowledge and Understanding.
Where employment alone is insufficient or the researcher has since left, institutions may point to supporting evidence such as:
- Internal research support, including funding for materials, technical assistance, or conference attendance.
- Evidence of work-in-progress presentations, internal or external.
- An external grant supporting a relevant programme of research held during the employment period.
Outputs cannot be claimed where the substantive link occurred only after the output was made public and the author was subject to compulsory redundancy — a safeguard REF 2029 added following sector feedback on the risk of institutions retaining the outputs of staff they had made redundant.
How does this differ from REF 2014 and REF 2021?
Decoupling is not new to REF 2029; it extends a direction of travel set out in the 2016 Stern Review of the REF, which recommended non-portability of outputs to reduce “poaching” incentives that favoured wealthier institutions. Each REF cycle has progressively loosened the tie between researcher identity and institutional claim.
| REF cycle | Output–researcher link | Portability on staff move |
|---|---|---|
| REF 2014 | Output captured entirely by the institution employing the researcher at the census date | Full transfer with the researcher |
| REF 2021 | Output could be captured by both the origin and destination institution on a move | Partial (dual claim) |
| REF 2029 | Output captured by the institution demonstrating a substantive link; no named author attached | Restricted; long-form outputs (e.g. monographs) retain five-year portability |
The volume of outputs an institution must submit is unchanged in formula terms: it remains the institution’s staff volume measure (FTE) multiplied by 2.5 at UoA level, consistent with the REF 2021 approach reported by REF 2029 planning guidance published by the University of Reading and others.
What does decoupling signal for evidencing contribution?
REF 2029’s decoupling addresses institutional-level attribution — which organisation gets credit for an output — but it does not answer a separate, longstanding question: which individuals, and in what capacity, actually contributed to producing it. That question sits squarely in the domain of contributorship taxonomies rather than research assessment exercises.
This is where the REF 2029 shift and the contributor-role movement intersect without colliding. CASRAI originated the CRediT contributor role taxonomy in 2014 to make individual contribution to scholarly outputs explicit and machine-readable; the standard is now stewarded by NISO as ANSI/NISO Z39.104-2022. Where REF 2029 deliberately removes the researcher’s name from the assessment record, CRediT statements retained in the published output itself remain the mechanism by which an individual’s specific role — conceptualisation, methodology, writing, data curation — stays evidenced and citable, independent of how any national assessment exercise chooses to allocate institutional credit.
For institutions, the practical implication is that internal recognition, promotion, and workload evidence can no longer lean on REF submission data as a proxy for individual contribution, because REF 2029 submissions will not carry that data. Institutions building internal case files for tenure, promotion, or grant applications need contribution evidence that exists independently of the REF submission — structured CRediT role statements attached to outputs, ORCID-linked publication records, and clear internal documentation of the “substantive link” evidence (funding, supervision, work-in-progress records) that REF 2029 itself now requires institutions to compile.
Guidance on research administration practice and on the underlying CRediT taxonomy is a reasonable starting point for research offices building this parallel evidence base ahead of the REF 2029 submission window.
Answer-first questions on REF 2029 outputs
What are the changes in REF 2029 for outputs?
REF 2029 renames outputs as Contributions to Knowledge and Understanding (CKU), weighted at 55% of the overall score, removes individual minimum and maximum output requirements, reinstates a recommended cap of five outputs per researcher, and requires a substantive link rather than a named author for eligibility.
Why is REF 2029’s decoupling of outputs important?
It marks a formal shift in what REF measures: institutional research environment and support, not individual researcher performance. Funding allocation logic follows institutions, so REF 2029 aligns assessment evidence with who receives the funding — the institution — rather than the individual author of an output.
What outputs are eligible for REF 2029?
Eligible outputs must be brought into the public domain between 1 January 2021 and 31 December 2028, meet REF 2029’s open access requirements, and demonstrate a substantive link to the submitting institution. Outputs solely authored by PhD students or teaching-only staff are generally not eligible.
Are REF 2029 outputs portable when staff move institution?
Portability is now restricted rather than automatic. Long-form outputs, such as monographs, retain five-year portability so they stay attached to the author after a move; most other outputs are captured by whichever institution holds the substantive link at the point of submission.
Implications and outlook for institutions
Research offices preparing for REF 2029 face two parallel evidencing tasks rather than one. The first is REF-facing: documenting the substantive link — employment records, internal research support, grant funding — for every output an institution intends to submit. The second is internal: maintaining contribution records that support promotion, recognition, and researcher career narratives now that REF submissions themselves will not do this job.
Sector commentary, including analysis from Wonkhe’s research and innovation desk, has framed this as REF revealing its true purpose: an institutional funding mechanism rather than a personal-merit exercise. Institutions that treat the two evidencing tasks as genuinely separate — REF eligibility on one track, individual contribution recognition on another — are better placed to avoid a governance gap where good research goes on the REF return but the people who did it go unrecorded anywhere durable.