Definition · Plain-language
FAIR Data Principles
The FAIR data principles — Findable, Accessible, Interoperable and Reusable — define how research data should be managed so both humans and machines can discover, access and use it effectively, maximising its value for science and society.
The step most authors miss
Doing CRediT right? Don’t stop at the statement.
A CRediT statement credits you inside one paper. The recognition CRediT was built for happens when those roles are tied to you, persistently. Sign in with your ORCID — free — and claim your CRediT contributions on casrai.org, the home of the standard. They become a verified, portable part of your identity, not a line that disappears into one PDF.
Free: claim your contributions, then export a journal-ready CRediT statement, schema.org structured data, JATS XML, CSV or BibTeX — and preview your public profile. A membership publishes that profile publicly and verifies the journals you serve.
The four FAIR principles explained
Findable requires that data be assigned a globally unique, persistent identifier (such as a DOI), that rich metadata describe the data and include the identifier, and that metadata be registered in a searchable resource. Accessible requires that the identifier resolve via a standardised, open protocol; that the protocol allows authentication where necessary; and that metadata remain accessible even if the data itself is no longer available. Interoperable requires that data use a formal, accessible, shared and broadly applicable language for knowledge representation, with vocabularies that themselves follow FAIR, and that datasets include qualified references to other data. Reusable requires rich metadata including accurate provenance, a clear and accessible data-usage licence, and conformance to domain-relevant community standards. Together, the 15 facets create a graduated, implementable framework rather than a binary pass/fail test.
FAIR is not the same as open
A common misconception is that FAIR requires open access to data. In fact the "A" in FAIR specifies that data "should be accessible" under defined conditions — which may include authentication, authorisation or formal agreements. The Accessible principle explicitly states that "metadata are always accessible, even when the data are no longer available". Sensitive data — such as patient records or commercially restricted datasets — can therefore be FAIR: they can have persistent identifiers, machine-readable metadata and clear licence terms, while access itself is controlled. The maxim in European Open Science policy is "as open as possible, as closed as necessary".
Implementation and maturity
FAIR compliance is not binary; it exists on a continuum. The GO FAIR initiative, the FAIR Maturity Indicators framework and the EOSC (European Open Science Cloud) FAIRness Evaluation tools all provide structured ways to measure how FAIR a dataset is. Implementation relies on several building blocks: persistent identifier infrastructure (DOI, ORCID, ROR), metadata standards appropriate to the domain (DataCite, Dublin Core, Darwin Core), discipline-appropriate repositories, and community vocabularies. For software and workflows, the FAIR for Research Software (FAIR4RS) principles extend the framework. Funders including the European Commission, NIH and Wellcome Trust reference FAIR in their data policies.
Key facts
At a glance
- Origin: Wilkinson et al. 2016, Scientific Data (doi:10.1038/sdata.2016.18)
- Findable: persistent identifier, rich metadata, registered in searchable resource
- Accessible: open protocol, authenticated if necessary, metadata persists when data gone
- Interoperable: formal shared vocabularies, qualified links to other FAIR data
- Reusable: provenance, community standards, explicit usage licence
- FAIR ≠ open: controlled-access data can still be FAIR
- Framework: 15 specific facets (not a binary pass/fail); measured by maturity indicators
- Adopted by: European Commission (EOSC), NIH, Wellcome Trust, Horizon Europe
Common misconceptions
What people often get wrong
Often heard: Making data FAIR means making it fully open and publicly available.
Actually: FAIR requires that data be findable, accessible under defined conditions, interoperable and reusable — but "accessible" explicitly allows for authentication and controlled access. Sensitive, confidential or commercially restricted data can still be FAIR if it has persistent identifiers, machine-readable metadata and clear licence terms.
Often heard: FAIR is a certification you receive once a dataset passes a checklist.
Actually: FAIR is a graduated framework: datasets can be more or less FAIR across 15 specific facets. Maturity indicator tools from GO FAIR and EOSC allow assessment on a spectrum; FAIR is an ongoing implementation goal, not a one-time certification.
Often heard: Any dataset deposited in a repository automatically becomes FAIR.
Actually: Repository deposit is one step, but FAIR compliance depends on the quality of metadata, the use of persistent identifiers, standard vocabularies and clear licence terms — all of which require deliberate effort beyond simply uploading a file.
Common questions
FAQ
Where did the FAIR principles come from?+
The FAIR Guiding Principles were published in Scientific Data in 2016 by Wilkinson et al., following a workshop at the Lorentz Center in the Netherlands in 2014. They were developed to address the challenge of making data machine-readable and reusable, not just human-understandable.
What is GO FAIR?+
GO FAIR is an international grassroots initiative providing an implementation network for the FAIR principles. It coordinates three pillars — GO CHANGE (policy), GO BUILD (infrastructure) and GO TRAIN (training) — to help organisations adopt the principles in practice. It is not a standards body but a community of practice.
Going deeper
