Most researchers think about their data only until the paper is accepted, then let it rot on a laptop. That is a waste — and, increasingly, a compliance problem. Well-managed data can be verified, reused and cited long after the paper, and a growing list of funders and journals now expect it. The framework everyone has converged on is FAIR: data that is Findable, Accessible, Interoperable and Reusable.
What FAIR actually means
| Principle | In practice |
|---|---|
| Findable | The dataset has a persistent identifier (a DOI) and rich metadata, so people and search tools can locate it. |
| Accessible | Once found, it can be retrieved by a clear method — open where possible, or a documented request process where not. |
| Interoperable | It uses open, standard formats and vocabularies so other tools and datasets can work with it. |
| Reusable | It is well described and carries a clear licence, so others know what they may do with it. |
Notice that FAIR is not a synonym for "open". Data can be FAIR and still restricted — the point is that it is properly described and its access conditions are clear.
Where to put your data
Do not leave data as a spreadsheet attached to an email. Deposit it in a recognised repository: a discipline-specific one if your field has it, or a general repository such as Zenodo or Dryad otherwise. A good repository gives your dataset a DOI (typically registered with DataCite), which makes it citable in its own right — so you get credit when others use it.
Describe it so a stranger could use it
The reusable part of FAIR lives or dies on documentation. Include a README that explains what each file and variable is, the units, how the data were collected, and any processing you applied. A dataset no one else can interpret is not reusable, however open it is. Add a licence (a Creative Commons or open-data licence) so users know their rights.
What not to share
Openness has limits. Never publish identifiable personal or patient data. De-identify it, and where the raw data are genuinely sensitive, share them under controlled access — but still publish the metadata so the dataset is findable and others can request it. "As open as possible, as closed as necessary" is the working rule.
Close the loop in your paper
Finally, tell readers where the data are. A data availability statement — a short paragraph naming the repository and identifier, or explaining any restriction — is now standard and often required. It is the link between your paper and the data that supports it.