A practical walkthrough for putting your paper, dataset or thesis into a repository — covering versions, licences and metadata.
Before You Upload Anything
The most common mistake researchers make is uploading the wrong version of their work. Publishers distinguish carefully between three things: the submitted manuscript (what you sent before peer review), the accepted manuscript (your own Word or LaTeX file, post-peer-review, pre-typesetting) and the version of record (the publisher's formatted PDF). Most publishers who permit green open access self-archiving will only allow one of these in a repository — almost always the accepted manuscript, and sometimes only after an embargo period of six to twenty-four months.
Before depositing, look up your publisher's policy. SHERPA/RoMEO is the standard tool: search your journal title and it will tell you which version you can share, where, and when. If you signed a publishing agreement, re-read it. This is not exciting, but it takes five minutes and saves you a rights complaint later.
If you are depositing a dataset or thesis rather than a journal article, the landscape is simpler. Theses are usually owned outright by their authors. Datasets generated from your own research belong to you unless a funder or institution agreement says otherwise — check that first, but in most cases you are free to proceed.
Choosing Where to Deposit
Pick a repository suited to your material. Institutional and subject repositories serve different needs: your university's institutional repository captures anything you produce regardless of discipline, while subject repositories reach your specific research community.
For preprints and accepted manuscripts in quantitative fields, arXiv is the long-established home, with versions tracked publicly from the moment of first submission. Life scientists lean on PubMed Central, which is also the mandatory deposit route for research funded by the US National Institutes of Health and the UK's Wellcome Trust. For datasets, software and mixed outputs, Zenodo — run by CERN and integrated with GitHub — accepts almost anything up to fifty gigabytes per record and assigns every deposit a DOI automatically. DOAJ (the Directory of Open Access Journals) is worth knowing too, though it catalogues journals rather than individual deposits.
If your institution has a repository, deposit there as well. Funders' open access policies increasingly require it, and it means your work appears in your institution's research profile.
Licences: Choosing the Right Creative Commons Label
When you upload, you will be asked to choose a licence. This is not bureaucratic noise — it tells every future reader exactly what they may do with your work. Creative Commons licences are the standard, and they stack permissions in a clear hierarchy.
CC BY (Attribution) is the most open: anyone may reuse, adapt and redistribute, including commercially, as long as they credit you. Most funders — including those operating under Plan S — require CC BY for journal articles. CC BY-NC restricts commercial use. CC BY-ND prohibits adaptation. CC BY-SA requires derivative works to carry the same licence. CC0 waives all rights, releasing work into the public domain — useful for datasets where ownership claims would obstruct reuse.
For most peer-reviewed articles where a funder mandates open access, CC BY is the right choice. For datasets intended for broad computational reuse, CC0 or CC BY. For theses where you are uncertain about third-party content you have quoted or reproduced, seek guidance from your library before choosing a licence — you cannot license away rights you do not hold.
Filling In the Metadata (and Why It Matters)
A deposit without good metadata is a book shelved spine-inward. Discoverability depends entirely on what you type into those fields. Fill in every one that applies: title, authors (with ORCID identifiers where possible), abstract, keywords, publication date, journal or conference name, funder, grant number, related DOIs. If the repository offers a subject classification or field-of-research code, use it.
The DOI your repository assigns — or one you link from the published version — becomes the permanent, citable address for that record. Include the published version's DOI in your metadata so the two records point to each other.
Check your entry before hitting submit. Typos in author names, wrong publication years and missing abstracts all reduce the chances that anyone finds your work through a search or aggregator. Most repositories let you edit metadata after deposit, but first impressions propagate quickly to harvesting services.