Data management plans

A DMP answers, before the data exists, the questions that hurt when answered too late: what data will the project produce, where will it live, who may see it, what does that cost, and what survives the grant. Funders require one with most proposals; the useful version is not the compliance document but the honest operating plan - and because it describes practices this pack already engineers, most of a good DMP can be drafted from, and checked against, the project itself. The software counterpart (rseng-management-planning's SMP practice) shares this design; write the two consistently and cross-reference them.

What a plan covers

The standard sections, each answered from project reality:

1. Data description: types, formats, expected volumes, sources - name open formats deliberately (rseng-scientific-file-formats) and reused third-party datasets with their licenses and terms.
2. Documentation and metadata: data dictionaries, READMEs, community metadata standards for the domain (rseng-data-management owns the practice; the DMP states which).
3. Storage and backup during the project: where working data lives, replication, who administers it - institutional storage beats lab improvisation; sensitive data goes where the steward says (rseng-regulatory-compliance).
4. Access, sharing and legal: who can access what and when, embargo plans, consent and GDPR constraints, anonymization strategy - with honest limits stated (rseng-regulatory-compliance); "as open as possible, as closed as necessary" is the working frame.
5. Preservation and sharing after the project: which data is deposited, where (domain repository first, Zenodo-class otherwise - rseng-archiving), with what identifiers and licenses (rseng-licensing for data licenses), and what is deliberately discarded (retention has costs; keeping everything is not a plan).
6. Responsibilities and resources: named roles (who curates, who deposits), storage and curation costs as budget lines - data work is fundable work; say so in the proposal.

Proportionality: a simulation project with regenerable outputs needs a lean DMP centered on code and configs (rseng-reproducibility); a project collecting human-subject data needs the full treatment. Match depth honestly.

Machine-actionable DMPs

DMPs are becoming structured data, not prose PDFs:

• The RDA DMP Common Standard defines the maDMP schema - a JSON model of datasets, distributions, hosts, licenses and costs that tools exchange.
• The Data Stewardship Wizard (DS-Wizard) builds DMPs from questionnaire knowledge models and exports funder formats plus maDMP JSON; DMPonline carries the major funder templates. When the user's institution runs one of these, draft THERE (or produce content ready to paste), so the plan lands in the system reviewers and stewards actually use.
• The agent-friendly consequence: a structured DMP is checkable - datasets listed in the plan can be diffed against datasets the project actually has, licenses in the plan against LICENSE files, deposit promises against archive records (rseng-archiving).

The DMP as a living document

Plans drift: new instruments, bigger volumes, a dataset that cannot be shared after all. Treat the DMP like the SMP:

• Version it with the project (repository or the DMP platform's versioning); update at milestones, reporting deadlines and whenever data reality changes - a plan contradicted by practice is a liability at review and audit time (rseng-management-planning owns the cadence).
• Run a drift check when revisiting: promised repositories vs actual deposits, promised metadata vs delivered, promised retention vs disk reality. Report gaps as actions with owners.
• Record AI assistance in drafting or revising the plan in aidecl.yaml (rseng-ai-declaration).

Working with this skill

This skill is source-independent: its authority is the RDA common standard, the platform documentation and the RDMkit guidance linked below. It is the data twin of the SMP practice in rseng-management-planning; rseng-data-management holds the underlying practice.

Learn more (verified):

• https://rdmkit.elixir-europe.org/data_management_plan - RDMkit on data management plans
• https://github.com/RDA-DMP-Common/RDA-DMP-Common-Standard - RDA maDMP common standard
• https://ds-wizard.org - Data Stewardship Wizard
• https://dmponline.dcc.ac.uk - DMPonline funder templates

Related skills

Check whether any of these applies before moving on:

• rseng-archiving - deposit and preservation promises
• rseng-data-management - the practice the plan describes
• rseng-licensing - data license choices in the plan
• rseng-management-planning - SMP twin, shared drafting cadence
• rseng-regulatory-compliance - GDPR, consent and anonymization sections
• rseng-scientific-file-formats - naming open formats deliberately