In brief
Preparation is mostly documentation. Audit your rights, remove or account for personal data, write a datasheet, prepare a small rights-cleared sample description, and decide your licensing preferences before you contact buyers.
On this page
1. Rights audit
- List every source the data came from
- Locate contributor agreements, assignments or licenses
- Flag anything with ML/AI restrictions in its terms
- Note any third-party content embedded in the data
2. Privacy and sensitive data review
- Identify personal data, including indirect identifiers
- Confirm consent scope or other lawful basis
- Remove or de-identify where consent is missing
- Exclude special-category data (health, biometrics and others listed in GDPR Article 9) unless specifically authorized
3. Documentation
Use the questions from Datasheets for Datasets or a dataset card template:
- Motivation: why the data was created
- Composition: what each record contains, counts, formats
- Collection: how, when and by whom
- Preprocessing: cleaning, filtering, labeling
- Known limitations and biases
- Licensing and restrictions
4. Sample and packaging
- Define a small, representative, rights-cleared sample — share only under NDA
- Choose delivery formats (see formats)
- Estimate the effort to deliver the full set
- Decide exclusivity preference and timing