To make your data easier to find, understand and reuse, you need to describe it clearly. This is called documenting your data or adding metadata.
What metadata is
Metadata is data about your data. It explains what your data is, how it was created, and how to use it. Think of it as an instruction manual for your dataset.
Why metadata matters
Adding documentation and metadata helps:
- you understand your own data more clearly, especially over time
- others find and use your data correctly
- meet funder and publisher requirements for data sharing
- increase your data’s visibility and reuse
- boost your reputation and citations
When to describe your data
Start documenting your data when your project begins. Keep it up to date as your research progresses. Include your plans for documentation in your Data Management Plan.
How to describe your data
There are two methods for describing your data. These are known as ‘embedded’ and ‘supporting’ documentation.
Embedded documentation
Embedded documentation is when you add information directly into your dataset or files. This could include:
- code, field or label descriptions
- summaries at the top of a file
- information in the file’s ‘properties’ section
You should also create a README file. This explains:
- file and folder structure
- naming conventions
- what each file contains
- software needed
- any licences or restrictions
Read Cornell University's guidance on writing README files.
Supporting documentation
Supporting documentation is when you use separate files to explain or provide context for your data. This could include:
- lab books or working papers
- interview questions or survey forms
- project reports or publications
If you're sharing your data through a repository, metadata is often added as part of the upload process.
Machine readability
To improve discoverability, use machine-readable formats such as:
- CSV, JSON, XML or Excel
- clear codes (not colours) in spreadsheets
Avoid using free text or locked formats like PDFs where possible.
What to include in your metadata
At a minimum, your metadata should include:
- titles – clear and meaningful to someone outside your project
- dates – when the data was created
- contributors – who collected and analysed the data
- file types and software – formats used and tools needed to open them
- folder and file naming – clear naming and structure
- rights and licence – who owns the data and how it can be reused
- contact details – who to contact for more information
- persistent identifier – such as a Digital Object Identifier (DOI)
You can also include:
- variable definitions
- units of measurement
- assumptions made
- explanations of codes or abbreviations
Write your metadata so that someone in your field (but outside your team) can reuse your data without contacting you.
Metadata standards and tools
Using metadata standards makes your data easier to find and compare. Some disciplines have their own standards. You can check these using resources such as:
- FAIRsharing – a searchable portal of data standards
- Research Data Alliance Metadata Standards Catalogue
You can also use tools to help create and track your metadata. For example:
- ISA Tools – for life sciences, environmental and biomedical data
- Digital Curation Centre – has a list of metadata tools and examples