In most cases data without a description (metadata) are useless. Thus a measurement data series in the form "10.2,10.5,10.1,10.9" is useless if there are no descriptive metadata.
Metadata increase the comprehensibility of data for other scientists, they serve the reproducibility of research and the retrievability of data. For you too, metadata is indispensable as a documentation of your data in order to understand your research after months or years.
What is required at least?
- Where does the data come from? Used methods?
- Which measuring instruments were used, which sources, which literature?
- Who carried out the surveys?
- Data format and units
- When and where were the data collected?
- Use clear and speaking names for the description and variables!
- If metadata should be used for your discipline, use it!
The Australian National Data Service has prepared a recommendable guide for the topic: Metadata Guide. The UK Digital Curation Centre (DCC) also provides a metadata catalog and the Research Data Alliance (RDA) has extensive metadata material on its website.