Provenance describes the origin of an object. Data provenance refers to the knowledge of where data originate, where they were collected, by whom, for what reason, and similar aspects that help to understand how the data were originally gathered, processed and altered. In daily use, the term “data provenance” refers to a record trail that accounts for the origin of a piece of data (in a database, document or repository) together with an explanation of how and why it got to the present place (Encyclopedia of Database Systems, pp 608-608). You can also call it the process of keeping records of changes in the data. The need for Data Provenance increases as the reuse of datasets becomes more common in research. The term was originally mostly used in relation to works of art, but is now used in similar senses in a wide range of fields (Wikipedia).
Researchers regularly use a lab notebook or a journal to document their hypotheses, experiments and initial analysis or interpretation of these experiments. If you manually change data in a dataset, this should also be documented. Sometimes records of changes in data can be kept by adding notes to programmes or scripts that are used.
Finally, when a dataset contains personal data, data provenance can help researchers to understand the specifics and the context in which the data were gathered, also to be able to assess whether or not the informed consent given for the first research, is applicable.
For every step of your data analysis, good data documentation is necessary.