What is the difference between data provenance and data lineage?

Data lineage traces how data moves and is transformed through a pipeline, table by table. Provenance is broader: it asks where something originated, who or what produced it, and whether it can be shown to be unchanged. Lineage maps the plumbing; provenance establishes a chain of custody.

Lineage maps the plumbing

Data lineage describes how data moved and was transformed on its way through a system: this table fed that job, which produced this column. It is mostly derivable from the pipeline itself, which is why warehouse tools such as Databricks, Snowflake and Collibra can generate it automatically.

Provenance establishes a chain of custody

Provenance asks a question lineage does not: where did this originate, who or what produced it, and can anyone show it has not changed since. It reaches back past the pipeline to the experiment, the instrument or the person, and forward to whether the bytes still match.

Why research needs the second one

A warehouse pipeline runs the same way every night, so its history can be inferred. A research result comes from a one-off experiment, a collaborator's spreadsheet and increasingly an automated agent. There is no repeating pipeline to derive anything from, and the person who could explain it leaves when the contract does. That is why research borrows the word lineage but almost always needs provenance.

Keep your research in a lab

Sign up with your email and create a lab in a few minutes.