Data lineage answers 2 questions: where did this number come from, and what depends on it?
It matters more as pipelines grow, since a single report can depend on dozens of tables.
Without lineage, tracing an error means reading code and asking people.
Table-level lineage shows which tables feed which, such as raw orders feeding a sales table.
Column-level lineage goes further, showing which source columns produce each output column.
Lineage can be captured from SQL, since queries name the tables they read and write.
Orchestrators and transformation tools often record lineage automatically as pipelines run.
Lineage usually appears as a diagram, with sources on the left and reports on the right.
Tracing upstream from a wrong figure narrows down where the error entered.
Tracing downstream from a change shows every report the change could break.
Rules on personal data make lineage important, since teams must know everywhere a person's data flows.
Lineage also helps remove unused tables, since nothing downstream depends on them.
Lineage is only as complete as what is captured, so manual spreadsheets and exports often create blind spots.
In practice, lineage sits in a data catalogue alongside descriptions, owners and quality checks.
The usual mistake is treating lineage as documentation written once, when it must update automatically to stay true.