Data Lineage View
A view showing how data moves and is transformed through a system, independent of the components that do the moving.
Component diagrams answer "what talks to what". A data flow diagram answers "where does this data come from, what happens to it, and where does it end up" — a different question with different stakeholders.
The audiences that need it specifically: security, because data flow crossing a trust boundary is where the threats are; privacy and compliance, because obligations attach to personal data wherever it travels; and data engineering, because lineage determines what can be trusted and what breaks when a source changes.
What it should show: sources, transformations, stores — including the ones people forget: caches, logs, analytics platforms, backups and third-party services — flows with their direction, and trust boundaries.
The exercise's most reliable output is a list of places personal or sensitive data has ended up that nobody had inventoried. Derived data is the recurring surprise: embeddings, model training sets, aggregates and denormalised read models all carry information from their sources and are rarely considered.
The related discipline it enables: data classification annotated on the flow, so that a reviewer can see at a glance where regulated data crosses a boundary and what control applies there. That combination is what turns a diagram into a compliance artefact rather than an illustration.