What is Column-Level Lineage?

Column-Level Lineage is the ability to trace the journey of an individual field or column through the entire data ecosystem, from its origin to its final consumption. While table-level lineage shows relationships between complete datasets ("table A feeds table B"), column-level lineage shows how each field is transformed, combined, or propagated.

For example, the "revenue_total" field in a finance dashboard may come from the "amount" column in the transactions table, multiplied by the "exchange_rate" column in the currency table, filtered by the "status" column = "completed." Column-level lineage documents exactly this chain of transformations, field by field.

This level of granularity is essential for answering questions such as: "Where does this number come from exactly?", "What fields are affected if I change the data type of this column?", or "Which dashboards display data derived from this PII field?" Without column-level lineage, these questions can only be answered manually, by reviewing code and queries one by one.

According to industry research, 70% of data incidents originate from column-level changes: renaming a field, changing its data type, modifying a calculation logic, or removing a column that other processes depend on. Table-level lineage does not capture these changes with the precision needed to prevent problems.

For regulatory compliance, column-level lineage is particularly valuable. GDPR requires knowing exactly where personal data fields flow, field by field. If the "email" field from the customers table is copied to 15 downstream tables, you need to know this to ensure that all storage points comply with retention and access policies.

Column-level lineage is built by analyzing SQL transformations, view definitions, ETL/ELT pipeline configurations, and BI tool metadata. Tools such as SQL parsers extract the relationships between source and destination columns, including operations such as JOINs, UNIONs, aggregation functions, and calculations.

The result is a directed graph at the field level where each node is a specific column in a specific table, and each edge represents a transformation or propagation. This graph allows navigating both upstream (where does this field come from?) and downstream (where does this field go?) with absolute precision.

Linedat offers column-level lineage integrated into the data catalog, visualizing the relationships between individual fields across the entire ecosystem. Combined with automatic PII detection, it allows tracing exactly where personal data flows, and the Impact Analysis feature shows the effect of proposed changes at the field level before they are executed.