Data Transformation

Converting extracted data into the structure, format, or model required by the target archive or system.

What Is Data Transformation?

Data transformation is the process of converting extracted data into the structure, format, and values required by its destination, whether that destination is a new application, an archive, or another storage environment. It’s the middle stage of the extract-transform-load pipeline: extraction gets the data out of the source, transformation reshapes it according to the mapping specification, and loading puts the transformed data into its target.

Transformation covers a range of operations depending on how different the source and target structures are. At its simplest, it’s a data type conversion; a date stored as text in the source becomes a proper date field in the target. At its more involved end, it includes reassembling data spread across many source tables into the reconstructed, human-readable business objects that make an archive actually usable, combining an invoice header table with its line-item table into a single coherent record, translating internal status codes into their business meaning, and normalizing inconsistent formatting across records that were entered by different users over many years.

The quality bar for transformation in an archiving context is higher than in a typical reporting or analytics pipeline, because there’s no opportunity to go back and re-transform something later once the source system is gone. A transformation error in a live analytics pipeline gets caught and corrected the next time the pipeline runs; a transformation error in an archival migration becomes a permanent, silent defect in the only remaining copy of that data.

Key Points

The reshaping step between extraction and loading. Transformation takes the raw extracted data and its mapping specification and converts it into the format and structure the target actually requires.
Ranges from simple conversion to full reconstruction. Transformation can be as simple as a data type conversion or as involved as reassembling multiple related source tables into a single reconstructed business object.
Normalizes inconsistency. Years of manual data entry across a legacy system typically produce inconsistent formatting and values that transformation is expected to clean up and standardize before loading into the archive.
Errors are effectively permanent in an archival context. Unlike a recurring analytics pipeline where a transformation bug gets caught and fixed on the next run, an archival transformation error becomes a lasting defect once the source system is retired.
Directly enables business object reconstruction. The readable, complete records that make an archive genuinely usable, rather than a pile of disconnected raw tables, are a direct output of correctly executed transformation.

Common Use Cases

🧾 Reconstructing business objects from raw tables

Extracted invoice header, line-item, and approval tables from a legacy ERP are transformed and joined back together into complete, readable invoice records in the archive, matching how they appeared in the original application.

🔤 Translating coded fields into business language

Internal status and category codes from a legacy system are transformed into their documented business meaning — a numeric code becomes “Approved” or “Rejected” — so archived records are understandable without institutional knowledge of the original system.

🧹 Normalizing inconsistent legacy data entry

Years of inconsistently formatted address fields, date formats, or free-text categories from manual data entry are standardized during transformation, so the archive supports reliable filtering and reporting rather than inheriting decades of formatting inconsistency.