Reviewed imports before production changes

Data operations

TL;DR
  • External data could affect customer-visible records, so direct imports were too risky.
  • I needed changes to be visible before they touched production data.
  • I staged, normalized, diffed, reviewed, and applied records in dependency order.
  • The result was a safer import path that supported speed without blind trust.

Imports are risky when they move directly from an external source into production records. Bad data can become live before anyone sees what changed.

Where it started

External records could change publishing, search, or customer-visible details. Applying those records directly would make mistakes expensive to find and reverse.

What needed to change

The import workflow stages external data first. It can compare incoming records against existing records, show differences, and apply changes in a safe order.

What I built

Data is fetched, normalized, staged, diffed, and reviewed before it is applied. Dependent records are created or updated in the order the domain requires.

This keeps review close to the import, not after the damage is already done.

What changed

Teams can move faster without treating every import as blind trust. The workflow makes change visible before it becomes permanent.

What I would watch

Staging adds time and state to maintain. It is worth it when external data can affect publishing, search, or customer-visible records.