Engineering · Published July 1, 2026
Data Migration: Understand the Old Ledger First
Migrations rarely fail on technology. They fail because nobody can describe what the legacy data actually contains.
In replacement projects, data migration is usually scheduled last and estimated most optimistically. Across several migrations, the lesson is consistent: the time goes into understanding legacy history, not into writing scripts.
Three tables before migration
A field mapping table from old to new, including type conversion and null handling.
A data quality table counting missing, duplicated and malformed values, which yields a cleaning estimate.
A business sign-off table listing what must be kept and what can be archived, signed by the business owner.
Execution strategy
Build the full script, then run it three times on a test database: once for errors, once for row counts, once for financial fields.
Rehearse a switchover including rollback; on launch day only sync the delta to keep downtime short.
Keep the old database read-only for three months so the business can still look up history.
Migration is about keeping old data trustworthy in a new system. One number that does not reconcile erodes confidence in everything.
Want to apply these ideas to your project?
We are happy to go deeper into implementation details, or to discuss the specific questions you are facing.
Start a conversation →