Data migration checklist for IT teams
A data migration moves the records a business runs on into a new system, usually while people keep working. Shortcuts surface after cutover, when they are expensive: orders missing the history support relies on, an integration still writing to the old database, and a finance team reconciling two systems by hand at month-end.
This data migration checklist covers the work from a fixed source and target to the old system decommissioned, including field mapping, staging runs, the cutover itself, and reconciliation. It is written for the IT lead who owns the migration and the business owner whose team lives in the data.
Frequently asked questions
How long does a data migration take?
Plan 6 to 12 weeks for a business system like a CRM or ERP. Most of that goes on mapping, cleaning, and staging runs; the production migration itself often fits in a weekend. Timelines slip when user validation starts late, so book the staging validation dates at the same time you set the cutover window.
Who should own a data migration?
IT or the systems team runs the migration, and the business owner of the data decides what migrates, how fields map, and whether staging looks right. Both names go in the migration brief. When IT makes those calls alone, fields land in places the team that uses them never looks, and the errors surface weeks later.
Should you migrate all historical data?
Migrate what the business still acts on and archive the rest as a readable export. A common line is two to three years of transactional history plus all open records, with the archive kept for the retention period. Migrating everything spends mapping and cleaning effort on records nobody will open, and it carries the old system's duplicates into the new one.
What should you do when reconciliation counts do not match?
Stop before go-live and account for every gap. Rejected records, filtered records, and merges each explain part of the difference; whatever remains unexplained is the real problem. Fix and rerun in staging until the counts reconcile or every difference is documented and accepted by the business owner. Do not launch on counts that are close.
Do you need a migration tool, or are built-in importers enough?
Built-in importers cover a straightforward move between mainstream systems; a dedicated ETL tool is worth it for complex transformations or millions of rows. Either way the tool only moves data. Freeze announcements, validation sign-offs, and integration repointing are tasks owned by people. Chaser tracks those steps in Slack and chases each owner as the cutover date approaches.
