Data migration is the riskiest moment of an HRIS project: a bad transfer and January’s payroll is wrong for 500 people. Our 3-wave method (sandbox, parallel run, cutover) secures this critical step.
Data to migrate #
| Category | Typical volume | Sensitivity |
|---|---|---|
| Employee profiles | 1 line / person | Very high |
| Active contracts | 1-2 / person | Very high |
| Leave balances | 1 per absence type | High |
| Historical payslips | 12-60 / person | High |
| Signed documents | 5-50 / person | Medium |
| Past reviews | 1-5 / person | Medium |
Step-by-step (3 waves) #
-
Wave 1 – Sandbox (weeks 1-3) #
Import into a test tenant. Validation on 10-20 typical cases. Identify format gaps, null values, encodings.
-
Wave 2 – Pre-prod (weeks 4-6) #
Full import in pre-production. Consistency tests: totals, counts, qualitative sampling. Reconciliation with old system.
-
Wave 3 – Production with parallel run (weeks 7-10) #
Production import. Month M processed in parallel in old and new system. Line-by-line comparison.
-
Official cutover #
Cutover date announced. Old system goes read-only. New system becomes single source.
-
Hypercare (4 weeks) #
Reinforced support after cutover: Illizeo team + in-house team on 24/5 hotline during first month.
FAQ #
What about ghost data (employees gone 5 years ago)?
Dedicated archive, anonymisation per GDPR policy. No systematic migration.
What if a gap is found during parallel run?
Joint investigation. Either the source system was wrong, or the import has a bug. Cause is tracked and fixed.
Cutover window?
Ideally a weekend. Avoids conflict with HR/payroll operations during the week.
What if cutover fails?
Documented Plan B: rollback to old system, re-import after fix. Always prepared even if never used in practice.
