How we prove a migration: run both until the difference is 0.00

The checklist

1. 1

   Run the old system and the new system side by side, on the same live data.

2. 2

   Compare the same report from both systems, period by period.

3. 3

   Investigate every difference before it is signed off — don't average them away.

4. 4

   Only cut over once the difference is 0.00.

This is how Case A's month-end reports matched during the parallel run.