1. Inventory the running system.
Map code, deployment, data, jobs and external connections. Confirm access owners, backups and what can be restored.
Process
Understand the important behavior before changing the code.
The work begins with current ownership, a controlled copy and the tasks the business depends on.
2. Capture useful examples.
Record important inputs and expected outputs with a responsible business reviewer. Include the exceptions staff currently handle by hand.
3. Choose a limited first step.
Compare the options and document the assumptions, exclusions and checks. Keep investigation separate from a promise to complete a rewrite.
4. Rehearse and verify.
Make the change in a suitable controlled environment. Repeat key tasks and a failed path, then compare records and outputs.
5. Release with a recovery decision.
Agree the switching point, later data changes and conditions for stopping or restoring. Document current ownership and the next improvement separately.