2. The same or similar data structures between legacy and the new software system is a dream, but the less differences between those two systems you have, the less processing errors you will have (e.g. address in the first system defined as a one field, in the new system can be defined as a 5 or 6 fields)
3. As little data sources as possible. Many Word files, Excel sheets, databases or TXT files makes the process more manual to evaluate each file and create a batch function to parse and process data automatically. It’s not always possible, and as fewer differences in all these files you have, the less manual work will be required.