By Neil Dyke, chief technology officer, Phoebus
Lenders are often nervous to move their mortgage book from one software provider to another because of what might go wrong. Therefore, it is vital to ensure that migrations are carefully planned.
So what are the pitfalls which may mean that onboarding or migration of accounts do not proceed as envisaged? Here are some key issues which need to be addressed.
No credible plan
It may be stating the obvious but it is imperative to have structure and a workable plan with time frames and who is doing what. What are the goals, objectives and testing methods for the migration?
Unclear data migration strategy
Which accounts should be migrated? Is it all accounts in the source system, just open accounts or only redeemed accounts? Is it open and redeemed accounts within the last x number of years?
There are many considerations when moving data with many stages along the way that could lead to delays, if you are not careful. Here are 10 of the most common problems.
- Data loaded into the new system does not balance with data in the source system
- It is not possible to accurately map source and destination data
- Data provided to the target system is not in the format required by its data load capability
- The source system does not provide useful error reporting for accounts which fail to load
- The target system does not have a data or loan account migration capability
- A secure data transport mechanism is not in place between the source and target system. A Virtual Private Network (VPN) or Secure File Transfer Protocol (SFTP), can be expensive and take significant time to set up.
- Data in the source system is incorrect or incomplete versus the destination system requirements. For example, the destination system requires a postcode for every address or a date of birth for each borrower.
- There is a lack of data quality or incomplete data in the source system. For example, loan opening balance plus transactions does not equal the loan closing balance.
- Incomplete data – often the data sources are multiple, and data is not available from all source systems at the same time
- There is a lack of domain knowledge to explain various aspects of the data and data structures in the source system.
Lack of effective collaboration and communication
Often in a lengthy project there are changes of personnel and the collaboration and communication are affected resulting in delays and more expensive data migration. Communication between parties across the whole timeframe of the migration project is essential to avoid errors and misunderstanding.
Testing the functionalities on self-generated data is a very common pitfall for most migrations. The testing strategy and plan should clearly reflect carrying out the activity on the real (or anonymised) migrated data.
If the testing and dry runs take too long to ensure a “clean” migration, deadlines are missed and the project is delayed.
Untested ETL is not acceptable. The Extract, Transform and Load routines should be thoroughly tested and preferably been through dry runs followed by a final dress rehearsal. Without these even the best of mapped data can fail to load.
The last pitfall is that the technical migration takes too long and you lose some of the working day or days.
Don’t let these pitfalls get in the way of moving mortgage books from legacy systems to an efficient platform. By working closely with a new provider and taking steps to mitigate potential challenges, migrations can, and indeed should, run smoothly.