CRM Application Implementation - Data Migration Strategy
In one of my earlier posts, importance of people, process and technology in any CRM application implementation is mentioned (ref). Some of the comments received for this post include on the strategy of data/information. I certainly liked this point and being in data modeling for quite some time and responsible for the data modeling in my present consulting work, I want to share a few points with my experience or knowledge in this post.
It is certain that any CRM application implementation involves data migration either from excel worksheets to the application chosen or from one CRM application to another. I generally observed that the migration team starts mapping the source data to the destination data and coming up with transformation rules for ETL tool. But I think there are a few steps to be done ahead of the data mapping exercise.
These steps include
1) Create data dictionary and identify the entities
2) Identify the source of reference data - Who provides the source data and how they provide/manipulate it. Is it the CRM business administrators or database administrators of IT applications? Who authorizes the reference data items? Eg. Customer Types - dividing the customers into different types and their categories.
3) Confirm with the business how do they think that the data is presently modeled/represented. The business community might be unhappy with particular set of data items or categorization or attribute information. It might be because of the poor data model in the present application and we need to standardize that in the new application.
4) Draft the different actions on the data/information by different process steps to understand how the data is being used and manipulated. This will determine the quality of the data. which is another important factors.
5) Obtain the data models from the product vendors and then start next level of data model (extensions of data base tables and creation of customized tables/columns) in the destination application's database. This is to be done in two layers (logical and physical).The logical models should be represented with UML diagrams. Recently I was involved in a data modeling discussion in the Tele Management Forum which lead to 79 threads and where the experts were arguing on a modeling point. Please refer the link (might need login access).
6) Design the database based on the above inputs and start normalizing this model to ensure that the business is comfortable in viewing/modifying the data, functionality and process steps are not changed and not impacted. In this phase it is always better to come up with ER diagrams of the customizations.
Post these steps comes the actual data migration steps where the confirmed and signed-off data architecture/design is to be mapped from the source to target. This means the design stage of any implementation stage should think about data design and part of it should be on migration also.
How far these design steps are given importance in a CRM implementation project's life cycle? I would say less importance when compared the functionality and process steps/tasks being designed, documented and tracked. To avoid any major data issues during testing stage it is always better to have a complete design and follow the above steps mentioned.
It would be great to know the experiences from you also and share proactive design steps to be taken.
