By John Tuttle, Senior Director of MDM


Data architecture is the heart of all Master Data Management systems. Different industries have specific needs when capturing customer data and therefor, one-size-fits-all data models must be tailored for the best fit to an organization’s requirements.

At Paradigm, we are fanatics about data modeling within the MDM practice. Paradigm means model after all! With our MDM clients we often have early discussions regarding data modeling approaches. One decision point is whether to use an existing model or build custom based on the organization’s enterprise model.

The great news is that Informatica’s Multidomain MDM system is extremely flexible when it comes to building out logical data models, alibi the fact that it’s a relational database system. However, this flexibility begins to harden as you build integrations, user interfaces, workflow and data cleansing into the MDM hub applications. Various components of the system are configured on top of the constructed model. Changes become harder as more is built. Think of the difficulty of changing the foundation of a house after the walls are completed, for instance. Additions or extensions to the model are easy, changes to the structures in place are very difficult to achieve. The data model should fit the organization, industry and use cases of the MDM program before you build on top of it.

But what about the pre-built model approach? The Master Data products in Informatica’s portfolio offer inbuilt model options with their 360 fueled applications. The advantage is that the application can be defined and built with many of the features that are most useful to data stewards. The lengthy period of definition and build, QA and test can be dramatically shortened. If the model fits the usage.

With the introduction of Customer 360 the debate was always around a Party-centric model or a Party Role-centric model. Starting with a Party Role model, Informatica moved to the Party model in later versions. They now support both within the C360 application. However, we find that some industries don’t align to either of these party models. While the C360 model is suitable for many industries and use cases, it is best for analytics and marketing purposes.

We find that manufacturing and Consumer Packaged Goods (CPG) companies implementing large scale ERP systems such as SAP Hana have modeling needs that don’t align to the C360 data model. Given the difficulty of structural changes, heavily modifying the underlying C360 data model becomes more difficult than a build to suit model. The good news is that we don’t have to start from square one with a data model in this situation. We have extensive experience implementing MDM systems at manufacturing and CPG companies over the years. The SAP model is also one that is known, and therefor we can get a jump on the time-consuming task of data modeling.

Our approach is to use an SAP specific model for the manufacturing and CPG industries as a starting point to the data architecture step in our methodology. All SAP implementations are not alike (even within organizations, SAP instances can have setup differences) so we begin with a model, instead of pre-built components you might find with some accelerators or pre-built industry models (Informatica has a number of these industry models pre-built, but tailoring them is the hard part). During the modeling phase, we can adjust the logical model to client specific needs before any configuration in the system. The build period becomes shorter the better defined the model (less re-work or modifications). With an existing model for manufacturing use cases, we can get to that point faster and still have a custom-built system specific to a client’s requirements and use cases.

Any MDM program must decide what model to use, whether the choice is pre-built or custom, when starting on the customer engagement journey. With all respect to Elvis Costello, “This year’s (data) model” is the model you start with and it will be the foundation of the journey your organization takes for year to come.