Recently Aaron Zornes and I hosted a webinar on the “Top 10 Best Practices for Successful MDM Implementations.” Almost 700 people registered for it – one of the highest number of registrations I’ve seen for a webinar, ever! This number tells me that many practitioners are eager to learn the best ways to ensure that their Master Data Management (MDM) implementations are successful.
The number one best practice was the need to use multiple data domains (such as customer, product, channels, and so on) to solve a business problem. After all, any business problem doesn’t constrain itself to a single domain like customer or product. Take for example a common business problem that many companies aim to solve – to increase cross-sell of other products to existing customers. At its simplest level, this process involves two steps (1) knowing which customers have bought what products (and for that matter knowing what products they have not have bought), and (2) through which channels did they buy (online, mail order, or store). So, in order to increase their ability to cross-sell, a company needs to implement three interlinked data domains – customers, products, and channels.
Aaron Zornes kicked off the webinar with his initial strategic planning assumption (SPA): “through 2012, corporate MDM platform teams will assume (and insist) that all MDM software platforms targeted for enterprise-level deployment or major role in mission-critical systems fully support both PARTY & THING data types” (customers and channels belong to PARTY data type, while products belong to THING data type). After Aaron completed explaining his SPA, I cited an example of a Fortune 50 pharmaceutical giant that implemented five different data domains in their MDM (customers, products, affiliations, sales personnel, and pricing) “to improve their efficiency in order-to-cash process.”
After my presentation of the example, a poll was conducted among the participants during the webinar and the results emphasized the need for multiple data domains. We asked the question, “For your current MDM project are you planning to master one data domain or many?” Forty percent of the attendees who answered selected “three or more domains” as the answer, while only 14% opted for a single domain. This result corroborates the fact that a majority of the MDM practitioners are requiring the ability to manage multiple data domains within their MDM system to solve a business problem.
Another best practice, that is closely aligned with the first one, was the ability to “start small and scale” – that is, to start small with a single domain or a smaller use case, prove the results, and then expand to other domains or solving the rest of the business problem. After all, no IT has an appetite, or the resources, these days to perform a multi-year, big-bang MDM implementation. To aid the “start small and scale” implementation approach, Aaron emphasized the need for an MDM product to have a flexible data model to enable IT to start with some data domains and then add other domains and solutions. Then, I pointed out in the above example, the pharmaceutical giant started their MDM implementation with the customer and affiliations domains, and then expanded to product, sales personnel, and pricing.
Since a flexible data model is a fundamental requirement to start small and scale, we conducted another poll, in which we asked the question “Is it important for you to use your own data model or industry specific data models?” A quarter of the attendees wanted the ability to “use their own data model,” while an overwhelming majority of more than two-thirds people wanted the ability to “use a combination of the two.” This result implies that the MDM practitioners are looking for a quick start with a predefined industry data model, but need the ability to customize it to their unique needs.
The other best practices were “automatically generate web services and user interface”, “create best version of the truth”, “support reference data management”, and so on. To listen to a recording of the webinar, please click here.
Are there any other best practices that you would like to share with our readers? If so, please post them in the comments box.