According to a recent 2013 Gartner report entitled “Big Data Adoption in 2013 Shows Substance Behind the Hype”, 64% of corporations surveyed specified that they were moving toward a Go-Live or are planning a Big Data/MDM project. The article points out that in many MDM projects, stakeholders still don’t know what to do with their data after it’s been matched, cleansed and standardized, causing new Big Data asylums.
To help you develop an MDM strategy and an MDM project plan, here are four starting points for an MDM implementation:
Define your MDM utopia and align with the business visions: Decide what you would like to achieve with MDM in an ideal world before taking the first steps towards adopting it. With each MDM project, define the vision with key management in the corporation to get a feel for how much compromise might be needed.
- What data domains and usage cases will be addressed?
- What are the milestones to a complete MDM vision?
- Does the scope define business benefit justification?
- Is the MDM vision singular and aligned with the business vision?
Define the MDM strategy and roadmap: The MDM strategy should focus on how the MDM vision will be comprehended as well how to manage master data assets within the organization. The following questions should be answered as part of the strategy, by the MDM program management:
- Where in the flow of data from Legacy/admin source to MDM can the data be authored?
- Can the MDM Hub enable users to author its data?
- How and who will validate changes to data through the MDM data flows?
- Will the MDM hub have a governance board or Data Governor who will validate matches and merges of duplicate customers in the hub?
- Will the data be standardized and enriched and by what process (Address Doctor, Third Party Processes)?
- Who will consume the hub data?
- Will it be stored in other locations as replica databases?
- Will the data be called in real-time through WSDLS or SQL calls?
Reinforce the data quality (DQ) process: MDM without a robust approach to data quality can be dangerous. The well-used IT cliché “garbage in, garbage out” applies to building an MDM hub solution. Without a Data Quality processes in place it will be difficult for a corporation to master or run analytics on its MDM hub. Filters on the type of data entering the hub, both during the initial load and during daily deltas, as well as data transformations can help the quality of the information coming out of the hub.
- Data profiling; assessing the data to understand its overall degree of accuracy, uniqueness and nullability.
- Data standardization; utilizing a business rule engine to guarantee that all data entering the hub conforms to specific level of quality, i.e., a Company Name must be filled in to enter a customer MDM hub.
- Address Doctor/USPS Standardization; automated pattern matching tools for fixing name and address data and applying postal standards.
Determine who the ‘owner’ of the master data is: It is essential to understand the needs of the business in relation to master data. This will include who within the organization needs to be able to access and control the information. Examples might be the sales teams will determine future sales; the analytic teams and credit groups can only access the information. Each requestor of the master data within the hub can affect the technology and internal processes required as part of an MDM project.
Ryan R Hartley