Best Practices: Master Data Management

Following are some of the best practices for adopting (note - not implementing) Master Data Management solutions within an Enterprise.

Understand the Business Context (semantics) prior to picking a solution

As per my earlier blog on EA, BPM, SOA and MDM it is very important to understand the business context, including the semantics of what each of business units, departments and entities mean when they refer to MDM entities such as Customers or Products. For example marketing deals with leads, sales with opportunities/accounts and services with paid customers. Should all the entity be referred to as customers throughout the business process? or should the master entity be referred to as Organization? Common vocabulary goes hand-in-hand with master data.For example what does this sign on a building mean? Is the building fully occupied and available for purchase? or does it mean that the entire building is unoccupied and available for rent? The business context needs to be clearly defined in terms of business units, departments and business processes.

Develop a comprehensive entity model.

Not only should one define the common attributes for the master data, one should map the entire set of attributes and where it is used. For example, in customer services the customer is used as a reference whereas in order management it is transactional data.
A set of models and documents that can be used both by Business and IT.

Establish data governance process right up front.

It is important to establish data governance process right upfront, especially as this may required dedicated resources from all the business units as well as from IT to manage the master data. The approach that worked for us is as follows:

• Executive Leadership Team Data Leadership Team that meets on a periodic basis (once a month) to establish data policies, standards, establish priorities for the data quality team
• Data Stewardship (Quality) Team are the business operations people who manage the data quality on a day to day basis. The business team enforces and ensures data quality across the enterprise.
• Enterprise Data Program team is responsible for developing and managing the programs and business rules in the various technology tools.

Integrate with 3rd party data providers

It is important for enterprises to consider bringing in 3rd party data providers such as D&B and Factiva for better understanding of the market. For example on a typical morning between 9:00am and 11:00am

• 706 businesses will move
• 578 businesses will change their phone numbers
• 60 businesses will change their name
• 49 businesses will shut down
• 10 businesses will file bankruptcy
• 1 business with change ownership

Source: Dun & Bradstreet (2004)

3rd party data from D&B could be used for leveraging legal name as the name of the organization, providing knowledge management systems that also provide news feeds as well as market statistics by industry, geography and demography and also mapping it to existing sales.Picking the right data standardization and matching engine:

This is a key technology that will either make or break the quality of your data. I agree that developing the business rules and configuring the matching tool is more important - however from a technology point of view - I would dedicate substantial amount of time evaluating and testing the tools with existing data before picking a technology. My preference would be to use one of the following matching engines:

Trillium, First Logic (now SAP), IBM or SAS.

In cases where a MDM product is packed with a different matching engine, I would be tempted to externalize the data quality to one of the above mentioned data matching engines. Just my preference - maybe the other quality engines may have got better. Do you own evaluation.

Expose all MDM functionality as services
Expose all functionality such as data (address) standardization, data matching, update master and propagate master key.

As usual please do feel free to drop me line with your comments and/or feedback.

- Yogish

Vendors need to focus on providing practical (real) advice

Since the end of last year - I have had the opportunity to talk to a number of CIOs, Chief Architects and business executives for large enterprises and one of the constant frustration I have heard from the end-user community (IT organizations) is the lack of practical (real) advice from the vendor community.

Following is an example of a simple Business scenarios:

• In order to increase it's revenue Business Operations would like to learn more about the customer - to identify opportunities for cross-sell/up-sell.
• They approach their partners - typically their preferred SI and/or their preferred software vendors.
• Depending on the vendor (including SI - their resell relationship with the ISV) they would recommend a Data Warehouse, Business Intelligence, EAI and/or Master Data Management Solution. In addition, they will also be willing to guarantee, typically as a fixed bid, implementation within weeks/months (typically 3 to 6 months).
• Business buys into it - spend the money and a year later had not yet achieved the end-result they were promised.

Problem:

• Vendors focus on delivering an IT solution (generate the revenue and move on to the next project) instead of solving the business problem

Recommended Approach:

• As part of every engagement - both the SI and ISVs need to insist on conducting a Business workshop (prior to starting the IT implementation).
  
• Provide real-practical advices - example: for a customer master the vendor typically looks for decision from the business to provide them with direction on where to add the customer - Lead Management, Opportunity Management or Order Management. As the vendors have done multiple implementations - shouldn't they know the best practices in the vertical and recommend the approach?
• SI typically have separate organizations that deal with business transformations and IT (CRM/ERP )implementation teams. Shouldn't every engagement have folks from both the practices at the customer site during the initial stages? Agreed the engagement cost will be a bit higher for the customer - but the value received will be substantial greater than the up-front cost. I for one - would be willing to pay. My problem was that no one approached me with such a proposal. However, I was lucky - my CIO recommended (and funded) that I have a full-time architect(s) from an SI on my team.
  

Thanks to SOA - I am starting to see changes in the SI engagement models. They seem to these days more focus on Architecture & Approach first - before bringing in their implementers. However, I am yet to see this transition happening in the ISV community.

The above example is based on my own experience and the case study is available here. As the case study indicates - I had to solve the same problem twice. First implement and learn from my mistakes and then repeat it again later. If we had real advice - we would have got it right the first time.

- Yogish

Blueprinting Information Architecture and BPM

Following my blog on Key Learning: Blueprinting Information Architecture is key to successful adoption of SOA one of the feedback was Don't we need to define the detailed business process to identify potential service candidates?

My response to that is No and explained it in this edition of the Strategic IT Update. Please click here for all the Strategic IT Update presentation and transcripts.

Reference: Customer Data Integration: BEA-IT Case Study (my presentation at CDI Summit in 2006)