No matter how well thought out the vision of a project is, nor how committed higher management is to achieving that vision, it is the key stakeholders that control the ultimate success of the project. Within an MDM implementation, the need for consensus is heightened. An MDM project is like the United Nations, bringing together business users and technologists, salespeople and foremen. Everyone must put aside their differences and reach a common understanding of the key concepts within a business. Providing the proper framework for data governance and an effective communication strategy are essential to a successful organization-wide implementation.
Determining the scope
Unlike most processes, where it is valuable to limit the initial scope, MDM projects are most successful when planned with an eye to the entire organization. Planning top-down from the start allows the organization to create a comprehensive solution for each master data set even though the actual implementation is incremental. It is imperative that your understanding of the customer model contains all facets of that customer.
Accounting sees a customer as an Accounts Receivable (AR) balance
o Credit limit
o Credit risk
o Secure information such as
Routing Number for direct bill
Social Security Numbers
Sales sees customers more deeply
o Contact Information
o Birthday
o Product interests
o Privacy Information
o Households
Support sees only the issues with a customer
o Support calls
o Open issues
o Complaints
It is important to take all of these views into account when building the customer model. It is also important to delineate what should be stored in an MDM solution. Transactional records such as a customer’s open support issues should remain in the source system. What may be valuable as master data is whether the customer has ever contacted support and whether the customer has a support agreement.
Before we go further, I would like to define a term that will provide value for the rest of this article. Implementing an MDM solution never happens in a vacuum. We have never seen an instance, where a company springs to life, ready to handle their data issues fresh. There are a web of processes and applications that are allowing all of the organization's systems to currently function in a synchronized fashion. For any functioning system within an organization, there is always at least one user who is responsible for the cleansing and reconciliation of the master data used. These individuals may report to IT or they may be business users. For simplicity’s sake, we will identify all of these individuals as "intermediate data stewards."
Data governance
A vital component of any MDM strategy is data governance. Data governance sets the rules for data management in an organization. Examples of data governance rules are required fields, standard values, encryption or security requirements, data retention policies, and so on. Organizations with mature data governance processes within their organization will only need to incorporate some strategies around master data management to solve many of the common process issues. But those organizations that have dealt with data issues in a much more distributed, ad hoc, laissez fair manner will need to develop some comprehensive policies to effectively deal with the data management requirements that MDM requires.
As master data is integrated across the organization a number of governance issues must be addressed:
How will the organization manage changes to the master data?
What rights does each group have over attributes of the models they maintain and consume?
How will dependencies be monitored and managed?
When data inconsistencies arise, how will these be addressed?
You must have the right tools to implement an MDM solution, but, without the proper processes in place, master data will begin to degrade in some of the following ways.
Master entities are updated in such a way that they break some supported system
Supported systems do not provide changes to synchronization keys
New systems are brought online without using the master-data integration strategies
Creating maintainable processes that address most common issues requires an organization to be proactive in identifying possible issues of contention. The following steps need to be completed for each master data set before any size organization is prepared to implement an MDM solution. Answering the global questions early in the project should not significantly increase the scope of your initial project. You should use the results of the following process to identify any potential issues early in the process, but not address every item in the initial phases. Attempting to be all things to all people is the quickest way for a project to spiral out of control.
1) Determine all groups that rely on this dataset.
Once you have selected an entity to review, the first step is identifying all groups within the organization that rely on this data. Identifying all of the stakeholders early in the process ensures that early work in the process is able to meet the needs of the larger organization in the future.
2) Determine all systems that assist these groups that have a dependency on this dataset.
After identifying the groups with a vested interest in your entity, you need to identify those systems that rely on the dataset.
3) Identify key decision-makers for each group or application.
Depending on the structure of your organization and the application in question, key contacts should be identified. Larger enterprise-wide applications may have dedicated teams that manage the application across the organization. Smaller, ancillary applications may be represented by a single person determined by the business area.
4) Model system attributes relationships, dependencies, and dependent processes.
Analysis of each system’s data needs should create a complete view of each master data entity. Common attributes across multiple systems should be identified during this step. Look at any issues with data types and sizes that may create issues during integration efforts.
5) Identify the data elements to be managed within the MDM system and any data quality rules that can be applied at the master data source.
Now that the data containers have been identified, data should be reviewed to determine which records will be managed within the MDM system. This may require decisions to be made on the definition of each master data entity.
i. Customer
Does customer only include active customers?
When does a prospect become a customer?
ii. Product
Do we include parts or SKUs?
How do we manage sets?
When reviewing the data, any data quality checks should be identified. What fields are mandatory? What fields are unique? What are the Business Rules for this dataset?
6) Manage initial inconsistencies and rule interference.
As the data is initially loaded into the system, look for those inconsistencies and issues that have been missed over time. I have never done an MDM implementation where the business users were not shocked by at least three major rules that were not being maintained.
7) Formalize processes for ongoing data issues.
Now that data has been loaded into the system, it is essential that a formalized plan is developed to manage changes to the data over time. This process should proactively address dispute resolution between multiple business divisions. Determining the lifecycle of master data members should be documented so that all stakeholders can be aware of the composition of these datasets.
8) Create policies for new system integration.
One of the great benefits of MDM is to provide clean data for new system-integration projects. Whether comparing an acquired company’s customer list or adding a new accounting system, clean data can provide efficiencies. Generic processes to identify these systems and their MDM integration potential can formalize this process and provide a roadmap for later projects.
Organizations need to create formal review processes that allow each stakeholder’s concerns to be heard. Depending on the size of the organization, the responsibilities associated with these processes may be a minor task or a full-time job. It is important that these MDM processes work within the larger data governance framework, as many of the key issues with master data will have far reaching effects across systems of the organization. In larger organizations, data governance requires far more knowledge and understanding than is imparted here. Data governance encompasses all facets of the organization’s data processes. It’s very important wing of MDM which most of times ignored by many organizations.
Loving P&C
DC*
No comments:
Post a Comment