There are many tasks you can perform to make your MDM implementation go smoother. Many people think being prepared is about the technical things but there are a lot of organizational challenges, data and system related challenges that must be addressed to have a successful MDM implementation. Some of the following preparation tasks may seem to be project work, but once you start, you will appreciate the ground work done as part of your preparation. An MDM integration is a business driven initiative, and the business needs to be involved and properly prepared for the tasks ahead. Consider the following organizational points before you start your implementation. There are many business drivers for wanting to implement an MDM solution: It is key to identify what your business goals are for implementing an MDM strategy and prioritize if many benefits are to be realized. Understanding the business drivers will help focus the activity when decisions are to be made. Simplifying MDM On The Cloud - Is it Right for You? Download Here An MDM implementation is a program, not a project, and as such benefits from having a strategy. Most organizations will takes years to complete their MDM journey and recognizing this early on will help set the proper expectation. Create an MDM assessment strategy and a roadmap to answer these questions to help keep things on track: It is important to recognize when you have had success and be able to measure and communicate that success to your stakeholders. Your MDM implementation will run over many phases and each phase is designed to achieve a business goal. Having defined your criteria for success and being able to measure your rate of success will assist in keeping the program on track, and assist with future funding to achieve the ultimate goal. Having concrete milestones is essential in order to measure and demonstrate progress in such a large undertaking. An MDM implementation is a program that will require many years to execute in most organizations. If you are considering an MDM implementation you probably already have numerous systems collecting, massaging, and reporting on customer information, and suffering from the diversity, volume, duplication, and data quality issues that come with it. It took years to create the problem, and it will take some years to straighten it all out. Get an executive sponsor to back the initiative to have success. Many organizations are project oriented, and even funding can be limited to a project by project basis. Long term initiatives like MDM require a longer vision and support in order to be ultimately successful, and only an executive can provide that backing. Many organizations operate in a series of silos. The very definition of master data indicates it is a shared, common interest set of information - which is at odds with a silo based organization. Executive sponsorship assists in having stakeholders work together. Some organizations do not have the concept of application ownership. A successful MDM implementation benefits from having an application owner whose sole responsibility is to look after the MDM application. MDM is a foundational application, which means that the business value and function that MDM supplies is directly tied to the clients that use it. MDM, being a foundational application, will be used by many client applications in the organization, and typically each one of those clients will have an agenda for customer information. The application owner can make sure that any decisions related to the MDM application are in the best interests of all of the stakeholders. Establish a data governance body to oversee the master data. If you are new to master data management, you are probably looking at mastering your Party, or person and organization information. Master data by definition is information which is of interest to a cross section of the enterprise. Interest means more than just having the information, but it also embodies how information is to be sourced and used, which can differ amongst the various stakeholders in the company. Governance gives you a process where you can resolve issues in the definition and use of information to ensure that the master data can service the needs of all the stakeholders in a way acceptable to all. Governance is important to establish things like: Some organizations practice data stewardship and not data governance. The big difference between the approaches is whether you really need to comply or not. Governance implies rules and enforcement, and this is what master data management really needs. In many organizations the master data is distributed amongst many different systems, is not centrally managed with a common set of data integrity and quality rules, and copied and modified in multiple locations. This is the MDM problem you are trying to solve and after all the hard work of implementing an MDM solution you do not want to end up with the exact same problem on a shiny new technology platform.
Organizational Preparation
Define the Business Goal
Whitepaper
Create a Strategy and Roadmap
Define your Success Criteria
Get an Executive Sponsor
Get an Owner
Bring Governance to the Party
Establish the rules and enforce them
Establish rules for:
- Where master data comes from
- When is it acceptable to deviate from a rule
- Provide an exception process with regular review and forced compliance based on defined timelines
- Process to add new attributes
- Process to control replication and when it's acceptable
Work Together
Many organizations are divided into silos and few initiatives span these silos. Your MDM implementation will span silos. Start early to establish the working relationship between the silos. Identify your stakeholders and get buy-in for the program.
SOA you need Governance
Modern MDM technology solutions are based on a Service Oriented Architecture approach and integrated into your systems using a service bus. MDM being a foundational application will provide a set of services that should be designed to meet the needs of the entire organization and thus require careful planning and governance to ensure that your services are strategic and you don't end up with JBOWS, (Just a Bunch Of Web Services).
A SOA governance function is an important organizational body in an IT organization. Services need to be well defined, have business rules that are well documented and understood, and be able to service the needs of a wide variety of clients. Commercial MDM products can offer hundreds of services out of the box contributing a significant number of services to the service catalogue.
System and Data Preparation
With the organization ready to embark on your MDM journey there are preparation tasks that will assist in keeping things moving along, and gain a better understanding of the real challenge before you.
Identify your Master Data
Not all information about a customer (or whatever domain your MDM implementation covers) is master data. Master data is information that is of interest to more than a single area of the organization. Master data is the key information that many people are interested in and can benefit from having a common view.
Your master data can be centrally managed and your other related domain information can exist in the systems that own or master that information. The Enterprise Service Bus is the common vehicle used to integrate the different sources of information to provide a consolidated view to the clients in a transparent manner.
Take a first cut at what your master data for the domain you are implementing will be comprised of. During the implementation you can refine the definition. By going through the process of the initial cut you will also get to exercise your Data Governance committee and processes, providing valuable experience for when timetables are tighter due to implementation targets.
Identify your Consumers and Producers
Identify the applications that produce master data and those that consume master data. Document how these applications:
- Store their master data. In a database or file or other
- How they update the master data. Do they create a key?
- Do they share master data?
- How is it shared?
- How often?
- Are additions and updates made by other applications sent to the current application to keep it up to date?
- How are they sent?
- How often?
- What volume of master data exists?
- How many add, update and inquiry transactions are executed?
- What information is created used or stored?
- What are your sources of data enrichment?
You may be looking at this list and thinking we should already have all this information. The truth is, yes you should but in reality you probably don't have it readily available.
You may also be looking at this information thinking that this is all project work. These questions however are very high level and will serve as only the starting place for when the implementation starts. This high level information will be required to fully understand the scope of your MDM implementation, the number of systems affected and the kinds of processing that will be impacted.
While executing the program, significant details about each one of these questions will need to be answered in order to solve the technology problems and ensure solutions account for the various categories of use cases.
Embrace the Metadata
Your MDM implementation will require you to gather significant information about your master data and systems that produce and consume it. Make sure that you retain this intellectual property by capturing the information in a metadata repository.
The MDM implementation will require you to capture:
- Where master data comes from
- Who uses it
- What is it used for
- What data is used
- What rules are associated with production and consumption
- Descriptions and definitions of the data
- Fit for purpose
And numerous other bits of information. This is a good base of important information that will be useful to you going forward.
Consider the canonical model
The canonical model is a definition of what your master data is to contain, how elements are named and defined and how it is organized. The major MDM products offer a predefined data model that is significant and well thought out. Having your own data model protects you from dependency on a vendor’s definition and provides you an opportunity to establish your own naming standards and rules for the data.
Your customer information, or other domain information you are implementing, will have additional attributes which are not necessarily part of your master data. The canonical model provides the opportunity to create a comprehensive model that can service the larger need of the organization.
The enterprise service bus serves as the vehicle to translate requests for master data information from your canonical model to the technical requirements of your solution, and responses from the solution back to your clients.
Profile your Data
Profile your data to gain an understanding of the anomalies and problems that will need to be solved when you attempt to create a master store of the information.
As part of your MDM implementation you will want to establish a set of rules designed to address data quality issues. It will be important to understand the significance of those rules on your existing data so come the time for your first production load you don’t end up excluding vast amounts of data because it didn’t fit the rules.
In Summary
It may seem like a lot of things need to be done just to prepare for an MDM implementation, but in reality it is in line with the scope of the work to come. No organization embarks on a multiple year business and IT initiative without significant planning and preparation.
The size of the problem to be solved will vary in each organization.
The existence and maturity of organizational processes such as governance will vary in each organization.
The timelines will vary in each organization.
The requirements to be successful are the same in each organization.
Tags
Data-as-an-Asset
Abhishek Kamboj
Project Manager With nearly 7 years of experience, Abhishek excels in overseeing and successfully delivering complex MDM projects.
