What steps can be taken to address the challenges associated with implementing master data management?
You've already been given excellent advice. There's a multitude of challenges and things that are important to get right.
However in my experience I've found that there can be the best intentions and plans for success but in the end I have too often witnessed significant resources for planning and committees but in the end fail to deliver anything tangible. You need governance but what that looks like is highly contextual to the problem you're solving and your organization. So my advice is, think big but start small. Triage data based on business impact and level of complexity. Find something simple but with potential high value/impact to the business and just do it. Don't be afraid to fail - you learn by failing. Just fail small not big.
I harp on this a lot - at the end of the day if you can't do something drop dead simple then you're just wasting your company's time. Pilot and operationalize the simple and show results. For example, a common set of geographical codes such as ISO standards that solves the ability to solve a problem with mapping your customers and sales. Then you will know if its possible. Don't start with managing and integrating operational customer hub. If you can't integrate simple code set what makes you think you could integrate your customer data? As logical as this is, I'm afraid many organizations get bent on MDM and related data governance programs, throwing significant resources and money at it with lots of assumptions. And in doing so missing the basics of do we have the political will to enforce stewardship? Can our applications and our infrastructure even support integrating this kind of data? Do we have a vision for a simple model of how it would work technically? Prototype with simple tool like Microsoft's MDS which is most likely free to your organization. It's very simplistic but functional for basic reference data or hierarchies. Then graduate to more complex data and technology when you're ready.
BTW...before you start, answer this question, "Is this operational or analytical MDM (meaning is it for feeding EDW or integrating same data across multiple operational systems). Or is it a combination? The answer to that will greatly drive the level of complexity. MDM for augmentation to an EDW is vastly simpler than attempting to integrate MDM data with multiple operational systems.
What are the biggest challenges in master data management implementation?
The biggest challenges I have experienced are:
Data Quality of the source systems, that prevent to quickly identify, and consolidate master entities.
Diversity of sources with different levels of normalization that require rework on the data integration processes.
Nonstructured data required to be integrated to feed the MDM master entities, from on-premise and cloud systems.
Multiple Excel spreadsheets containing business rules required to be integrated into a repository in order to be included in the MDM process. For example to identify a unique customer or business units.
Special characters not being identified in the integration processes nor data quality, which causes the process to crash if there is not a code page standardization to use Unicode in all the systems.
Establishing a data quality strategy before implementing the MDM, that will support to have a clear data integration process and identify in a faster way the key information to create the Master Entities, avoiding rework and further headaches.
Create an Operational Data Integration repository to consolidate the diversity of sources with different levels of normalization.
Extend Integration strategy with a Data Virtualization solution to include non-structured data required to be integrated to feed the MDM master entities, and provide flexibility for APIs.
This is also covered with the Data Virtualization solution.
Establish a Governance Strategy to standardize code pages thru Unicode in all the systems.
Integration of multiple source data, managing and addressing the data quality challenges, de-duplication of data, address matching, standardization of data etc.. identifying the domains for which MDM needs to be implemented. Business should drive this initiative and not IT. Choosing the right tool and its implementation partner. Clear understanding of the attributes for which the MDM will be done. Realistic expectation and a program to run data quality & data governance with right stewardship
There are multiple steps to address the challenges.
Key challenges include:
Funding - understanding how to calculate the dollars required to adequately plan, design, build, test, and implement an MDM solution.
Tool selection: understanding which toolset is right for your organization is crucial.
Communication: communication the benefits of MDM are difficult as it's back end capability that doesn’t offer hard tangible benefits to the user as such. And those that benefit directly from it are a very narrow cross-section of the user community.
Endorsement: securing an endorsement from the business and convincing the business to support the investment of multi-million dollars for a toolset that has little visibility and plays a silent role in the background is a tough proposition
Industry alignment: ensuring your industry will align with other industries You partner with so the data can be easily and readily be available and in a better format than it was prior to the implementation. Ensuring clear dialogue with all parties to make the right decisions upfront or it will be costly to retrofit and re-engineer a solution post-implementation.
Data classifications: having a clear understanding of your data estate is crucial when the selection of an MDM solution is being undertaken.
Skills and experience: it’s very important to assess the current skills and experience in the team that have MDM skills, this is a major consideration
Infrastructure: select the toolset that will integrate with your current infrastructure to avoid introducing layers of complexity into your enterprise environment.
One of the best things that you can do is identify the stakeholders for Master Data Management. Once that group is identified you have that group assign a leads to be on the governance committee. Assign one person on this committee to be the overall lead so there is one voice to communicate to the stakeholders. Once full organized you can then setup monthly meetings to go over changes and gathering of approvals. There are a lot of tools out there that can help in this process to put restrictions and rules around the creation and use of Master Data. Be careful on which one you choose and it can create unneeded complexities in some cases. hardest part of it all is getting people to believe that there is a need for governance. Be safe, wash your hands, and wear a mask.
What are key differences between MDM and Data Governance?
What are the practical differences in which each of these solutions is applied?